401(k) Pension Plans: Loan Provisions Enhance Participation but May
Affect Income Security for Some (Letter Report, 10/01/97, GAO/HEHS-98-5).

Pursuant to a congressional request, GAO: (1) determined the effects of
pension-plan borrowing on participation in and contributions to 401(k)
pension plans; (2) described the demographic and economic
characteristics of workers who borrow from their pension accounts; and
(3) identified the potential consequences for participants who borrow
from their pension accounts.

GAO found that: (1) plans that allow borrowing have a somewhat higher
proportion of employees participating than other plans, all other
factors being equal; (2) in addition to employer matching, allowing
borrowing increases participation among eligible employees, especially
lower-income employees; (3) allowing pension-plan borrowing also
significantly affects how much employees contribute; (4) participants in
plans that allow borrowing contribute, on average, 35 percent more to
their pension accounts than participants in plans that do not allow
borrowing; (5) based on individual financial data GAO examined,
relatively few plan participants--less than 8 percent--have one or more
loans from their pension accounts; (6) this is true for a point in time
and would not include those who had repaid a past loan or who might
borrow in the future; (7) blacks and hispanics, lower-income
individuals, participants who have recently been turned down for a loan,
and workers who also are covered by other pension plans are more likely
to borrow from their pension account than other participants; (8) plan
borrowers, on average, have fewer assets than nonborrowers and have more
nonhousing debt relative to income than nonborrowers; (9) while
borrowing provisions may reduce retirement income, they also can
encourage workers to save for their retirement; (10) loan provisions of
many pension plans provide for repaying the loan at favorable interest
rates, which may be lower than the investment yield that could have been
earned had the money been left in the pension account; (11)
consequently, the borrower will have a smaller pension balance at
retirement, since the interest paid to the account is less than what the
account balance could have earned form investing in equities; however,
other potential effects of borrowing could outweigh these disadvantages;
(12) if loan provisions influenced the employee's decision to
participate in the pension plan, the employee's retirement income would
likely have been even less had there not been such provisions; (13)
allowing participants to borrow from their defined-contribution pension
plan, therefore, may be a double-edged sword; and (13) there are both
advantages and disadvantages to borrowing from other voluntary
retirement savings accounts, such as individual retirement accounts,
however, few of the positive effects of pension-plan borrowing would be
realized in mandatory retirement programs like Social Security.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  HEHS-98-5
     TITLE:  401(k) Pension Plans: Loan Provisions Enhance Participation 
             but May Affect Income Security for Some
      DATE:  10/01/97
   SUBJECT:  Retirement pensions
             Employee retirement plans
             Tax law
             Employee benefit plans
             Disadvantaged persons
             Loans
             Loan interest rates
             Social security benefits
             Blacks
             Hispanics
IDENTIFIER:  Social Security Program
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** 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.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to the Chairman, Special Committee on Aging, and the Honorable
Judd Gregg, U.S.  Senate

October 1997

401(K) PENSION PLANS - LOAN
PROVISIONS ENHANCE PARTICIPATION
BUT MAY AFFECT INCOME SECURITY FOR
SOME

GAO/HEHS-98-5

401(k) Pension Plans

(207446)


Abbreviations
=============================================================== ABBREV

  IRA - individual retirement account
  IRS - Internal Revenue Service
  EBRI - Employee Benefit Research Institute

Letter
=============================================================== LETTER


B-277050

October 1, 1997

The Honorable Charles E.  Grassley
Chairman
Special Committee on Aging
United States Senate

The Honorable Judd Gregg
United States Senate

Since they were established in 1978, 401(k) pension plans have grown
rapidly and now account for about one quarter of all pension plans
and pension assets.\1 Through 401(k) plans, workers can affect their
retirement savings, depending on the contribution level they choose
and the investment decisions they make.  Most 401(k) pension plans
also allow participants to borrow against their pension accounts at
relatively low interest rates.  Pension-plan borrowing is not limited
to 401(k) plans.  For example, to encourage saving in individual
retirement accounts (IRA), recent legislation allows early withdrawal
from these accounts for certain purposes, such as for a first-time
home purchase. 

Proponents of pension-plan borrowing argue that allowing such access
to retirement savings before retirement increases participation in
pension plans where participation is voluntary.  They also argue that
pension-plan borrowing increases contributions to plans where
participants determine their contribution levels.  However, critics
argue that pension-plan borrowing often results in lower pension
income at retirement.  In addition, opponents to the creation of
individual Social Security accounts fear that if borrowing from such
accounts were allowed, Social Security income could be jeopardized. 

Given the importance of pension-plan borrowing and possible
implications for Social Security reform and individual retirement
accounts (IRA), you asked us to (1) determine the effects of
pension-plan borrowing on participation in and contributions to
401(k) pension plans, (2) describe the demographic and economic
characteristics of workers who borrow from their pension accounts,
and (3) identify the potential consequences for participants who
borrow from their pension accounts. 

To conduct our work, we analyzed two databases.  The first is a
nationally representative sample containing detailed demographic and
employment information about American workers in 1992; the second
contains pension plan information for plans with 100 or more
participants in 1992.  These data were the most recent available at
the start of our study, and we have no reason to believe that worker
characteristics have significantly changed.  We also reviewed the
economic and pension literature and solicited comments from outside
pension analysts.  We did our work between October 1996 and August
1997 in accordance with generally accepted government auditing
standards.  (For further details on our scope and methodology, see
app.  I.)


--------------------
\1 There are two basic types of pension plans:  defined benefit and
defined contribution.  Pension benefits in defined-benefit plans are
generally calculated with a formula based on years with the firm, age
at retirement, and salary averaged over some number of years.  In
defined-contribution plans, employers generally promise to make
guaranteed periodic contributions to workers' accounts, but
retirement benefits are not specified.  Many employers also match
workers' contributions to these accounts.  A 401(k) plan is a
defined-contribution plan. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

Plans that allow borrowing have a somewhat higher proportion of
employees participating than other plans, all other factors being
equal.  In addition to employer matching, allowing borrowing
increases participation among eligible employees, especially
lower-income employees.  Allowing pension-plan borrowing also
significantly affects how much employees contribute.  Participants in
plans that allow borrowing contribute, on average, 35 percent more to
their pension accounts than participants in plans that do not allow
borrowing. 

