Alternative Market Mechanisms for the Student Loan Program	 
(18-DEC-01, GAO-02-84SP).					 
								 
This report reflects the results of a collaborative effort	 
between GAO and representatives of the Secretary of Education. As
required by the Higher Education Amendments of 1998, GAO formed a
study group to identify and evaluate a means of establishing a	 
market mechanism for the delivery of student loans. This study	 
group consisted of representatives of the Department of the	 
Treasury, Office of Management and Budget, Congressional Budget  
Office, entities making Federal Family Education Loan Program	 
(FFELP) loans and other entities in the financial services	 
community, and other participants in the student loan market. The
group met as a whole four times before the public release of a	 
draft of this report, and various group members corresponded with
GAO and Education between group meetings as well. The mandate	 
called for the evaluation of at least three different market	 
mechanisms relative to 13 criteria. In consultation with the	 
study group, GAO selected five general models for further	 
evaluation--adjustments to the current system and four additional
market mechanism models. Adjustments to the current system, in	 
which information would be collected from current market	 
transactions for use in determining the appropriate level of	 
lender yield and which Congress or some independent entity would 
still set through statute or regulation, would involve the least 
change from the current FFELP. The loan origination rights	 
auction model would involve lenders bidding for the right to	 
originate loans. In the loan sale model, the government or a	 
government-designated entity would originate loans. Private	 
lenders would then bid in an auction to purchase these loans	 
after after they have been originated. The federal funding model 
affords lenders the opportunity to borrow funds from the federal 
government to make FFELP loans at a predetermined interest rate  
or at an interest rate determined by some type of bidding	 
process. Lastly, the market-set rate model allows lenders and	 
borrowers to negotiate their own interest rates and perhaps other
loan terms.							 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-02-84SP					        
    ACCNO:   A02625						        
  TITLE:     Alternative Market Mechanisms for the Student Loan       
Program 							 
     DATE:   12/18/2001 
  SUBJECT:   Higher education					 
	     Lending institutions				 
	     Loan repayments					 
	     Student financial aid				 
	     Student loans					 
	     Federal Family Education Loan Program		 
	     Free Application for Federal Student Aid		 
	     Health Education Assistance Loan			 
	     Income Dependent Education Assistance		 
	     Parent Loan for Undergraduate Students		 
	     William D. Ford Federal Direct Loan		 
	     Program						                                                                 
	     Federal Perkins Loan Program			 
	     Dept. of Education Stafford Student Loan		 
	     Program						 								 
	     Dept. of Education Income Contingent		 
	     Repayment Program					 
								 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-02-84SP
     
GAO-02-84SP

Alternative Market Mechanisms for the Student Loan Program

A Report by the U.S. Department of Education and the U.S. General Accounting
Office

December 18, 2001

CONTENTS

Page

Overview 1 Chapter 1: Program Background and the 1998 HEA Mandate 8 Chapter
2: Adjustments to the Current FFELP System 25 Chapter 3: Loan Origination
Rights Auction 32 Chapter 4: Loan Sale 42 Chapter 5: Federal Funding 49
Chapter 6: Market-Set Rates 55 Chapter 7: Income-Contingent Repayment 63

Appendixes

Appendix I: Mandate From the Higher Education Amendments of 1998 69 Appendix
II: List of Study Group Members 71 Appendix III: Technical Aspects of
Auction Design 75 Appendix IV: Details on Income-Contingent Repayment in
FDLP 82 Appendix V: Call Options as a Mechanism for Determining Net

Lender Yields 93 Appendix VI: Additional and Dissenting Views From Rene
Champagne 96 Appendix VII: Additional and Dissenting Views From Michael
Hershock and

Richard Pierce 97 Appendix VIII: Additional and Dissenting Views From Paul
Tone and

Richard Pierce 103 Appendix IX: Public Comments From Career College
Association 131 Appendix X: Public Comments From CBA, NCHELP, and SLSA 134
Appendix XI: Public Comments From FinAid Page 137

Appendix XII: Public Comments From National Student Loan Program 138
Appendix XIII: Public Comments From Oklahoma Lender Advisory Council 140
Appendix XIV: Public Comments From Philip Schrag 143 Appendix XV: Public
Comments From Patricia Smith, Study Group Member 145 Appendix XVI: Public
Comments From Texas Guaranteed Student Loan

Corporation 147

Glossary 150

Tables

Table 1: The Four Sets of Evaluation Criteria 4 Table 2: General Differences
Between the Five Models 6 Table 3: Outstanding Loan Balances at End of
Fiscal Year 2000 12 Table 4: Loan Originations in Fiscal Year 2000 12 Table
5: Concentration of FFELP Outstanding Loan Balances at the End of

Fiscal Years 1994-2000 15 Table 6: Concentration of FFELP Loan Originations
in Fiscal Years 1994-2000 15 Table 7: Federal Costs for the FFELP Fiscal
Year 2000 18 Table 8: The Four Sets of Evaluation Criteria 21 Table 9:
Summary of Analysis for Adjustments to the Current System Model 31 Table 10:
Summary of Analysis for Loan Origination Rights Auction Model 41 Table 11:
Summary of Analysis for Loan Sale Model 48 Table 12: Summary of Analysis for
Federal Funding Model 54 Table 13: Summary of Analysis for Market-Set Rates
Model 62 Table 14: Direct Loans in Repayment by Loan Type 83 Table 15:
Direct Loans in Repayment by Plan 83 Table 16: Source of Direct
Consolidation Loans in Repayment 84 Table 17: Direct Consolidation Loans in
Repayment by Plan 84 Table 18: Direct Consolidation Loans in Repayment by
Plan: Defaulted Loans

Formerly Held by DCS 85 Table 19: Direct Consolidation Loans in Repayment by
Plan: Regular

Consolidation of Non-DCS Loans 85

Figure

Figure 1: Cash Flows for an FFELP Loan From Origination Through Repayment 19
Figure 2: Cash Flows for an FFELP Loan in Default 20

Abbreviations

AGI adjusted gross income CBO Congressional Budget Office CP Commercial
Paper DCS Debt Collection Service EPA Environmental Protection Agency FAFSA
Free Application for Federal Student Aid FCC Federal Communications
Commission FDLP William D. Ford Federal Direct Loan Program FFELP Federal
Family Education Loan Program FHA Federal Housing Administration GSE
government sponsored enterprise HEA Higher Education Act of 1965 HEAL Health
Education Assistance Loan HHS Department of Health and Human Services ICR
income-contingent repayment IDEA Income Dependent Education Assistance IRS
Internal Revenue Service OMB Office of Management and Budget PLUS Parent
Loan for Undergraduate Students PRC Postal Rate Commission SAP special
allowance payment

Page 1

OVERVIEW

In the Federal Family Education Loan Program (FFELP), lenders annually make
more than $22 billion in loans to eligible student borrowers who attend
postsecondary institutions and their parents. Additionally, through the
William D. Ford Federal Direct Loan Program (FDLP), the federal government
makes more than $11 billion available to these borrowers. The federal
government insures FFELP loans against default and assures lenders of a
specified yield that adjusts with interest rates. 1 Borrowers pay interest
at a variable rate up to a maximum rate established by law. The federal
government pays a ?special allowance? to lenders-the difference between a
borrower?s rate and the specified yield-when the borrower?s rate is lower.
The Congress occasionally adjusts both the borrower?s rate and the lender?s
yield.

In setting lender yield, the Congress attempts to ensure a yield high enough
to maintain lender participation in the program but not so high as to
require spending more taxpayer dollars than necessary. While the Congress
considers information from Education, other federal agencies, and program
participants in setting the lender yield, it often lacks critical
information on the costs to lenders of making and servicing loans. Thus,
setting lender yield is difficult, as illustrated by the extensive
deliberations surrounding the 1998 reauthorization of the Higher Education
Act of 1965 (HEA).

When the Congress considered the reauthorization of HEA in early 1998, both
House and Senate committees expressed concern about the process of setting
lender yield. The House Committee on Education and the Workforce stated in
its report:

?Currently, the Federal Family Education Loan (FFEL) program is a
market-based program with private sector participation. However, to a large
extent lender returns are set through a political process rather than a
market process. This is disturbing for two reasons. First, if lender yield
is set too low, private capital will become unavailable, and the student
loan programs will collapse. Second, if the rate of return is set too high,
the Federal Government forgoes savings that could be put to better uses or
returned to the taxpayer.? 2

The report of the Senate Committee on Labor and Human Resources expressed
similar concerns when it stated that the committee ?has wrestled with its
desire to balance the twin objectives of reducing the interest rate paid by
borrowers and preserving access to loans under the FFEL program.? 3

1 FFELP loans are guaranteed by one of 36 nonprofit agencies designated by
the Secretary of Education. If a loan goes into default, the guaranty agency
generally pays the lender 98 percent of principal and accrued interest. The
federal government then generally reinsures 95 percent of the guaranty
agencies' payments to lenders (and also pays them fees for loan processing
and issuance, account maintenance, and default aversion). The lender?s yield
is the face value of the interest rate on the FFELP loan. Currently, this is
specified in legislation as the 90-day commercial paper rate plus 2.34
(1.74) percentage points for loans (not) in repayment. 2 H.R. No. 105-481,
at 154-5 (1998). 3 S.R. No. 105-181, at 53 (1998).

OVERVIEW Page 2

The 1998 amendments to HEA required that GAO and Education jointly convene a
study group to identify and evaluate means for establishing a market
mechanism for the delivery of student loans. See appendix I for the full
text of the mandate. This study group consisted of representatives of the
Department of the Treasury, the Office of Management and Budget (OMB), the
Congressional Budget Office (CBO), entities making FFELP loans, other
entities in the financial services community, other participants in the
student loan programs, and other individuals designated by GAO and
Education. See appendix II for a list of study group members. GAO and
Education, in consultation with the study group, were charged with
identifying at least three different potential market mechanisms and
evaluating them with respect to 13 criteria laid out in the mandate.

In January 2001, a draft report prepared by GAO and Education was released
for public comment. The draft report analyzed several models based on
research and analysis, discussions with study group members, and comments
received from others not on the study group. Serving on the study group does
not constitute agreement either in whole or in part with the proposed models
or the analysis of models presented in this report. Some study group members
have provided additional or dissenting views, which appear here as
appendixes.

BACKGROUND AND METHODOLOGY

Lenders in FFELP make loans to student and parent borrowers and receive a
yield-an interest rate on the loans-that is set by legislation. Borrowers
choose a lender, typically from a list their schools maintain. A guaranty
agency reviews the loan application and issues a guarantee on the loan. As
long as the loan is serviced properly, the guarantee is maintained, and the
guaranty agency repays the lender if the borrower defaults on the loan.
Education oversees lender, school, and guaranty agency participation in the
program and reimburses the guaranty agencies for default payments provided
to lenders. In addition, Education makes payments to lenders to make up the
difference, if any, between the interest rate the borrower pays and the
yield the lender is entitled to receive.

The yield that lenders receive from FFELP loans has been adjusted
occasionally since 1977. 4 For Stafford loans and their predecessors, the
Congress set lender yield generally at the 91-day Treasury bill (T-bill)
rate plus a markup of 3.5 percentage points in 1977. The Congress made
downward changes in the markup over the T-bill rate in 1986 and 1992. In
1995, different rates for loans in different stages were established, so
that the markup was reduced for loans that were in school, grace, or
deferment periods relative to the markup for loans in repayment. Both

4 Before 1977, a committee composed of the Secretary of Health, Education,
and Welfare, the Secretary of the Treasury, and the Director of OMB decided
the level at which lender yield should be set. Since 1977, the yield has
been established by using a formula set in legislation rather than by a
committee?s determination, and the Congress adjusts this formula
periodically.

OVERVIEW Page 3

markups were reduced further in 1998. Finally, in 1999, the Congress changed
the basis for the yield from the 91-day T-bill to the 3-month Commercial
Paper (CP) index. 5

In making these changes, the Congress has generally tried to set a yield
that would maintain lender participation in the program without spending
more than is necessary. Lenders require a reasonable return on their
investment in the program to continue to take part in it. Without a
reasonable return, they would be likely to devote their resources to more
profitable investments, jeopardizing the continued availability of loan
capital for FFELP. However, if the yield is set too high, federal funds
could be wasted.

In order to explore ways of bringing market information to bear on the
yield-setting process, the Congress, in the 1998 HEA reauthorization,
mandated a study of the potential use of market mechanisms in FFELP. The
mandate calls for evaluating at least three different market mechanisms
relative to 13 criteria. We grouped the evaluation criteria into four sets,
as shown in table 1.

5 The lender?s yield is the face value of the interest rate on the FFELP
loan. Currently, this is set in legislation at the CP rate plus 2.34
percentage points for loans in repayment. It is the lender?s total interest
revenues as a percentage of the value of the loans. A lender?s net yield is
the lender?s yield less all costs, which include the costs of (1) raising
funds to make the loans, (2) servicing the loans, (3) defaults, and (4)
other administrative expenses. Loan servicing functions include maintaining
contact with borrowers, billing for repayments, and taking steps to avoid
defaults if loans become delinquent.

OVERVIEW Page 4

Table 1: The Four Sets of Evaluation Criteria

Set Related criteria from the 1998 HEA amendments

Description of model, including variations A description of how the
mechanism will be administered and operated

(12) The proposed federal and state role in the operation of the mechanism
(11) Transition procedures (13) a

Costs, savings, and effects on subsidies for program participants

The cost or savings of loans to or for borrowers, including parent borrowers
(1) The cost or savings of the mechanism to the federal government (2) The
cost, effect, and distribution of federal subsidies to or for participants
in the program (3)

Effects on lender participation, loan availability, and service quality

The effect on the diversity of lenders, including community-based lenders,
originating and secondary market lenders (7) The availability of loans to
students by region, income level, and categories of institutions (10) The
effect on loan availability during a transition period (13) a The effect on
investment in human capital and resources, loan servicing capability, and
the quality of service to the borrower (6) The degree to which the mechanism
will provide market incentives to encourage continuous improvement in
delivering and servicing loans (9)

Simplicity, regulatory burden, and program integrity The effect on the
simplicity of the program, including the effect of the plan

on the regulatory burden on students, institutions, lenders, and other
program participants (5) The effect on program integrity (8)

Note: The numbers in parentheses refer to the number assigned to the
specific criteria in the legislative mandate. The analysis related to the
fourth criterion-the ability of the mechanism to accommodate the potential
distribution of subsidies to students through an income-contingent repayment
(ICR) option-was similar across all models. Rather than repeat our analysis
of this criterion for each model, we discuss it separately in chapter 7.
Therefore, this criterion does not appear in this table.

a We split the thirteenth criterion, on transition procedures, into a
descriptive part and a part related to loan- availability issues.

THE FIVE MODELS

To identify different market mechanisms for analysis, we reviewed reports
and asked study group members and others to submit proposals. We looked
primarily for models in which some type of market process either determines
the lender yield or provides information used to set the yield.

We grouped the proposals into five general models because similar
characteristics emerged among some of the proposals. In study group meetings
and other conversations with group members, including those with staff from
the other government agencies, we discussed the various models and their
implications. We also researched comparable programs where they

OVERVIEW Page 5

existed, such as other federal auctions and the experiences of student loan
programs in other countries.

The five models would make a variety of changes to FFELP. The first of the
five models we identified-adjustments to the current system-would involve
the least change from the current FFELP. Information would be collected from
current market transactions for use in determining the appropriate level of
lender yield, which the Congress or some independent entity would still set
through statute or regulation. The loan origination rights auction model
would involve lenders bidding for the right to originate loans (either for
the right to procure a certain volume, with which they could originate loans
at any school, or for the right to originate loans at particular schools). A
lender?s bid might consist of a specific yield level, so the yield would be
directly determined by this mechanism. Alternatively, the bid might consist
of a dollar amount for the right to originate loans, with the lender?s yield
on these loans set outside the process. In the loan sale model, in contrast,
the government or a government-designated entity would originate loans. 6
Private lenders would then bid in an auction to purchase these loans after
they have been originated. Again, the lender yield would be set outside the
process, and lenders? bids to purchase these loans would determine their net
yield. In the federal funding model, lenders would have the opportunity to
borrow funds from the federal government to make FFELP loans. They would
borrow either at a predetermined interest rate or at an interest rate
determined by some type of bidding process. The lender?s yield would still
be set by the Congress, but by changing lenders? funding costs, this model
would determine the net yield for lenders. Finally, in the market-set rate
model, lenders and borrowers would negotiate their own interest rates and
perhaps other loan terms. Regulatory limits, such as a limit on the range of
interest rates a lender would be allowed to offer, could be imposed. This
process would set both the lender yield and the borrower interest rate.

Table 2 shows some of the major differences between the models, although
most of the models have several variations that would allow for further
differentiation.

6 Schools could potentially originate loans as currently practiced in FDLP.

OVERVIEW Page 6

Table 2: General Differences Between the Five Models

Market mechanism model Question to differentiate the models

Adjustments to the current

system Loan origination rights auction

Loan sale Federal funding Market-set rate

How does the model work? The Congress

or its designee would use market information to set lender yield

Lenders would bid for the right to make (1) a certain volume of FFELP loans
or (2) loans at specific schools

Lenders would bid to purchase loans after the government or some designated
entity originated the loans

Lenders would be allowed to borrow funds from the government to make FFELP
loans

Lenders and borrowers, or schools on the borrowers? behalf, would negotiate
interest rates

What rates or costs does the market determine?

Rates or costs are determined only indirectly, based on market information

Either version above could determine (1) the lender?s yield or (2) lender?s
cost to receive a given yield

Lender?s cost to acquire the loan and ensure a given yield

Lender?s funding cost Lender?s yield

and borrower?s interest rate

Who originates loans? Private lenders Private lenders Federal government,

through contractor or other entity

Private lenders Private lenders

Are private lenders restricted in how much, or at which schools, they
originate?

No Yes, in some cases Not relevant No Unclear

Is the borrower?s rate or ability to choose the lender changed?

No Yes, for lender choice Yes, for lender choice No Yes, for both

rates and choice of lender

How does the role of schools change?

No change Schools may lose choice of lenders

Schools may serve as the originator of the loan

No change Schools would play a stronger role in determining which lenders to
use

THE STRUCTURE OF THIS REPORT

Chapter 1 of this report includes background material, a brief history of
FFELP, and a description of the study group?s work. Chapters 2 through 6
contain the analysis of each of the five models, relative to the 13 criteria
set out in the mandate. Chapter 7 then discusses ICR as it might be applied
to any of the models.

OVERVIEW Page 7 PUBLIC COMMENTS

We made a draft of this report available to the public for comment as
mandated by section 801 of the 1998 Amendments to HEA. Eight sets of written
and/or electronic comments were submitted by eight organizations and two
individuals during the 60-day comment period of January 19, 2001, to March
19, 2001 (see apps. IX-XVI). These comments were made available to the study
group for review.

In summary, several of the respondents specifically recommended that FFELP
continue without change and that the current formula for lender yield remain
in place. Several also expressed the opinion that maintaining the
availability of loans to borrowers and student access to higher education
was of greater importance than reducing federal program costs. One
respondent suggested that continued participation of local lenders helped
ensure access for risky or less profitable borrowers.

Several respondents expressed concern about potential consequences
associated with introducing one or more of the market mechanism
alternatives. Some stated that adjustments to the current system would cause
the least disruption, while others stated that all market models could cause
disruptions. While a number of respondents expressed concern that market-set
rates had the potential for discrimination and denial of access to funds,
one respondent stated that market-set rates may cause less disruption than
other market models.

Two respondents commented mainly on the ICR option. They expressed concern
regarding Education?s payment calculator, used by borrowers to compare the
monthly payments of ICR with other available loan repayment options, because
it projects the total payments of repayment plans by adding up the total
dollars to be repaid. They commented that adding up the total dollars can
prove misleading. To remedy this, they recommended using a repayment
calculator that calculates the present value of total dollars to be repaid
and noted that such a payment calculator is available on the Internet. The
respondents also referred interested readers to a forthcoming book on ICR.
Additionally, some respondents believed that the draft report exaggerated
the cost advantages of FDLP over FFELP. Several respondents also provided
technical comments that we incorporated where appropriate.

Page 8

CHAPTER 1 PROGRAM BACKGROUND AND 1998 HEA MANDATE

FFELP, established by HEA, as amended, provides more than $22 billion of
loans annually to students pursuing a postsecondary education and their
parents. Since 1965, FFELP has made higher education affordable for more
than 40 million Americans. Access to FFELP loans is nearly universal among
students enrolled at least half time at postsecondary education institutions
eligible to receive at least an unsubsidized loan. More than 21 percent of
all undergraduate students receive FFELP aid. 7 Lenders make loans to
eligible borrowers, and the federal government guarantees the loans against
default. Borrowers attending schools that participate in FFELP are legally
free to choose their own lender. Many borrowers select a lender from a list
provided by the school. Borrowers receive loans with differing levels of
subsidies, which are based on financial need and length of time in school.
The federal government ensures that lenders receive a certain yield,
currently based on a CP interest rate, on FFELP loans. The federal
government pays lenders the difference between the borrower?s interest rate
and the lender yield. The Congress adjusts this yield occasionally, but
because critical information on lenders? costs and profitability is not
available, the Congress cannot be sure that a particular yield will ensure
sufficient lender participation without spending more than necessary. To
help the Congress consider market-based alternatives to legislation, the
Congress directed that GAO and Education form a study group with government
and non-government experts to analyze ways in which lender yield in FFELP
could be set through a market mechanism for the delivery of student loans.
In accordance with the 1998 amendments to HEA, each mechanism was to be
evaluated relative to 13 criteria set out in the mandate. The study group
considered a number of proposals, and GAO and Education conducted an
analysis of five models that involve market processes to different extents.

HOW THE FEDERAL LOAN PROGRAMS OPERATE

HEA has created two major federal loan programs for students pursuing
postsecondary education. 8 In 1965, the original HEA established the
guaranteed student loan program (subsequently renamed FFELP), in which loans
are originated by lenders, guaranteed by state- chartered guaranty agencies,
and reinsured by the federal government. In 1993, HEA authorized FDLP, a
second major loan program, in which the federal government provides loan
capital, schools originate loans directly to students, and outside
contractors perform loan origination and servicing functions. Since that
time, competition has increased in the student loan market and student
borrowers have benefited from reduced fees and lower interest rates.

7 Based on an analysis of the 1995-96 National Postsecondary Student Aid
Study sponsored by the National Center for Education Statistics. 8 HEA also
authorized the Perkins Loan program, a relatively small program that
provides funds to postsecondary

institutions that are used to establish a revolving fund from which loans to
students are made. The Perkins Loan program-previously known as National
Defense and National Direct Student Loan programs-predates HEA by 7 years.

CHAPTER 1: BACKGROUND Page 9

Types of Loans

FFELP and FDLP both include four major types of loans: subsidized Stafford,
unsubsidized Stafford, PLUS, and consolidation loans. 9 Student borrowers
receive subsidized and unsubsidized Stafford loans. Parent borrowers receive
PLUS loans. Borrowers may take out consolidation loans before or after
entering their repayment period.

Subsidized Stafford Loans Needs-based subsidized Stafford loans are
available to eligible students at participating institutions. The guaranteed
amount of a subsidized loan may not exceed a student?s ?unmet financial
need.? 10 The maximum loan amount is also subject to annual and aggregate
loan limits.

A key feature of the subsidized Stafford loan is its interest subsidy while
the student is not in repayment. The government pays, on the borrower?s
behalf, all interest accruing on the outstanding principal while the
borrower is attending school at least half-time, for 6 months after
attendance (the ?grace period?), and for periods of authorized deferment.
Stafford borrowers participating in the standard repayment plan have a
repayment period of 10 years, not counting periods of authorized deferment
and forbearance. 11

FFELP and FDLP lenders may also offer graduated and income-sensitive or
income-contingent repayment plans to borrowers. Borrowers have the option of
picking a different repayment plan each year. If borrowers waive their
selection, rules of the standard repayment plan apply.

As previously noted, the interest rate on subsidized Stafford loans is
currently the 91-day T-bill rate plus 1.7 percentage points when the
borrower is in school or in other nonpayment periods and the T-bill rate
plus 2.3 percentage points when the borrower is in repayment. The rate is
reset on July 1 each year, based on the T-bill rate from the last Treasury
auction conducted before June 1.

Unsubsidized Stafford Loans Unsubsidized Stafford loans differ from
subsidized Stafford loans in three key respects: They are not need-based,
the government does not pay interest during in-school periods, and loan
limits are higher. The unsubsidized Stafford loan is not need-based; the
approved amount is not limited

9 Originally, ?PLUS? stood for the official name of the program, ?Parent
Loans for Undergraduate Students.? Under current law, the program is simply
named PLUS. Formula interest rates charged to students under FDLP were set
in law generally equal to the formula maximum rates on corresponding FFELP
loans. 10 ?Unmet need? is, in its simplest terms, the borrower?s cost of
attendance minus estimated family contribution

minus estimated financial aid from other sources. 11 A deferment is a period
during which borrowers do not need to pay principal and the federal
government pays interest. Borrowers are eligible for a deferment under
certain conditions, such as going on to further schooling. A forbearance is
a period during which borrowers do not need to pay principal but are
responsible for any interest that accumulates. The borrower?s eligibility
for a 10-year repayment schedule is conditioned by the requirement that the
borrower repay at least $600 per year. The 1998 amendments to HEA provide
for a repayment schedule of up to 25 years for ?new borrowers? who
accumulate more than $30,000 in FFELP loans.

CHAPTER 1: BACKGROUND Page 10

by the same financial need formula used for subsidized Stafford loans.
Instead, the expected amount of the unsubsidized Stafford loan may be
considered all or part of the estimated family contribution. Interest on the
unsubsidized Stafford loan is paid entirely by the borrower and is not
subsidized by the federal government. The borrower is not required to make
interest payments during in-school, grace, and other deferment periods.
However, interest accrued during such periods may be capitalized (added to
principal) when the loan enters repayment for the first time or when it
returns to a repayment status following a period of deferment. Higher loan
limits are available on unsubsidized Stafford loans to independent students.
12 The maximum borrower?s interest rate is the same as for subsidized
Stafford loans.

PLUS Loans The PLUS loan is available to parents for their student
dependents. For a school to certify a PLUS loan, both parent and student
must meet program eligibility requirements. For the PLUS loan, HEA requires
the lender to determine that the parent borrower does not have an ?adverse
credit history? before making the loan and to use as a minimum the
guidelines for determining adverse history in the student financial aid
regulations. If the lender discovers adverse credit history in the
applicant?s credit bureau report, it can still make the loan if it documents
?extenuating circumstances.? Even without extenuating circumstances, the
lender can still make the loan if the borrower obtains a creditworthy
endorser.

Like the unsubsidized Stafford loan, the PLUS loan is not need-based and may
replace all or part of the student?s estimated family contribution. Also
like the unsubsidized Stafford loan, payments of interest on the PLUS loans
fall upon the borrower entirely, with no federal subsidy. Payments of
interest are not required during periods of authorized deferment but may be
capitalized upon expiration of the deferment.

Unlike both types of Stafford loans, there is no fixed annual or aggregate
limit for PLUS loans. A loan may not be made, however, for an amount greater
than the student?s cost of attendance less his or her estimated financial
aid. Also unlike Stafford loans, the PLUS loan has no grace period-it enters
repayment upon full disbursement, unless the borrower happens to qualify for
a deferment. Unless payments are deferred, the first payment due date for a
PLUS must be established within 60 days of final disbursement.

Consolidation Loans Consolidation loans are new originations that do not
contribute to the increases in outstanding balances because they refinance
already existing loans. Borrowers may consolidate FFELP loans, FDLP loans,
and Perkins loans authorized by legislation other than HEA.

12 Independent students are students who, by meeting certain regulatory
criteria, are presumed to receive no financial support from their parents. A
student is considered ?independent? who is at least 24 years old and who is
a graduate or professional student, a veteran of the U.S. armed forces, or
married or has dependents other than a spouse. A financial aid administrator
may also classify a student as independent under special circumstance, even
if none of these criteria is met.

CHAPTER 1: BACKGROUND Page 11

The purpose of the consolidation loan program is to give a borrower who has
multiple loans- possibly from different lenders, different guarantors, and
even from different loan programs-the opportunity to have them combined into
a single debt. The consolidation loan offers two unique advantages. First,
the borrower who has been dealing with multiple servicers and repayment
schedules can deal with a single servicer and single repayment schedule.
Second, a combined debt of more than $7,500 qualifies the borrower for a
repayment term longer than the maximum 10 years generally available on
Stafford and PLUS loans. For example, consolidation loan borrowers with at
least $20,000 of combined qualifying debt may qualify for a repayment period
of 20 years, borrowers with more than $60,000 for the maximum 30 years. 13
Periods of deferment and forbearance are not included.