Based on individual financial data we examined, relatively few plan
participants--less than 8 percent--have one or more loans from their
pension accounts.  This, of course, is true at a point in time and
would not include those who had repaid a past loan or who might
borrow in the future.  Blacks and Hispanics, lower-income
individuals, participants who have recently been turned down for a
loan, and workers who also are covered by other pension plans are
more likely to borrow from their pension account than other
participants.  Plan borrowers, on average, have fewer assets than
nonborrowers and have more nonhousing debt relative to income than
nonborrowers. 

While borrowing provisions may reduce retirement income, they also
can encourage workers to save for their retirement.  Loan provisions
of many pension plans provide for repaying the loan at favorable
interest rates, which may be lower than the investment yield that
could have been earned had the money been left in the pension
account.  Consequently, the borrower will have a smaller pension
balance at retirement, since the interest paid to the account is less
than what the account balance could have earned from investing in
equities.  However, other potential effects of borrowing could
outweigh these disadvantages.  For example, borrowing for purposes
such as education or training could increase lifetime family income
and, hence, retirement income.  In addition, if loan provisions
influenced the employee's decision to participate in the pension
plan, the employee's retirement income would likely have been even
less had there not been such provisions. 

Allowing participants to borrow from their defined-contribution
pension plan, therefore, may be a double-edged sword.  Our results
also suggest that there are both advantages and disadvantages to
borrowing from other voluntary retirement savings accounts, such as
IRAs.  However, few of the positive effects of pension-plan borrowing
would be realized in mandatory retirement programs like Social
Security. 


   BACKGROUND
------------------------------------------------------------ Letter :2

The number of defined-contribution pension plans, especially 401(k)s,
has been growing, and by 1993, they accounted for 88 percent of all
pension plans and 61 percent of all active pension-plan participants. 
Many participants' defined-contribution plans supplement another
pension plan. 

A 401(k) pension, or salary-reduction, plan is a defined-contribution
plan that allows participants to contribute, before taxes, a portion
of their salary to a qualified retirement account.  Investment income
earned on 401(k) account balances accumulates tax-free until the
individual withdraws the funds at retirement.  However, participation
in 401(k) is voluntary, and contribution levels are determined by the
individual but can be no larger than $9,500 per year. 

About 85 percent of 401(k) pension plans are the only pension plan
sponsored by the employer, although the majority of 401(k) plan
participants are covered by another pension plan.  A recent study of
selected 401(k) plans shows that worker participation rates for these
plans varies from about 50 percent to over 90 percent.\2 Participant
contributions to 401(k) accounts, on average, are about 7 percent of
earnings.  To encourage participation in and contributions to these
pension plans, plan sponsors may wholly or partially match employee
contributions and provide education on the importance of retirement
saving.  In addition, over half of all 401(k) pension plans allow
participants to borrow from their pension accounts. 

Borrowing from 401(k) pension plans is legally permissible and allows
plan participants to borrow the lesser of $50,000 or one half of
their vested account balance.  The term of the loan cannot exceed 5
years, unless the loan is used to purchase a primary home. 
Furthermore, the loans are generally offered at very favorable
interest rates.  A recent survey of 401(k) plans found that about 70
percent of the plans that allow borrowing charge an interest rate
equal to or less than the prime rate plus one percentage point, while
less than 10 percent charge an interest rate equal to the local
bank's lending rate.\3

The repayment of loan principal and, unless the primary residence is
used to secure the loan, interest payments are not tax-deductible. 
Failure to repay the loan results in the outstanding loan balance
being considered a taxable pension distribution.  The borrower is
then responsible for paying all taxes on the distribution plus a
10-percent early withdrawal penalty if the borrower is under 59-1/2
years old. 

Overall, about half of all firms with 100 or more employees that
offer savings and thrift plans permit participant loans.\4
Previously, GAO reported that 57 percent of the firms with 100 to 499
employees that offer a 401(k) plan permit participant loans.\5
Similarly, 80 percent of firms with 500 to 4,999 employees and 46
percent of firms with 5,000 or more employees permit loans against
401(k) accounts.  In addition, over 95 percent of 401(k) plans that
offer loans have at least one plan participant with an outstanding
loan. 

Pension-plan loan provisions, however, are controversial.  Advocates
argue that loan provisions are an incentive to lower-income workers
to participate in pension plans where participation is voluntary. 
Furthermore, many 401(k) plan administrators think loan provisions
also have a somewhat positive impact on participants' contribution
rates to pension accounts.  A survey conducted by the Employee
Benefit Research Institute (EBRI) suggests that most workers think
that participants should be able to withdraw retirement funds to pay
for financial emergencies, to buy a house, or to pay for a child's
education.\6 Workers may be more willing to save for retirement if
they can have access to their savings for emergencies before
retirement. 

Opponents of loan provisions argue that permitting participants to
borrow from their retirement accounts works against the policy
objective of enhancing retirement income.  Almost half of the
companies that do not permit 401(k) loans surveyed by William M. 
Mercer say that loan programs are contrary to plan philosophy.\7
Almost 60 percent of employed respondents to a recent EBRI survey
think about using their own retirement funds only at the time of
retirement.\8


--------------------
\2 Robert L.  Clark and Sylvester J.  Schieber, Factors Affecting
Participation Rates and Contribution Levels in 401(k) Plans
(Washington, D.C.:  Watson Wyatt Worldwide, May 1996). 

\3 Buck Consultants, 401(k) Plans:  Employer Practices and Policies
(New York:  Buck Consultants, Sept.  1996).  The report does not
indicate the bank lending rate used. 

\4 Department of Labor, Bureau of Labor Statistics, Employee Benefits
in Medium and Large Private Establishments, 1993, Bulletin No.  2456
(Nov.  1994). 

\5 401(k) Plans:  Incidence, Provisions, and Benefits
(GAO/PEMD-88-15BR, Mar.  29, 1988). 

\6 EBRI, Public Attitudes on Investment Preferences, EBRI Report No. 
G-44 (Washington, D.C.:  EBRI, May 1993). 

\7 William M.  Mercer, Inc., Fax Facts Survey Results:  Savings Plan
Loans (New York:  William M.  Mercer, Inc., Aug.  9, 1995). 

\8 EBRI, Public Attitudes on Investment Preferences, EBRI Report No. 
G-61 (Washington, D.C.:  EBRI, Nov.  1994). 


   LOAN PROVISIONS, AMONG OTHER
   FACTORS, ENCOURAGE
   PARTICIPATION IN AND LARGER
   CONTRIBUTIONS TO 401(K) PLANS
------------------------------------------------------------ Letter :3

Our analysis of the two databases on worker characteristics and
pension-account activity shows that pension-plan borrowing increases
participation in 401(k) plans (see app.  II).  However, a number of
other factors, such as employer matching and size of firm, also
influence participation and contribution amounts. 