Borrowers who consolidate their loans give up certain benefits tied to the
underlying loans subject to consolidation. These benefits include certain
deferments and all service and employment cancellations available to
borrowers under the Federal Perkins Loan Program. Borrowers also give up the
variable interest rate of the FFELP and FDLP loans consolidated in lieu of a
rate fixed for the life of the loan.

In recent years, consolidation loans have been used to resolve defaults. A
borrower who has defaulted on an FFELP or FDLP loan may pay that defaulted
debt in full by consolidating it. This has two effects. First, the
borrower?s credit history is improved-in some cases immediately and in
others after payments have been made. Second, resolution of the default
qualifies the borrower for additional student financial assistance. A
borrower qualifies for consolidation of a defaulted loan by making a
satisfactory repayment arrangement with its holder, as defined in
regulations.

The availability of loan consolidation to borrowers must be considered as
part of the analysis of any market mechanism proposal. Loan consolidation
allows borrowers to change lenders. However, the ability to switch lenders
affects the perceived federal benefit that could result from some of the
market mechanism models discussed in this report.

Student Loan Volume and Default Rates

FFELP loans represent more than half of federal student loans outstanding
and newly originated, and in both FFELP and FDLP, Stafford loans make up the
bulk of overall outstanding loans. At the end of federal fiscal year 2000,
the total amount of outstanding FFELP loans was about $166 billion, while
about $58 billion of FDLP loans were outstanding. During fiscal year 2000,
FFELP loans accounted for about 63 percent of new loan volume, FDLP for
about 37 percent. Tables 3 and 4 detail loan balances and new loan
originations, respectively, for fiscal year 2000.

13 This is based on the outstanding balances of all loans in the qualifying
underlying loan programs.

CHAPTER 1: BACKGROUND Page 12

Table 3: Outstanding Loan Balances at the End of Fiscal Year 2000

Total loans Loan Federal Family Education Loan

Program Federal Direct

Loan Program Amount Percent Stafford

Subsidized $85,084 $23,976 $109,059 48.7% Unsubsidized 37,410 15,487 52,897
23.6

PLUS a 14,552 3,609 18,160 8.1

Consolidation 29,082 14,643 43,725 19.5

Total $166,128 $57,714 $223,841 100.0%

Note: Dollars are in millions. Totals do not sum because of rounding. a
Includes loans from Supplemental Loans for Students, a program that no
longer exists.

Source: Department of Education. Table 4: Loan Originations in Fiscal Year
2000

Total loans Loan Federal Family Education Loan

Program Federal Direct

Loan Program Amount Percent Stafford

Subsidized $11,259 $5,785 $17,044 37.8% Unsubsidized 9,126 4,240 13,366 29.6

PLUS 2,326 1,318 3,644 8.1

Consolidation 5,695 5,369 11,065 24.5

Total $28,406 $16,712 $45,119 100.0%

Note: Dollars are in millions. Totals do not sum because of rounding.
Source: Department of Education.

The cohort default rate for federal student loan programs hit its highest
level-22.4 percent-in fiscal year 1990 but has since greatly declined. The
reduction in the default rates has been attributed to a variety of
circumstances. Efforts by FFELP participants, actions by both the Congress
and Education, the availability of additional repayment plans, and the
strong economy have all contributed to decline. The most recent student loan
default rates-the 1998 cohort default rates-showed that overall, the direct
and guaranteed student loan programs had similar default rates-6.6 percent
for FDLP and 6.7 percent for FFELP.

Participants in the Student Loan Programs

FFELP involves many participants, including borrowers, schools, lenders,
loan servicers, guaranty agencies, and Education. Some of these participants
also have roles in FDLP.

CHAPTER 1: BACKGROUND Page 13

Borrowers Every eligible student pursuing postsecondary education at least
half-time at a participating school meeting certain requirements may obtain
a loan. In addition, in some cases parents of undergraduates may also borrow
through FFELP or FDLP. The loans can be used to pay for tuition and
education- related expenses at eligible 4-year colleges and universities,
2-year community colleges, private colleges and universities, and for-profit
trade and technical schools (sometimes referred to as proprietary schools).

For Stafford loans, student borrowers do not have to make payments while in
school or in other authorized periods of nonpayment. Depending on the
borrower?s income, and the income of his or her family, he or she may be
responsible for the interest that accrues during these periods of
nonpayment, in which case it is added to the loan principal at the beginning
of the repayment period. The maximum interest rate a borrower may be charged
is set in legislation as the 91-day T-bill rate plus an add-on of 1.7
percentage points when the borrower is in school or in other nonpayment
periods and 2.3 percentage points when the borrower is in repayment. The
borrower rate is adjusted annually, based on the new value each year of the
T-bill rate, but it is capped at 8.25 percent no matter how high the T-bill
rate.

Schools Eligible schools decide whether to participate in FFELP, FDLP, or
both; most schools choose to participate in only one of the programs. In
both programs, schools make various certifications necessary for a borrower
to obtain a loan, and in FDLP they perform certain loan origination
functions as well. Borrowers in FFELP schools are legally free to choose
among all eligible FFELP lenders. Schools often provide borrowers a
recommended list of lenders that is based on the services, loan terms (such
as rate discounts for good performance), and other key services offered by
the lenders. 14 Most borrowers use a lender that their school recommends.

Lenders and Secondary Markets Lenders originate and hold FFELP loans. HEA
limits eligibility to originate and hold these loans primarily to (1) banks
and certain other savings institutions, (2) pension funds, (3) insurance
companies, (4) one state or private, nonprofit agency for each state, and
(5) with certain limitations, schools. Lenders pay the government an
origination fee of one-half percent for each loan and, in the case of
consolidation loans, a fee of 1.05 percent. The government offsets interest
and special allowance payments owed to lenders to collect a 3 percent
borrower origination fee and authorizes lenders to charge the borrower for
this fee. Although lenders may discount the fee to the borrower, so that the
borrower may not pay the full 3 percent, the lender must still pay the full
fee to the government. The 1998 Amendments and regulations further specified
the circumstances under which lenders may charge borrowers a reduced
origination fee.

14 Service competition by lenders involving ?inducements,? such as mailing
unsolicited loan applications or paying schools for referrals of loan
applicants, is not permitted.

CHAPTER 1: BACKGROUND Page 14

An eligible lender can approve and originate loans. Once the loan is
originated, the lender can keep the loan on its books and earn either a
positive or negative return and interest spread based on the lenders? yield
and its own interest expenses and other expenses, sell the loan to a
purchasing lender and record the gain or loss on sale, or securitize the
loan by selling the loan to a trust that has beneficial ownership of the
loans and funds its holdings by selling debt to investors and book a gain or
loss, depending on the terms of the transaction.

In the FFELP student loan market, secondary markets refer to financial
institutions that purchase student loans from lenders and provide liquidity
to the student loan market. In 1972, the Congress chartered a national
secondary market, Sallie Mae, as a shareholder-owned government-sponsored
enterprise (GSE) to provide liquidity for the student loan market. Some
other lenders also serve as secondary markets for FFELP loans. In addition,
under HEA, each state can designate a not-for-profit secondary market to
help ensure that every student at every eligible institution can receive a
loan and to provide liquidity to originators.

Financial institutions other than those defined as eligible by HEA are not
eligible for direct participation as lenders in FFELP. However, HEA
authorizes the use of trustees as eligible lenders to hold loans for the
benefit of others without regard to the latter?s own eligibility. Certain
secondary markets, those that are not designated by their state to be the
state?s eligible lender, as well as other nonbank financial institutions,
use trustees to originate loans.

Sallie Mae and some secondary markets also effectively manage or service an
additional amount that was held by originators with which these secondary
markets have ?pipeline? arrangements. Some of the funding for loans in the
pipeline is borrowed by the originating lender from the subsequent
purchaser. 15

In 2000, nearly 4,000 lenders took part in FFELP. Both loan holdings and
loan originations were concentrated in larger institutions, as shown in
tables 5 and 6. The top 10 loanholders held 68 percent of outstanding loan
balances and the top 10 loan originators originated 52 percent of the loan
volume in 2000. Both loan holdings and loan originations became increasingly
concentrated, as shown in tables 5 and 6.

15 Under such arrangements, the secondary market performs all or some of the
marketing, origination, funding, or servicing functions for the originating
lender of record while the student is in school. The loan is transferred to
the secondary market when or before the loan enters repayment pursuant to a
forward purchase agreement, often at a predetermined price.

CHAPTER 1: BACKGROUND Page 15

Table 5: Concentration of FFELP Outstanding Loan Balances at the End of
Fiscal Years 1994-2000

% Share held by Fiscal year end Top 10 loanholders Top 50 loanholders

1994 54 79 1995 55 82 1996 57 85 1997 58 87 1998 58 87 1999 62 89 2000 68 91

Source: Department of Education. Table 6: Concentration of FFELP Loan
Originations in Fiscal Years 1994-2000

% Share originated by Fiscal year Top 10 originators Top 50 originators

1994 37 67 1995 37 69 1996 40 73 1997 45 78 1998 52 79 1999 52 82

2000 52 82

Note: Originations exclude consolidation loans. Source: Department of
Education

FFELP lenders? yield and eligibility requirements are set by federal statute
and are administered by Education. This yield is based on a variable
short-term index, which closely tracks changes in the market cost of funds.
The formula for lenders? yield for new Stafford loans in FFELP-the largest
component of FFELP-is now based on a CP index. Lender yield is the 90-day CP
rate plus 1.74 percentage points while the borrower is in school or during
other nonpayment periods and it is CP plus 2.34 percentage points when the
borrower is in repayment. The lender yield is adjusted quarterly with new
values of the CP rate. This rate applies to new FFELP loans made after
January 1, 2000.

After July 1, 2003, the Stafford formula is scheduled to change to a markup
of 1 percentage point over the rate for ?comparable maturities? of Treasury
securities, as determined by the Secretary of Education in consultation with
the Secretary of the Treasury. This new rate is both the

CHAPTER 1: BACKGROUND Page 16

borrower?s rate and the lender?s yield. Such a change may narrow the
interest margins earned by lenders and will likely increase their funding
risks as well as increase the costs of hedging. Several study group members
expressed concern that lenders may respond by exiting from FFELP, which
could potentially limit borrower access to student loan funds and disrupt
ongoing working relationships among schools, lenders, and students.

A special allowance payment (SAP) is a quarterly payment that the federal
government makes to FFELP lenders. It equals the difference between the rate
a borrower pays and what the current formula provides for lender yield.
Thus, if the lender yield exceeds the maximum borrower rate, the government
pays the difference to lenders in the form of the SAP. If the difference is
negative, the lender receives the borrower rate and no SAP. The SAP is
intended to maintain the incentive associated with the yield for the lender
if interest rates change, while permitting borrowers to pay lower rates if
interest rates fall. If CP declines, the quarterly lender yield declines,
and so does the SAP generally. The minimum or ?floor? yield to lenders is
the borrower?s rate for the year. The SAP is designed to make sure that the
lenders receive a reasonable rate of return on average. However, because
risks and costs can vary across borrowers and lenders, so can returns.
Therefore, implementing a market mechanism that results in a variety of SAPs
that reflect actual risks and costs could potentially lower the federal cost
of the program.

The borrower?s interest rate set in legislation is a maximum, and FFELP
lenders may offer lower interest rates to borrowers. Most often lenders
charge FFELP borrowers the maximum rates at the outset but then offer some
rate reductions during repayment. For example, they may reduce rates after
the borrower makes a certain number of payments on time or if the borrower
chooses to make payments through electronic funds transfers. Many lenders
now also offer rate reductions upon loan origination or discounts on the 3
percent origination fee.

Loan Servicers Loan servicers undertake the processing necessary to ensure
that cash flows of the loans are recorded and transferred to and from
lenders, guaranty agencies, and Education. Loan servicing is more
concentrated than are loan holdings or loan originations, with the top two
loan servicers accounting for about 50 percent of the loan servicing market.
Holders of student loans can service their own loans or they can contract
out for loan servicing to be performed by another entity; this is known as a
third-party servicer arrangement. 16 If servicing does not conform to
procedures Education established by regulation, the lender may not be
reimbursed if the borrower defaults. Better servicing can reduce the risk of
default and thus lower government costs. Additionally, schools benefit
because higher defaults could threaten some schools? eligibility to
participate in FFELP.

16 Third-party servicers are not subject to eligibility limitations by
organizational type. However, they must meet federal requirements of
administrative capability and financial responsibility, and they are subject
to audit by Education. Eligible lenders remain responsible for the
performance of their legal duties, despite any delegation of functions to
third-party servicers, and they must monitor their servicers? activities.

CHAPTER 1: BACKGROUND Page 17

Guaranty Agencies Guaranty agencies administer the federal guarantee on
FFELP loans, confirm borrower eligibility, monitor the status of loans,
provide delinquency and default aversion counseling, and provide claims
adjustments. FFELP loans are guaranteed by one of 36 nonprofit or state
agencies designated by the Secretary of Education. If a loan goes into
default, the guaranty agency generally pays the lender 98 percent of
principal and accrued interest. The federal government then generally
reinsures 95 percent of the guaranty agencies? payments to lenders and also
pays them fees for loan processing and issuance, account maintenance, and
default aversion. Guaranty agencies also retain a portion of the collections
they are able to make after a loan has gone into default. The government
also directly guarantees or reinsures FFELP lenders against the inability of
guaranty agencies to fulfill their guarantees because of insolvency.

As long as the lender complies with the regulations, the guarantee
substantially limits FFELP lenders? losses due to borrower default. Federal
reinsurance is available only if the guaranty agency correctly enforced
federal regulations and attempted to collect from delinquent borrowers. If
the loan servicing and collections are not done in accordance with federal
regulations, the reinsurance can be voided and can create losses for the
guaranty agency.

Guaranty agencies are authorized to collect a single insurance premium from
FFELP borrowers of not more than 1 percent of the principal amount of their
loans. Before 1998, some guaranty agencies had selectively reduced or
eliminated this insurance premium. Since reauthorization eliminated the fee
as a source of operating revenue, the elimination of guarantee fees has
become widespread because of market pressures. Any fees collected go into an
agency?s federal reserve account, which consists of federal funds that the
guaranty agency maintains to pay default claims.

The Department of Education The Department of Education role in
administering FDLP and FFELP differs significantly between the two programs.
Under FFELP, also known as the guaranteed student loan program, money is
borrowed from private lenders, such as banks, and the federal government
guarantees repayment if the borrowers default. Under FDLP, students or their
parents borrow money directly from the federal government through the
schools the students attend, which include vocational, undergraduate, or
graduate schools.

HEA provides the structure of FFELP program requirements and then authorizes
the Secretary of Education to administer the program. Among the Secretary?s
responsibilities is the promulgation of regulations to provide detail on how
requirements will be implemented. Sometimes, the statutory requirement is
very specific, in which case, the regulation simply restates the statutory
language (e.g., loan limits). In other cases, the statute expresses the
requirement in only the broadest terms and gives extensive authority to the
Secretary to define standards of compliance such as due diligence with
respect to loan collection.

Education is also involved in two types of cash flows relevant to the topic
of this report. First, it reimburses a guaranty agency after the guaranty
agency pays a lender for a defaulted loan.

CHAPTER 1: BACKGROUND Page 18

Second, it makes certain payments to lenders, including the difference
between the borrower?s interest rate and the lender?s yield, for all
borrowers, and interest during in-school and other authorized periods, for
borrowers with subsidized Stafford loans. Education is also responsible for

determining a student?s eligibility to receive federal student financial
assistance; gatekeeping, monitoring, and enforcement activities for
postsecondary schools; recognizing accrediting agencies; monitoring the
participation of guarantors, lenders, secondary markets, and third-party
servicers in FFELP; managing FDLP; collecting and resolving defaulted FFELP
and FDLP loans; maintaining a centralized database on individuals who apply
for and receive federal student financial assistance; managing the financial
aspects of the Federal Student Financial Aid Programs, such as receipt,
disbursement, accounting, and financial reporting for federal funds;
developing and disseminating information about the federal student loan
programs; developing cost estimates for the student loan programs; and
providing technical support and information for financial aid
administrators.

The federal costs of FFELP include borrower defaults and loan discharges,
borrower interest subsidies, and payments to lenders and guaranty agencies.
The federal costs associated with FFELP for fiscal year 2000 are presented
in table 7. Figure 1 depicts the cash flows in an FFELP program loan to and
from the student borrower, school, lender, loan servicer, guaranty agency,
and Education. Figure 2 depicts the cash flows associated with a FFELP
program loan in default.

Table 7: Federal Costs for the FFELP Fiscal Year 2000

Item Amount

Interest benefits $1,943,067,250 Default claim reimbursement 1,646,483,901
Special allowance payments 975,825,865 Death, disability, and bankruptcy
309,476,408 Account maintenance fees 180,000,000 Loan processing and
issuance fees 149,799,369 Collection costs 78,558,740

Total $5,283,211,533

Source: Department of Education.

CHAPTER 1: BACKGROUND Page 19

Figure 1: Cash Flows for an FFELP Loan From Origination Through Repayment

CHAPTER 1: BACKGROUND Page 20

Figure 2: Cash Flows for an FFELP Loan in Default Determining the actual
financial performance and the true costs of the FDLP at present is
difficult. Because FDLP is a relatively new program, little repayment
activity and historical data are available. This lack of historical FDLP
data has caused Education to rely heavily on data from FFELP to develop
estimates for most key cash flow items. Also, estimating the subsidy cost of
FDLP is difficult because of their dependency on interest rate projections
and corresponding fluctuations in subsidy costs which depend on the extent
of changes in interest rates. 17

MANDATE IN THE 1998 AMENDMENTS TO HEA AND THE STUDY GROUP?S WORK

The Study Group and Its Objectives, Scope, and Methodology

The 1998 amendments to HEA mandated that GAO and Education form a study
group to identify and evaluate a means of establishing a market mechanism
for the delivery of student loans. This study group consisted of
representatives of CBO, OMB, Treasury, entities making FFELP loans, other
entities in the financial services community, other participants in the
student loan programs, and other individuals designated by the Comptroller
General and the Secretary of Education. (See app. I for the full text of the
mandate and app. II for a list of study group members.) The group met as a
whole four times before the public release of a draft of this report, and
various group members corresponded with GAO and Education between group
meetings as well.

17 GAO recently reported that the estimated cost savings associated with
FDLP are sensitive to changes in the student borrower rate and the rate at
which Education borrows from Treasury. See Department of Education: Key
Aspects of the Federal Direct Loan Program's Cost Estimates (GAO-01-197,
Jan. 12, 2001).

CHAPTER 1: BACKGROUND Page 21

The mandate calls for the evaluation of at least three different market
mechanisms relative to 13 criteria. We grouped the evaluation criteria into
four sets, as shown in table 8.

Table 8: The Four Sets of Evaluation Criteria

Set Related criteria from the 1998 HEA amendments

Description of model, including variations A description of how the
mechanism will be administered and operated

(12) The proposed Federal and State role in the operation of the mechanism
(11) Transition procedures (13) a

Costs, savings, and effects on subsidies for program participants

The cost or savings of loans to or for borrowers, including parent borrowers
(1) The cost or savings of the mechanism to the Federal Government (2) The
cost, effect, and distribution of Federal subsidies to or for participants
in the program (3)

Effects on lender participation, loan availability, and service quality

The effect on the diversity of lenders, including community-based lenders,
originating and secondary market lenders (7) The availability of loans to
students by region, income level, and categories of institutions (10) The
effect on loan availability during a transition period (13) a The effect on
investment in human capital and resources, loan servicing capability, and
the quality of service to the borrower (6) The degree to which the mechanism
will provide market incentives to encourage continuous improvement in
delivering and servicing loans (9)

Simplicity, regulatory burden, and program integrity The effect on the
simplicity of the program, including the effect of the plan

on the regulatory burden on students, institutions, lenders, and other
program participants (5) The effect on program integrity (8)

Note: The numbers in parentheses refer to the number assigned to the
specific criteria in the legislative mandate. The analysis related to the
fourth criterion-the ability of the mechanism to accommodate the potential
distribution of subsidies to students through an ICR option-was similar
across all the models. Rather than repeat our analysis of this criterion for
each model, we discuss it separately in chapter 7. Therefore, this criterion
does not appear in this table.

a We split the thirteenth criterion, on transition procedures, into a
descriptive part and a part related to loan availability issues.

The group solicited proposals for market mechanisms that might meet the
intent of the mandate. Group members, as well as outside observers,
submitted proposals for consideration. GAO and Education considered all
submissions and grouped them where we saw some similarities within the
proposals. For example, under the general idea of conducting an auction for
the right to make FFELP loans, there were two main possibilities as to how
such an auction might be conducted, as well as many different considerations
at the level of specific auction design. Rather than selecting one of the
main possibilities and a specific set of auction design decisions, we
analyzed this type of model in general, weighing the advantages,
disadvantages, and tradeoffs in the different possible choices.

CHAPTER 1: BACKGROUND Page 22

In study group meetings and other conversations with group members,
including staff from the other government agencies, we discussed the various
models and their implications. Two of the group meetings were devoted to
generating ideas about possible changes to FFELP, including some general
factors to consider and some specific proposals. In the third meeting, the
group discussed preliminary analysis results, and in the fourth meeting, the
group discussed a draft of this report.

GAO and Education also researched comparable programs in existence, such as
other federal auctions and the experiences of student loan programs in other
countries. We used previously published reports, material available from the
Internet, and contributions from study group members for much of this
research.

The Five Models Discussed in This Report

The potential changes to FFELP discussed in this report are grouped under
five general models: adjustments to the current system, loan origination
rights auction, loan sale, federal funding, and market-set rate.

A model involving adjustments to the current system would change the process
of setting lender yield by building upon competition for loan originations,
purchases, and servicing that is a part of FFELP today. Proposals within
this model include (1) making periodic incremental downward adjustments in
the lender yield while monitoring any resulting changes in lender
participation; (2) using information about lenders? cost of funds and
servicing costs to set the yield; (3) establishing an independent blue
ribbon commission to gather information, analyze data, and either set lender
yield or recommend a yield level to the Congress; and (4) using information
from secondary market loan sales transactions to determine the value lenders
place on loans, and establishing a one-time payment to lenders based on this
value to replace the quarterly SAP.

A loan origination rights auction would require lenders to bid to
participate in FFELP. The borrower?s interest rate and other loan terms
would be set through legislation or regulation and could remain the same as
today. Lenders would generally submit bids to participate in FFELP loan
origination, and bidders offering the best terms to the government would win
the allocations. Variations within this type of model differ as to how
restrictive the outcomes would be. A volume procurement auction would
require lenders to bid on the right to an amount of loan volume at a
particular interest rate (or price, for a predetermined interest rate),
perhaps for

CHAPTER 1: BACKGROUND Page 23

volume within different sectors of schools. 18 An actual origination rights
auction, in contrast, would require lenders to bid for the right to
originate at groups of schools.

The loan sale model presumes that private lenders would purchase loans
sometime after they are originated by the government or by some entity under
government contract. The loan would be government property until sold at
auction. The government would assemble packages of loans, and private
lenders would evaluate the packages and submit bids. Loans would carry the
same borrower terms as today, and the lender yield would be set
legislatively and could be the same as the borrower?s interest rate. Loan
origination could be performed by the federal government; schools, private
lenders that bid to originate and service the loans but operate under
government contract, or state agencies or some other entity. Loans could be
sold immediately after origination or held until a borrower graduates or
otherwise completes schooling, at which time a borrower?s loans could be
sold as part of a package. Loan servicing could be performed by the
purchaser, if servicing responsibilities are sold with the loan, or by some
other entity.

The federal funding model would adjust funding costs for FFELP lenders,
ultimately determining lender?s net yield. Private lenders would continue to
originate loans, except that they would be allowed to borrow funds from the
federal government to make FFELP loans. The interest rate at which they
borrow, in conjunction with the lender yield on FFELP loans, would determine
lenders? net yield. This process could be made mandatory for lenders who
want to participate in FFELP, or it could be offered as an option, with
lenders still being allowed to fund student loans through traditional
methods. Once lenders receive these funds, or if they choose not to
participate, everything else would proceed functionally as in the current
system. The funds lenders borrow from the government could be payable on
either a regular amortization schedule or a schedule tied to borrower
repayments on the FFELP loans made with the federal funds. Finally, the
markup over a Treasury-based interest rate, such as the 91-day T-bill rate
or the 10-year Treasury bond rate, could be fixed, or lenders could bid on
the markup.

In a market-set rate model, the borrower or the school chooses among lenders
for the best attainable interest rate. Unlike with the other models, the
borrower?s interest rate would no longer be set by legislation. Instead, it
would emerge from negotiations between lenders and borrowers or likely
between lenders and schools. 19 The resulting rate would thus be both the
borrower?s rate and the lender?s yield. Loans would still be federally
guaranteed, and lenders would face no limitations on the amount of loans
they could originate and hold. Under this model, some borrowers might pay a
higher interest rate. Possible variations could limit the potential effects
on disparate groups of borrowers. For example, one alternative would place a
limit on the maximum range of rates or fees a lender could offer. 20 Another
possibility would be

18 For instance, volume for 2-year and 4-year schools, or public, nonprofit
private, and proprietary schools, could be offered in separate auctions. 19
Lenders would be likely to assume that enough common characteristics exist
across borrowers at a school to

enable them to give the same rate to all a school?s borrowers rather than to
try to negotiate with each borrower. 20 This would be similar to a limit
that exists in the Federal Housing Administration (FHA) insurance program.
FHA limits the variation between the highest and lowest ?mortgage charge
rates? (analogous to FFELP origination fees) that a lender may offer to any
borrower within a given time period to no more than 2 percentage points
within well- defined markets, as defined by geographic region and ?risk
characteristics,? among other factors.

CHAPTER 1: BACKGROUND Page 24

for the government to provide subsidies to borrowers who cannot otherwise
obtain low interest rates from lenders.

The models we discuss might or might not be accompanied by changes in the
role of guaranty agencies in FFELP. To analyze such changes, the models
would have to be specified in more detail than is possible in this report.
For this reason, we do not discuss possible changes in guaranty agencies?
role.

The models we discuss may present budgetary issues. The Credit Reform Act of
1990 changed the budget treatment of federal credit programs to put the
scoring of direct and guaranteed loans on an equal footing. Under credit
reform, loan program budget estimates for a given fiscal year reflect the
estimated long-term cost to the government when the loan is disbursed. The
cost to the government is the net present value of all cash flows to and
from the government, including loan disbursements, repayments of principal,
interest payments, recoveries or proceeds of asset sales, and payments by
the government to cover defaults, interest subsidies, origination and other
fees, penalties, and recoveries. (Federal administrative costs are excluded
from these calculations.) For the student loan programs, Education generates
these estimates for the administration using a governmentwide credit subsidy
model developed by OMB. Projected program cash flows for the life of each
loan cohort are generated by an Education model and discounted by an OMB
credit subsidy model using a discount rate based on Treasury instruments
with a maturity comparable to that of the underlying loans. For the student
loan programs, this comparable maturity rate is the 10 to 20 year bond rate.
Shifts in the timing of payments and receipts would need to be considered
within the context of credit reform requirements. To analyze such changes,
or "score" the proposals for budgetary purposes, would require more detailed
specifications than is possible in this report. For this reason, we do not
calculate or compare the budgetary costs of models we present in this
report.

Page 25

CHAPTER 2 ADJUSTMENTS TO THE CURRENT FFELP SYSTEM

The proposals to adjust the current FFELP system focus on lenders?
sensitivity to interest rates, secondary market trends, and lender costs.
Whether or not the Congress maintains its legislative authority to set the
lender yield, each of these adjustments would modify the process, making
FFELP a system more responsive to student loan market conditions. Each of
the four proposals would preserve the existing relationships between
borrowers, schools, lenders, and other FFELP participants, while relying on
the acquisition of new or better information to adjust lender yields.
Current options for borrowers to consolidate loans would continue to be
available under these proposals. These proposals would affect lender yields
similarly to the proposals discussed in other chapters. However, the
proposals discussed in this chapter might take longer to affect lender
yields than would the proposals we discuss elsewhere in this report. Each
adjustment to the current system could reduce federal costs but possibly at
the expense of discounts that lenders currently offer to some borrowers. The
extent to which each proposal can realize these savings depends on choices
regarding design and implementation.

DESCRIPTION OF MODEL AND VARIATIONS

Each of the proposals maintains the structure of FFELP while changing the
mechanism for determining lender yield. The proposed adjustments to the
system include

incremental adjustments proposal, cost of funds proposal, proposed
blue-ribbon commission that would set lender yields, and loan transaction
proposal that would use data from secondary market transactions to improve
the yield-setting process.

Incremental Adjustments

Under the incremental adjustments proposal, the Congress would reduce the
lender yield incrementally, either annually or as needed, while carefully
monitoring lender participation after each adjustment. If such a reduction
in lender yield caused lender participation to drop to a dangerously low
level or caused an erosion of service quality, the Congress could raise the
lender yield until it believed that lender participation and service quality
were once again at acceptable levels. Legislation would be necessary for
each adjustment, as it is in today?s program, unless the Congress
specifically delegated the rate-setting role to an executive agency or to an
independent government commission.

Cost of Funds

The cost of funds proposal relies on lenders? cost data to determine lender
yield. Data on lenders? costs would include the actual expense to lenders
associated with raising funds, loan

CHAPTER 2: ADJUSTMENTS TO THE CURRENT SYSTEM Page 26

origination, and loan holding, as well as an estimation of servicing costs
deemed reasonable to maintain high-quality service to borrowers. On the
basis of these factors, the Congress could determine an appropriate
differential between the cost of funds and the cost of servicing. This
differential could be applied to the determination of a market-based lender
yield. The Congress, an executive agency, or an independent government
commission could administer the cost of funds approach.

Blue-Ribbon Commission

There are two distinct ways in which a blue-ribbon commission could be
organized. It could be an advisory commission whose members could include
expert staff from the executive branch, representatives from the student
loan or banking industries, and recognized authorities on higher education
finance. This type of commission could consider the costs of funds, loan
transaction cost, or other similar information. The commission could also,
for example, recommend different methods for paying or calculating special
allowance payments. The Congress would retain responsibility for setting the
rate and could accept or reject the commission?s recommendation as it saw
fit. Alternatively, the Congress could establish the commission as an
executive branch agency or as an independent federal entity and could give
it the responsibility of determining the lender yield.

Loan Transaction Information

This proposal would use data from loan transactions to gauge the underlying
value that participating FFELP lenders place on loans. The government would
collect information on the terms and conditions of secondary market
transactions in the existing FFELP. This information could serve as the
basis for a modified determination of lender yield.

Another component of this proposal is its modification of the SAP that
lenders receive. Under this plan, the SAP would be made to the loan holder
only once, either when the loan was initiated or when and if consolidation
or refinancing occurred, rather than quarterly as in the current FFELP. The
proposal empowers an executive agency or an independent commission to
establish this single supplemental payment for loans made each year rather
than having the Congress continue to legislate the SAP. The amount of the
single supplemental payment would be influenced by data gathered from
transactions in the secondary market. Additionally, the amount of a modified
SAP could vary according to lender size, student loan activity, type of loan
or type of school, or any other characteristics deemed relevant.

COSTS, SAVINGS, AND EFFECTS ON SUBSIDIES FOR PROGRAM PARTICIPANTS

Each of the proposed adjustments to the current system could potentially
reduce or increase the federal operating costs of FFELP. Further, each of
these proposals makes two assumptions regarding the current FFELP: (1)
lenders? participation in the student loan program is contingent on profit
sustainability, and (2) the current lender yield results in profits beyond
those necessary

CHAPTER 2: ADJUSTMENTS TO THE CURRENT SYSTEM Page 27

to maintain sufficient active lender participation. These assumptions, and
the corresponding reforms these proposals envision, drive the cost and
savings effects for program participants, potentially saving money for the
taxpayers.

If the lender yield is lowered incrementally and lender participation is
maintained, the action might cost lenders a portion of their profits while
saving the government some of what it otherwise would have spent on payments
to lenders. As lender yield is reduced, profits for participating lenders
could fall and lenders could begin to leave the program. Smaller lenders
(those with low volume or high cost) would likely be the first to exit the
program. However, remaining lenders might be likely to make up for this loss
by acquiring the assets of the outgoing firms or simply by increasing
lending activity. Thus, remaining lenders would grow larger, further
concentrating the market. Larger lenders, able to operate at a lower cost
per dollar of loan originated, might pass savings on to borrowers or might
make improvements to service delivery.

Like the incremental adjustments proposal, the cost of funds approach could
result in either costs or savings to the federal government and program
participants. The cost of funds approach calls for a mechanism by which
lenders can detail and report their costs and the government can form
objective judgments to analyze and justify these estimates. Although some
savings might be realized through reduced federal payments to lenders, this
proposal would impose additional administrative costs on the federal
government because cost data are not readily available. If these figures
were accessible, the government could assess cost data by using a number of
criteria. These could include the types of students or schools served.

An additional question is the role that the government should have in
determining the accuracy of cost estimates under this proposal. The outcome
of these decisions directly affects both the federal administrative cost of
implementing the model and the federal payments to lenders in the program.
If, for example, lenders document substantial costs for raising funds and
the government accepts these figures, then the lender yield could rise,
increasing the government?s cost. In contrast, if the government chooses to
contest these figures, it may become embroiled in a political and
controversial investigation that could stall FFELP loan delivery.

The use of loan transaction information could result in decreased costs for
the taxpayer. For example, while the federal government would incur some
costs associated with the pursuit and analysis of information regarding
secondary market loan sales, these costs might be small compared to any
potential savings from reduced payments to lenders.

EFFECTS ON LENDER PARTICIPATION, LOAN AVAILABILITY, AND SERVICE QUALITY

The adjustments to the current system would be likely to affect the current
FFELP structure less than the models discussed in other chapters. These
proposals could preserve or possibly improve the quality of loan service.
There is consensus among the study group members that FFELP lenders have
responded to the competitive pressure introduced by FDLP.

CHAPTER 2: ADJUSTMENTS TO THE CURRENT SYSTEM Page 28

The incremental adjustments proposal might reduce the number of lenders in
the program. For example, the adjustment process might initially set the
lender yield so low that it would force out lenders with high costs or low
volume. If these lenders leave the program, they may find re- entry costs
prohibitively high should the yield be raised in a subsequent adjustment.
Determining what number of lenders is sufficient is a value-laden
assessment. Mergers and acquisitions resulting in fewer lenders but the same
loan volume and more efficient service provision may be desirable. However,
a decline in lender participation might reduce loan availability or disrupt
services to students and schools.

The cost of funds proposal relies more heavily on lender information to set
the lender yield by considering lenders? operating costs as a component of
the yield determination. Thus, the cost of funds proposal has the potential
to improve the quality of service that lenders provide. If lenders recognize
that the yield will reflect their funding costs as well as their service
costs, they may have the incentive to reduce costs for borrowers or to
enhance service delivery by purchasing or developing new technology. While
these actions would have clear benefit to borrowers, higher lender costs
would force the government to increase the yield, raising federal FFELP
costs.

Merely establishing a blue-ribbon commission would not affect lender
participation, loan availability, or lender diversity. Establishing a
blue-ribbon commission is a procedural reform, and commissioning an
independent entity alone to implement adjustments to FFELP would do little
to alter the provision of loans or the quality of service. Consequently, the
extent to which loan availability, service quality, or lender diversity
would change hinges on the rate-setting strategy the commission chose to
employ.

The implementation of the loan transaction proposal relies on
lender-provided information. This proposal requires the government to
collect secondary market sales data. According to Education, this new data
collection would impose additional reporting burdens on lenders, but the
extent is unclear.

SIMPLICITY, REGULATORY BURDEN, AND PROGRAM INTEGRITY

Each proposal presents logistical and regulatory challenges to the Congress.
The incremental adjustments approach calls for carefully monitoring lender
activity, while the cost of funds proposal represents a shift toward
oversight and evaluation of lender operations. Lenders? potential
unwillingness to release cost information, as well as the difficulty
inherent in analyzing it, would present challenges to implementing this
proposal. While the blue-ribbon commission approach aims to remove rate
setting from the congressional arena, the extent to which an independent
entity could effectively oversee a program of this size remains unknown.
Although a blue-ribbon commission might be empowered to set the yield, it
would be subject to administrative requirements that might leave it open to
the same budget and political pressures that the Congress faces. Finally,
collecting data from loan sale transactions might complicate FFELP
administration, requiring the government to track the prices associated with
secondary market loan sales. Since lender yield would still have to be set
by the Congress or some other entity under each approach rather than
emerging from a market process, the rate-setting

CHAPTER 2: ADJUSTMENTS TO THE CURRENT SYSTEM Page 29

mechanism still might not be free from the political process. Any of these
approaches could result in either more or less regulation, while all would
impose new administrative burdens.

The incremental adjustments proposal could entail additional regulatory
burden for FFELP. Specifically, the incremental adjustments proposal would
require carefully monitoring lender response each time an adjustment was
made. The need for timely information and for prompt action to counteract
excessive adjustments requires special considerations. Some mechanism would
be required to halt further downward adjustments from negative consequences
such as lender departure. In addition to regulatory burden, this proposal
could create significant administrative burden. If this proposal were to be
implemented, it might require delegation to an independent entity such as
the blue-ribbon commission. In any case, some entity would have to monitor
the program and make recommendations as to further changes.

Concerns exist regarding the implementation of the cost of funds proposal as
well. Administering a cost of funds model would introduce additional data
collection needs. This proposal requires the annual development of two sets
of proprietary lender information (costs of raising funds and an estimation
of servicing expenditures) for use in setting lender yield. Collecting and
verifying lender cost information could present new challenges. The data
collection process itself might be impeded by the complexity of lender
transactions (such as securitizations and forward purchase agreements) and
by the variation in lender structure (free- standing, part of a holding
company whose main focus is student loans, part of a larger bank for which
student loans may be a small business). Inaccurate and inadequate data would
limit the government?s ability to manage this program. The proposal would
also require the subsequent modification of lender yield if lender cost data
indicated that the current rate were inappropriate. Some of these
assessments, however, require normative judgments about the degree to which
costs should vary by the types of schools and students lenders serve.

Many of these challenges to simplicity and regulatory burden apply also to
the blue-ribbon commission proposal. Panel selection might become
politicized and controversial. Additionally, while establishing the
commission would reduce the direct congressional role in setting yield, it
remains unclear whether the commission would be advisory or authoritative,
how the panel would access data, and whether consensus would be required for
commission decisions. The commission would need authority to gather relevant
data as well as the resources to analyze them. Implementation decisions
would also center on the type of legislation necessary to establish the
commission?s tasks and objectives and on the delegation of congressional
staffers for administrative and substantive assistance.

The blue-ribbon commission might or might not remove the setting of the
lender yield from the political process. If the commission were advisory,
then the Congress would still be responsible for setting the rate and could
still face political pressure from FFELP participants who disagreed with its
recommendation. If the commission were a federal agency, then the Congress
would not face such pressure directly but the commission might. Political
pressure on the commission?s decision could be especially great if the
Congress required it to publish a public notice of its proposed decision and
respond to public comments on its proposal before implementing it. The
Congress has imposed such a requirement on most federal agency rulemaking
decisions. If the Congress decided to exempt the commission from this
requirement, then the commission would

CHAPTER 2: ADJUSTMENTS TO THE CURRENT SYSTEM Page 30

be less subject to political pressure but would also be less accountable to
the public and might lose the benefit of information that public comments
could provide.

One federal agency that functions like a blue-ribbon commission is the
Postal Rate Commission (PRC). PRC is an independent executive agency that
recommends postal rates to the U.S. Postal Service?s board of governors.
When the Postal Service requests a rate change, it must provide PRC with
data about the relevant costs. Before making a rate recommendation, PRC must
hold public hearings, which include testimony and written submissions from
the Postal Service, its competitors, and others that are interested. In
making a recommendation, PRC must consider nine legislatively established
criteria as well as the record of its hearings. The postal ratemaking
process is lengthy and complicated. There is a process for modifying or
rejecting PRC recommendations, but those recommendations have usually been
accepted without modification.

The loan transaction information proposal requires the government to monitor
lender participation and corresponding secondary market transactions.
However, collecting and verifying private secondary market transaction
information might present challenges, especially considering the proprietary
nature of agreements between secondary market participants. Carefully
monitoring the secondary market will demand administrative oversight,
forcing decisions about what information will be gathered and how it will be
analyzed.

The federal government, or its designated blue-ribbon commission, could
periodically auction a limited amount of direct loans as an alternative
source of secondary market information. Existing law authorizes sales of
direct loans. However, selling a portion of the FDLP portfolio could present
budgetary challenges. Under federal credit reform, such sales could possibly
be recorded as a loss, which would require offsets under current budget
rules. Other ways to offset the subsidy cost of selling direct loans would
be to securitize some of the FDLP loans in Education's portfolio.

Finally, implementing an auction of FDLP loans could increase administrative
burden. Insufficient competition could threaten the integrity of the bidding
process. It would also involve decisions on who would facilitate the auction
and how competition would be preserved. Auction design issues and the
importance of competition are discussed in detail in appendix III. The
results of the analysis for the adjustments to the current system are
summarized at a broad level in table 9.

CHAPTER 2: ADJUSTMENTS TO THE CURRENT SYSTEM Page 31

Table 9: Summary of Analysis for Adjustments to the Current System Model

Description of model, including variations All proposals maintain current
roles and relationships for FFELP

participants. Proposals use information from the marketplace-for example,
data on lenders? cost of funds or secondary market sales-to set the yield
level. Might involve establishing a blue-ribbon commission to either set the
yield or recommend a yield level to the Congress.

Costs, savings, and effects on subsidies for program participants

Federal FFELP costs could increase or decrease but are more likely to
decrease. Savings to the taxpayer, if any, could come at the expense of some
discounts that lenders currently offer to some borrowers.

Effects on lender participation, loan availability, and service quality

Loan availability should not be greatly affected if adjustments are
incremental. The number of lenders in the program could fall if adjustments
force out lenders with low volume or high cost. Service quality would be
subject to the same pressures.

Simplicity, regulatory burden, and program integrity

Relationships between FFELP participants would change little. Oversight of
lender operations, and required data gathering and analysis, might prove
difficult. Blue-ribbon commission would require administrative framework.

Page 32

CHAPTER 3 LOAN ORIGINATION RIGHTS AUCTION

The origination rights auction model would require lenders to bid for the
right to originate and obligate them to lend FFELP loans. Auctions rather
than the legislative process would set the yield that lenders would receive.
If the auctions were sufficiently competitive, bidding would eliminate any
excess in what the government currently pays lenders and would maintain
lender incentives to reduce their costs. Therefore, auctions have the
potential to reduce federal FFELP costs. The effects of auctions on borrower
interest rates, the distribution of federal payments to lenders, the quality
of service, and the burden on borrowers and schools are uncertain. The
effects depend on the details of auction design as well as on a number of
assumptions about markets that may or may not be valid in relation to the
student loan program. The diversity of lenders would likely decline unless
special provisions were made to ensure the participation of small lenders.
Students? and parents? access to loans could be reduced during the
transition to auctions and, depending on auction design details, perhaps
permanently. The federal government would have to develop new regulations to
govern the conduct of the auctions and, in particular, to preserve
competition in bidding. Students, lenders, and schools would all bear the
burden of adjusting to the auction system.

DESCRIPTION OF MODEL, INCLUDING VARIATIONS

In origination rights auctions, lenders would bid for the right to originate
student loans. The federal government would develop regulations to implement
the auctions and would maintain its current subsidies to borrowers. Major
choices that must be made in designing the auctions include

whether the auctioned rights are rights to originate all loans at particular
schools or groups of schools or rights to originate a specified volume of
loans at any of a number of schools, whether origination rights or volumes
are grouped by school, whether each student borrower is allowed to keep all
loans with a single lender, and which bidding method is used, including
whether lenders bid on the interest rate they would receive on loans they
originated or on the price they pay to originate loans at a predetermined
interest rate.

Basic Model and Role of Participants

All versions of the origination rights auction model share several features.
The borrower?s maximum interest rate, in-school subsidy policy, and some
other loan terms would be set through federal legislation or regulation, as
at present. Lenders would bid for the right to originate loans. They would
indicate the prices that they were willing to pay or the interest rates they
were willing to accept for the right to originate at particular schools or
to originate a specified loan volume. In each auction, a lender could submit
a single bid or multiple bids that incorporated different prices or interest
rates. The government would sort bids by price or interest rate, and

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 33

the most favorable bids would win the right to originate. It would be
possible to set aside some origination rights for nonbidders. However, most
rights would be determined through the bidding process. Lenders who
submitted winning bids would be allowed to originate only at the schools
where they won rights or to originate only the volume they won. Except for
nonbidders who received set-asides, nonwinning lenders would not be
permitted to originate loans. Lenders could buy and sell origination rights
after the auction but before loans were originated. They could also buy and
sell loans after origination, as they do today.

The federal government could be responsible for conducting the auctions and
ensuring borrowers access to loans. The government would determine whether
lenders were eligible to participate in FFELP. Education might also act as a
loan consolidator, as in the current FDLP. It might need to establish
procedures to ensure that eligible borrowers received loans and that winning
lenders performed as promised. The government would also need to establish
procedures to ensure access to loans when loan demand exceeded auctioned
loan volume. In addition, it might need to re-auction loan origination
rights for lenders who could not or would not perform as promised, although
it would be possible instead for these rights to be resold in the secondary
market. During the transition from the current system of setting FFELP
lender yield to an auction system, the government would have to develop new
regulations to govern the conduct of auctions, develop the expertise to
conduct auctions itself or contract with a nongovernment entity to conduct
the auctions, educate lenders and borrowers about auctions, and act as
lender of last resort to ensure that all eligible borrowers have
uninterrupted access to loans. No state role is envisioned beyond the one
that now exists in FFELP.

Major Variations on the Model

Several high-level decisions must be made in designing an origination rights
auction model. The items to be auctioned could be either the rights to
originate all loans at particular schools or groups of schools or the rights
to originate a specified volume of loans at any of a number of schools.
Origination rights or volumes may or may not be grouped by school. The right
to lend includes the obligation to lend. A variety of bidding methods could
be used. Finally, each student borrower may or may not be allowed to keep
all loans with a single lender.

The study group considered three alternative designs for an origination
rights auction. In the ?rights auction,? lenders would bid for the right to
originate all loans at a particular school or group of schools. In the
?volume procurement? design, lenders would bid for a specified loan volume
that could be used at any school and either all winning lenders would pay
the same price or each would pay the price it bid. The Income Dependent
Education Assistance (IDEA) version resembles volume procurement, except
that each winning bidder would pay the price bid by the next highest bidder
and there would be an explicit ICR provision. These three proposals
illustrate but do not exhaust the range of available alternatives. Other
combinations of auction characteristics are possible. For example, the right
to originate at specific schools could be combined with each winner?s paying
the next highest bid price. Any origination rights auction could either
include or exclude income-contingent repayment.

The three versions of the model illustrate a fundamental choice that must be
made in the design of an origination rights auction. In the rights auction
version, lenders bid on the right to originate

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 34

loans to students at a particular school or group of schools. In the volume
procurement and IDEA versions, they bid on a specified volume of loans,
which they can lend to students at any of a number of schools.

Under any alternative, schools may or may not be grouped together when
origination rights are auctioned. In rights auctions, grouping would mean
that lenders would bid for the right to originate at particular groups of
schools. In volume procurement, it would mean that lenders would bid on a
volume of loans that they could use only at particular groups of schools.
Under either alternative, the rights to originate at all groups of schools
could be sold at the same auction, or else a separate auction could be held
for each group. At the opposite extreme, the government could decide not to
group schools but to require lenders to bid for the right to make loans to
all FFELP-eligible students, regardless of school.

A variety of bidding methods could be used. One important issue is whether
lenders would bid on the interest rate they would receive on loans they
originated or on the price they would pay to originate loans at a
predetermined interest rate. If lenders bid on an interest rate, then that
rate would likely be expressed as a markup over a reference rate that bears
some relationship to lenders? cost of funds, such as a T-bill rate or the CP
rate. If lenders bid on a price, then their bids might or might not be
allowed to be negative. (A negative bid would be one in which the lender
asked the federal government to pay it a specified price in order to make it
willing to lend.) Issues related to auctions, including the grouping of
schools and the choice of bidding method, are discussed further in appendix
III.

A final important design choice is whether to allow each borrower to borrow
from a single lender, at least for the duration of a particular degree
program. For example, a student entering an undergraduate program could be
given the right to borrow from one lender until he or she completed that
program, even if that lender did not win the right to originate at that
student?s school for all the years in which the student was enrolled in that
program. Alternatively, a student could be required to switch lenders
whenever his or her previous lender lost the right to originate at his or
her school.

COSTS, SAVINGS, AND EFFECTS ON SUBSIDIES FOR PROGRAM PARTICIPANTS

Auctions have the potential to reduce federal FFELP costs, but their ability
to realize this potential depends on whether there is sufficient competition
in bidding. The frequency with which auctions were held, the definition of
auctionable rights, and the choice of bidding method could also affect
federal costs. The distribution of federal payments to lenders would be more
uneven in some types of auctions and unchanged in others. Auctions could
reduce all borrowers? interest rates, raise some borrowers? rates, or leave
all borrowers? rates unchanged.

In an ideally competitive auction, bidding competition among lenders would
eliminate any excess profits that lenders would otherwise receive. The
bidding process would ensure each lender just enough profit to make that
lender willing to remain in FFELP. Furthermore, the winning lenders would be
the ones with the lowest loan origination costs. Appendix III describes an
ideally competitive auction, explains how such an auction would produce
these

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 35

results, and shows how the less-than-ideal auctions that would be likely
could produce most of the same results.

Regardless of whether lenders currently receive excess profits, lenders
might reduce their costs by developing less expensive methods of originating
loans to secure volume by lower bids. Lower bids would mean reduced federal
payments to lenders. However, origination rights auctions will only reduce
federal FFELP costs if there is sufficient competition among lenders.
Whether the student loan origination market at present is sufficiently
competitive for origination rights auctions to reduce federal costs is
unclear.

Evidence from other federal auction programs suggests that origination
rights auctions could produce savings, which could be retained by the
federal government or passed on to the borrower. In the Health Education
Assistance Loan (HEAL) program, under which lenders bid for the right to
originate student loans to students of the health professions, the federal
government kept the volume artificially low by capping loan volume. This
resulted in a drop in the spread over T-bills from 3 percent in 1993 to 1.5
percent in 1997. 21 The Federal Communications Commission?s (FCC) auctions
of wireless spectra have yielded revenues exceeding their costs, but these
auctions took several months on some occasions, and auctions on certain
spectra had to be recompeted because of the lack of competition.

The current student loan market, however, may not be competitive enough to
enable origination rights auctions to produce savings for the federal
government. As chapter 1 showed, a limited number of large lenders hold a
large share of loans in the student loan market; furthermore, it may be
costly for new lenders to enter the market. If the limited number of lenders
is the result of natural scale economies, the associated cost savings have
the potential to benefit all participants. If the limited number of lenders
results in less competition and excess lender profits, this market structure
is undesirable, with or without auction mechanisms, and is best addressed by
our nation?s laws regarding monopolistic practices. In either case, because
the number of lenders is limited, the dominant lenders? influence on the
winning bids could conceivably lead to FFELP costs that were as high as or
higher than current costs. Some believe that lenders might also collude in
setting their bids. The relatively small number of lenders in the market
would make collusion easier.

FCC?s wireless spectrum auctions illustrate this problem and some solutions
to it. In some FCC auctions, bidders were able to signal their bidding
strategies to other bidders by using the last three digits of the amount bid
or by strategically withdrawing from the bidding. FCC corrected the former
problem by setting minimum bid increments that bidders were required to use
in raising bids. It solved the latter problem by limiting the number of
withdrawals a bidder was allowed to make during an auction.

A competitive secondary market in loan origination rights could at least
partially compensate for insufficiently competitive auctions. Such a market,
which does not now exist, would enable lenders to buy and sell rights to
originate student loans outside the auctions and would add a new

21 In contrast to FFELP or FDLP, the HEAL program never exceeded more than
$500 million in insurance authority for a single year. The experiences under
the HEAL program may or may not indicate the potential for savings through
the adoption of an auction mechanism in a larger program.

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 36

layer of complexity to the program. 22 The Environmental Protection Agency
(EPA) sulfur dioxide emission allowance auctions provide an example.
Allowances sold at those auctions have been concentrated in the hands of a
relatively few firms. However, the secondary market, on which allowances are
traded outside auction, is much larger than the auction market. After the
first few years of EPA auctions, the secondary market, rather than the
auctions, appears to have determined the price of allowances. Because the
auction price reflected the secondary market price, insufficient competition
in the auction might have been less of a problem than it would have been in
the absence of a secondary market. 23

Although sufficiently competitive auctions could reduce federal payments to
lenders, their overall effect on FFELP costs would also depend on details of
the auction design. The frequency with which auctions were held, the
definition of auctionable rights, and the choice of bidding method could
influence FFELP costs. They could do so by affecting either federal payments
to lenders or the cost of conducting auctions.

At present, all lenders receive a uniform federal payment equal to the
difference between the lender yield and the maximum borrower interest rate,
both set by law. Some types of origination rights or volume procurement
auctions would lead to an unequal distribution of federal payments to
lenders, while others would not change the current pattern of equal payments
to all lenders. Both the choice of bidding method and the grouping of
schools would affect the distribution of federal payments to lenders. A full
discussion of competition, auction design and frequency, bidding methods,
and the grouping of schools is in appendix III.

In addition, if borrowers do not have the option of borrowing from a single
lender until they complete their education program, these proposals could
increase the attractiveness of consolidation to borrowers. If lenders expect
the use of consolidation to increase, they will bid less, and federal
revenues from the FFELP auctions could decline.

EFFECTS ON LENDER PARTICIPATION, LOAN AVAILABILITY, AND SERVICE QUALITY

Auctions could either reduce the quality of service or have no effect on it.
In the absence of special provisions, they could reduce the diversity of
lenders. The definition of auctionable rights, the choice of bidding method,
and the frequency with which auctions were held could affect borrower and
lender access to FFELP as well as the quality of service. Some students and
parents could have difficulty obtaining loans during the transition to an
origination rights auction. Some transitional challenges might be
anticipated and thus minimized through the use of a pilot auction program,
although results from a pilot might be difficult to interpret if
participating lenders are uncertain as to the likelihood of the program?s
continuation.

22 A secondary market in origination rights would differ from the existing
secondary market, in which loans are bought and sold after they have been
originated. 23 FCC and EPA auction models are dissimilar in many ways from a
loan origination rights auction. For example,

end users of wireless spectrum communication often have readily available
alternatives and thus are not likely to suffer if an auction winner does not
provide such service. Further, consumers may actually benefit if the holders
of emission rights are not exercised. Caution is necessary in using these
auctions as models for an origination rights auction.

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 37

Auctions may or may not reduce service quality but are unlikely to improve
it. The number of lenders permitted to serve each school, the frequency with
which auctions were held, and the method by which bids were evaluated could
also affect service quality.

Auctions might shift the balance of competition in the student loan market
toward price competition and away from service competition. At present,
FFELP lenders compete to a limited extent on the basis of price, through
discounts to preferred borrowers. They also compete on the basis of service.
Auctions would force lenders to place more emphasis on price competition,
since the major (and, in most variants, the sole) criterion for determining
auction winners would be the price or interest rate bid. It is possible,
although not certain, that lenders would pay less attention to service as
they paid more attention to price.

Service quality could be higher if multiple lenders were allowed to serve
each school than if there were only one lender per school. If more than one
auction winner could serve each school, then students could choose among
lenders. Lenders would have an incentive to compete for students on the
basis of service quality. This incentive would not exist if only one lender
could win the right to serve each school.

Finally, service quality could be improved if the government were allowed or
required to take service quality into account in choosing auction winners,
but there are important drawbacks to this. One option would be to evaluate
bids on the basis of service quality as well as price or interest rate.
Another would be to prequalify bidders using service quality criteria and
then conduct the auction on the basis of price or interest rate alone. These
procedures would make auctions more complex than if bids were evaluated
solely on the basis of price or interest rate. The added complexity could
deter small lenders from participating in FFELP. Loans that had service
quality features as legally enforceable terms would also be more difficult
to sell in the secondary market than loans without such features.