Participation rates in plans with loan provisions are about 6
percentage points higher than plans with no loan provisions (see fig. 
1).  Employer matching also increases participation rates by about 20
percentage points depending on the match rate.  These findings are
consistent with the results of other research.\9 Under the typical
situation--where the employer contributes about half of what
participants contribute--borrowing provisions plus employer matching
increases participation by about 28 percentage points--from 55
percent to 83 percent.\10

   Figure 1:  Predicted
   Participation Rates in a Firm
   With a 401(k) Plan

   (See figure in printed
   edition.)

Note:  Participation rates are for a firm offering no other pension
plan. 

Our analysis also indicates that smaller firms tend to have slightly
higher participation rates than larger firms.  This may be due to
smaller firms more effectively targeting benefits to employee needs. 
In addition, a recent study found that the type and quality of
information provided to employees on 401(k) plans may also be an
important factor in encouraging employee participation in 401(k)
pension plans.\11 The impact of providing high-quality information
appears to be greatest on workers with lower earnings. 

In our analysis of 401(k) plans, we also found that average annual
employee contribution amounts are 35 percent higher in 401(k) plans
with loan provisions than in 401(k) plans with no loan provisions. 
Employer matching also increases average contribution amounts in
401(k) plans but not to the same extent as loan provisions. 
Depending on the employer match rate, we estimate that average annual
employee contribution amounts are typically 10- to 24-percent higher
than with no employer matching.  The effect of both loan provisions
and employer matching can be dramatic--an increase in average
contribution amounts of over $600 per year (see fig.  2). 
Furthermore, one study suggests that providing high-quality
pension-plan information to plan participants may also increase
contribution levels to 401(k) plans.\12

   Figure 2:  Predicted Average
   Employee Contribution Amounts
   to a 401(k) Plan

   (See figure in printed
   edition.)

Note:  Average contribution amounts are for a firm offering no other
pension plan. 

These results are further corroborated when we examined individual
participant contributions to 401(k) pension accounts.  We estimate
that a typical 401(k) participant\13

covered by a pension with loan provisions and receiving an average
employer match rate will contribute a higher proportion of earnings
to his or her 401(k) account than an identical participant covered by
a plan with no loan provision or employer matching--8.6 percent
versus 4.9 percent (see fig.  3). 

   Figure 3:  Predicted
   Contribution Rates to a 401(k)
   Plan

   (See figure in printed
   edition.)


--------------------
\9 Clark and Schieber, Factors Affecting Participation Rates and
Contribution Levels in 401(k) Plans, and William E.  Even and David
A.  Macpherson, Factors Influencing Employee Participation in 401(k)
Plans (Oxford, OH:  Miami University, Dec.  1996). 

\10 Some research suggests that employer matching does not play a
critical role in increasing participation in 401(k) plans.  It may be
that the match-rate variable in our analysis is reflecting plan
characteristics that are not reported in the database.  See Leslie E. 
Papke, "Participation in and Contributions to 401(k) Pension Plans,"
Journal of Human Resources, Vol.  30, No.  2 (1995), pp.  311-325,
and Andrea L.  Kusko, James M.  Poterba, and David W.  Wilcox,
Employee Decisions with Respect to 401(k) Plans:  Evidence From
Individual-Level Data, Working Paper No.  4635 (Cambridge, MA: 
National Bureau of Economic Research, Feb.  1994). 

\11 Clark and Schieber, Factors Affecting Participation Rates and
Contribution Levels in 401(k) Plans. 

\12 Clark and Schieber, Factors Affecting Participation Rates and
Contribution Levels in 401(k) Plans.  However, it may also be true
that workers who are more likely to contribute to a 401(k) plan are
more likely to attend pension-plan training sessions. 

\13 The typical 401(k) participant is married, college educated,
white, 45 years old, and male; has an annual family income of $61,469
and a net worth of $146,182; and has participated in the pension plan
for 4 years.  The relative impacts of pension-plan borrowing and
employer matching, however, do not depend on the demographic
characteristics of the participant. 


   BORROWERS MAY BE LESS
   ECONOMICALLY SECURE THAN
   NONBORROWERS
------------------------------------------------------------ Letter :4

Plan participants with no outstanding plan loans are in a better
financial position than borrowers.  Plan borrowers, on average, have
less family income, lower net worth, and more nonhousing debt than
nonborrowers.  Total family income of borrowers is 83 percent of that
of nonborrowers (see table 1).  The total net worth and nonhousing
net worth of borrowers is also considerably lower than that of
nonborrowers.  In addition, retirement-account borrowers have about
$1,500 more in nonhousing debt and have much higher
nonhousing-debt-to-income ratios than nonborrowers. 



                          Table 1
          
            Average Assets, Debts, and Income of
           Defined-Contribution Plan Participants
          Who Can Borrow From Retirement Accounts,
                            1992

                                               Outstanding
                               No pension-    pension-plan
                                 plan loan          loan\a
--------------------------  --------------  --------------
Loan balance from defined-               0         $2,963\
 contribution account
Nonhousing debt                    $16,911        $18,509\
Nonhousing net worth              $113,021        $77,672\
Net worth                         $161,827       $119,304\
Total family income                $62,139        $51,422\
Ratio of nonhousing debt      25.2 percent    42.5 percent
 to total family income
----------------------------------------------------------
\a Significantly different from the no-loan group at the 1-percent
level. 

Source:  GAO analysis of the 1992 Survey of Consumer Finances
database. 

Nevertheless, our analysis indicates that 401(k) plan participants
who also are covered by another pension plan are 50-percent more
likely to have an outstanding loan than other participants (see app. 
II).  Those with only a 401(k) pension plan--and, thus, with the most
to lose by borrowing from their pension accounts--are less likely to
do so.  But participants who have recently been turned down for a
loan from another source are almost 40-percent more likely to borrow
against their pension account than other plan participants, holding
all else equal. 

Black and Hispanic pension-plan participants are almost twice as
likely as white participants to borrow against their pension account
(see app.  II), after controlling for income and assets.  Minorities
may have more difficulties obtaining commercial loans, including
mortgages.\14 Our results also indicate that other characteristics of
an individual, such as age, gender, and marital status, do not
significantly affect pension-plan borrowing. 

Pension-plan borrowers may use their pension-plan loan for living
expenses, an automobile purchase, or housing (rather than borrowing
from a commercial source to finance a home purchase), all of which
could be considered necessities.\15 A smaller proportion of
pension-plan borrowers report having housing debt than nonborrowers,
but a larger proportion report having education loans (see table 2). 



                          Table 2
          
          Proportion With Specific Types of Loans
             and Attitudes Toward Borrowing for
                      Specific Reasons

                                               Outstanding
                               No pension-    pension-plan
                                 plan loan          loan\a
--------------------------  --------------  --------------
Outstanding housing debt             70.0%           64.7%
Outstanding education                13.2%           19.7%
 loans
All right to borrow for
education expenses                   91.2%           89.4%
automobile                           86.7%           95.8%
living expenses                      35.3%           49.2%
vacation                             13.8%            8.4%
luxury goods                          9.0%            7.6%
----------------------------------------------------------
\a In all cases, the difference in proportions between the no-loan
and the loan groups is statistically significant at the 1-percent
level. 

Source:  GAO analysis of the 1992 Survey of Consumer Finances
database. 

Attitudes toward borrowing money also differ between plan borrowers
and nonborrowers.\16 A larger proportion of plan borrowers think it
is all right to borrow to finance an automobile, but a slightly
smaller proportion think it is all right to borrow to finance
education expenses.  Almost half of the plan borrowers say it is all
right to borrow money to cover living expenses compared to about a
third of nonborrowers.  Less than 10 percent of each group think it
is all right to borrow to finance luxury goods, such as jewelry, and
less than 10 percent of plan borrowers think it is all right to
borrow to cover the expenses of a vacation.  This suggests that
relatively few participants--whether borrowers or nonborrowers--would
elect to borrow against their pension accounts to finance the
purchase of nonnecessities. 


--------------------
\14 Alicia H.  Munnell, Geoffrey M.  B.  Tootell, Lynn E.  Browne,
and James McEneany, "Mortgage Lending in Boston:  Interpreting HMDA
Data," American Economic Review, Vol.  86, No.  1 (1996), pp.  25-53,
and Geoffrey M.  B.  Tootell, "Redlining in Boston:  Do Mortgage
Lenders Discriminate Against Neighborhoods?" Quarterly Journal of
Economics, Vol.  111, No.  4 (1996), pp.  1049-79. 

\15 The Survey of Consumer Finances does not ask how pension-plan
loans are used. 

\16 The question on the Survey of Consumer Finances that is asked by
interviewers reads, "People have many different reasons for borrowing
money which they pay back over a period of time.  For each of the
reasons I read, please tell me whether you feel it is all right for
someone like yourself to borrow money..."


   BORROWING FROM PENSION ACCOUNT
   MAY REDUCE RETIREMENT INCOME
------------------------------------------------------------ Letter :5

Pension-plan participants who borrow from their pension accounts risk
having substantially lower pension balances at retirement.  Under
reasonable assumptions about pension-plan savings and borrowing
behavior, a borrower could have 2- to 28-percent less pension income
at retirement (see app.  II).  Many 401(k) participants have a
substantial amount of their pension balances invested in the stock
market and earn a relatively high rate of return.  Pension-plan
loans, however, generally have a favorable interest rate, which may
be much lower than the return on the pension-account investments. 
Consequently, a borrower may earn less on the loan balance because he
or she is making interest payments to the account at the relatively
low interest rate rather than earning higher returns from
investments, such as equities. 

How much pension income is lost depends on the size of the loan, the
interest rate of the loan, the rate of return of pension account
investments, and whether or not the borrower continues to make
pension contributions while repaying the loan.  For example, if a
borrower decides to forgo making pension-plan contributions during
loan repayment, he or she could have over 20-percent less retirement
income.  Continuing pension-plan contributions while repaying the
loan, on the other hand, could lead to a relatively small retirement
income loss of less than 7 percent. 


   OBSERVATIONS
------------------------------------------------------------ Letter :6

People save for many reasons, including retirement, emergencies, home
purchase, and a college education.  Saving for retirement receives
favorable tax treatment, but in the past, it was at the cost of being
virtually inaccessible until late in life.  Since retirement savings
could not be used for other purposes, people were reluctant to save
in retirement accounts.  Allowing participants to borrow against
their 401(k) pension accounts for reasons unrelated to retirement can
increase both participation in these plans and participant
contributions.  However, pension-plan borrowing is a two-edged sword: 
Individuals who were prompted to participate because of the borrowing
provision increase their retirement savings, but individuals who opt
to borrow lose some of the tax advantages to retirement savings and
risk having less income at retirement. 

Our findings have implications for other sources of retirement
income.  Since participation in IRAs is voluntary,\17 our results
suggest that early access to IRA funds may increase both
participation in and contributions to these accounts but at the risk
of lower retirement income.  On the other hand, individual Social
Security accounts--if created--would require participation, and
contribution levels would be set by law.  Consequently, individual
Social Security accounts would not benefit from the positive aspects
of borrowing provisions, but the borrowing provisions would increase
the risk of reduced retirement income. 


--------------------
\17 While participation in IRAs and 401(k) plans is voluntary,
different rules apply to these two retirement savings accounts. 
Maximum contribution limits for IRAs are much lower than for 401(k)
plans, spouses are allowed to make contributions to IRAs but not to
401(k) plans, and IRA contributions for some people are made with
before-tax dollars. 


   COMMENTS
------------------------------------------------------------ Letter :7

We asked pension plan experts to comment on a draft of this report. 
They generally agreed with the study approach and results.  They made
a few technical suggestions, which we incorporated where appropriate. 


---------------------------------------------------------- Letter :7.1

We are sending copies of this report to the Secretary of Labor,
relevant congressional committees, and other interested parties.  We
will make copies available to others on request. 

This report was prepared under my direction.  Please contact Francis
P.  Mulvey, Assistant Director, at (202) 512-3592 or Thomas L. 
Hungerford, Senior Economist, at (202) 512-7028 if you or your staff
have any questions concerning this report. 

Jane L.  Ross
Director, Income Security Issues


SCOPE AND METHODOLOGY
=========================================================== Appendix I

To determine how pension-plan borrowing affects workers'
participation in and contributions to a pension plan and retirement
income, we addressed the following questions: 

  -- Does the ability to borrow from defined contribution pension
     accounts increase participation in and contributions to 401(k)
     pension plans? 

  -- What are the demographic and economic characteristics of workers
     who borrow from their pension accounts? 

  -- What are the potential consequences for participants who borrow
     from their retirement accounts? 

To conduct our work, we analyzed two data sources.  The first, the
1992 Survey of Consumer Finances prepared by the Federal Reserve,
provided a nationally representative individual-level sample.  The
second, the 1992 research database of Internal Revenue Service (IRS)
Form 5500 reports, which are maintained by the Pension and Welfare
Benefits Administration of the Department of Labor, provided a
nationally representative plan-level sample.  We also reviewed the
relevant technical literature and talked to pension experts. 