Auctions could increase, decrease, or have no effect on lender diversity in
the short term but might reduce it in the long term. 24 Provisions to
protect small bidders? access to FFELP could be built into the auction
design. Small-bidder protections may make auctions more competitive as well
as ensure small lenders? participation in FFELP. The choice of bidding
method, the number of lenders allowed to originate at each school, and the
frequency with which auctions were held could also affect lender diversity.

In the short term, auctions may have any of several effects on the number of
FFELP lenders. They may increase lender participation by enabling new
lenders to enter the market on relatively equal terms with existing lenders.
Or they may have no effect on participation if entry into the origination
market is so expensive that no new lenders are willing to enter. It is also
possible that auctions would increase large lenders? dominance of FFELP
market. Large lenders would have several advantages over small lenders in
origination rights auctions. Because they can usually obtain loan funds more
cheaply and spread their costs over a larger volume of loans, large lenders
would generally be able to outbid their smaller competitors. In addition,
auctions

24 As we showed in chapter 1, the student loan industry is highly
concentrated and is becoming more concentrated. If the trend toward reduced
competition in the industry were expected to continue regardless of FFELP
rules, then auctions? effects on lender diversity would have to be evaluated
relative to this trend.

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 38

would necessarily make lenders uncertain about their prospects of continuing
to participate in FFELP. This uncertainty may be more likely to deter small
lenders than it would be to deter large lenders. In the long term, it is
possible that auctions would gradually reduce the number of lenders,
although we do not know how likely this outcome is. Lenders who did not win
origination rights in one auction might leave FFELP permanently because it
might be expensive to re-enter the origination market. If losing bidders
dropped out permanently after each auction, then the number of bidders would
gradually decline. Only a few large lenders might be left to originate
student loans.

Evidence from the HEAL program neither clearly supports nor clearly refutes
this argument. The number of HEAL bidders declined during the last 4 years
of the program. 25 (These are the only years for which we were able to
obtain comparable data.) However, this decline may have stemmed from lenders
withdrawing from the auctions after the Congress decided to end the program.
Moreover, the number of lenders receiving loan origination rights did not
fall continuously. There were four or five large HEAL lenders in the last
few years of the program until the very last year. Furthermore, lenders
entered and re-entered rather than dropping out permanently after losing at
auction.

Small-bidder protection measures and the choice of bidding method could
affect the diversity of lenders. Auctions can be designed to enhance small
bidders? ability to compete. More lenders might participate in auctions if
all auction winners paid the same price than if each paid a different price.
Lender participation might also be greater in an auction with multiple
rounds of bidding than in one with a single round. However, a single-round
auction might be more attractive to small lenders because it would be
simpler than a multiple-round auction. Lender diversity would probably be
greater if more than one lender were allowed to originate at each school
than if there were only a single lender per school. More lenders would be
likely to participate in FFELP if multiple lenders could serve each school.
For more details on these issues, see appendix III.

An additional complication associated with the volume procurement proposal
would be the challenge of estimating the annual borrowing needs of students.
An underallocation of loan volume could result in delays in funding and in
loan access problems for student loan providers. The Congress might want to
include additional requirements to ensure that loans were available to all
eligible borrowers. For example, it could require the federal government to
serve as a lender of last resort or design a mechanism to increase the
allocation for lenders who were approaching their allocated limit.
Alternatively, the federal government could pay lenders to serve borrowers
who would otherwise not receive loans. (Allowing lenders to submit negative
price bids at auction-that is, requesting that the federal government pay
them to serve certain groups of schools-would be one way of implementing
such payments.)

Auctions should be held when borrowers can receive loans when they need
them. After an auction is completed, time is needed to process bids and
notify the winning lenders. The winning lenders, in turn, need time to
disburse loans after they are notified. For borrowers to

25 Most HEAL lenders were also large FFELP lenders and considered HEAL
lending complementary and marginal in importance to their FFELP lending
activities. Many schools found the HEAL program unsatisfactory, believing it
unstable and unpredictable.

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 39

receive loans when they need them, all these things have to be done before
the start of the school term. If there were a secondary market in
origination rights, then trades on this market would also have to be timed
to enable lenders to receive funds and disburse loans in time for the
beginning of school. Schools? academic calendars vary widely, and academic
programs can start at any time of year. Therefore, it would not be possible
to link the timing of auctions to that of loan disbursements. However, it
would be possible for loans to be disbursed continuously between auctions.
For example, auctions could be held once every 3 to 5 years, and the winners
of each auction could use the origination rights they won at any time
between that auction and the next year?s auction.

Finally, the transition to auctions could temporarily disrupt both student
and parent access to FFELP loans. Lenders would require some time to become
accustomed to the auction system. Some lenders might at first miss deadlines
or fail to submit payments that the government might require before the
auctions. Some might decide not to participate in the initial auctions.
Those who did participate might not bid on as many origination rights as
they would if they were more familiar with the auction process. The
government might experience delays in processing bids and notifying auction
winners. For all these reasons, auctions could initially reduce FFELP loan
volume. The federal government could maintain student and parent access to
loans during the transition period by acting as the lender of last resort or
by paying private lenders to do so. In addition, the federal government
could identify potential access problems during the transition by conducting
an auction pilot program before implementing an auction system for all FFELP
loans. Using knowledge gained from the pilot program, it could then take
steps to minimize those problems.

SIMPLICITY, REGULATORY BURDEN, AND PROGRAM INTEGRITY

Because insufficient competition could threaten the integrity of the bidding
process, auctions would require special rules to preserve competition. The
definition of auctionable rights, the choice of bidding method, and the
frequency with which auctions were held could affect the simplicity and
integrity of FFELP and the burden that auctions would place on lenders,
students, and schools. Lenders, students, and schools would all bear the
burden of adjusting to new regulations and market practices. Some
transitional challenges could be anticipated and thus lessened through the
use of a pilot program.

To preserve the integrity of auctions, the federal government would have to
adopt and enforce regulations to maintain competition in bidding. Rules to
prevent bidders from colluding would be necessary, and their nature and
complexity would depend on the kinds of collusion to which a particular type
of auction could be vulnerable. FCC, for example, set minimum bid increments
to prevent bidders from signaling their bidding strategies by using the
final digits of their bids.

Collecting the data needed to administer this program could present new
challenges. Tracking the results of auctions and administering a program
with origination rights auctions or volume procurement auctions requires
data not currently available. At a minimum, the government would require the
ability to monitor eligible lenders? loan portfolios by school and loan
volume. Insufficient data would limit the government?s ability to manage
this program.

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 40

All auctions would require rules to ensure that bidders were able to pay the
amounts they bid. Without such rules, nonserious bidders could distort
competition. Other federal auctions provide examples of rules that could be
adopted in an origination rights auction. EPA requires each bidder in sulfur
dioxide emission allowance auctions to send a certified check or letter of
credit to cover its bid before the auction. (This rule is feasible only for
sealed-bid auctions, in which each bidder submits a single bid.) FCC
requires bidders in its wireless spectrum auctions to submit refundable
deposits to cover the cost of placing bids. A final option, not used in any
federal auction, is to require all potential bidders to show some evidence
of their ability to pay. One way of implementing this option is to use the
current FFELP eligibility criteria as evidence of ability to pay. Another
alternative is to require FFELP-eligible lenders to pass additional
ability-to-pay tests before allowing them to participate in the auction.

The greater the total number of auction winners, the longer and more complex
the process of determining winners is likely to be. Length and complexity
will probably be greater, for example, the larger the number of groups into
which schools are divided and the greater the number of winners per group or
per school. The longer and more complex the bid evaluation process, the
greater the burden on lenders. If lengthy and complicated bid evaluation
methods caused delays in the disbursement of loans, then they could also
impose burdens on borrowers.

Bidding methods differ in their simplicity, speed, and vulnerability to
collusion. An auction that has a single round of bidding is faster than one
with multiple rounds, and the bid evaluation process is simpler. Allowing
lenders to bid on groups of schools that they define complicates the
evaluation of bids. (See app. III for details.) It is simpler and faster to
evaluate bids on the basis of price or interest rate alone than to include
service quality criteria in the evaluation. Finally, as appendix III
explains, a multiple-round auction may be more vulnerable to collusion than
a single-round auction, and an auction in which each bidder pays a different
price could be either more or less susceptible to collusion than one in
which all bidders paid the same price.

Frequent auctions could impose substantial burdens on lenders, students, and
schools. Because participating in auctions would cost lenders both money and
time, more frequent auctions would be more burdensome to lenders. In
addition, according to some school representatives in the study group,
students and schools value the ability to deal with a single lender.
Therefore, more frequent auctions could impose a greater burden on schools
and students by disrupting long-term student-lender and school-lender
relationships.

Any origination rights auction system would require lenders, students, and
schools to adjust to new regulations and changes in the student loan market
that resulted from auctions. The nature of the regulations adopted to
implement auctions would depend heavily on the specifics of the auction
design. However, those regulations would probably have to be very detailed,
and monitoring lender compliance could be costly. Regardless of the
particulars, lenders would have to learn how to participate and might be
subject to new regulations to ensure their ability to use the origination
rights they won. All origination rights auctions would restrict students?
and schools? ability to work with the lenders of their choice, and those
with a single winning lender per school would eliminate their ability to do
so. Lenders, students, and schools would all have to adapt to the changes in
loan availability, lender and borrower interest rates, and service quality

CHAPTER 3: LOAN ORIGINATION RIGHTS AUCTION Page 41

that could result from auctions. Table 10 summarizes the analysis for the
loan origination rights auction model at a broad level.

Table 10: Summary of Analysis for Loan Origination Rights Auction Model

Description of model, including variations Lenders submit bids to originate
loans. Some win right to originate and

others do not. Auction could be conducted for the right to originate at
specific schools or groups of schools or for a certain loan volume that
could be used at any of a number of schools. Other major design options
include grouping schools, choice of bidding method, whether a borrower keeps
all loans with one lender.

Costs, savings, and effects on subsidies for program participants

Federal costs could lessen if competition is sufficient. Distribution of
payments to lenders would be unchanged or more uneven, depending on auction
type. Borrowers? interest rates could decline for all, remain unchanged for
all, or increase for some.

Effects on lender participation, loan availability, and service quality

Loan origination costs could lessen if competition is sufficient. Service
quality may decline or remain unchanged. Diversity of lenders is likely to
decline unless auctions included small- lender protections. Schools? ability
to work with preferred lenders could be reduced. Loans would not be
available to all students in some types of auctions.

Simplicity, regulatory burden, and program integrity Program integrity would
require new rules to preserve competition.

Burden on lenders, students, and schools would differ, depending on auction
type. Lenders, students, and schools would bear the burden of adjusting to
new auction regulations developed by the government.

Page 42

CHAPTER 4 LOAN SALE

Under the loan sale model, the federal government, or some entity other than
private lenders, would originate all student loans, and the government would
later sell them at auction in a secondary market to the highest bidding
lenders. Preferably, the government would elect to sell loans immediately
following origination or later when the borrower enters repayment. Because
the federal government would be responsible for loan origination, and the
loan guarantee terms could change, the federal role in FFELP would be vastly
different from today. Purchasers of the secondary market loans could include
today?s FFELP lenders as well as new participants, however some current
lenders, including lenders who currently originate loans but subsequently do
not hold them, might not choose to participate. The federal costs of the
loan program could also be lower, partly because the government?s lower cost
of securing the funds to make loans and reduction in fee payments to
guaranty agencies for services related to verifying lender eligibility. If
there is sufficient competition, then the auction among secondary market
participants could reduce federal FFELP costs. Secondary market
concentration would be likely to increase with this mechanism, perhaps
reducing the participation of small community-based lenders. Loan
availability would be universal, with the federal government being the only
program lender. Centralizing the loan origination and distribution functions
may simplify the loan process. Schools and borrowers would deal only with
the federal government or the designated entity to obtain loan aid.
Adjustment to the new system could prove burdensome for some lenders.

DESCRIPTION OF THE MODEL, INCLUDING VARIATIONS

The loan sale model and variations we discuss in this chapter share several
features. Generally, some entity other than private lenders would originate
student loans. The terms of the loan, such as the borrower?s maximum
interest rate and the in-school subsidy policy, would be set through
legislation, as now. Packages of student loans-with or without a
guarantee-would be purchased privately at periodic sales or auctions. While
the purchaser buys loans, either the purchaser or some other entity may
perform the servicing functions. In each auction, a lender could submit a
single bid or multiple bids that incorporated different prices. The
government would sort bids by price, and the highest bids would win the
right to purchase loans. The government could set a minimum price below
which it would not sell any of the loans it originated. It would be possible
to set aside some loan packages for nonbidders. However, most purchase
rights would be determined through the bidding process. A variety of bidding
methods could be used, as described in appendix III.

Proposals for this model vary as to who originates loans, how loans are
grouped when they are sold, who is responsible for servicing loans, whether
the loans sold are federally guaranteed, and whether payment is due in full
at the completion of the loan sale. Origination could be carried out by a
federal entity or by private contractors, as is in FDLP. The government
could sell these loans immediately or it could hold off selling the loans
until a borrower enters repayment. If the loans are sold with government
servicing, the purchaser is buying the stream of payments and the

CHAPTER 4: LOAN SALE Page 43

government needs to contract for servicing. 26 If the loans are bundled and
auctioned off without government servicing, purchasers can arrange for
servicing or possibly they can retain the services of the government
contract servicer. Loan bundles are less valuable if they are sold without a
guarantee than if they are sold with a federal guarantee. Allowing lenders
to make payments in installments could foster competitive participation in
loan sales.

Several options are available for loan origination. One alternative would be
to give a government entity (such as a new government-sponsored enterprise)
the authority to originate federally guaranteed student loans. The
government could maintain the origination procedures in place for FDLP,
using private contractors to originate loans. Another option would be for
the government to require the student-lending industry to charter a mutually
owned corporation that would originate and provide short-term funding of
loans. The objective would be to structure the entity?s ownership and
transactions to give it an incentive to keep down its costs. Achieving that
objective could be difficult without competition, however.

Loans from particular groups of schools may or may not be bundled together
when loans are auctioned. Grouping by schools would mean that lenders would
bid for packages of loans that were originated at particular groups of
schools. The loans from each grouping of schools could be sold at the same
auction, or else a separate auction could be held for each group. In
defining groups of schools, several alternatives are possible. An
Internet-accessible interface could inform potential purchasers of the
characteristics of loans bundled as a package. Issues related to the
grouping of schools at auctions are presented in appendix III.

There are several options for servicing loans after they are sold at
auction. One would be that the government could keep the responsibility of
servicing loans that were later sold at auction. This could possibly
increase auction proceeds by attracting bidders who do not have servicing
capacity but are interested in receiving interest income and loan principal.
Auction proceeds would also vary, depending on whether the lenders expect
the government?s servicing to be more or less costly than other servicing
options. If the purchaser services loans, combining the loans of all
borrowers reduces the burden on each borrower. Some provisions would need to
be made for loan consolidation and loan deferment. Either the government or
the private lender could consolidate the loans if the borrowers return to
school.

Another issue is whether the loans or securities are sold with federal
guarantee. If the loans are sold with a guarantee, then the guarantee could
be set at the current 98 percent of principal or it could be set higher or
lower. The government would have to decide on the extent of the guarantee
and would have to define the conditions under which the guarantee applies,
such as death or bankruptcy. One variation is to have no explicit federal
guarantee or due diligence requirements. In another variation, lenders could
have the right to resell the loans to Education in the event of death,
disability, bankruptcy, or the borrower?s election of ICR. Other variations
assume that federal loan guarantees continue, as with the current program.

26 A variation on loan sales with government servicing is selling securities
backed by student loans. Securitization is the process of selling debt
securities to investors with groups of loans serving as collateral for the
debt.

CHAPTER 4: LOAN SALE Page 44

One variation would allow lenders to pay for bundled loans in installments.
Installment financing arrangements could be tied to the loan payment
revenues received by the purchasing lender or to a fixed repayment schedule,
and the appropriate interest rate would have to be determined. Either
arrangement reduces the initial funds needed to purchase loan bundles,
thereby potentially increasing lender participation.

Treasury developed an additional variation of this model and presented an
outline of it to the study group. The study group did not review this and
this outline is reproduced in appendix V.

COSTS, SAVINGS, AND EFFECTS ON SUBSIDIES FOR PROGRAM PARTICIPANTS

Loan sales could potentially raise enough revenue to reduce overall federal
FFELP costs, but their ability to realize this potential depends on
sufficient competition in secondary market bidding. The federal government
(or some entity acting on its behalf) will originate all loans, taking
advantage of the government?s ability to raise funds at a lower interest
rate. The grouping of loans into packages, the choice of bidding method, and
the frequency of auctions could also affect federal costs. If competition
prevails, there could be a lower net lender yield. However, insufficient
competition could threaten the integrity of the bidding process. Auctions
are not likely to affect borrower costs. However, schools participating in
FFELP could incur costs in implementing a new delivery system.

Reduced federal costs could result from eliminating federal payments to
originating lenders and financing loans with low-cost Treasury financing.
Removing originating lenders would eliminate their current federal SAP as
well as the need to pay private lenders the in-school interest subsidy.
Separating loan origination from the rest of the student loan process might
enable the federal government to take advantage of any savings because of
greater specialization in student loan origination. An entity responsible
solely for student loan origination would not have to be concerned with
raising necessary funds competitively or with servicing loans efficiently.
However, removing competition from loan origination could eliminate
incentives to continuously improve the efficiency of the origination
process.

Further federal savings could accrue if loan sales foster competition in the
secondary market for loans, but savings would depend in part on auction
design decisions. Ideally, competitive bidding for loans among secondary
market lenders would result in loans being held and serviced by those with
the lowest cost of funds and the most cost efficient at servicing loans.
Increasing concentration in the existing secondary market for student loans
raises questions about the extent to which competition will operate in loan
sales and the extent to which potential cost savings will accrue to the
federal government. If auctions are not sufficiently competitive, then
increased lender net yield could result, potentially reducing federal
revenues from the sales and thereby increasing federal FFELP costs. Although
sufficiently competitive auctions could reduce net lender yield, their
overall effect on FFELP costs would also depend on the details of the
auction design. Auction frequency, the choice of bidding method, and the
definition and grouping of auctionable packages could either lower or
increase FFELP costs.

CHAPTER 4: LOAN SALE Page 45

The terms of sale affect the price that bidders pay for loans in an auction.
Lenders bid more for loans that are sold with a guarantee. 27 The net effect
on federal costs would then depend on a comparison of the additional revenue
raised by guaranteeing loans to the federal cost of providing a guarantee.
Also, federal costs could depend on whether winning bids are payable in full
immediately after the close of the auction or whether a schedule of
installment payments is allowed. Allowing installment payments extends
federal funding to secondary market participants, possibly enabling small
bidders to participate more easily. Small bidders would have to come up only
with a down payment rather than with the full purchase price, enabling
bidders with less funding to compete with better-funded large bidders.
However, allowing loan purchasers to pay in installments could result in new
risks and potential costs. The most significant of these risks would be that
lenders obtaining federal funding might be unable to repay the money they
borrow from the government. Several considerations might mitigate this risk.
For example, lenders could be made to meet additional criteria, related to
financial soundness and creditworthiness, before being allowed to borrow
from the federal government. In addition, legislation or regulation could
treat the auctioned loans as collateral for the borrowing. The cost of
administering an installment payment option would be an additional source of
increased federal government costs for FFELP.

Ultimately, borrower costs are unlikely to change much after the sale of
loans at auction is implemented. Some lenders now offer origination fee or
interest rate discounts off of the maximum rate to some borrowers (for
example, those who attend schools with low default rates). If lenders are
providing discounts from profits, then auctions could eliminate the
discounts by reducing those profits. As a result, some borrowers who now
receive discounts may no longer receive them. However, if discounts do not
come from lenders? profits, then auctions would not eliminate the discounts
and would have no effect on borrowers. Furthermore, if auctions reduced
federal FFELP costs, the Congress could decide to pass those savings on to
borrowers by lowering the maximum borrower interest rate. In that event, all
borrowers would benefit from lower interest rates. Other discounts that
loanholders currently provide occur after the loan has entered repayment.
Lenders have provided these discounts for automatic electronic payment and
demonstrated timely payment experience. Loan holders will continue to have
an incentive to encourage timely or electronic repayments which result in
lower costs by offering discounts. Issues related to competition, auction
frequency, auction design, bidding methods, grouping, and payment methods
are presented in detail in appendix III.

EFFECTS ON LENDER PARTICIPATION, LOAN AVAILABILITY, AND SERVICE QUALITY

With loan sales, the federal government or some entity other than private
lenders would originate all student loans. Loans would be universally
available to all eligible borrowers. Concentration in the secondary market
would be likely to continue to increase, resulting in reduced diversity of
lenders. As lenders focus more on price paid to the government, they might
pay less attention to service quality.

27 The budget-scoring effects for loans sold with a guarantee are beyond the
scope of this report.

CHAPTER 4: LOAN SALE Page 46

In the short term, auctions may have any of several effects on the number of
FFELP lenders. They may increase lender participation by enabling new
lenders to enter the market on relatively equal terms with existing lenders.
It is also possible that auctions would increase large lenders? dominance of
the secondary market. Large lenders would have several advantages over small
lenders. Because they can usually obtain loan funds more cheaply and can
spread their costs over a larger volume of loans, large lenders would
generally be able to outbid their smaller competitors. In addition, auctions
would necessarily make lenders uncertain about their prospects of continuing
to participate in FFELP. Small lenders are more likely than large lenders to
be deterred by this uncertainty.

Over the long term, lenders who do not win in one auction might leave FFELP
permanently, if they find it expensive to reenter the market. If losing
bidders drop out permanently after each auction, then the number of bidders
could gradually decline, leading to higher federal costs. However, it is
also possible that the losing bidders-especially those participating by
holding and servicing loans from prior years-will not leave FFELP. With a
loan sale, a participant who has been a winner for 3 years but does not win
in year four will probably not leave FFELP, because that lender still has
revenue from the previous years? loans. That lender may find it easy to
reenter in the next year, and this may be less a concern than in the loan
origination rights auction model.

One way the Congress could ensure lender diversity and pervasive
participation would be to include special provisions, such as installment
payment payments, that would protect small bidders? ability to compete.
Although small-bidder protections may restrict the ability of larger firms
to compete, they may also improve the overall competitiveness of the market.
Another way of promoting competition in auctions is to adopt rules to ensure
that all bidders are able to pay the amounts they bid. Without such rules,
nonserious bidders could distort the auction results by placing bids that
were unrelated to their valuations of the rights. Details of auction
provisions regarding these features are presented in appendix III.

Loan sale auctions may affect the service quality of loan origination. In
some models, such sales could remove competition from loan origination and
could eliminate existing incentives to continuously improve the efficiency
of the origination process. A single originating lender may not feel
competitive pressure and therefore might be less likely to introduce new and
innovative loan options.

There is some concern that auctions could shift the nature of competition in
the student loan market toward price competition and away from service
quality competition. At present, FFELP lenders compete to some extent on the
basis of price through discounts to preferred borrowers. They also compete
to some extent on the basis of service quality. Auctions would force lenders
to place more emphasis on price competition, since the major (and, in most
variants, the sole) criterion for determining auction winners would be the
price bid. It is possible, although not certain, that lenders would pay less
attention to service quality as they paid more attention to price. 28

28 Service quality could be higher if multiple servicers rather than only
one servicer were allowed. If students have the opportunity to select
servicers or to choose a lender for consolidating loans, servicing lenders
would have

CHAPTER 4: LOAN SALE Page 47 SIMPLICITY, REGULATORY BURDEN,

AND PROGRAM INTEGRITY

Loan sales could simplify the origination process for borrowers and schools
by centralizing the origination and distribution of all loans. Some
transitional challenges could be anticipated and thus lessened through the
use of a pilot program. Lenders choosing to continue their FFELP
participation would face an adjustment period during which they would need
to learn how to participate in auction sales. While the integrity of the
student loan program is not expected to suffer, auctions would require the
federal government to establish special rules to preserve competition.

Loan sales could extend FDLP?s origination and distribution channels.
Borrowers and schools would no longer be able to choose a lender. This
change would require adjustments in the short run by schools not currently
participating in FDLP. Since required transitions primarily affect loans
entering repayment, the effect on student borrowers should be minimal. It is
possible that the current distinction between FFELP and FDLP would be
eliminated if all loans were sold at auction.

Lenders who continue in the program would have to learn how to participate
in the new program. Regardless of the way auctions are designed, they
require participating lenders to develop skills and procedures that they do
not now have. Initially, lenders would need to make decisions without
experience as a guide. In addition, the federal government could identify
some potential logistical problems during the transition by conducting a
pilot program before implementing a loan sales system. Using knowledge
gained from the pilot program, it could then take steps to lessen those
problems.

The government would have to adopt and enforce regulations to maintain
competition in bidding and ensure that participating lenders were
creditworthy. Rules to prevent bidders from colluding would be necessary,
especially if the trend toward increased secondary market concentration
continues. The nature and complexity of these rules would depend on the
types of collusion to which a particular type of auction could be
vulnerable.

Operating and administering a loan sales program require additional federal
data collection. An installment payment option tied to borrower repayments
would require tracking the payments received by participating lenders from
borrowers and matching this information to lender installment payments. As
previously discussed, obtaining and using accurate data are critical to
effectively managing the program. Results of the analysis for the loan sale
model are summarized at a broad level in table 11.

incentive to compete for students on the basis of service quality. A
proposed borrower Web interface would better inform borrowers of their
repayment options, where and to whom to send their payments, and other
related matters. This could serve to limit or reduce delinquency and reduce
servicing issues.

CHAPTER 4: LOAN SALE Page 48

Table 11: Summary of Analysis for Loan Sale Model

Description of model, including variations The government or a
government-designated entity originates loans.

Private entities bid to purchase packages of loans, either after origination
or upon borrower?s graduation. Loans could be sold with or without a
guarantee. Purchase could be paid up front or financed and paid over time.

Costs, savings, and effects on subsidies for program participants

Federal costs could be lower; effect would depend on how loans are packaged.
No federal payment would be made to originating lenders. Federal payments to
guaranty agencies could be reduced or delayed. Borrower costs would be
unaffected.

Effects on lender participation, loan availability, and service quality

Loans would be available to all. The concentration of loanholders could
increase. Schools could serve as loan originators. The service quality in
repayment could decline.

Simplicity, regulatory burden, and program integrity Simple for borrowers;
federal role more complex.

Continuing lenders must learn auction mechanism. Program integrity is not
expected to suffer.

Page 49

CHAPTER 5 FEDERAL FUNDING

The federal funding model would give lenders the opportunity to borrow from
the federal government funds with which to make FFELP loans. Borrowing could
be provided as an option to lenders or could be made mandatory for lender
participation in FFELP. Lenders would be required to bid on the interest
rate they would pay to the government for the use of the federal funds. The
total amount of federal funds available to FFELP lenders would be limited.
Lender yield would be set legislatively and could be the same as the maximum
borrower interest rate, which is currently based on Treasury rates. This
model would affect lenders? cost of funds rather than lender yield. Federal
costs could increase, decrease, or remain the same. Some roles of FFELP
participants could change, depending on how the model is implemented,
including major changes for the role of lenders. The model could also
potentially affect the ability of schools and students to work with lenders
of their choice, as well as the availability of loans and the quality of
service to borrowers.

DESCRIPTION OF MODEL, INCLUDING VARIATIONS

In the federal funding model, the federal government would auction to
lenders a predetermined volume of federal funds set aside for them to use to
make FFELP loans. Each bid would include an interest rate and the volume the
bidder wanted to borrow at that interest rate. The interest rate at which
each winning lender borrowed, in conjunction with a legislatively set lender
yield that could equal the FFELP maximum borrower interest rate, would
determine that lender?s net yield. Borrowing from the federal government
could be made mandatory for lenders who want to participate in FFELP, or it
could be offered as an option, with lenders still being allowed to fund
student loans by traditional methods. Finally, lenders could repay the funds
they borrowed from the government on either a regular amortization schedule
or a schedule tied to borrower repayments on the FFELP loans made with the
federal funds.

Basic Model and Roles of Participants

The federal government would estimate the volume of loans that students will
demand in a given year. Lenders would submit one or more bids for a portion
of the total volume. For example, a lender could bid for a certain volume at
one interest rate, and then additional volume at a different rate, for as
many different rates as desired. The agency operating the bidding system
would collect all bids, sort the interest rates from high to low, and total
the volume that all bidders desired at each interest rate. The interest rate
emerging from the auction would be the lowest interest rate bid at which the
sum of the desired volumes just exhausted the volume of federal funds being
allocated. All lenders whose bids were above or equal to this rate would
receive the allocation they bid. To prevent any one lender from gaining an
unacceptably large share of the FFELP market, the government might limit the
volume of federal funds that any lender could receive.