   SURVEY OF CONSUMER FINANCES,
   1992
--------------------------------------------------------- Appendix I:1

The Survey of Consumer Finances randomly sampled 3,906 households
regarding current and past employment by family members, assets and
debts, and demographic information.  Included in the current
employment portion of the survey were detailed questions about
pension participation.  From the survey, we created a database
containing information on respondents and their spouses who were
working and between the ages of 18 and 64 at the time of the survey. 
We did not independently verify the accuracy of the Survey of
Consumer Finances database because it is commonly used by
researchers.  We used the Survey of Consumer Finances to determine
the effects of pension-plan borrowing on participation in and
contributions to 401(k) pension plans and to describe the demographic
and economic characteristics of workers who borrow from their pension
accounts. 

For the analysis of the impact of borrowing on contributions to
pension accounts, the subsample of the survey contained information
on 477 workers who participate in a 401(k) pension plan.  Since the
dependent variable is a continuous variable, which can be no less
than zero, the multivariate regression estimation technique used is a
tobit model.  A tobit model takes into account the fact that the
participation rate can be no less than 0 percent, and the results
from this model will not predict a participation rate of less than 0
percent.\18 Let C\* be an individual's desired contribution rate,
which is affected by the individual's characteristics.  If the
desired contribution rate is greater than zero, then the individual
contributes to his or her pension account.  If it is less than or
equal to zero, then the individual does not contribute to his or her
account.  Formally, the model is written as

C\* = 'X+,

where the X vector contains the variables, the  vector
contains the parameters to be estimated, and the last term is the
random error that captures the unobserved factors affecting the
desired contribution rate.  The dependent variable--that is, the
observed contribution rate--is

C = C\* if C\* >0
C = 0 if C\* ï¿½0. 

To describe the demographic and economic characteristics of workers
who borrow from their pension accounts, the subsample we used for our
analysis contained information on 769 workers with
defined-contribution pension plans that allowing borrowing.  We were
interested in determining how participant characteristics affect the
likelihood or probability that an individual has an outstanding loan
against his or her pension account.  The dependent variable for this
analysis is a variable that is equal to one if the individual has an
outstanding pension-account loan and equal to zero if he or she does
not have an outstanding loan.  The multivariate estimation technique
used for the analysis is a logit model, which will prevent
predictions from being outside the probability range of 0 to 1.  In
the logit model, the probability that an individual will have an
outstanding pension plan loan is a function of the individual's
characteristics:

P = f('X),

where P is the probability, the X vector contains the variables or
characteristics used in the estimation, the  vector contains
the parameters to be estimated, and f is the cumulative logistic
probability function.  The parameter vector is estimated using
maximum likelihood techniques.\19


--------------------
\18 William H.  Greene, Econometric Analysis, 3rd ed.  (Upper Saddle
River, NJ:  Prentice Hall, 1997), ch.  20. 

\19 The logit model can be stated in more formal terms.  Let L\* be
an individual's unobserved propensity to borrow from his or her
pension plan.  An individual will borrow if L\* is greater than zero
and will not borrow if L\* is less than or equal to zero.  The model
can be written as

L\* = 'X + 

where the X vector contains the variables used in the estimation, the
 vector contains the parameters to be estimated, and the
last term is the random error that captures unobserved factors
influencing the borrowing propensity.  The dependent variable is
defined by

L = 1 if L\* >0
L = 0 if L\* ï¿½0.

See Greene, Econometric Analysis, ch.  19, for the derivation of the
likelihood function. 


      CONSTRUCTION OF VARIABLES
------------------------------------------------------- Appendix I:1.1

The primary variables of interest are whether or not a worker can
withdraw funds from his or her pension account, the proportion of
salary contributed to the defined-contribution pension plan account,
and whether or not the worker has an outstanding pension-plan loan. 
The Survey of Consumer Finances asks respondents who have
defined-contribution pension plans,\20 "Can you borrow against that
account?" and "If you needed money in an emergency, could you
withdraw some of the funds in that account?" If the answer to either
of these questions was "yes," we considered that plan as allowing
participants to withdraw funds from their account before retirement. 
Respondents to the survey also were asked how much they contribute to
their pension account.  The contribution rate is the ratio of the
respondent's contribution to his or her salary.  Other variables used
in the analysis include sex, race, income, net worth, education,
recent loan experiences, and whether or not the individual is covered
by another pension plan (see table I.1). 



                         Table I.1
          
             Variables Used in the Multivariate
           Regression Analysis of the 1992 Survey
                of Consumer Finances Sample

Variable                      Definition
----------------------------  ----------------------------
Female                        Indicates whether or not the
                              worker is a female

Age 35 to 44 years            Indicates whether or not the
                              worker is between the ages
                              of 35 and 44 years

Age 45 to 54 years            Indicates whether or not the
                              worker is between the ages
                              of 45 and 54 years

Age 55 to 64 years            Indicates whether or not the
                              worker is between the ages
                              of 55 and 64 years

Married                       Indicates whether or not the
                              worker is married

Dropped out of school before  Indicates whether or not the
the 12th grade                worker has less than a high-
                              school education

1 to 4 years of college; no   Indicates whether or not the
degree                        worker has some college
                              education

4 or more years of college;   Indicates whether or not the
college degree                worker is a college graduate

Black                         Indicates whether or not the
                              worker is black

Hispanic                      Indicates whether or not the
                              worker is Hispanic

Covered by another pension    Indicates whether or not the
plan                          worker is covered by another
                              pension plan

Natural logarithm of          The natural logarithm of one
employer match rate           plus the ratio of the
                              employer contribution to the
                              worker's pension account to
                              the worker's salary

Can withdraw funds from       Indicates whether or not the
pension account               worker is able to borrow
                              against his or her pension
                              account or withdraw funds in
                              an emergency

Family income $25,000 to      Indicates whether or not the
$34,999 per year              worker's family income is
                              between $25,000 and $34,999
                              per year

Family income $35,000 to      Indicates whether or not the
$44,999 per year              worker's family income is
                              between $35,000 and $44,999
                              per year

Family income $45,000 to      Indicates whether or not the
$59,999 per year              worker's family income is
                              between $45,000 and $59,999
                              per year

Family income $60,000 to      Indicates whether or not the
$74,999 per year              worker's family income is
                              between $60,000 and $74,999
                              per year