CHAPTER 5: FEDERAL FUNDING Page 50

As in the current FFELP, the Congress would continue to set the maximum
borrower interest rate. The reference rate for the maximum borrower interest
rate (and the lender yield) could be any interest rate. However, because
lenders would be likely to borrow from the government at Treasury-based
rates (that is, at rates that are markups over the interest rate paid on a
Treasury debt instrument), the Congress might consider keeping the maximum
borrower rate Treasury- based as well and changing the lender yield back to
a Treasury-based rate. This rate could change from the 91-day rate to a
longer-term rate. By giving lenders a Treasury-based source of funds, this
model could allow the borrower rate to remain Treasury-based and give
lenders the ability to match-fund. 29

The bidding is likely to be based on the same Treasury rate as the basis for
the maximum borrower interest rate, to eliminate concerns about basis risk.
30 For example, the relevant Treasury rate could be the 91-day T-bill rate
or the 10-year Treasury bond rate. The interest rate for bidding could then
be expressed as a specified number of percentage points above or below the
Treasury rate. 31

The lender yield for FFELP loans would continue to be set legislatively. One
option is for the Congress to set the lender yield equal to the maximum
FFELP borrower interest rate so that the government makes no SAP. 32
Lenders? net yield, which depends on the difference between the
legislatively set yield and their funding and operations costs, would be
affected in this model by changes to funding costs.

To ensure that loans were available to students, lenders who won at auction
would be required to use the entire volume they borrowed from the government
to originate FFELP loans. Lenders would be required to return the unused
portion of funds borrowed. To avoid having the government lend more funds
than necessary, requiring a process to accommodate the return of funds,
lenders could apply for federal funds after originating the FFELP loans.
Otherwise, the government would need to match the timing of lender borrowing
and payment to the corresponding student loan disbursements and repayments.
In other respects, this model would leave the roles of FFELP participants
unchanged. In addition, FDLP could remain as it is today, with terms for
students the same as in FFELP.

29 ?Match funding? refers to lenders matching the basis of the interest rate
at which they borrow to finance a loan with the basis of the interest rate
that they receive from the loan. If they are able to do so, then changes in
the interest rate affect their costs and revenues identically and do not
affect their net profits. If their funding costs and their revenues are
based on different interest rates and those rates do not move in tandem,
then their net profits could fluctuate. 30 Basis risk is the risk created by
a mismatch between the interest rate at which an entity borrows and the rate
at

which it lends. For example, borrowing at a CP-based rate and making loans
at a Treasury-based rate introduces risk because, if the CP rate were to
rise relative to the Treasury rate, borrowing costs would rise relative to
income from the loans. This would reduce profits. 31 Another option would be
to establish a borrowing rate and then conduct bidding in terms of the price
that lenders

are willing to pay to borrow at that rate rather than bidding on the rate
directly. For example, the rate could be set at the maximum borrower rate
plus a small markup, and lenders could bid a dollar amount for the right to
borrow a certain volume of federal funds at that rate. Lenders willing to
pay the most would win the right, and enough winners would be selected to
exhaust the volume that was put up for auction. 32 The government would
presumably still pay lenders a SAP if the borrower rate were high enough to
hit its cap.

CHAPTER 5: FEDERAL FUNDING Page 51

Major Variations on the Model

Participation in this system could be made optional or mandatory for FFELP
lenders. Lenders could be offered the option of borrowing from the federal
government or funding their FFELP loans and competing as they do now. The
lender yield would have to be sufficiently high that nonparticipation was a
realistic option and nonparticipants could make a reasonable profit. If the
lender yield were too low, participation would effectively become mandatory.
Alternatively, participation could be made mandatory so that only lenders
who took part in this borrowing would be allowed to make FFELP loans.

The volume of federal funds that the government would have to make available
at auction would depend on whether federal funding were mandatory or
optional for lenders. If lenders were required to use federal funds for
FFELP loans, or if federal funding were mandatory in effect even though
optional in principle, then the government would have to set aside
sufficient funds to meet the loan volume that it expected students to need.
If lenders? use of federal funding were truly voluntary, then the government
would have to set aside less than the total loan volume expected.

Auction design details, such as the choice of bidding method, could affect
the results. These design issues are common to all the auction models we
discuss in this report. For further details on auction design issues, see
appendix III.

COSTS, SAVINGS, AND EFFECTS ON SUBSIDIES FOR PROGRAM PARTICIPANTS

Federal FFELP costs under this proposal could increase, decrease, or remain
the same. Costs could decline as a result of a bidding process if auctions
were sufficiently competitive. The frequency with which auctions were held
and the choice of bidding methods could also affect federal FFELP costs.
(See app. III for an explanation of these issues.) However, the federal
government would also face new risks and potential costs in a federal
funding model. The most significant would be that lenders who obtain federal
funding might be unable to repay the money they borrowed from the
government, although the likelihood of this might be quite low. Requiring
lenders to meet certain criteria for borrowing federal funds could mitigate
this risk. The cost of distributing funds to and collecting repayments from
lenders would be an additional source of increased federal government costs
for FFELP. Some of a cost increase might be recovered by reducing or
eliminating the SAP. 33

In addition, the federal government could face conflicting incentives in the
guarantee process. If the federal government is both reinsuring student
loans and providing funds for them, then strict enforcement of due diligence
rules could create a loss risk for the government. For example, if an FFELP
loan loses its guarantee because of improper servicing, the lender might
have difficulty repaying a portion of the funds it borrowed from the
government. If the guarantee

33 If due diligence regulations are not followed or something else threatens
the guarantee, one recourse that Education has now is to withhold SAP
payments to the lender. If the SAP were eliminated, the government might
have less leverage over the lender than it now has.

CHAPTER 5: FEDERAL FUNDING Page 52

were maintained, the lender would receive the insurance payment for the
defaulted loan and would then have funds available to repay the government.
In effect, the government could be seen as guaranteeing its own loans.
However, the federal guarantee could operate in different ways. Rather than
paying a default claim, the government could simply reduce the amount that
the lender owes by the amount of the default. Or, if a student defaults, the
guaranty agency might pay the government directly (with that portion of the
lender?s debt to the government being erased) rather than the guaranty
agency?s paying the lender.

This model would change net yield for lenders by determining their funding
costs rather than affecting lender yield. Net yield depends on lender yield
and costs, such as funding and operating costs. Most of the market mechanism
models use the mechanism to explicitly set the lender yield, but this model
uses it to set one portion of costs. Either approach ultimately has an
effect on net yield.

Costs could decline for lenders who used federal funds. If these federal
funds bear a low enough interest rate, then lenders? costs could decrease
and their net yield could increase. Additionally, if the model gives lenders
a source of funds tied to Treasury rates, potentially allowing lenders to
match their revenues from FFELP loans, lenders would not bear the risk and
costs of interest rate mismatches. Lender costs could be lowered as lenders
gained the ability to match the movement in their cost of funds to the
movement in the lender yield. As noted, both would be likely to be based on
Treasury rates, so lenders would face no basis risk.

If lenders were not required to use federal funds to originate FFELP loans,
then participating lenders might be forced effectively to give up all or
some of the savings they realized in order to make competitive bids. This
could, in turn, affect discounts offered to borrowers. Participating
lenders? costs might remain unchanged or decline. Therefore, those lenders?
gross profit margins might remain unchanged or increase.

If lenders were required to use federal funds to originate FFELP loans, then
lenders might be forced to give up all or some of the savings they realized,
or even more than the amount of those savings, to make competitive bids.
This could also affect discounts to borrowers. Lenders? costs could
increase, remain unchanged, or decrease. Lenders? gross profit margins
could, therefore, decline, remain unchanged, or rise.

An implicit subsidy to borrowing lenders would be present if lenders had the
use of federal funds before student loan disbursements and after loan
repayments by student borrowers.

EFFECTS ON LENDER PARTICIPATION, LOAN AVAILABILITY, AND SERVICE QUALITY

This model could potentially reduce the diversity of lenders, particularly
if lenders were required to use federal funding to originate FFELP loans.
Additionally, the relationships among FFELP participants could substantially
change. The effects of federal funding on loan availability would depend on
whether lenders? use of federal funding were optional or mandatory and on
whether and sufficient federal funding were available. The service quality
could decline if federal funding were mandatory for FFELP.

CHAPTER 5: FEDERAL FUNDING Page 53

If lenders using this process secured significant funding cost advantages
relative to other lenders, and if other aspects of FFELP remained
competitive, then over time competition could squeeze out nonparticipating
lenders or force them to participate. Lenders with different costs of funds,
such as those who have access to tax-exempt funding compared with those who
do not have access to it, would be affected differently. The process might
effectively become mandatory to ensure profitability, at least for certain
types of lenders, and the number and diversity of lenders could decrease.

Like origination rights auctions, federal funding auctions can reduce the
participation of small lenders in FFELP, especially if the use of federal
funding is mandatory. Small lenders, who usually cannot borrow funds in the
private market at rates as low as those paid by large lenders, might not be
able to bid as high for federal funds as large lenders could. If maintaining
a large number of lenders or a diversity of lender types is desired, certain
design features could be built into the model. Appendix III describes these
options.

Several questions remain about lenders? relationships to the federal
government. The process might be seen as reducing the lenders? role to that
of a contractor for a system that appeared similar to FDLP. In contrast,
lenders would own the loans they originate, and they would still compete on
price and service quality after getting their funding. A further
consideration is whether, after a change from private to federal funding for
FFELP, private sector funding could be brought back easily if it were deemed
necessary.

The effects of federal funding on loan availability would depend on whether
lenders? use of federal funding were optional or mandatory and on whether
sufficient federal funding were available. If federal funding were mandatory
and the volume of federal funds auctioned were insufficient to meet student
needs, then loan availability would be reduced. If federal funding were
mandatory but the federal government auctioned a sufficient volume of funds
to meet student needs, then some students might have temporary difficulty
finding a lender who had federal funds available from which to lend, but
permanent effects on loan availability would be unlikely. If federal funding
were optional, then loan availability might be less likely to be affected
because FFELP lenders would have access to alternative sources of funds.

Under this model, service quality might decline or remain unchanged. If a
particular lender did not win any federal funding in a particular year (or
even over a longer time period), then the borrowers and schools that were
accustomed to dealing with that lender would have to go to other lenders
during that year. Borrowers might have more difficulty meeting their
repayment obligations if they had to deal with multiple lenders than if they
could each deal with a single lender. Some borrowers might be more inclined
to consolidate their loans rather than deal with multiple lenders. In
addition, schools could lose the ability to work with nonwinning lenders and
could be required to adapt their information technology systems and student
financial aid procedures to accommodate winning bidders. These concerns,
including fixed investments in computer technologies specific to particular
lenders, could be important for high-quality service to schools.

CHAPTER 5: FEDERAL FUNDING Page 54 SIMPLICITY, REGULATORY BURDEN,

AND PROGRAM INTEGRITY

Bidding on the interest rate lenders would pay the federal government would
require its own set of administrative decisions and a regulatory framework
for how the bidding process would be conducted. If criteria for lender
participation in the auction were established, beyond criteria that lenders
must currently meet to participate in FFELP, these would add to
administrative burden. As with origination rights auctions, lenders could
face difficulties during the transition to federal funding, and both the
transition and the timing of auctions could affect the availability of loans
to student borrowers. Some transitional challenges could be anticipated and
thus lessened through the use of a pilot program.

Operating and administering a federal funding program would create
additional data collection needs. The government would need to keep track of
the funds borrowed by participating lenders and match lender draw-downs to
the timing of student loan disbursements and match lender loan payments to
loan repayments by student borrowers. As mentioned previously, collecting
sufficient and accurate data is key to managing the program and reducing the
risk of loss of government funds. Executive agencies have expressed concern
about the federal funding model. In addition, the federal government could
identify some potential logistical problems during the transition by
conducting a pilot program before implementing a federal funding system for
all FFELP loans. Using knowledge gained from the pilot program, it could
then take steps to lessen those problems. The results of the analysis for
the federal funding model are summarized at a broad level in table 12.

Table 12: Summary of Analysis for Federal Funding Model

Description of model, including variations Lenders would borrow funds from
the federal government to make FFELP

loans. The process could be made optional or mandatory for FFELP
participation. Lenders would be charged an interest rate determined through
a bidding process.

Costs, savings, and effects on subsidies for program participants

Federal FFELP costs could increase, decrease, or remain the same. Funds
borrowed from the federal government might not be repaid. Lenders could
potentially eliminate risks from interest rate mismatches.

Effects on lender participation, loan availability, and service quality

If federal funding were optional, lenders could continue to participate in
FFELP and not change behavior. If participating lenders secured significant
cost funding advantages, all lenders might have to use federal funding, and
lender diversity might eventually decline. Loan availability should not be
threatened if sufficient federal funds are made available. Service quality
could decline or remain unchanged.

Simplicity, regulatory burden, and program integrity Regulations would be
required to govern the bidding process.

Additional requirements might be possible to allow lenders to participate in
auctions.

Page 55

CHAPTER 6 MARKET-SET RATES

A FFELP guaranteed student loan program with market-set borrower and lender
interest rates would use competition to determine the interest rates and
other terms and conditions of student loans. The primary distinguishing
feature of this model is that the borrower rate, as well as the lender
yield, would be determined in the marketplace and rates could vary across
borrowers and lenders. Student borrowers, or schools, would shop for the
most favorable loans. Lenders would charge the interest rates that borrowers
agree to pay. Other possible changes might include no legislatively mandated
rate ceiling for borrowers, no federal subsidies to lenders, and no federal
payment of interest while borrowers were in school. Because of this, federal
costs would be likely to decline in a simple version of the model. Lenders
could offer interest rates and other loan features based on the needs of
different students at different schools. Competition might lead lenders to
offer different loan packages to different students on the basis of lenders?
perceptions of risks and costs. Because a federal guarantee on student loans
would remain, the range of rates might not be exceptionally wide. In
addition, the increased competition might induce improvements in loan
origination and processing that could improve the delivery system for both
schools and students. Some study group members believe all students should
receive a comparable interest rate set by the Congress. Allowing the market
to set the rate is likely to affect students with the greatest need.
Market-set rates could reduce or eliminate some borrowers? access to loans.
However, there are variations to this model that could ameliorate this
problem. The market-set rate model would also place a greater burden of
adjustment on all FFELP participants than would the other models we discuss
in this report.

DESCRIPTION OF MODEL, INCLUDING VARIATIONS

In the model we describe in this chapter, student borrowers, or schools
negotiating on their behalf, would shop among lenders for the best available
interest rates. Market competition would determine the interest rates
lenders would receive, which would be the same as the interest rates
borrowers would pay. Particular versions of the model differ according to
how, if at all, they limit variation in interest rates among borrowers.

Basic Model and Roles of Participants

Market competition would determine interest rates for both borrowers and
lenders. Lenders would charge the interest rates that borrowers agree to
pay. The current federal loan guarantee would remain, but the federal
government would provide no other subsidies to borrowers or lenders. Lenders
could offer different interest rates to different students at different
schools on the basis of their evaluations of how default risks and servicing
costs differ among students and schools. Students or their representatives
would be able to shop among lenders for the best available interest rates,
possibly using interest rate information that would become available over
the Internet. However, because students are often too inexperienced to shop
for loans, schools would be likely to shop on their behalf. In the market,
schools would be likely to be able to

CHAPTER 6: MARKET-SET RATES Page 56

negotiate better interest rates for students and to choose a limited set of
lenders with whom to conduct business.

Whether the student or the school does the shopping for the loan and loan
terms could change the loan terms actually provided to the student. It is
logically possible that the school will have concerns that do not line up
exactly with the interests of the student. For example, ease of
administration and support for the school?s lending operations may be more
important to the school than to the student. In contrast, the student may be
more interested in the interest rate and the ease of dealing with the lender
after graduation.

The current mix of federal and state roles in operating the mechanisms could
change. In the simplest version of this approach, the federal government
would no longer set the rate ceiling or other terms and conditions of FFELP
loans. However, because the federal guarantee would remain, due diligence
rules for servicing would still be required, and they would affect loan
terms and conditions. The role of the states need not change. State guaranty
agencies and secondary markets could continue to function.

The market-set rates model would require student borrowers, or schools
acting on their behalf, to be well-informed about the characteristics of
loans available in the market. Without accurate and timely information,
students or schools would be unable to evaluate loans offered to them or
negotiate effectively with lenders. Schools or students would need
information about loan terms offered to all types of students, for all types
of schools, and in all regions of the country. With such data available,
students and schools could compare their loan offers and rates with offers
and rates made in other circumstances. These comparisons would not eliminate
all rate and service differentials. However, such information would
facilitate students or schools in finding the best option.

Major Variations on the Model

The extent to which interest rates vary among borrowers could be limited in
several ways, or it might not be limited at all. One option would be to
allow market competition to determine interest rates without any restriction
on the rates that lenders could charge. If the Congress believed that this
option was unacceptable because it would make loans unaffordable for some
borrowers, resulting in unacceptably high average borrower interest rates,
or produce too much variation in borrower interest rates, then it could
choose one of three methods of limiting variation in borrower rates. It
could

designate a lender of last resort, limit the extent to which a lender could
offer different rates to borrowers presenting similar risks and costs, or
provide an interest rate subsidy to all FFELP borrowers, which need not be
constant.

The federal government could serve as a lender of last resort, using either
FDLP, a newly established lending program, or designated private lenders.
The private lender or lenders would have to be paid because the lenders
would presumably not otherwise lend at the designated rate.

CHAPTER 6: MARKET-SET RATES Page 57

The Congress would need to determine the interest rate that the lender of
last resort would be allowed to charge, decide on the compensation to be
paid to the lender of last resort, and set criteria for borrowers to be
eligible to borrow from the lender of last resort.

Another option would be to limit variation in the interest rate or
origination fees that a lender could offer to a group of borrowers. The
federal government could define groups of borrowers by student or school
characteristics that are associated with varying levels of servicing cost or
default risk. For example, borrower groups could be based on average loan
balances, type of degree sought, or type of school attended. The federal
government would limit the interest rate or origination fee differences that
would be allowed within each group but would not restrict these differences
between groups. FHA imposes this type of limitation on the variation in
?mortgage charge rates,? analogous to FFELP origination fees, among
recipients of the home mortgages it guarantees. Market competition
determines the interest rates and fees on FHA- guaranteed mortgages, and
different borrowers pay different rates and fees. However, FHA divides
mortgages into groups on the basis of location and characteristics affecting
prices and charges, such as fees or costs. FHA limits the difference between
the highest and lowest mortgage charge rate within a group in any time
period to 2 percentage points.

A final option would limit the rate level by setting a federal subsidy for
borrowers for all FFELP loans. Such a subsidy would lower the interest rate
for all borrowers by the same number of percentage points. It would also
preserve borrowers? incentive to shop for the lowest interest rates, since a
borrower would always benefit if he or she could find a lower rate.
Furthermore, this subsidy could be uniform for all borrowers or could vary
by the same student or school characteristics-average loan balances, type of
degree sought, or type of school attended-as designated above. For example,
borrowers in short-term programs or schools with high default rates, who
might face the highest rates in the marketplace, could receive larger
subsidies than those in other programs or schools, to try to ensure a more
uniform rate for all borrowers.

COSTS, SAVINGS, AND EFFECTS ON SUBSIDIES FOR PROGRAM PARTICIPANTS

Market-set rates are likely to reduce federal costs in the simplest version
of the model, in which government payments to lenders would be eliminated.
Subsidies to borrowers would be reduced in most versions of the model.
Average borrower interest rates would probably rise in the short run but
could increase, decrease, or remain unchanged in the long run. However,
interest rates are likely to vary more among borrowers than they do today,
increasing for some borrowers and possibly declining for others; thus,
federal cost savings might come at the expense of borrowers who pay higher
rates, potentially making some schools unaffordable for some borrowers.
Lower initial interest rates for the lowest-cost, lowest-risk borrowers
might replace the discounts that some lenders currently offer to those
borrowers. A shift to market-set borrower interest rates in FFELP would also
be likely to necessitate a change in the method by which FDLP interest rates
were determined.

In the simple version of the model, federal FFELP costs under a system of
market-set rates are likely to be lower than they are now. Federal costs
include the cost of the federal loan guarantee,

CHAPTER 6: MARKET-SET RATES Page 58

payment of interest while borrowers are in school (for subsidized Stafford
loans), and payments to lenders to cover the difference between the lender
interest rate and the maximum borrower rate. By eliminating the federal
government?s payments of in-school interest and its payments to lenders, the
market-set rate model would lower federal FFELP costs. (The cost of the loan
guarantee would not change, assuming that the guarantee structure did not
change and that these program changes did not affect default rates.) If the
federal government had to serve as lender of last resort or pay private
lenders to do so, or subsidize borrowers, then it would incur extra costs,
which would reduce or eliminate the federal savings from switching to
market-set rates. If the federal subsidies or lender-of-last-resort costs
were great enough, federal costs for FFELP could increase.

Market-set rates would eliminate all federal payments to lenders in the
simplest version of the model. Under most options, there would be no federal
subsidies to borrowers. Subsidies to borrowers would continue to exist only
if the federal government chose to limit differences in borrower interest
rates, or reduce the rates for all borrowers, by paying a subsidy to all
borrowers. Eliminating the rate cap or making it less binding for all
borrowers would be a major change in federal policy, which historically has
set the same rate cap for all borrowers.

In the short run, the interest rates that borrowers pay, on average, would
be likely to be higher than they now pay. In the long run, the average
borrower rate could be higher than, lower than, or the same as at present.
In the simplest version of the model, in the short run, eliminating the
federal cap on borrower interest rates and federal subsidies to lenders
would probably cause the average borrower rate to rise. In variants of the
model that included methods of limiting borrower interest rates, borrower
rates could increase by a lesser amount, or remain unchanged, in the short
run. However, the long-run effect on borrower rates in any version of the
model would depend on whether lender participation in FFELP increased or
decreased. If lenders? freedom to set interest rates induced entry by new
lenders and increased participation by current lenders, then those rates
might fall. If some current lenders exited the FFELP and no new ones
entered, then borrower rates could increase.

Market-set rates would probably increase the variation in the interest rates
different borrowers pay, raising some borrowers? rates while lowering
others?. The increase in variation may not be large, however, since the
federal government would still guarantee 98 percent of the principal of each
loan. Currently, the legislatively set borrower interest rate cap limits the
amount of variation in borrower rates. The market-set rate model would allow
some borrowers? rates to exceed the current ceiling. Borrowers that lenders
perceived as having high servicing costs or high default risks (for example,
those who had low loan balances or who attended schools with high student
loan default rates) could face higher rates. These borrowers could end up
paying interest rates exceeding the current ceiling. In the long run,
interest rates for borrowers with low servicing costs or low default risks
could decrease, if market competition eventually caused their average
borrower rate to fall. In that event, low-cost, low-risk borrowers, who now
receive discounts that give them interest rates below the ceiling, could pay
even less.

Low-cost, low-risk borrowers might pay lower up-front interest rates instead
of receiving discounts. Currently, lenders often give those borrowers
interest-rate discounts after repayment

CHAPTER 6: MARKET-SET RATES Page 59

begins. With increased competition at the front end of the loan, lenders
might offer those borrowers lower contractual interest rates, leaving less
flexibility for rate discounts during repayment. However, lenders might
still offer discounts for good performance or for automatic electronic
payment if offering those discounts increased profits for lenders.

In addition to affecting FFELP borrower interest rates, a system of
market-set rates would probably require a change in the method of setting
borrower interest rates in FDLP. Currently, FDLP rates are based on FFELP
borrower rates. If FFELP borrower rates were determined through market
competition, then they would probably vary more across borrowers and over
shorter periods of time than they do now. It would become more difficult to
base FDLP rates on FFELP borrower rates because of the increased variation
in the latter. It would be possible to set FDLP rates on the basis of an
average of FFELP borrower rates over a designated period of time. 34
Alternatively, the Congress could set an FDLP interest rate independently of
FFELP borrower rates. This option, however, would be likely to make
determining FDLP interest rates at least partially a political decision and
potentially set a ceiling for FFELP loans.

EFFECTS ON LENDER PARTICIPATION, LOAN AVAILABILITY, AND SERVICE QUALITY

Market-set rates could give lenders a continuing incentive to reduce their
costs. If student loan markets are somewhat competitive, lower-cost lenders
can offer lower rates and increase their market shares and profits. In
addition, lenders setting higher rates will lose customers unless they are
providing superior levels of service and other loan features attractive to
borrowers. They might reduce the quality of service or leave it unchanged
but might have less incentive to improve it than in the current system.
Effects on lender diversity are uncertain. Some borrowers would be likely to
lose access to FFELP. This problem would be somewhat mitigated if there were
a lender of last resort. During a transition period, some borrowers who
would ultimately have access to FFELP loans could be temporarily unable to
obtain them, while other borrowers who would ultimately lose access might be
able to borrow under FFELP.

Lenders? costs would also be likely to continue to decline after a
market-set rate system was put in place. If lenders had to compete for
business from students or schools that wanted to obtain low interest rates,
then they would face a continuing incentive to cut interest rates. To cut
interest rates and maintain a sufficient profit to make it worthwhile for
them to remain in FFELP, lenders would have to reduce their financing,
origination, or servicing costs. Because any lender who could gain a cost
advantage over competitors would gain borrowers at their expense, all
lenders would be under continuing pressure to keep costs low.

Market-set rates might reduce service quality or have no effect on it. The
balance of competition in the student loan market might shift toward price
competition and away from service quality competition. At present, FFELP
lenders compete to a limited extent on the basis of price, through discounts
to preferred borrowers. They also compete on the basis of service quality.
Under market-set rates, lenders might place more emphasis on price
competition. It is possible,

34 The method chosen for setting the direct loan rate would also affect
budget scoring for both programs.

CHAPTER 6: MARKET-SET RATES Page 60

although not certain, that lenders would decrease their emphasis on service
quality as they paid more attention to price.

It is also possible, although not certain, that market-set rates would
reduce service quality by weakening long-term relationships between lenders
and schools. At present, schools maintain lists of ?preferred lenders.?
Financial aid officers become knowledgeable about the practices of those
lenders. Some schools? computer systems are programmed to deal with lenders
on their preferred lists. Since schools often serve as intermediaries
between lenders and students, school- lender relationships can improve
service quality if they help schools assist students in dealing with
lenders. In a system of market-set rates, schools might choose lenders on
the basis of price and sacrifice these long-term relationships with lenders.
If students chose lenders directly, they might not choose the lenders with
which their schools previously had ongoing relationships. However, it is
also possible that students and schools would see value in lender-school
relationships and be unwilling to sacrifice these relationships to obtain
slightly lower interest rates.

The effect of market-set rates on lender diversity is uncertain. If lenders?
freedom to set interest rates attracted new lenders into FFELP in the long
run, then lender diversity could increase. Alternatively, with increased
price competition, large lenders who could invest in the human capital and
servicing systems that would enable them to offer low interest rates might
drive other lenders out of FFELP and prevent new lenders from entering.
Similarly, secondary market lenders could be affected, if competition drove
less efficient lenders and secondary markets out of the market via mergers.
However, consolidations of lenders and secondary markets are already
occurring in FFELP. It is possible that those consolidations would continue
under a system of market-set rates and that market-set rates would have no
effect on them.

Without a lender of last resort or a substantial federal interest rate
subsidy, some borrowers could be priced out of the FFELP market. In a system
of market-set rates, lenders would face incentives to charge high interest
rates to borrowers they saw as having high risks of default or high
servicing costs. For example, they would be likely to charge high rates to
students in low- income regions, those from low-income families, and those
attending proprietary vocational schools. Lenders would be likely to
consider those kinds of students as presenting especially high default
risks. Similarly, they might offer higher rates to borrowers in short-term
programs or schools at which they are unlikely to end up with high loan
balances, such as community colleges. Since loan servicing costs are
relatively fixed, borrowers with low loan balances are more expensive to
service on a per-dollar basis. If the interest rates that lenders were
willing to offer some students were so high that those students were
unwilling to pay them, then those students would not be able to obtain FFELP
loans. Although the federal loan guarantee would reduce the extent to which
different students posed different risks to lenders, it might not be
sufficient to prevent some students from losing access to FFELP. A lender of
last resort or a substantial federal subsidy could eliminate this access
problem. 35

35 Regulations restricting variation in interest rates lenders could charge
might not eliminate the problem because lenders might find it unprofitable
to lend to the highest-risk or highest-cost students at any rate that the
regulations would permit.