Family income $75,000 or      Indicates whether or not the
more per year                 worker's family is $75,000
                              or more per year

Natural logarithm of number   The natural logarithm of the
of years covered by this      number of years the worker
defined-contribution plan     has been covered by his or
                              her pension plan

Family net worth $50,001 to   Indicates whether or not the
$100,000                      worker's family net worth is
                              between $50,001 and $100,000

Family net worth $100,001 to  Indicates whether or not the
$250,000                      worker's family net worth is
                              between $100,001 and
                              $250,000

Family net worth $250,001 to  Indicates whether or not the
$1,000,000                    worker's family net worth is
                              between $250,001 and
                              $1,000,000

Family net worth over         Indicates whether or not the
$1,000,000                    worker's family net worth is
                              over $1,000,000

Outstanding loan              Indicates whether or not the
                              worker has an outstanding
                              loan against his or her
                              pension account

Contribution rate             The ratio of the worker's
                              contribution to his or her
                              pension account to the
                              worker's salary
----------------------------------------------------------

--------------------
\20 About 60 percent of the respondents who can borrow from their
pension plans participate in a 401(k) pension plan, and another 30
percent participate in either a thrift savings plan or a
profit-sharing plan. 


   IRS FORM 5500 RESEARCH
   DATABASE, 1992
--------------------------------------------------------- Appendix I:2

We used IRS' Form 5500 research database for 1992 to determine the
effects of pension-plan borrowing on participation in and
contributions to 401(k) pension plans.  Under the Employee Retirement
Income Security Act of 1974, private employers must annually file a
separate Form 5500 with the IRS for each employee's pension plan. 
Each report contains financial, participant, and actuarial
information.  We did not independently verify the accuracy of the
Form 5500 research database because this database is commonly used by
researchers.  The 1992 Form 5500 research database was obtained from
the Pension and Welfare Benefits Administration of the Department of
Labor. 

The plans selected for analysis are plans that had 100 or more
participants and offered defined-contribution plans with 401(k)
features as the primary plan.  All plans that were terminated during
the year or where there was a resolution to terminate the plan are
not included in the sample.  Furthermore, we selected only plans that
had one or more active participants, that is, those with pension
accounts.\21 The final sample used in the analysis contains 7,245
plans with an average of 337 active participants.  The analysis
consists of estimating two multivariate statistical models. 

The first model estimated examined the impacts of firm and plan
characteristics on participation in the plan.  The dependent variable
is the percent of employees eligible to participate who participate
in the plan.\22 The second model examines average employee
contributions to the plan.  Ordinary least squares regression
techniques were used to estimate both models.  Formally, the models
can be expressed as

Y = 'X+,

where Y is the dependent variable, which is either the participation
rate or the natural logarithm of average contribution amounts; the X
vector contains the independent variables; the  vector
contains the parameters to be estimated; and the last term is the
random error that captures the unobserved factors influencing the
dependent variable. 


--------------------
\21 Participants of a pension plan include all current employees who
are covered by the plan (active participants plus those eligible to
participate but do not) plus retired and former employees who are
covered by the plan. 

\22 Some pension plans mistakenly reported active participants on
line 7a(4) of Form 5500, which should contain the total number of
workers eligible to participate in the plan.  Consequently, we could
not determine which plans had 100-percent participation and which had
made mistakes filling in the Form 5500.  Therefore, all plans with a
participation rate calculated to be 100 percent were deleted from the
sample for the regression analysis of the participation rate (about
45 percent of the sample).  We did, however, estimate the
participation rate model with the full sample, and the results were
quantitatively and qualitatively the same. 


      CONSTRUCTION OF VARIABLES
------------------------------------------------------- Appendix I:2.1

The first dependent variable is the ratio of active participants to
the number of all employees eligible to participate in the plan.\23
The second dependent variable is the natural logarithm of the average
contribution rate.  The average employee contribution variable is the
ratio of total contributions to the plan to the number of active
participants.  The independent variables used in the analysis are
variables used by other researchers,\24

such as employer matching and firm size, plus a variable denoting if
the plan participants had any outstanding loans (see table I.2). 



                         Table I.2
          
             Variables Used in the Multivariate
          Regression Analysis of the 1992 IRS Form
                   5500 Research Database

Variable        Definition
--------------  ------------------------------------------
Participation   Ratio of active participants in the
rate            pension plan to all employees eligible to
                participate in the plan

Natural         Natural logarithm of the ratio of
logarithm of    contributions to active participants
average
employee
contributions

Firm size       Number of employees of the firm divided by
                1,000

Firm size       Firm size squared
squared

Match rate\a 1  Indicates whether or not the employer
                match rate\ is greater than zero and less
                than or equal to 0.1

Match rate 2    Indicates whether or not the employer
                match rate is greater than 0.1 and less
                than or equal to 0.2

Match rate 3    Indicates whether or not the employer
                match rate is greater than 0.2 and less
                than or equal to 0.3

Match rate 4    Indicates whether or not the employer
                match rate is greater than 0.3 and less
                than or equal to 0.4

Match rate 5    Indicates whether or not the employer
                match rate is greater than 0.4 and less
                than or equal to 0.5

Match rate 6    Indicates whether or not the employer
                match rate is greater than 0.5 and less
                than or equal to 0.6

Match rate 7    Indicates whether or not the employer
                match rate is greater than 0.6 and less
                than or equal to 0.7

Match rate 8    Indicates whether or not the employer
                match rate is greater than 0.7 and less
                than or equal to 0.8

Match rate 9    Indicates whether or not the employer
                match rate is greater than 0.8 and less
                than or equal to 0.9

Match rate 10   Indicates whether or not the employer
                match rate is greater than 0.9 and less
                than or equal to 1.0

Match rate 11   Indicates whether or not the employer
                match rate is greater than 1.0 and less
                than or equal to 1.5

Match rate 12   Indicates whether or not the employer
                match rate is greater than 1.5

Can borrow      Indicates whether or not there are
                outstanding loans from the plan

Sole pension    Indicates whether or not employer has
plan            other qualified pension plans
----------------------------------------------------------
\a The match rate is the ratio of employer contributions to
participant contributions. 


--------------------
\23 The Pension and Welfare Benefits Administration adds a variable
to the Form 5500 research database that indicates the number of
current employees participating in the plan and calls this variable
"active participants."

\24 See, for example, Leslie E.  Papke, "Participation in and
Contributions to 401(k) Pension Plans," Journal of Human Resources,
Vol.  30, No.  2 (1995), pp.  311-25. 


   SIMULATION MODEL
--------------------------------------------------------- Appendix I:3

To determine the potential consequences of borrowing from a 401(k)
account, we prepared a simulation model.  We created a 35-year annual
earnings series with a starting salary of $25,000.  Annual earnings
were allowed to grow with age\25 and with inflation (assumed to be 3
percent).  The contributions to the 401(k) account are 6.8 percent of
annual earnings.\26 We assumed that the 401(k) account balance earns
an annual rate of return of 11 percent.  The simulation involves a
$40,000 loan against the pension account made in the 15th year and
paid back over a 10-year period in equal installments.  Pension
account balances were determined for several different loan interest
rates.  We created simulations under two extreme scenarios:  (1) the
borrower continues to make contributions to the 401(k) account while
repaying the loan and (2) the borrower suspends making contributions
to the 401(k) account while repaying the loan. 


--------------------
\25 Thomas Hungerford and Gary Solon, "Sheepskin Effects in the
Return to Education," Review of Economics and Statistics, Vol.  69,
No.  1 (1987), pp.  175-77. 

\26 This is the average contribution rate for 401(k) participants
reported in 401(k) Pension Plans:  Many Take Advantage of Opportunity
to Ensure Adequate Retirement Income (GAO/HEHS-96-176, Aug.  2,
1996). 


SUPPLEMENTARY ANALYSIS
========================================================== Appendix II

This appendix contains supplementary tables of multivariate
statistical results from the two databases that we used to conduct
our work. 


   PARTICIPATION RATES IN 401(K)
   PLANS
-------------------------------------------------------- Appendix II:1

The coefficient estimates from the regression model of the
participation rate are shown in table II.1.  The coefficient
estimates indicate the effect of a change in an independent variable
on a plan's participation rate holding the values of all other
independent variables constant.  For example, the coefficient
estimate of 0.0591 for the borrowing variable indicates that plans
that allow participant borrowing have participation rates that are
about 6 percentage points higher than plans that do not allow
borrowing. 



                         Table II.1
          
            Regression Results for Participation
          Rates (Dependent Variable: Participation
                           Rate)

                                      Coefficient estimate
Variable                                  (standard error)
----------------------------  ----------------------------
Constant                                            0.5716
Firm size                                        -0.0020\a
                                                  (0.0006)
Firm size squared (x1000)                         0.0050\a
                                                  (0.0015)
Can borrow                                        0.0591\a
                                                  (0.0063)
Sole pension plan                                -0.0206\b
                                                  (0.0090)
Match rate 1                                      0.1035\a
                                                  (0.0141)
Match rate 2                                      0.1410\a
                                                  (0.0101)
Match rate 3                                      0.1759\a
                                                  (0.0108)
Match rate 4                                      0.1975\a
                                                  (0.0112)
Match rate 5                                      0.2226\a
                                                  (0.0124)
Match rate 6                                      0.2612\a
                                                  (0.0164)
Match rate 7                                      0.2600\a
                                                  (0.0194)
Match rate 8                                      0.2818\a
                                                  (0.0206)
Match rate 9                                      0.2943\a
                                                  (0.0252)
Match rate 10                                     0.3183\a
                                                  (0.0256)
Match rate 11                                     0.3341\a
                                                  (0.0177)
Match rate 12                                     0.3210\a
                                                  (0.0256)
Adjusted R\2                                        0.2102
Number of observations                               4,006
----------------------------------------------------------
\a Significant at the 1-percent level. 

\b Significant at the 5-percent level. 

Source:  GAO analysis of IRS 1992 Form 5500 research database. 


   CONTRIBUTIONS TO 401(K) PLANS
-------------------------------------------------------- Appendix II:2

The regression results of the effects of pension-plan characteristics
on average employee contribution levels are reported in table II.2. 
The coefficient estimates indicate the effect of a change in an
independent variable on average contribution levels holding the
values of all other independent variables constant.  For example, the
coefficient estimate of 0.3682 for the borrowing variable indicates
that borrowing provisions increase average employee contribution
levels by 36.8 percent. 



                         Table II.2
          
          Regression Results for Average Employee
          Contribution Levels (Dependent Variable:
           Natural Logarithm of Average Employee
                       Contribution)

                                      Coefficient estimate
Variable                                  (standard error)
----------------------------  ----------------------------
Constant                                            7.0208
Firm size                                        -0.0043\a
                                                  (0.0015)
Firm size squared (x1000)                         0.0119\a
                                                  (0.0044)
Can borrow                                        0.3682\a
                                                  (0.0167)
Sole pension plan                                -0.1300\a
                                                  (0.0231)
Match rate 1                                      0.1608\a
                                                  (0.0410)
Match rate 2                                      0.1042\a
                                                  (0.0302)
Match rate 3                                      0.1205\a
                                                  (0.0313)
Match rate 4                                      0.2427\a
                                                  (0.0317)
Match rate 5                                      0.1222\a
                                                  (0.0343)
Match rate 6                                      0.1779\a
                                                  (0.0412)
Match rate 7                                      0.1460\a
                                                  (0.0446)
Match rate 8                                      0.1838\a
                                                  (0.0464)
Match rate 9                                      0.1615\a
                                                  (0.0529)
Match rate 10                                     0.1211\b
                                                  (0.0569)
Match rate 11                                     0.0991\a
                                                  (0.0374)
Match rate 12                                       0.0551
                                                  (0.0493)
Adjusted R\2                                        0.0745
Number of observations                               7,245
----------------------------------------------------------
\a Significant at the 1-percent level. 

\b Significant at the 5-percent level. 

Source:  GAO analysis of IRS 1992 Form 5500 research database. 