CHAPTER 6: MARKET-SET RATES Page 61

During a transition to market-set rates, access to FFELP loans could be
reduced from its current level but might differ from what it would be after
lenders, schools, and students adjusted to the new rate-setting system.
Because lenders would charge the highest-risk and highest-cost students
rates that those students might not be willing to pay, those students could
lose access to loans. However, differences in interest rates and student
access during a transition period need not be the same as those that would
ultimately exist in a system of market-set rates. Students, or schools
acting on behalf of students, might need time to develop expertise at
shopping for low interest rates. During a transition period, their lack of
expertise might keep rates higher than they would ultimately be, and some
students could initially be priced out of FFELP. Lenders would also need
time to develop expertise in assessing the risks and costs of lending to
different kinds of students. The criteria they initially used to determine
interest rate offers might not be the same as those they would later use.
Therefore, some students could be priced out of FFELP during a transition
period but regain access to the program later. Likewise, some students who
were able to obtain loans during the transition might later be priced out of
the program. A lender of last resort or substantial federal subsidy could
eliminate access problems that arose during the transition.

SIMPLICITY, REGULATORY BURDEN, AND PROGRAM INTEGRITY

Eliminating most existing program regulations dealing with interest rates
charged borrowers would likely simplify FFELP for lenders but would place
additional burdens on borrowers and schools, and new regulations could make
the program more complex for all participants. 36 The extent to which
regulations become more complex or simpler will depend on program design
features dealing with the exact extent of interest rate variability
permitted and the extent to which borrower protection on servicing remain in
place. In the simplest version of this model, the federal government would
no longer set maximum borrower interest rates, pay interest while borrowers
were in school, or pay subsidies to lenders. Eliminating these features of
FFELP would probably make the program simpler for lenders. However,
borrowers and schools may find the program more complex. In addition, the
loss of access to loans, if it occurred, could burden some students. Any new
regulations to preserve participation of small lenders could burden large
lenders. New regulations designed to preserve both student access to loans
and student or school incentives to shop could burden various program
participants. The nature of those burdens depends on the type of regulation.
For example, if the program limited the range of rates that a lender was
allowed to charge different students, then lenders would face an additional
burden. If students had to satisfy a complex set of criteria to be eligible
to borrow from a lender of last resort, then borrowers who had difficulty
obtaining loans from other lenders would face an additional burden.

Students, schools, and lenders would all have to adjust to changes in
regulations and market practices. The adjustments that would be required are
probably greater than those that the other

36 Even if regulations dealing with borrower rates were eliminated or
modified, regulations dealing with servicing loans could remain unchanged.

CHAPTER 6: MARKET-SET RATES Page 62

models in this report require. If schools changed their preferred lender
lists whenever lenders changed their interest rates, then students could
have to deal with multiple lenders, and financial aid officers would have to
deal with different lenders each year. This might require financial aid
officers to face the confusion and administrative burden of constant changes
in procedures and forms that different lenders required. The same results
could occur if students rather than schools choose lenders. The confusion
and difficulties of dealing with multiple lenders each potentially offering
different interest rates and terms (unlike the present rates and terms that
do not vary by lender) might increase the attractiveness of FDLP to schools.
Lenders would have to determine their cost structures and provide mixes of
prices and services that keep them attractive to students or schools.
Students or schools would have to learn how to choose lenders. Students
might also face difficulties dealing with multiple lenders. Guaranty
agencies would have to deal with more lenders per student than they do now.
Finally, the procedures for consolidating loans might need to change in
order to accommodate the variety of interest rates and terms of FFELP loans.
The results of the analysis for the market-set rates model are summarized at
a broad level in table 13.

Table 13: Summary of Analysis for Market-Set Rates Model

Description of model, including variations Lender and borrower interest
rates would be set by market competition.

Borrowers would pay the rates lenders charged, unless the model included
explicit subsidy for borrowers. Loans retain a federal guarantee. Options
exist to limit variation in borrower interest rates Rate cap for students
might be eliminated or less binding, representing a major federal policy
change.

Costs, savings, and effects on subsidies for program participants

Federal costs could decline but could rise or remain unchanged under some
options. Costs will likely rise for some borrowers and fall for others.
Average borrower costs would be likely to rise in the short run; long-run
effect is uncertain. Government payments to lenders and borrowers would be
eliminated or reduced under some options.

Effects on lender participation, loan availability, and service quality

Lenders have continuing incentive to reduce costs. Service quality may
decline or remain unchanged. Effect on lender diversity is uncertain. Loans
would probably not be available to all students and schools and their
ability to negotiate could be limited unless ameliorative policies were
adopted.

Simplicity, regulatory burden, and program integrity All FFELP participants
would bear a substantial burden of adjusting to a new

system. Borrowers or schools would have a greater burden to shop for loans.
Eliminating some existing regulation would simplify FFELP for lenders, but
new regulations could burden all FFELP participants.

Page 63

CHAPTER 7 INCOME-CONTINGENT REPAYMENT

The Student Loan Reform Act of 1993 expanded the range of loan repayment
options available under the federal student loan programs. Of interest in
considering market mechanisms in these programs is the availability and use
of the ICR plan in FDLP. Any of the proposed models for introducing market
mechanisms into FFELP could include an ICR option or requirement. Under ICR,
the amount of a borrower?s monthly loan repayment varies with income over
time. Some borrowers with low postgraduation incomes or high debt levels
would make lower monthly payments with ICR than with other loan repayment
options. With the monthly installment payment based on the borrower?s
income, it is also far more difficult in theory for the borrower to default.
The availability of ICR expands options for student borrowers to choose
educational programs or careers that are less likely to provide a high
income. Student borrowers who earn less make lower monthly repayments and
are therefore also relieved of some of the risk associated with an uncertain
future.

ICR DESIGN ISSUES COMMON TO MODELS

ICR design issues common to all market mechanism models include how
borrowers? repayment terms should be determined, whether the federal
government or private lenders should hold ICR loans, who should be able to
choose or require ICR, and how borrowers? incomes should be verified.

Repayment Terms Under FDLP ICR Now and Alternatives

Any ICR plan must specify how a borrower?s loan repayments are to be
determined. Under the ICR option that now exists in FDLP, loan repayments
depend in a complex way on income and other factors. Some borrowers may have
portions of their principal or interest forgiven. Examples from other
countries that use ICR illustrate alternative methods of determining
repayment obligations in an ICR system.

In FDLP, borrowers may choose an ICR option under which monthly payments
depend on income and, for some borrowers, loan principal and interest rate
and marital status. Under this option, the borrower?s monthly payment is the
smaller of two amounts: (1) the amount the borrower would repay annually
over 12 years in the absence of ICR, multiplied by an income percentage
factor based on the borrower?s and spouse?s adjusted gross income (AGI) and
marital status, or (2) 20 percent of the borrower?s income in excess of the
poverty level. The ICR repayment obligation is recalculated each year to
reflect changes in the interest rate, income percentage factor, and poverty
level. Appendix IV explains the repayment formula in more detail.

CHAPTER 7: INCOME-CONTINGENT REPAYMENT Page 64

Under FDLP?s ICR plan, the federal government subsidizes some borrowers by
forgiving part of their principal or interest payments. Two kinds of
subsidies are available to ICR borrowers in addition to those available to
Stafford borrowers in general. First, if the borrower?s monthly payment is
less than the monthly interest accrued on the borrower?s loans, then the
unpaid interest is capitalized up to a limit of 10 percent of the original
principal balance on each individual loan. The federal government writes off
any unpaid interest that exceeds this limit. Second, the federal government
forgives any unpaid loan balance remaining after 25 years in repayment. This
forgiveness includes any remaining unpaid principal, unpaid interest, or
charges on the borrower?s loans.

Other countries that use ICR in their student loan systems illustrate
alternatives to the FDLP method of determining ICR repayment obligations.
For example, some countries have income thresholds below which loan
borrowers are not required to make any repayments. Once the borrower has
reached or exceeded the set threshold, payments are determined according a
percentage of income, as in Australia, or of income above the threshold, as
in New Zealand. In Sweden, payments are simply a percentage of total income,
with no threshold. Loan forgiveness provisions vary widely. For example,
Sweden forgives all student loan debt that the borrower has not repaid by
age 65. New Zealand partially forgives interest payments for certain low-
income borrowers. Countries with ICR systems also vary as to whether ICR is
voluntary or mandatory. In some countries, such as Sweden, all borrowers use
ICR. In New Zealand, ICR is the default option but borrowers may choose to
repay their loans at a faster rate than required under ICR. In Australia,
students indicate whether they wish to use ICR or pay their entire
obligation up front. Numerous and significant differences between loan
programs run by other nations make the relevance of these program features
debatable.

ICR Loans Held by the Federal Government or Private Lenders

Either the federal government or private lenders could hold ICR loans in
FFELP. One option that could be incorporated into all market mechanism
proposals is to have the federal government, rather than the FFELP lender,
hold the loans of borrowers who have chosen the ICR plan. For example, the
government could buy at face value (?par?) all privately held loans that
were being converted from a conventional repayment plan to ICR. If the
federal government initially held all FFELP loans (as under the loan sale
proposal), then it could retain ICR loans and sell only non-ICR loans. In
either case, the federal government could use a loan servicing contractor to
service loans in the ICR plan. Borrowers could select ICR before or after
entering repayment.

Private lenders could hold FFELP ICR loans under any of the market mechanism
proposals. If lenders originated all loans, then they could continue to hold
them or sell them to other lenders, regardless of whether a borrower chose
ICR. If the federal government initially held all FFELP loans, then it could
sell both ICR and non-ICR loans to private lenders. Privately held ICR loans
raise special issues related to the verification of borrower incomes.
Financing subsidies to borrowers could also differ between privately held
ICR loans and publicly held ones. These

CHAPTER 7: INCOME-CONTINGENT REPAYMENT Page 65

issues are discussed later in this chapter. An additional option would be
for private lenders to hold ICR loans but have them serviced by the
government of its contractor.

Borrowers could elect to leave ICR, regardless of whether the federal
government or private lenders hold the loans. As with the current ICR
program, the option of changing repayment plans could be incorporated into
ICR under any of the market mechanism proposals.

Choosing or Being Required to Use ICR

In the current FDLP, a standard (non-ICR) repayment schedule is the default
option but any borrower may choose to use ICR instead. Under a market
mechanism approach, ICR in FFELP could continue with this freedom of
borrower choice. Another possibility would be to allow only delinquent and
defaulted borrowers to choose ICR as a last resort. If ICR were limited to
delinquent and defaulted borrowers, another option would be to allow the
lender to put a loan into ICR status. Alternatively, the federal government
could mandate ICR for delinquent or defaulted borrowers, for borrowers whose
student loan debts were large in comparison with their incomes, or even for
all FFELP borrowers. A final possibility would be to make ICR the default
option for all FFELP borrowers but to allow borrowers to choose a standard
repayment schedule instead of ICR.

Income Verification

Regardless of whether the federal government or private lenders held ICR
loans, the extent to which the Internal Revenue Service (IRS) should be
involved in the verification process is a major policy decision that the
Congress would face if ICR were included in FFELP. Income information on
borrowers? federal income tax returns filed with IRS may be more accurate
than income information that borrowers supply (such as pay stubs, bank
statements, and borrowers? own copies of their tax returns). If borrowers
supplied income information to the holders of ICR loans, then some borrowers
might underreport their incomes, especially if the ICR plan includes a
future discharge of indebtedness. Even if loan holders required ICR
borrowers to present certified copies of their tax returns, fraud could
still be an issue. Education?s Office of Inspector General found that
student aid applicants, even when required to provide a copy of their tax
returns as part of the Free Application for Federal Student Aid (FAFSA)
verification process, cannot always be relied on to provide accurate
information. 37 For these reasons, IRS involvement in the verification
process may be desirable. However, IRS has, in general, argued that the use
of tax return information for ?non-tax collection? purposes undermines
public confidence in the tax system and, therefore, reduces voluntary
taxpayer compliance with tax laws.

More specific policy decisions about IRS involvement in income verification
depend on whether the federal government or private lenders would hold ICR
loans under FFELP. If the federal

37 Education, Office of the Inspector General, Accuracy of Student Aid
Awards Can Be Improved by Obtaining Income Data from the Internal Revenue
Service, ACN: 11-50001 (Washington, D.C.: Jan. 29, 1997).

CHAPTER 7: INCOME-CONTINGENT REPAYMENT Page 66

government held the loans, then one possible method of verifying income
would be the one currently FDLP?s ICR plan uses. In FDLP ICR, two methods of
income verification are used: IRS data-matching and borrower-supplied
?alternative documentation? of income. Alternative documentation consists of
a pay stub, dividend statement, or canceled check or, if none of these is
available, a signed statement that explains the borrower?s income sources
and provides its address. Alternative documentation of income is required
for the first year a borrower is in repayment and, for certain borrowers,
for the second year. Otherwise, IRS data-matching is used to verify income.
Appendix IV explains the FDLP income-verification procedure in more detail.

If the federal government held FFELP ICR loans and if IRS information were
used in verifying borrowers? incomes, then a second policy issue might be
whether or how federal contractors involved in administering ICR should be
able to obtain information from federal income tax returns. Contractors now
handle loan operations, such as origination and servicing. Education
requires FDLP ICR borrowers to sign a consent form giving contractors
permission to obtain information about their incomes from IRS. 38 Either the
Congress (through legislation) or Education and Treasury (through
regulation) may wish to consider whether a similar method of borrower
consent should be used for ICR in FFELP.

For the current FDLP ICR plan, Education and IRS have devised a system in
which consent forms are transmitted to IRS electronically for review. (See
app. IV for details.) Education then transmits to IRS the items of
information to be verified for those taxpayers. Treasury estimates that
100,000 consents are processed each year. Education and Treasury have been
working toward a possible statutory or regulatory solution to this
paper-intensive approach. The separate consent form is arguably a
significant paperwork obstacle to borrowers completing the ICR application
process. In addition, the form itself may heighten borrower concern about
the federal use of tax return data.

If private lenders held ICR loans, then the main income-verification issue
would be whether lenders should be able to obtain borrower income
information from IRS or should be required to rely on income information
that borrowers supply. As discussed above, borrower-supplied information may
be less accurate than IRS information. Some members of the lending community
have suggested that lenders could collect income information from FFELP ICR
borrowers and share it annually with Education. Education would then compare
the self-reported data with actual IRS tax return data through some type of
data exchange to ensure that the self- reported income was accurate within a
predefined tolerance. Treasury has stated that in order for this to occur,
either taxpayer consent or an amendment to the IRC would be necessary.

38 The Internal Revenue Code authorizes Treasury to disclose information
about FDLP ICR borrowers? identities, tax filing statutes, and AGI to
officers and employees of Education for the purpose of determining the
appropriate ICR amount (26 U.S.C. 6103(l)(13)). Because Education uses
contractors to administer FDLP, these disclosures are made on taxpayer
consent forms filed with IRS. Subject to certain limitations, the Internal
Revenue Code allows Treasury to disclose a taxpayer?s federal income tax
return information to any person if the taxpayer requests this disclosure
(26 U.S.C. 6103(c)).

CHAPTER 7: INCOME-CONTINGENT REPAYMENT Page 67 ICR VARIATIONS IN THE
DIFFERENT

MARKET MECHANISM MODELS

ICR could be incorporated into any of the market mechanism proposals
discussed in this report. Some features of ICR would vary, depending on the
market mechanism proposal chosen. In particular, the federal acquisition of
ICR loans could differ for some market mechanisms, federal payments to
private lenders that held ICR loans would exist in only one market
mechanism, and any subsidies to ICR borrowers would probably have to be
financed differently for different market mechanisms.

Federal Acquisition of ICR Loans

In all the market mechanism models except loan sales, if the Congress
decided that the federal government should hold FFELP ICR loans, then the
federal government would have to buy from private lenders all loans for
which borrowers chose or were required to use ICR. In the loan sale model,
in which the federal government would originate all loans and then sell
loans to private lenders, it would not auction loans that were designated
ICR loans. Under loan sales, if a borrower chose or were required to use ICR
after a private lender had bought his or her loan, then the federal
government would have to repurchase that loan from the lender.

Federal and Borrower Payments to Private Lenders Holding ICR Loans

Under all the market mechanism models, if private lenders held ICR loans,
then lenders would receive the income-contingent payments that borrowers
made. In addition, under adjustments to the current system only, they would
continue to receive a SAP, based on the difference between the lender yield
and the maximum borrower interest rate.

Financing Subsidies to ICR Borrowers

If private lenders held ICR loans, then the federal government might or
might not reimburse lenders for the costs of ICR borrower subsidies (that
is, for uncapitalized interest and for loan balances forgiven after 25
years). Under all market mechanism models, the federal government would pay
the costs of the subsidies if it reimbursed lenders for the costs of these
subsidies. However, a decision by the Congress not to require the federal
government to reimburse lenders for the costs of the subsidies could mean,
depending on the market mechanism model, that lenders, borrowers, or the
federal government actually paid the costs of the subsidies. Under
adjustments to the current system, if the federal government reimbursed
lenders for the costs of borrower subsidies, then federal FFELP costs would
be higher than if lenders received no reimbursement. If the federal
government did not reimburse lenders for the costs of borrower subsidies,
then FFELP lenders would pay for the subsidies. However, lender
participation in FFELP would probably be reduced, and loan availability
might also be reduced.

CHAPTER 7: INCOME-CONTINGENT REPAYMENT Page 68

Under the federal funding model, if lenders borrowed at a predetermined
interest rate, then lenders? costs would be higher if the federal government
did not reimburse them for borrower subsidy costs. Such a cost increase
would be likely to reduce lender participation in FFELP, reduce any
discounts currently available to borrowers, and possibly reduce loan
availability.

Under all auction models, including the auction variant of federal funding,
whether the federal government reimbursed lenders for the costs of borrower
subsidies would not affect whether the federal government actually paid
those costs. Lenders? bids would depend on whether the federal government
reimbursed them for those costs. If it did not reimburse lenders, then
lenders? bids would reflect the increase in expected costs attributable to
ICR. If loans (in the loan sale model) or origination rights (in the
origination rights auction or volume procurement model) were grouped, bids
would be lower for groups in which borrowers were more likely to receive ICR
subsidies. Alternatively, ICR loans or the right to issue ICR loans could be
auctioned separately from other origination rights. Under all auction
models, the government would bear the anticipated cost of ICR subsidies in
the form of lower revenue from bidding lenders. The net result could be
higher or lower federal costs associated with privately held ICR loans,
depending on whether the private lenders operating expenses were higher or
lower than those of federally operated ICR.

Under market-set rates, if the federal government did not reimburse lenders
for the costs of borrower subsidies, then lenders would probably charge
borrowers higher interest rates, which would reflect the expected costs
attributable to ICR. Therefore, borrowers would be likely to bear the
anticipated cost of ICR subsidies for privately held loans. The borrowers
whom lenders believed most likely to use ICR would bear the greatest share
of the costs of the ICR subsidies. Regulatory limits on the rates lenders
could charge may reduce the availability of FFELP loans, especially to
borrowers whom lenders thought most likely to use ICR.

Page 69

APPENDIX I MANDATE FROM THE HIGHER EDUCATION AMENDMENTS OF 1998

This appendix reprints section 801 of the Higher Education Amendments of
1998 (HEA), Pub. L. No. 105-244.

(a) Study Required.-The Comptroller General and the Secretary of Education
shall convene a study group including the Secretary of the Treasury, the
Director of the Office of Management and Budget, the Director of the
Congressional Budget Office, representatives of entities making loans under
part B of title IV of the Higher Education Act of 1965, representatives of
other entities in the financial services community, representatives of other
participants in the student loan programs, and such other individuals as the
Comptroller General and the Secretary may designate. The Comptroller General
and Secretary, in consultation with the study group, shall design and
conduct a study to identify and evaluate means of establishing a market
mechanism for the delivery of loans made pursuant to such title IV.

(b) Design of Study.-The study required under this section shall identify
not fewer than 3 different market mechanisms for use in determining lender
return on student loans while continuing to meet the other objectives of the
programs under parts B and D of such title IV, including the provision of
loans to all eligible students. Consideration may be given to the use of
auctions and to the feasibility of incorporating income-contingent repayment
options into the student loan system and requiring borrowers to repay
through income tax withholding.

(c) Evaluation of Market Mechanisms.-The mechanisms identified under
subsection (b) shall be evaluated in terms of the following areas:

(1) The cost or savings of loans to or for borrowers, including parent
borrowers.

(2) The cost or savings of the mechanism to the Federal Government. (3) The
cost, effect, and distribution of Federal subsidies to or for participants
in the program.

(4) The ability of the mechanism to accommodate the potential distribution
of subsidies to students through an income-contingent repayment option.

(5) The effect on the simplicity of the program, including the effect of the
plan

APPENDIX I: MANDATE FROM THE 1998 AMENDMENTS TO HEA Page 70

on the regulatory burden on students, schools, lenders, and other program
participants.

(6) The effect on investment in human capital and resources, loan servicing
capability, and the quality of service to the borrower.

(7) The effect on the diversity of lenders, including community-based
lenders, originating and secondary market lenders.

(8) The effect on program integrity. (9) The degree to which the mechanism
will provide market incentives to encourage continuous improvement in the
delivery and servicing of loans.

(10) The availability of loans to students by region, income level, and by
categories of institutions.

(11) The proposed Federal and State role in the operation of the mechanism.
(12) A description of how the mechanism will be administered and operated.
(13) Transition procedures, including the effect on loan availability during
a transition period.

(14) Any other areas the study group may include. (d) Preliminary Findings
and Publication of Study.-Not later than November 15, 2000, the study group
shall make the group?s preliminary findings, including any additional or
dissenting views, available to the public with a 60-day request for public
comment. The study group shall review these comments and the Comptroller
General and the Secretary shall transmit a final report, including any
additional or dissenting views, to the Committee on Education and the
Workforce of the House of Representatives, the Committee on Labor and Human
Resources of the Senate, and the Committees on the Budget of the House of
Representatives and the Senate not later than May 15, 2001.

Page 71

APPENDIX II LIST OF STUDY GROUP MEMBERS

APPOINTED BY EDUCATION AND GAO

Corye Barbour Legislative Director United States Student Association

Bill Beckmann President and Chief Executive Officer Student Loan Corporation

Mary F. Bushman Vice President, Government Relations AFSA Data Corporation

Kathy Cannon Senior Vice President Bank of America

Judy Case Director of Financial Aid University of Massachusetts Medical
School

Rene R. Champagne Chairman, President and Chief Executive Officer ITT
Educational Services, Inc.

Jacqueline Daughtry-Miller Vice President, Student Loan Department
Independence Federal Savings Bank

Anthony P. Dolanski Director, Sallie Mae Servicing Sallie Mae, Inc.

Ivan Frishberg Higher Education Project Director U.S. Public Interest
Research Group

APPENDIX II: STUDY GROUP MEMBERS Page 72

Richard D. George President and Chief Executive Officer Great Lakes Higher
Education Corporation

Jonathan Gruber Department of Economics Massachusetts Institute of
Technology

Michael H. Hershock President and Chief Executive Officer Pennsylvania
Higher Education Assistance Agency

D. Bruce Johnstone Department of Higher and Comparative Education University
at Buffalo State University of New York

James C. Lintzenich President and Chief Executive Officer USA Group

Claire J. Mezzanotte Senior Director, Structured Finance, Asset Backed
Securities Fitch IBCA, Inc.

David Mohning Director of Student Financial Aid Vanderbilt University

Deborah Mott Senior Vice President, Corporate Finance Ferris, Baker Watts,
Incorporated

Barmak Nassirian American Association of Collegiate Registrars and
Admissions Officers

Chalmers Gail Norris Executive Director Utah Higher Education Assistance
Authority

Richard H. Pierce President and Chief Executive Officer Maine Education
Services

APPENDIX II: STUDY GROUP MEMBERS Page 73

Susan L. Pugh Director, Office of Student Financial Assistance Indiana
University

Marilyn B. Quinn Executive Director Delaware Higher Education Commission

Robert A. Scott President Adelphi University

Patricia Smith American Association of State Colleges and Universities

Paul S. Tone Senior Vice President, Industry and Government Relations UNIPAC

Laurie Wolf Director, Enrollment Management Des Moines Area Community
College

Paul W. Wozniak Managing Director PaineWebber Incorporated

DESIGNATED BY FEDERAL AGENCIES

Nabeel Alsalam Principal Analyst Congressional Budget Office

Barbara Bovbjerg Director, Education, Workforce, and Income Security Issues
U.S. General Accounting Office

Robert Cumby Former Deputy Assistant Secretary, Office of Economic Policy
U.S. Department of the Treasury

Maureen McLaughlin Deputy Assistant Secretary, Office of Postsecondary
Education U.S. Department of Education

APPENDIX II: STUDY GROUP MEMBERS Page 74

Lorenzo Rasetti Program Examiner Office of Management and Budget

Page 75

APPENDIX III TECHNICAL ASPECTS OF AUCTION DESIGN

Several of the policy options we reviewed in this report, particularly those
in chapters 3 and 4, involve some sort of auction mechanism for setting the
terms on which student loans are originated. The purpose of this appendix is
to review the relevant theoretical and empirical literature on auctions.

IDEAL CONDITIONS FOR AUCTIONS TO WORK

Standard theoretical treatments of auctions in the economic literature
generally assume that the objective of the party conducting the auction is
pecuniary-that is, that a successful auction is one that provides the most
favorable price quotations by bidders. 39 Bidders are likely to make more
attractive bids, other things being equal, when the following conditions are
present: a large number of bidders, easy entry and exit conditions for
bidders, as much relevant information as possible available to bidders, and
the existence of secondary markets. However, in many practical situations,
the state of the market deviates from the ideal conditions described above.
For example, as we noted in chapter 1, the student loan market has been
characterized by the concentration of much of the business in the hands of a
few large lenders.

The competitiveness of a market can affect the success of auctions, but
auctions themselves can influence the competitiveness of the market. In the
short term, auctions either may make the market more competitive or may have
no effect on competition. They may increase competition by enabling new
participants to enter the market on relatively equal terms with existing
participants. Alternatively, auctions may have no effect on competition if
entry into the market is so expensive that no new participants are willing
to enter. In the long term, it is possible that auctions will gradually
reduce competition and, in the case of FFELP, increase federal costs,
although how likely this outcome is cannot be known.

The current student loan market, however, may not be competitive enough to
enable auctions to produce savings for the federal government. This may be
because a few large lenders dominate the student loan market and it is
costly for new lenders to enter the market. The dominant lenders? influence
on the winning bids could conceivably lead to FFELP costs that are as high
as or higher than current costs. Lenders might also collude in setting their
bids. The relatively small number of lenders in the market would make
collusion easier.

The student loan program is distinguished by several nonpecuniary policy
objectives, such as encouraging participation by small lenders, maximizing
access by student borrowers, and maximizing families? choices of
postsecondary education options.

39 As we noted in chapter 3, in the context of the student loan program,
this could be defined in one of several ways: the interest rate offered by
bidders, a markup over a reference rate such as a T-bill rate or the CP
rate, or a price to be offered by bidders. Unless otherwise noted, the
discussion in this appendix pertains to any pecuniary outcome.

APPENDIX III: TECHNICAL ASPECTS OF AUCTION DESIGN Page 76 OPTIONS FOR
AUCTION DESIGN

There are various ways of structuring the auction process. We review several
of the variants that have been used in financial markets or that have been
proposed for the student loan program. Where appropriate, we note aspects of
auction design that might be significant when the auction is being conducted
under less than ideal conditions. Although most of the available theory and
evidence pertains to the implications of auction designs for pecuniary
outcomes, we also note the implications of auction design for nonpecuniary
policy objectives. This list is by no means exhaustive. Although we indicate
whether a particular auction design may be more or less conducive to
attaining certain objectives than another auction design, we do not
generally answer the question of whether auctions as such are superior to
nonauction methods of setting interest rates and other terms for student
loans.

Uniform Versus Multiple Prices

In an auction for multiple identical items, where bidders can bid on
different quantities of the item, competitive bids state the amount and
price desired and are ranked from the highest to the lowest price. Bids are
accepted at successively lower prices until the desired level of funds has
been raised or all items have been sold. In a uniform-price auction, all
winning bidders then pay the same price-the cut-off price. 40 By contrast,
in a multiple-price auction, all accepted bids are filled at the price that
each bidder bid, so some wining bidders pay more than others.

Treasury recently made a transition from multiple-price auctions to
single-price auctions in its auctions of Treasury securities. A Treasury
official told us that the uniform-price auction was expected to produce
benefits, from Treasury?s point of view, in that it would encourage more
bidders to participate in bidding by reducing the importance of specialized
knowledge regarding market demand and the information costs associated with
its collection. It expected that the concentration of bidders would decline
and that there would be more revenue to the Treasury. Its analysis of data
from the initial experience with single-price auctions lends modest support
for these predictions.

Single Versus Multiple-Round Auctions

Rather than selling items one at a time, a large set of related items can be
auctioned simultaneously, with the auctioning continuing for multiple rounds
until the best possible price is attained for all items. FCC uses
multiple-round auctions for some of its auctions of licenses. The auction
closes when all bidding activity has stopped on all licenses. The principal
advantage of a multiple-round auction is the information that it provides
bidders about the value other bidders place on a license. This information
increases the likelihood that licenses will be assigned to the bidders that
value them the most and will generally yield more revenue than auctions
where there is much uncertainty about common factors that affect the value
of a license to all bidders-that is, who bid and how much was bid. In a
multiple-round auction, bidders

40 A variant of this technique involves accepting the second highest bid.

APPENDIX III: TECHNICAL ASPECTS OF AUCTION DESIGN Page 77

need not guess about the value the second highest bidder places on the
license because bidders have the opportunity to raise their bids if they are
willing to pay more than the current high bidder.

Multiple-round bidding is also more likely than single-round bidding to be
perceived by participants and observers as open and fair. Auction theory
shows that multiple-round bidding tends to increase revenue by reducing the
incentive for bidders to be overly cautious during bidding while trying to
avoid the winner?s curse-meaning the highest bidder would bid too much and
regret its purchase. Multiple-round bidding provides information about other
bidders? estimates of common information, thus reducing bidders? incentive
to bid cautiously to avoid falling victim to the winner?s curse and
regretting their purchases.

An auction that has a single round of bidding is faster than one with
multiple rounds, and the bid evaluation process is simpler. Also, a
multiple-round auction may be more vulnerable to collusion than a
single-round auction and an auction in which each bidder pays a different
price could be either more or less susceptible to collusion than one in
which all bidders paid the same price.

Sealed Bid Versus Open Outcry Auctions

In a sealed-bid auction, each bidder submits a single bid. The bids are then
all opened at once and the winner is determined. By contrast, in an
open-outcry auction, bidders submit bids publicly and then have the
opportunity to revise their bids in light of other bids. Open-outcry
auctions are generally not used in financial markets. A Treasury official
told us that for Treasury auction, it was important to minimize event
risk-that is, the risk that market conditions will change while the auction
is ongoing. An open-outcry auction for Treasury securities might increase
this risk.

Auction Frequency

In a potential auction in FFELP, whether for loan origination rights or for
loans that have already been originated, how frequently auctions would be
held would be an important decision. FFELP costs could be higher the more
frequently auctions were held. Because there would be some costs to
conducting each auction, the total administrative costs of auctions would be
higher the more frequently auctions were held. Similarly, the cost to
lenders of bidding in auctions would be higher the more frequently auctions
were held, which could reduce the number of bidders. More frequent auctions
could also result in fewer bidders because some lenders would not find it
worthwhile to enter the FFELP market if they could not be sure of remaining
in it for a long time. With less competition among lenders, federal payments
to lenders could be higher.

However, there are also reasons why FFELP costs could be lower with more
frequent auctions. Because lower-cost lenders can outbid those with higher
costs, lenders would have a greater incentive to reduce their costs the more
frequently auctions were held. Therefore, federal payments to lenders could
be lower with more frequent auctions. In addition, lenders might increase
their profits during the time between auctions if they were able to reduce
their costs

APPENDIX III: TECHNICAL ASPECTS OF AUCTION DESIGN Page 78

during that time. The federal government would capture more of these profit
increases and could therefore make lower payments to lenders the more
frequently auctions were held. Furthermore, more frequent auctions reduce
lenders? risk that the interest payments they would receive would be out of
line with their cost of funds. Therefore, more frequent auctions could make
origination rights more valuable to lenders, reducing federal payments to
them. Finally, more frequent auctions could make it more difficult for a few
lenders to become entrenched as perennial auction winners. Thus, auction
winners could face more competition at each auction. This competition could
lead to lower federal payments to lenders.

Lender participation could be either higher or lower the more frequently
auctions were held. Because the cost to lenders of bidding in auctions would
be higher the more frequently auctions were held, more frequent auctions
could mean fewer bidders. Small lenders might be especially sensitive to the
costs of bidding. However, because more frequent auctions reduce lenders?
risk that the interest payments they would receive would be out of line with
their cost of funds, they could make origination rights more valuable to
lenders, inducing more lenders to participate. Also, more frequent auctions
could make it more difficult for a few lenders to become entrenched as
perennial auction winners. Thus, past auction winners could face more
competition at each subsequent auction.

Frequent auctions could impose substantial burdens on lenders, students, and
schools. Because participating in auctions would cost lenders both money and
time, more frequent auctions would be more burdensome to lenders. Because
both students and schools value the ability to deal with a single lender,
more frequent auctions could impose a greater burden on them by disrupting
long-term student-lender and school-lender relationships. Less frequent
auctions could have a negative effect on service quality. Lenders whose
rights to originate loans were secure for many years might pay less
attention to service. Students might then become confused about their
repayment responsibilities and the default rate might rise. However, less
frequent auctions could also improve service. Students value the ability to
borrow from a single lender. Having this ability could make it easier for
them to keep track of and repay their debts, thereby reducing the default
rate. The more frequently auctions were held, the more likely it would be
that a student would have to change lenders, especially if each auction had
only one winning lender per school. This effect of auction frequency on
service quality could be eliminated in several ways, though. Each student
could be allowed to remain with one lender throughout his or her educational
program, even if a different lender subsequently won origination rights at
his or her school, or students who had more than one lender could retain the
right to consolidate all their loans. Less frequent auctions could also
improve service quality by facilitating long-term relationships between
lenders and schools. The more frequently auctions were held, the more likely
it would be that a school would have to change lenders, especially if each
auction had only one winning lender per school.

Grouping

It is possible that auctioning origination rights or loan volume for small
groups of schools would produce higher federal FFELP costs than auctioning
rights or volume for larger groups or conducting auctions without grouping
schools. However, there is no information available to

APPENDIX III: TECHNICAL ASPECTS OF AUCTION DESIGN Page 79

enable us to determine how likely this outcome would be. If the per-student
cost of originating loans declines as the number of students served
increases, then it is less expensive for each lender to serve a large number
of students than to serve only a few students. Under these conditions, total
lender costs are lower when there are a few lenders, each of which serves
many students, than when there are many lenders, each of which serves a few
students. Therefore, federal payments to lenders would be greater when each
of a large number of auction winners served a small number of students than
when each of a small number of winners served many students. If schools were
bundled into small groups when auctions were held, then it is possible that
the auctions would produce a large number of winning lenders, each of which
would serve relatively few students. This outcome could be less likely if
schools were bundled into large groups or if schools were not grouped at
all. Thus, federal payments to lenders could be higher the larger the number
of school groups. For example, federal payments to lenders could be greater
if schools were grouped by state than if origination rights for all
FFELP-eligible schools were auctioned as a single package.

Likewise, it is possible that FFELP costs would be higher if multiple
lenders were permitted to serve each school than if there were only a single
lender for each school. Once again, there is no information available to
enable us to determine the likelihood of this outcome. As the previous
paragraph showed, it is possible that federal payments to lenders would be
greater if each of a large number of auction winners served a small number
of students than if each of a small number of winners served many students.
If many lenders were allowed to serve students at each school, then there
could be a large number of winning lenders, each of which served relatively
few students. This outcome could be less likely if only one lender were
allowed per school. Federal payments to lenders could be greater with
multiple lenders per school than with a single lender per school.

How schools were grouped together in the auctions would be likely to affect
the distribution of federal payments to lenders. If schools were not grouped
or if the characteristics of each group of schools resembled those of
FFELP-eligible schools as a whole, then the distribution of those payments
would probably be similar to what it is at present. However, if schools in
each group were similar to one another but different from those in other
groups, then lenders would bid more for the right to originate loans in the
groups they perceived as more desirable (for example, those with higher
per-student loan amounts and lower default rates) than for the right to
originate in ?less desirable? groups. Federal payments to lenders who won
origination rights in the ?less desirable? groups would then be greater than
federal payments to lenders who won rights in the ?more desirable? groups.
The resulting pattern of unequal federal payments to lenders would give the
federal government better information about the costs of lending to students
at different schools, but in so doing it could also lead to an erosion of
political support for FFELP. A similar pattern of unevenly distributed
federal payments to lenders would likely result if lenders were allowed to
define their own groups of schools when they bid. If lenders defined their
own groups of schools, they would probably group schools according to such
characteristics as per- student loan amounts and default rates.

Students in low-income regions, from low-income families, or attending
proprietary vocational schools could lose access to loans if Education
grouped similar schools together or allowed

APPENDIX III: TECHNICAL ASPECTS OF AUCTION DESIGN Page 80

lenders to define their own groups of schools. In a volume procurement
auction, those types of students could lose loan access if winning lenders
were allowed to choose the students to whom they would lend. Lenders would
be likely to consider those kinds of students as presenting especially high
default risks. If they were allowed to, some lenders might refuse to lend to
students with those characteristics or to bid on groups of schools that
served such students. If similar schools were grouped together or if lenders
were permitted to group schools themselves, then groups of schools that
lenders perceived as serving students with high risks of default (for
example, schools with very high student loan default rates) might attract no
bids. In a volume procurement auction in which winning lenders were
permitted to choose their borrowers, lenders might refuse to lend to
students at those schools.

Ability to Pay

Auctions that involve payments to a government entity by a successful bidder
for specified rights, as opposed to offering an interest rate, would require
rules to ensure that bidders were able to pay the amounts they bid. Without
such rules, nonserious bidders could distort competition. Other federal
auctions provide examples of rules that could be adopted in an origination
rights auction. EPA requires each bidder in its sulfur dioxide emission
allowance auctions to send a certified check or letter of credit to cover
its bid before the auction, or else to specify a method of electronic
transfer or other payment method. (This rule is feasible only for sealed-bid
auctions, in which each bidder submits a single bid.) FCC requires bidders
in its wireless spectrum auctions to submit refundable deposits to cover the
cost of placing bids. A final option, not used in any federal auction, is to
require all potential bidders to show some evidence of their ability to pay.
One way of implementing this option is to use FFELP eligibility criteria as
evidence of ability to pay. Another alternative is to require FFELP-eligible
lenders to pass additional ability-to-pay tests before allowing them to
participate in the auction.

Federal costs could depend on whether winning bids are payable in full
immediately after the close of the auction or whether a schedule of
installment payments is allowed. Installment payments enable small bidders
to participate more easily. Small bidders would have to come up only with a
down payment rather than with the full purchase price, enabling bidders with
less funding to compete with better funded large bidders. However, allowing
loan purchasers to pay in installments could result in winning bidders not
paying off the installments. When FCC allowed some spectrum auction winners
to pay their bids in installments, it experienced many defaults by auction
winners. This raised a related issue of who owns the auctioned item (student
loans) if the purchaser declares bankruptcy. In the case of wireless
spectrum rights FCC auctioned, bankruptcy of winning bidders has left the
disposition of the auctioned item to be determined by a bankruptcy court.
Solving this issue with respect to student loan packages before
implementation will enable Education to reclaim and resell such loans in
subsequent auctions.

Small Bidders

Some federal auctions use several methods of ensuring small bidders? access.
In its auctions of federal debt, Treasury allows small buyers not to
participate in the auction and to agree in

APPENDIX III: TECHNICAL ASPECTS OF AUCTION DESIGN Page 81

advance to buy a limited, pre specified amount of debt at the price that
emerges from the auction. (Of course, only a small minority of buyers could
use this option. If many buyers decided to use it, then the auctions would
have few bidders and would not work properly.) Treasury also imposes a 35
percent limit on the market share of any winning bidder. FCC offers bidding
credits to small bidders. In addition, FCC formerly allowed auction winners
to pay their bids in installments, but this policy resulted in many defaults
by auction winners. Independent of its annual auction, EPA gives electric
power plants some sulfur dioxide emission allowances for free. 41 In the
early years of its auctions, EPA also sold allowances at a fixed price
outside auction. However, the price it set turned out to be much higher than
the market price, and it later abandoned this practice.

Evidence from the FCC wireless spectrum auctions suggests that small-bidder
protections can make auctions more competitive. In some early auctions, FCC
gave preferential treatment to certain categories of bidders (such as
women-owned and minority-owned businesses) for certain types of licenses.
This policy intensified bidding competition among the nonpreferred bidders.
At the same time, preferred bidders bid more for the licenses that were
subject to the preferences than nonpreferred bidders would have been willing
to pay for the same licenses. The preferences increased auction revenues.

Commercial Paper Versus Treasury Rates

FFELP costs could be lower if the maximum borrower interest rate were based
on the CP rate than if it were based on the 91-day T-bill rate. Under
current law, the maximum borrower interest rate is based on the 91-day
T-bill rate and the lender?s cost of funds is based on the CP rate. The
relationship between the 91-day T-bill rate and lenders? cost of funds is
relatively unstable. For this reason, federal payments to lenders that are
based on the CP rate are more valuable to lenders than federal payments that
are based on the 91-day T-bill rate, for loans made at the same
predetermined interest rate. A convenient way of setting a predetermined
lender interest rate would be to set it equal to the borrower interest rate.
Thus, if the borrower rate were based on the CP rate, lenders? bids would be
more favorable to the government than if the borrower rate continued to be
based on the 91-day T-bill rate. FFELP costs would be lower if the basis for
the borrower rate were changed to the CP rate.

41 In addition to protecting small lenders, free distribution of some
student loan origination rights before auction could be a method of phasing
in origination rights auctions.

Page 82

APPENDIX IV DETAILS ON INCOME-CONTINGENT

REPAYMENT IN FDLP

THE RATIONALE FOR ICR

ICR gives borrowers the opportunity to repay student loans based on their
income. One analyst has stated that current economic conditions place a
greater burden on borrowers than conditions of previous eras did. Low
inflation means that today?s graduates and borrowers will pay back, in real
terms, as much as 40 percent more than student loan borrowers who graduated
in the 1970s and 1980s. Graduates of earlier eras benefited from paying back
their loans in cheaper dollars. Current borrowers cannot depend on inflation
to reduce their ?real? debt burden. However, today?s graduates may have
higher inflation-adjusted postgraduation incomes than graduates of the 1970s
and 1980s. We are unable to determine whether today?s graduates face higher
debt burdens relative to their incomes than graduates of earlier eras.
However, if today?s graduates do face higher debt burdens than graduates of
the 1970s and 1980s, and if it is believed that today?s graduates should
have the same debt burdens as graduates of those earlier decades, then a
properly designed and administered ICR plan can be justified as one way to
reduce the burden for today?s graduates. ICR allows some borrowers with low
incomes or high debt burdens to maintain a good credit rating by making
lower monthly payments than they would have to make under a standard
repayment schedule. With the monthly installment payment based upon the
borrower?s income, it is also far more difficult in theory for the borrower
to default. The availability of ICR also expands the educational and
occupational options for student borrowers. It may be especially valuable
for some borrowers who are in the early stages of their careers.

DATA ON ICR

The Student Loan Reform Act of 1993 expanded the range of loan repayment
options available under the federal student loan programs. Of interest to
the study group on market mechanisms in the student loan programs is the use
of the various repayment plans available in FDLP, especially the ICR plan.
The use of the various repayment plans varies significantly by whether the
loan was a consolidation loan. Thus, we are providing an analysis of overall
ICR usage in the Direct Loan program as well as ICR usage among
consolidation borrowers.

All Loans in Repayment

Nearly 2.7 million borrowers have loans in repayment under FDLP. They have
loan amounts totaling nearly $34 billion. 42 Of the total loan amount, 31
percent is subsidized Stafford loans, 17 percent is unsubsidized Stafford
loans, 11 percent is PLUS loans, and 41 percent is consolidation loans.

42 All information is as of May 31, 2000.

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 83

Of the total amount in repayment, 56 percent is being repaid through the
standard repayment plan, 20 percent is being repaid through the graduated
repayment plan; 12 percent is being repaid through the extended repayment
plan, and 2 percent is being repaid through ICR.

Roughly half of the loans being repaid under the ICR option require payments
that are less than the interest due on the loan. See tables 14 and 15.

Table 14: Direct Loan in Repayment by Loan Type

Loan Number of borrowers Percent of

total borrowers

Principal balance outstanding Loan amount Average

loan amount Percent of total

loan amount

Subsidized 1,501,329 56% $9,627,369,029 $10,431,517,693 $6,948 31%
Unsubsidized 932,386 35 5,861,802,639 5,694,855,988 6,108 17 PLUS 372,453 14
3,335,602,873 3,788,560,419 10,172 11 Consolidation 709,786 26
13,447,855,619 14,070,935,315 19,824 41

Total a 2,684,031 100% $32,272,630,160 $33,985,869,415 $12,662 100%

a Unduplicated borrower count. Table 15: Direct Loans in Repayment by Plan

Repayment plan Number of borrowers Percent of

total borrowers

Principal balance outstanding Loan amount Average

loan amount Percent of total

loan amount

Standard 1,944,060 72% $17,315,051,937 $18,884,768,619 $9,714 56% Graduated
402,876 15 6,757,482,096 6,729,516,982 16,704 20 Extended 149,703 6
3,981,172,966 4,116,109,476 27,495 12 Income contingent repayment 193,289 7
4,041,457,111 4,049,233,342 20,949 12 Payments greater than or equal to
interest

101,725 4 2,033,145,308 2,057,503,793 20,226 6 Payments less than interest
91,564 3 2,008,311,803 1,991,729,549 21,752 6 Alternative plan 10,740 0
177,466,051 206,240,996 19,203 1

Total a 2,684,031 100% $32,272,630,160 $33,985,869,415 $12,662 100%

a Unduplicated borrower count.

Consolidation Loans in Repayment

As shown in table 14, consolidations are more than 40 percent of the Direct
Loan repayment portfolio by loan amount. Education tracks this closely, as
it allows borrowers to consolidate defaulted loans for the purpose of
obtaining an income contingent repayment arrangement. Defaulted loans that
are not being rehabilitated through the Direct Loan income contingent

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 84

repayment or another reasonable and affordable plan are normally pursued and
collected by collection agencies contracted by Education?s Debt Collection
Service (DCS).

Nearly 710,000 borrowers have consolidation loans in repayment totaling more
than $14 billion in original loan amounts. Almost all (95 percent) of the
$14 billion in original loan amounts are ?regular? consolidation loans. The
remaining 5 percent represent loans previously held by DCS. See table 16.

Of the total consolidation amount in repayment (table 17) 29 percent is
being repaid through the standard repayment plan, 27 percent is being repaid
through the ICR plan. 22 percent is being repaid through the graduated
repayment plan, and 20 percent is being repaid through the extended
repayment plan.

Roughly half of the loans being repaid under ICR require payments totaling
less than the interest owed on the loan. See tables 16 and 17.

Table 16: Source of Direct Consolidation Loans in Repayment

Consolidation type Number of

borrowers Percent of total

borrower count

Principal balance outstanding Loan amount Average

loan amount

Percent of total

loan amount

Regular consolidation loans

598,715 84% $12,762,306,108 $13,362,649,377 $22,319 95% Defaulted loans
formerly held by DCS

112,135 16 684,081,869 706,738,859 6,303 5

Total a 709,786 100% $13,446,387,978 $14,069,388,236 $19,822 100%

a Unduplicated borrower count. Table 17: Direct Consolidation Loans in
Repayment by Plan

Repayment plan Number of borrowers Percent

of total borrower

count Principal

balance outstanding

Loan amount Average loan amount

Percent of total

loan amount

Standard 301,817 43% $3,729,080,437 $4,145,517,660 $13,735 29% Income
contingent repayment 182,178 26 3,816,246,780 3,834,087,757 21,046 27
Payments less than interest 87,383 12 1,923,399,764 1,936,038,142 22,156 14
Payments greater than or equal to interest

94,795 13 1,892,847,016 1,898,049,615 20,023 13 Graduated 137,190 19
3,053,074,468 3,101,952,324 22,611 22 Extended 84,028 12 2,730,481,991
2,852,202,918 33,943 20 Alternative plan 5,337 1 117,197,409 135,216,042
25,336 1

Total a 709,786 100% $13,446,081,085 $14,068,976,702 $19,821 100%

a Unduplicated borrower count.

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 85

Of the loans being repaid through income-contingent repayment (see the
193,289 borrowers in table 15). Ninety-four percent are consolidation loans
(see the 182,178 borrowers in table 17).

Of the consolidation loan amounts previously held by DCS, 11 percent are in
ICR. Of the regular consolidation loan amounts, not previously held by DCS,
28 percent are in ICR. See tables 18 and 19.

Table 18: Direct Consolidation Loans in Repayment by Plan: Defaulted Loans
Formerly Held by DCS

Repayment plan Number of

borrowers Percent of total

borrowers Principal balance

outstanding Loan amount Average loan

amount Percent

of total loan amount

Standard 63,550 57% $318,715,398 $343,082,099 $5,399 49% Graduated 29,748 27
216,876,576 216,141,878 7,266 31 Income contingent repayment

11,778 11 80,719,328 77,778,778 6,604 11 Extended 6,905 6 66,899,517
68,730,851 9,954 10 Alternative plan 10 0 40,742 45,933 4,593 0

Total a 112,135 100% $683,251,561 $705,779,539 $6,294 100%

a Unduplicated borrower count. Table 19: Direct Consolidation Loans in
Repayment by Plan: Regular Consolidation of Non-DCS Loans

Repayment plan Number of

borrowers Percent of total

borrowers Principal

balance outstanding

Loan amount Average loan amount

Percent of total loan

amount

Standard 238,397 40% $3,409,473,808 $3,801,485,419 $15,946 29% Graduated
77,204 13 2,663,424,434 2,783,316,598 36,051 21 Income contingent repayment

107,603 18 2,835,914,828 2,885,506,253 26,816 22 Extended 170,798 29
3,735,395,565 3,756,175,778 21,992 28 Alternative plan 3,523 1 89,756,731
103,761,954 29,453 1

Total a 598,715 100% $12,733,965,367 $13,330,246,002 $2,265 100%

a Unduplicated borrower count.

ICR Usage Under FDLP

Although the ICR option appears to be more expensive for many borrowers over
the full repayment term than, for example, the standard repayment plan, no
ICR borrower ?overpays? interest because the current plan has been
established without an internal cross-subsidy. Analyses of possible
repayment options, as well as contemporary repayment calculators that are
provided to aid borrowers? choices among various repayment plans, are
typically described in total dollars rather than in terms of the net present
value of those dollars. This can make ICR seem to be more expensive for
borrowers than it really is. Some observers believe that this illusory
high-interest cost of ICR has discouraged financial aid administrators from
urging

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 86

borrowers to consider choosing it. Also, these analyses and repayment
calculators at best can only assume, and cannot predict, a borrower?s future
income. Such projections have typically specified the historical income
growth for college graduates (on average, 5 percent annually). However, ICR
is more valuable to borrowers with a flatter income trajectory. The fact
that ICR usage has not approached initial projections does not mean that ICR
should be discarded. While the currently operational ICR plan may in fact be
relatively unattractive to many borrowers, financial aid administrators?
poor familiarity with it has certainly affected its take-up rate.

MONTHLY PAYMENT CALCULATION FOR LOANS IN ICR

The monthly payment calculation under ICR is the lesser of the principal
balance and AGI based calculation or 20 percent of the borrower?s
discretionary income. These two calculations are explained below. 43

Principal Balance and AGI-Based Calculation

The monthly payment is the amount the borrower would repay annually over 12
years, using standard amortization multiplied by an income percentage factor
based on the borrower?s or couple?s AGI and whether the borrower is single
or married and head of the household:

12-year standard amortization = (principal balance)(monthly interest
rate)/[1 (monthly interest rate + 1) -n ].

The monthly interest rate is equal to the annual interest rate divided by
12, and n is the number of months remaining in the repayment term.

For this calculation, the principal balance is the original principal
balance plus any capitalized interest when the borrower first entered
repayment. This amount remains static except for the following situations:

1. Reporting new disbursements, disbursement adjustments, and cancellations.
The principal balance is adjusted to reflect these changes. 2. A new loan is
received after the new loan has entered repayment. In this situation, the

principal balance is the principal balance of the new loan when the loan
first entered repayment plus any capitalized interest when the borrower
first enters repayment plus the outstanding principal balance on the
existing loans (including capitalized interest) plus outstanding accrued
interest, collection charges, late charges, and any other charges. 3.
Borrowers request joint repayment. In this situation, the principal balance
is recalculated on

the basis of the borrower?s combined outstanding principal balance when the
borrower enters 43 See also ?Examples of the Calculations of Monthly
Repayment Amounts,? published annually in the Federal Register, most
recently at 65 Fed. Reg. 34,006 (25 May 2000).

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 87

joint repayment plus any capitalized interest when the borrower first enters
repayment plus the outstanding principal balance on the existing loans
(including capitalized interest) plus outstanding accrued interest,
collection charges, late charges, and any other charges. 4. For purpose of
the annual recalculation of the payment amount when new income

information is received, after periods in which a borrower makes payments
that are less than interest accrued on the loan. In this situation, the
principal balance is the highest outstanding principal balance (including
amounts capitalized) calculated for the borrower while paying under the ICR
plan if this amount is higher than the original principal balance when the
loan first entered repayment, plus any capitalized interest.

The monthly payment calculation is principal based and AGI based monthly
payment = (12-year standard amortization * income percentage factor).

The income percentage factor is obtained from a table published annually by
the Secretary in the

Federal Register. 44 If the borrower?s exact AGI is not found in the table,
the income percentage factor is obtained by linear interpolation between the
next higher and next lower AGIs. The linear interpolation method calculates
the income percentage factor on the basis of intervals between the incomes
and income percentage factors on the table.

20 Percent of the Borrower?s Discretionary Income

The borrower?s discretionary income is calculated on the basis of poverty
guidelines provided by the U.S. Department of Health and Human Services
(HHS): discretionary income = AGI poverty guideline. Thus, the monthly
payment calculation is discretionary income monthly

payment = (discretionary income * 0.20)/12.

Treatment of Married Borrowers Not in Joint Repayment

For a married borrower to be eligible for ICR, the borrower?s income
information and the borrower?s spouse?s income information are required,
even if the spouse files a separate tax return. Thus, both spouses are
required to sign the IRS consent form. In addition, both spouses are
required to provide the same required income type, AGI, or alternative
documentation of income. If both spouses have direct loans and are in joint
repayment, the servicer refers to the borrower who has the most recent
in-school period to determine income type.

If the borrower is separated from the borrower?s spouse, the borrower is not
required to provide the spouse?s income. The borrower is required to send
the servicer a self-certifying statement that indicates the change in
marital status. If the borrower is separated and the borrower filed a joint
income tax return, the borrower may submit alternative documentation of
income for only the borrower.

44 65 Fed. Reg. 34, 007 (25 May 2000).

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 88

Treatment of Married Borrowers in Joint Repayment

Married borrowers may choose to repay their loans jointly under ICR. The
decision to repay jointly under ICR does not make each borrower liable for
the other?s direct loan debt. Joint repayment is used as a method to
determine the borrowers? monthly repayment amounts. In order to be eligible,
both borrowers are required to request joint repayment on the Repayment Plan
Selection Form. The payment amount is based on the borrowers? combined
outstanding direct loan debt and the borrowers? combined income when the
servicer calculates the borrowers? joint repayment amount. The borrowers?
individual payment amount is determined so that each payment amount is
proportionate to each person?s level of debt. When payments are applied in
any joint repayment situation, they should be applied to interest on both
borrowers? accounts before they are applied to principal on either account.
This helps avoid negative amortization on either account. Obviously, if
payments are not high enough to cover interest on both accounts, there will
be negative amortization. In cases in which both spouses choose joint
repayment and the loans are serviced at different servicers, one of the
borrower?s loans will be transferred.

If the servicer holds the loans of only one borrower and the borrower is
selecting ICR for the first time and wants to repay jointly, the servicer
should wait until the spouse?s loans are transferred before calculating the
joint repayment amount. Therefore, the borrower should be put on interest
billing, even if the servicer has income information for borrower and
spouse. The borrower should be left on interest billing until the spouse?s
loans are transferred.

Married borrowers may select joint repayment, even if they filed their taxes
separately. The ICR plan does not assume that a borrower?s AGI is
proportionate to his or her debt. A joint payment amount is calculated on
the basis of the combined AGI and combined debt amounts. The borrower
receives a bill for the portion of the joint payment amount that is
proportionate to the borrower?s individual debt.

ALTERNATIVE DOCUMENTATION OF INCOME

The servicer is required to collect alternative documentation of income for
the first year a borrower is in repayment and for certain borrowers for the
second year. 45 All married borrowers must submit alternative documentation
of income for their spouses, unless the borrower is separated from the
borrower?s spouse. If the borrower has been in repayment for more than a
year, the servicer must determine whether the AGI data or alternative
documentation of income should be used to calculate the ICR monthly
repayment amounts.

45 Documentation supporting the borrower's income can be in the form of a
pay stub, a dividend statement, or a canceled check. If these types are not
available, the borrower can provide a signed statement that explains each
income source and provides its address. If the borrower provides the
servicer with other forms of supporting documentation, the servicer
determines whether the documentation is acceptable at its own discretion,
with guidance from Education's On-Site Monitor. Supporting documentation of
alternative documentation of income cannot be older than 90 days.

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 89

The servicer collects alternative documentation of income from borrowers if
the borrower?s AGI from IRS would be likely to reflect any period of time in
which the borrower was in school. If AGI data do not represent an in-school
period, the servicer uses the AGI data from IRS to calculate the monthly
payment amount. If the borrower receives an in-school deferment, the
servicer does not consider the time of the in-school deferment as an
in-school period. In situations in which the borrower has multiple loans
with the servicer, if any of the loans require alternative documentation of
income, the borrower is required to provide alternative documentation for
all the loans.

If the borrower is not required to submit alternative documentation of
income, the servicer calculates the borrower?s monthly payment amount using
the borrower?s AGI data from IRS. However, the servicer collects alternative
documentation of income in situations in which IRS cannot provide the
servicer with valid income information or if the borrower?s reported AGI
does not reasonably reflect current income. The servicer accepts the
borrower?s alternative documentation for the following situations:

the borrower has lost or changed employment or is undergoing some other
special circumstances, the borrower did not file a sufficiently recent tax
return, IRS provided income information but not currently enough to
determine the payment amount, or IRS did not provide AGI data during the
first AGI solicitation and the borrower did not request the servicer to
re-solicit IRS.

In addition to these situations, the servicer can use alternative
documentation of income from the borrower even when IRS is processing a
request for AGI data or if AGI data are received from IRS. However,
alternative documentation of income cannot be submitted or solicited solely
to reduce the amount of time a borrower is assigned ICR interest-only
monthly payments. The guideline is to be used only if the alternative
documentation most accurately reflects the borrower?s current ability to
repay the loan. If the servicer has both current AGI and alternative
documentation of income, the servicer uses the alternative documentation of
income to calculate the monthly repayment amount, unless the borrower
requests otherwise.

If a borrower and the borrower?s spouse have chosen to repay their Direct
Loans jointly, the servicer refers to the spouse who most recently left
in-school status in order to determine whether AGI or alternative
documentation of income should be solicited.

ANNUAL INCOME INFORMATION RENEWAL

Each year, the servicer renews the borrower?s income information, either by
soliciting AGI data from IRS with the taxpayer?s consent or by receiving
updated alternative income information from the borrower. The servicer
begins the annual income renewal process at the end of August. The servicer
does not request updated income information if the servicer has received
income information from the borrower within the past 6 months. For example,
if the borrower entered

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 90

repayment under ICR in March 1999, the servicer would not have renewed the
borrower?s annual income until August 2000.

If the borrower?s loans require renewal of income information, the servicer
must determine whether the borrower is required to submit alternative
documentation of income or whether AGI data from IRS can be used. Valid AGI
data cannot reflect an in-school period. If AGI data reflect an in-school
period, the servicer uses alternative documentation of income to recalculate
the monthly payment.

Renewal for Borrowers Who Require AGI

At the end of August each year, the servicer attempts to solicit IRS for AGI
data for the borrower and spouse. If the servicer cannot obtain the AGI
data, the servicer asks the borrower to submit alternative documentation of
income. Alternative documentation is acceptable to recalculate the payment
amount if the AGI data are not obtained from IRS or if the AGI data do not
accurately reflect the borrower?s current income. Once the servicer receives
updated income information, the servicer recalculates the borrower?s monthly
payment amount. After the recalculation, the servicer discloses the
borrower?s new monthly payment amount.

If the servicer does not receive updated income information from IRS, or the
alternative documentation of income form and supporting documentation from
the borrower by the end of the calendar year, the servicer removes the
borrower from the ICR plan. The servicer reverts the borrower?s loans to the
standard repayment plan, or to the borrower?s previous repayment plan if the
borrower switched to ICR from another repayment plan.

Renewal for Borrowers Who Require Alternative Documentation of Income

For borrowers required to submit alternative documentation of income, the
servicer asks at the end of August each year for the alternative
documentation of income form and supporting documentation for income renewal
purposes. The servicer uses the AGI data from IRS to verify the borrower?s
previous year?s alternative documentation of income information.

If the servicer does not receive the form and supporting documentation from
the borrower by the end of the calendar year, the servicer removes the
borrower from the ICR plan. The servicer reverts the borrower?s loans to the
standard repayment plan or to the borrower?s previous repayment plan if the
borrower switched to the ICR plan from another repayment plan.

IRS Consent Form

Before the servicer is able to solicit AGI data from IRS, the borrower and
the borrower?s spouse must provide the servicer with a valid and signed IRS
Consent to Disclosure of Tax Information form. A married borrower is not
eligible for ICR without an IRS consent form with both spouses? signatures
and IRS validation of the form. When the servicer receives the form, the

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 91

servicer ensures that the borrower and spouse have correctly completed it.
If the form is incomplete, the servicer either sends an imaged copy of it
with a new consent form or retrieves the original consent form and sends it
to the borrower and spouse to complete. The borrower and spouse must either
complete a new consent form in its entirety and submit it to the servicer or
complete the original consent form and submit it to the servicer.

Once the servicer receives a completed IRS consent form, the servicer images
the borrower?s IRS consent form and passes the image to IRS within 3
business days. If the form is invalid- for example, if IRS receives the
image of the form more than 60 days after the date when the borrower signed
the form-IRS rejects it. If IRS rejects the form, the servicer requests the
borrower to submit an updated form.

When IRS initially validates the form, the form is valid for 5 tax years.
The 5 tax years are defined on the IRS consent form. Any subsequent renewals
of the forms are also valid for 5 tax years.

Solicitation for IRS Consent Form Renewals

IRS consent form renewals are solicited in December of the year following
the last covered tax year indicated on the form. For example, at the end of
December 2000, the servicer has to begin the renewal cycle for all IRS
consent forms that expired after the 1999 tax year. If the servicer does not
receive a new IRS consent form from the borrower, the servicer requests an
IRS consent form at least two additional times before the end of June. If
the borrower does not provide an updated form, the servicer does not solicit
updated AGI data in August. The servicer removes the borrower from the ICR
plan the December following the initial renewal attempt if the borrower does
not provide the servicer with an updated IRS consent form and if updated
income information is not received.

If the borrower voluntarily revokes the IRS consent form, the servicer
immediately removes the borrower from the ICR plan. If the borrower does not
choose a new repayment plan, the servicer notifies the borrower that the
servicer placed the borrower on the standard repayment plan.

Treatment of Married Borrowers

A married borrower is required to have his or her spouse?s consent
authorizing IRS to disclose tax information for the purpose of determining
the borrower?s AGI. The servicer requires both spouses to sign the IRS
consent forms in order for a married borrower to participate in the ICR
plan. Both spouses must sign the IRS consent form, even if the couple filed
a joint tax return. Both signatures are required for all married borrowers,
unless they are separated.

Soliciting AGI Data From IRS

The servicer solicits IRS for AGI data once a month. The servicer solicits
the AGI data from IRS for a particular borrower either when initial or
renewal AGI data are required. Depending on the AGI data sent by IRS, the
following three situations could occur:

APPENDIX IV: DETAILS ON INCOME-CONTINGENT REPAYMENT IN FDLP Page 92

IRS mismatch. IRS is unable to recognize the borrower?s or spouse?s Social
Security number or name as provided by the servicer. No AGI data. The
borrower or spouse did not file a tax return the previous year covered by
the IRS consent form, or IRS provided outdated (or invalid) AGI data,
because sufficiently current data were not available. The servicer receives
sufficiently current AGI data. IRS provides the borrower?s or spouse?s
current AGI data.

If IRS cannot provide the servicer with AGI data because of a mismatch, no
AGI data are available, or IRS sent outdated AGI data, the servicer only
resolicits IRS for AGI data on the borrower?s request. (The borrower is
permitted to request that the servicer resolicit IRS for AGI data at any
time.) If IRS cannot provide income information, the servicer requests
alternative documentation of income from the borrower. When the servicer
receives valid AGI data from IRS, the servicer calculates the borrower?s
monthly payment amount on the basis of the AGI.

Eligible AGI Data and Alternative Documentation for Borrowers Entering ICR

AGI data used to calculate the repayment amount must be current data. Prior
or next-prior calendar year AGI data (or both, for a married borrower and
spouse) from IRS are acceptable if received before August 31 for borrowers
entering ICR who are required to provide AGI data. For information received
after August 31, the servicer accepts only prior-year AGI data from the IRS.
For example, 1998 or 1999 AGI data are acceptable if the servicer received
the information from IRS before August 31, 2000. If the servicer received
the information after August 31, 2000, only 1999 AGI data were acceptable.
If the borrower submits alternative documentation of income, the only
requirement is that the supporting documentation not be older than 90 days.
If married borrowers have income information from two different tax years,
the servicer designates the most current tax year to the income data.

Substitution for AGI Data

The servicer can accept a signed tax return with accompanying required
documentation (for example, W-2 forms or 1099 forms) from the borrower as
proof of AGI data, if the servicer cannot obtain the data from IRS. A signed
tax return from the borrower cannot be considered alternative documentation
of income but may be used temporarily to calculate the borrower?s monthly
payment. If the servicer receives a signed tax return from a borrower, the
servicer attempts to recollect AGI data from IRS, using the signed tax
return as proof that valid AGI data exist for the borrower. If the servicer
continues not to be able to obtain AGI data from IRS, the servicer collects
alternative documentation of income from the borrower to recalculate the
borrower?s monthly payment amount.

Page 93

APPENDIX V CALL OPTIONS AS A MECHANISM FOR DETERMINING NET LENDER YIELDS

Competitive forces in the market could help establish lender yield on loans
and remove any ?excess? lender profits through the inclusion of a ?call
option? on all new FFELP loans. The call option would give the federal
government the right to ?buy? the FFELP loan within a specified time period
at a predetermined price (?exercise price?) set by legislative formula. The
exercise price would be set when the loan originates and would include
consideration for a lender?s cost in making the loan.

Within the specified time period, the government would auction the call
option (the right to buy the loan). FFELP lenders willing to pay the highest
positive value to the government for the option would receive the right to
purchase the loan at the predetermined exercise price. The government would
receive the bid price on the option. If lenders were unwilling to bid a
positive value for the option, the federal government could purchase the
loan at the exercise price or the originator would keep the loan.

The call option model uses competitive secondary market forces to determine
the value of a loan. In determining whether and how much to pay for the call
option, buyers review the expected value of the loan (including special
allowance payments, interest payments, and so on) relative to the exercise
price. When the expected value exceeds the exercise price, potential buyers
would be likely to bid positive amounts for the option.

Assume solely for illustration that origination costs were 1 percent of the
loan amount. The exercise price would be 101 percent of the face value of
the loan and would just compensate the originator for expenses. If the
estimated income from the loan for the time it would be held were worth more
to the originator, or to other potential loan holders, than 101 percent of
the face value, then the option would have a positive value. The
government?s selling the option would recoup those excess returns.

EFFECT ON PROGRAM STRUCTURE

Under this mechanism, most aspects of FFELP could remain unchanged. As is
generally true of the other models discussed in the body of the report, the
federal government could provide special allowance payments if deemed
desirable as an addition to the lenders? return from a maximum statutory
student interest rate, and loans would remain guaranteed lenders. Lenders
would originate and service loans, and schools and borrowers could continue
to choose their loan originator. Additionally, loan originators could
continue to compete for loan volume through decreased interest rates and
fees to students.

APPENDIX V: CALL OPTIONS AS A MECHANISM FOR DETERMINING NET LENDER YIELDS
Page 94 EFFECT ON BORROWERS

Financial benefits to students would remain unchanged or might increase
under this model. It is unlikely that benefits to borrowers would decrease,
because concessions by lenders would be a way of getting additional
business; they would not be required to bid in advance for the additional
loans. The concession to borrowers would diminish the likelihood that they
would have to bid for the option to retain the loan.

The model could permit differential concessions to student borrowers but
would not require differential student rates to ensure loan availability.
Availability of loans to all borrowers could be assured in one of several
ways: (1) The special allowance payments and interest rates would provide a
return adequate to service the most-expensive-to-service loans, (2) The
government could exercise the options on loans expensive to service and
resell the loans at a loss, or (3) This model could be supplemented (as
suggested for other models discussed above) by a lender of last resort or
administrative requirements on FFELP lenders.

In a completely private exchange market, all options would not be sold for
the same price, even apart from discounts to students. Because some loans
are cheaper to service than others, as discussed earlier in this report, it
would be expected that options on these loans would sell at a higher price.
Where the government is on one side of the transaction, it would be
desirable to bundle options by loan characteristic in order to allow for
price variations to parallel these cost variations. The same issues of
design of option bundles would emerge as would arise in the rights auction
model. The possibility of private transactions and bundling of government
sales could mean that, unlike the current system in which special allowance
payments provide a one- size-fits-all gross return, this option model could
produce more efficient incentives of all types of loans, without
necessitating excess net returns for cheap-to-service loans.

BUDGETARY EFFECT

While it is unclear whether the government would realize any budgetary
savings under this model, federal costs are not expected to increase. Any
potential excess lender profits after the exercise date of the options would
either be provided to students in the form of discounts or recouped by the
federal government in the option price.

EFFECT ON LENDERS

The model would likely decrease or maintain existing lender returns in the
aggregate. It is unlikely that overall lender returns would increase.
Rather, if bundling of options by loan characteristics could be done
effectively, the model would foster an environment where income would more
closely match the characteristics of the loan. Loans with higher servicing
costs (lower estimated value) would demand lower call option prices.

The originator of a loan might have a slight advantage in bidding on the
options, because he or she would not bear the burden of exercising the
options and loading the loans on his or her own, different loan servicing
system. Combined with existing arrangements through which lenders

APPENDIX V: CALL OPTIONS AS A MECHANISM FOR DETERMINING NET LENDER YIELDS
Page 95

essentially extend a line of credit to students for their course of study at
a particular school, this advantage would tend to keep a student?s loans
with a given servicer. This would be efficient for both servicer and
student. The existing possibilities for loan consolidation also need not be
disturbed in this model.

MODEL CONSIDERATIONS

One key component of this program would be the call option exercise
price-that is, the price at which the government would have the right to
acquire the loans. It should be established near the cost to originate
student loans. If the exercise price were too low, loan originators would
not participate. If the exercise price were too high, there would not be
bidders for the options. The government would be required either to let them
expire (leaving loans with the originators, who might earn an excess return)
or to exercise the option and resell the loans at a loss (albeit, probably a
small loss unless the error on the exercise price was quite large).

The costs related to loan origination are very similar among all student
loans, regardless of delinquency or balance. Measuring the cost of
origination would require obtaining information that is readily available
about certain fixed and variable costs. The origination allowance built into
the exercise price could be a fixed amount of dollars per loan plus a
percentage of the borrower?s principal balance. Because many originators of
student loans are not the ultimate holders, contracts with secondary
marketers could also provide a source of information on origination costs.
In contrast, when the Congress (or its designee) is evaluating the net
return associated with a given SAP, it must measure a wide range of lender
costs: servicing costs over the likely 10-year life of loans, the effects of
differing loan balances, the likelihood of default costs, and the likelihood
that loans will be prepaid or consolidate. It should be much easier to
determine the market cost of originating student loans than to determine the
market rate of return for loans over their life, taking these other costs
into account.

Another key design decision involves determining the call option
availability period (the time when the option can be exercised). Lender
uncertainty regarding the option could be considerably reduced if the option
first became effective near the time when originators now most frequently
sell many student loans. This might be just as the loan entered repayment.
Alternatively, the effective period for the option might be set earlier, if
it were thought that a substantial portion of any excess profits on loans
were received early in the life of the loan. In either case, it would seem
desirable for the option to have a fixed expiration date, without an
excessively long window in which it could be exercised.

Page 96

APPENDIX VI ADDITIONAL AND DISSENTING VIEWS

FROM RENE CHAMPAGNE

This appendix presents an additional and dissenting view submitted by study
group member Rene Champagne in accordance with section 801(d) of the Higher
Education Amendments of 1998.

Received by e-mail December 12, 2000. I am particularly concerned with any
proposal that attempts to segregate federal student loans by type of student
or by type of institution because of the inherent opportunity for
discrimination provided under such formulas. Title IV programs have been
created by Congress to insure equal access to postsecondary education for
all students regardless of race, ethnic background, gender or income level
assuming the student met certain eligibility standards applied uniformly to
all students. The same holds true for institutions. Congress has made it
very clear that they will not tolerate the segregation of institutions
reflected but their support of Historically Black Colleges, Career Colleges
and Schools and Community Colleges as well as traditional four-year
institutions. I fear segregation by student type and institution type as
contained in certain proposals borders on ?redlining? and therefore must be
avoided. ?Blended? portfolios must continue to be used in the future to
insure that all students are properly afforded equal access to the
institution of their choice.

Page 97

APPENDIX VII ADDITIONAL AND DISSENTING VIEWS FROM MICHAEL HERSHOCK AND
RICHARD PIERCE

This appendix presents an additional and dissenting view submitted by study
group members Michael Hershock and Richard Pierce in accordance with section
801(d) of the Higher Education Amendments of 1998.

January 16, 2001 Dear Reader: Attached is a paper entitled ?Today's
Competitive Loan System Is Already Filled with Market Mechanisms,? prepared
in our capacity as Members of the Market Mechanisms Study Group. We were
assisted in the preparation of the paper by Harrison Wadsworth of the
Education Finance Council.

We have requested that this paper be appended to the report of the study
group because we believe that the paper presents a viewpoint that is not
thoroughly explored in the body of the study group report. It is our hope
that this paper will assist readers in better understanding the market
forces currently at work in the Federal Family Education Loan Program.

The presentation of this paper is not intended as a dissenting view, nor
should its presence in the report's appendix be interpreted as representing
any particular characterization of the study group report.

Michael H. Hershock Richard H. Pierce President and CEO President and CEO
Pennsylvania Higher Education Maine Education Services Assistance Agency

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM MICHAEL HERSHOCK AND
RICHARD PIERCE Page 98

Today's Competitive Loan System Is Already Filled With Market Mechanisms

A proposal prepared for the Study Group On Market Mechanisms In Federal
Student Loan Programs

Prepared by: Richard H. Pierce President and CEO Maine Education Services
Corp.

Michael H. Hershock President and CEO Pennsylvania Higher Education
Assistance Agency

Harrison M. Wadsworth Deputy Executive Director Education Finance Council

May 5, 2000

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM MICHAEL HERSHOCK AND
RICHARD PIERCE Page 99

Today's Competitive Loan System Is Already Filled With Market Mechanisms

We recommend that one of the three or more systems of market mechanisms
included in the General Accounting Office/Department of Education report to
Congress under Section 801 of the Higher Education Amendments of 1998 should
be the current student loan system. The current system involves intense
competition for business among thousands of private organizations, with
additional competition from state government agencies and the federal Direct
Loan program.

Under any scenario that preserves the basic mission of the Higher Education
Act, Congress will always be involved in setting prices for at least some
segments of borrowers. We believe the alternative is unacceptable - reducing
access to higher education for students. Today's prices are ceilings, which
Congress could lift if it wished. Given the level of competition, prices of
loans for most borrowers might not rise. Prices for some categories of
borrowers probably would - such as borrowers attending high-default schools
with low loan balances. Any federally supported student loan system will
always have intensive Congressional oversight. No system is perfect.
Imperfection will always invite attempts at improvement by lawmakers and
administrators.

Market mechanisms currently exist in every aspect of the FFEL Program.

Loan origination: Lenders compete to offer schools the fastest, most
efficient, most reliable origination of loans. Students demand this, and
school financial aid administrators in turn also demand the best quality
service. Loan delivery must be accomplished within hours of the receipt of
an application. Anything more will send students and schools elsewhere. A
characteristic of the FFEL Program is that there are many alternatives. Once
a school has entered direct lending, the only choice if there are service
problems is to convert to FFELP, a process that requires the school to
forsake the investment it has made in systems that process direct loans. The
federal government should not take steps to make FFELP more like direct
lending, winnowing participation down to a few huge players offering
basically the same products, leaving schools and borrowers with few choices.

Loan Servicing: Private lenders must follow due diligence procedures in
servicing loans or they lose the 98% federal guaranty. Some lenders, such as
EFC members, only exist to make student loans. If they displease students
and schools, they will be out of business. Even in large diverse lending
organizations, the student loan department is usually a specialized area
where the jobs of its employees depend on success in student lending.

In addition, a number of financial considerations motivate lenders of all
types to provide top quality service, in addition to the necessity of
complying with federal regulations. Lenders not only lose 2% of their
principal when loans default. They also

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM MICHAEL HERSHOCK AND
RICHARD PIERCE Page 100

have cash flow issues, because they have to keep repaying their investors
whether or not they are receiving payments on the underlying loans. When a
loan defaults, there is a delay before insurance is paid, causing cash flow
problems that, if severe, could cause a default on bond payments. Bond
issues are structured with reserves and other methods to make a default
highly unlikely for the majority of an issue. This brings a high rating from
credit rating agencies, reducing borrowing costs. Low borrowing costs
translate to low interest rates, lots of competition for business, including
borrower benefit programs for students and their families. One of the key
factors that rating agencies check is the track record of the company that
will service the student loans that form the collateral for a securities
issue.

High quality loan servicing holds a rank of highest importance in the
federally backed loan programs. The penalties for problems are severe.
Borrowers are tracked down and punished by seizure of income tax refunds,
wage garnishment, lawsuits, loss of professional licenses and destruction of
credit ratings. Bankruptcy usually doesn't help since a student loan
generally cannot be discharged. Schools face bad publicity at a minimum and
closure at a maximum if their students' default rates rise too high.
Guaranty agencies lose money at increasing rates if their borrowers' default
rates rise too high. The Education Department and Congress face the wrath of
the public and a potential loss of confidence in the loan programs if
default losses rise again, like they did in the late 1980's. In other words,
everyone involved in the loan program has an incentive to make sure that if
a loan does default, it is not because of servicing problems. There are
strong incentives to do everything possible to avoid loan defaults and to
cure them if they do default.

Liquidity: As loan volume has grown in recent years, there never has been a
problem with finding adequate funds for college. A healthy secondary market
for loans ensures liquidity in the student loan market. That means that
funds are available at reasonable cost. As it has become apparent that the
FFEL Program will survive and grow, more investors, including international
investors, have been willing to invest in student lending. This is good for
American students. It shows the health of the loan program. It is something
that should be encouraged. Proposals that disrupt the system or cause
uncertainty and instability will result in reduced interest in investing the
tens of billions of dollars needed every year to finance student loans. At a
minimum, this loss of liquidity will increase lending costs, meaning more
money will go into paying off bonds, leaving less for improvements in
technology and for borrower benefits.

Subsidies for Borrowers: Subsidies from various sources are in place to help
reduce borrowers' costs. As a result, student loans are some of the lowest
cost loans available anywhere. The federal government has put a ceiling on
interest rates and capped that ceiling to make sure they stay low. The
government lowers rates further by paying interest for borrowers at various
times. Lower-income borrowers have their Stafford Loan interest paid while
in school and during periods of deferment. All borrowers, including parents,
are guaranteed that their interest payments will not rise above a low level,
no matter what the Federal Reserve does. These caps have saved borrowers
millions of dollars. Recent Stafford Loan borrowers have a portion of their

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM MICHAEL HERSHOCK AND
RICHARD PIERCE Page 101

interest costs - 0.5 percentage point - paid by the government at all times.
On top of that, interest payments are deductible from federal income taxes
for many borrowers. The combination of subsidies leaves a net interest cost
for subsidized Stafford loans of less than 4% - the lowest-cost consumer
loans available. On top of that, most borrowers can reduce their costs even
further thanks to lender-offered borrower benefits programs. The system is
set up to keep borrowing costs as low as possible.

Access: Much of the discussion of market mechanisms in the student loan
programs and of ways to set interest rates seems to ignore the fundamental
purpose of the federally backed student loan programs - to ensure that all
Americans have a way to pay for higher education. A corollary purpose of the
programs, also important, is to keep the cost of borrowing as low as
possible. There are various philosophical points of view about borrowing.
Some, including many in other countries, believe that the nation should
provide a higher education to all who wish it or all who qualify. For many
years, the state of California followed that policy, not charging tuition to
in-state residents attending state universities, although significant "fees"
were charged. Others believe that governmental assistance to pay for higher
education should be concentrated on students with the greatest need. Another
approach treats assistance as a reward for students who demonstrate merit.
Policy in the United States has included all three approaches, with
need-based aid the predominant approach since 1965. Increasingly in recent
years, states and schools have put resources into merit-based grant aid. But
the loan programs have always been either based on need or open to all,
without regard to grades or other merit factors. Given the trend towards
merit-based grants, the FFEL Program will become more critical to ensuring
that every American can pay for college. Whether this trend changes or not,
this country will need a large loan program for the foreseeable future.

Summary and Conclusions: Experimentation with massive restructuring of the
FFEL Program could threaten its viability. Any major changes must be weighed
against the strength of the existing program. The best decision for Congress
may be to do nothing rather than continuing the turmoil that has kept
program participants on edge for the past 10 years. Constant upheaval is
costly and eventually wears down loan providers, financial aid
administrators, and others who are involved in the loan programs.
Competitive pressure and the technological revolution of the 1990's have
brought significantly reduced borrowing costs for students, their families
and taxpayers. This is a good thing. It should be encouraged.

Competition with direct lending is no longer a central issue for lenders,
although the competition with FFELP does seem to be a major concern for
schools in the Direct Loan Program. Lenders are much more focused on doing
battle in the marketplace with their FFELP competitors.

Currently, the largest lenders are increasing market share. Still, the
smaller state and regional lenders serve to provide alternatives to the big
national companies, keeping prices low and service high. Many times they are
able to meet local needs and provide specialized service to schools thanks
to longstanding relationships. This allows customization of products to fit
the needs of particular schools, particular students and

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM MICHAEL HERSHOCK AND
RICHARD PIERCE Page 102

particular states. It would be a mistake to encourage development of an
effective cartel or oligopoly by altering the loan program so that only the
largest participants remain viable.

Beware of arguments about competition being ?unfair.? Such statements always
suffer from tunnel vision -- only looking at one of many factors that affect
the competition. For example, should it be considered unfair for an
organization to have access to interest-free funds? Is it unfair for one
company to be able to offer its student loan customers other financial
products that a competing company (or program) cannot offer? Probably not.

Student loan programs should not be dumbed down, with the federal government
picking the lowest common denominator for loan programs and forcing everyone
else to reduce their services and raise their prices to that level. The
program should remain flexible enough to encourage excellence, innovation
and the use of the latest technology.

Market mechanisms already dominate the way prices are set and services are
provided in today's FFEL Program. Any study of market mechanisms should
include a look at how competition works in today's program, which is
effectively delivering $22 billion in new loans this year while servicing a
$150 billion portfolio with increasing efficiency and decreasing defaults.

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 103

APPENDIX VIII ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE

This appendix presents an additional and dissenting view submitted by study
group members Paul Tone and Richard Pierce in accordance with section 801(d)
of the Higher Education Amendments of 1998.

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 104

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 105

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 106

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 107

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 108

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 109

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 110

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 111

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 112

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 113

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 114

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 115

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 116

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 117

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 118

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 119

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 120

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 121

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 122

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 123

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 124

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 125

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 126

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 127

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 128

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 129

APPENDIX VIII: ADDITIONAL AND DISSENTING VIEWS FROM PAUL TONE AND RICHARD
PIERCE Page 130



            