The tobit-model results of individual pension-plan participants
reported in table II.3 examine the influence of participant
characteristics on contribution rates holding all other
characteristics constant.  When a variable changes, it will have two
effects on the overall contribution rate.  First, for individuals
already making a contribution, an increase in a variable with a
positive coefficient estimate will directly increase the contribution
rate.  Second, for individuals who are not making contributions to
their 401(k) accounts, an increase in this variable will increase the
likelihood that they contribute to their plan account.  The marginal
impacts of a variable change reported in table II.3 include both
these impacts on the expected value of the contribution rate.  For
example, the marginal impact of 3.0247 for the borrowing variable
indicates that, on average, contribution rates of participants in
plans with borrowing provisions are about 3 percentage points higher
than for participants in other plans.\27



                         Table II.3
          
          Tobit-Model Results for the Contribution
           Rate (Dependent Variable: Contribution
                           Rate)

                                 Coefficient
                                    estimate
                                   (standard      Marginal
Variable                              error)        impact
------------------------------  ------------  ------------
Constant                             -1.1724
Female                                0.2887        0.2327
                                    (0.3244)
Age 35 to 44 years                 -0.8662\b       -0.6981
                                    (0.4105)
Age 45 to 54 years                  1.4346\a        1.1562
                                    (0.4576)
Age 55 to 64 years                    0.4641        0.3741
                                    (0.5946)
Married                            -1.7866\a       -1.4399
                                    (0.4298)
Dropped out of school before          0.8815        0.7105
 the 12th grade                     (1.1460)
1 to 4 years of college; no          -0.6867       -0.5534
 degree                             (0.4776)
4 or more years of college;        -1.5289\a       -1.2322
 college degree                     (0.4462)
Black                                -0.5284       -0.4258
                                    (0.7617)
Hispanic                              1.7746        1.4302
                                    (1.0690)
Covered by another pension          1.3272\a        1.0697
 plan                               (0.3223)
Natural logarithm of employer       0.7589\a        0.6116
 match rate                         (0.1620)
Can withdraw funds from             3.7530\a        3.0247
 pension account                    (0.4430)
Family income $25,000 to              1.3454        1.0843
 $34,999 per year                   (0.7846)
Family income $35,000 to            4.8156\a        3.8811
 $44,999 per year                   (0.8101)
Family income $45,000 to            2.9547\a        2.3813
 $59,999 per year                   (0.7905)
Family income $60,000 to            4.1386\a        3.3355
 $74,999 per year                   (0.8635)
Family income $75,000 or more       4.5341\a        3.6542
 per year                           (0.8446)
Natural logarithm of number of       -0.3309       -0.2667
 years covered by this              (0.1900)
 defined-contribution plan
Family net worth $50,001 to         1.6156\a        1.3021
 $100,000                           (0.4819)
Family net worth $100,001 to        2.2099\a        1.7811
 $250,000                           (0.5198)
Family net worth $250,001 to        3.6329\a        2.9279
 $1,000,000                         (0.6248)
Family net worth over                 1.1595        0.9345
 $1,000,000                         (0.6746)
                           7.0546\a
                                    (0.1142)
Logarithm of likelihood              -7016.1
 function
----------------------------------------------------------
\a Significant at the 1-percent level. 

\b Significant at the 5-percent level. 

Source:  GAO analysis of the 1992 Survey of Consumer Finances
database. 


--------------------
\27 Several specifications were tried that deleted the income
variables and the wealth variables.  The parameter estimates of the
variables of interest were quantitatively the same in each case.  In
addition, the match rate was entered as a quadratic in one
specification with very little change in the coefficient estimates of
the other variables.  Furthermore, the overall impact of matching is
the same in both specifications. 


   CHARACTERISTICS OF PENSION-PLAN
   BORROWERS
-------------------------------------------------------- Appendix II:3

A logit model was estimated to determine the magnitude of the effects
of participant characteristics on the likelihood of having an
outstanding pension-plan loan (see table II.4).  The coefficient
estimates do not indicate the magnitude of the impacts on the
likelihood of having an outstanding loan due to changes in the
variables.  Consequently, the marginal impacts of changes in the
variables on the likelihood were calculated and are reported in the
third column of table II.4.  For example, the marginal impact of
0.0578 for black participants indicates that the likelihood of blacks
having an outstanding loan is 5.8 percentage points higher than for
whites.  Given that about 7.6 percent of plan participants have
outstanding loans, then blacks are about 5.8/7.6 times 100--or 76
percent--more likely to have an outstanding pension-plan loan than
whites. 



                         Table II.4
          
           Logit-Model Results for Whether or Not
          Participants Have Outstanding Loans From
                   Their Pension Account

                                   Parameter
                                    estimate
                                   (standard      Marginal
Variable                              error)        impact
------------------------------  ------------  ------------
Constant                             -3.7994
Female                               -0.2854       -0.0145
                                    (0.1550)
Age 35 to 44 years                    0.3341        0.0170
                                    (0.1920)
Age 45 to 54 years                    0.2815        0.0143
                                    (0.2184)
Age 55 to 64 years                   -0.1426       -0.0073
                                    (0.3006)
Married                               0.3298        0.0168
                                    (0.1976)
Dropped out of school before          0.0874        0.0045
 the 12th grade                     (0.3921)
1 to 4 years of college; no           0.1365        0.0070
 degree                             (0.2291)
4 or more years of college;         0.7249\a        0.0369
 college degree                     (0.2127)
Black                               1.1337\a        0.0578
                                    (0.2068)
Hispanic                            1.7382\a        0.0886
                                    (0.2934)
Covered by another pension          0.7651\a        0.0390
 plan                               (0.1515)
Recently turned down for loan       0.5829\a        0.0297
                                    (0.1855)
Family income $25,000 to              0.0192        0.0010
 $34,999 per year                   (0.2937)
Family income $35,000 to              0.0094        0.0005
 $44,999 per year                   (0.2975)
Family income $45,000 to             -0.3429       -0.0175
 $59,999 per year                   (0.3128)
Family income $60,000 to             -0.4968       -0.0253
 $74,999 per year                   (0.3411)
Family income $75,000 or more      -1.0754\a       -0.0548
 per year                           (0.3522)
Natural logarithm of number of      0.2608\a        0.0133
 years covered by this              (0.0900)
 defined-contribution plan
Family net worth $50,001 to        -0.9459\a       -0.0482
 $100,000                           (0.2722)
Family net worth $100,001 to         -0.0553       -0.0028
 $250,000                           (0.2251)
Family net worth $250,001 to         -0.3286       -0.0167
 $1,000,000                         (0.2913)
Family net worth over                -0.1263       -0.0064
 $1,000,000                         (0.3226)
Logarithm of likelihood              -776.81
 function
----------------------------------------------------------
\a Significant at the 1-percent level. 

Source:  GAO analysis of the 1992 Survey of Consumer Finances
database. 


   CONSEQUENCES OF BORROWING FROM
   RETIREMENT ACCOUNTS
-------------------------------------------------------- Appendix II:4

Our simulation results are presented in table II.5 and show the
pension account balance after 35 years for each scenario.  The
results show that as long as the interest rate of the loan is less
than the rate of return of the pension account balance (assumed to be
11 percent), borrowers will have a lower account balance at
retirement.  The actual reduction depends on the gap between the
account rate of return and the loan interest rate, and whether or not
pension contributions continue during the loan repayment period. 
Furthermore, these results hold only if the loan is repaid. 



                         Table II.5
          
                     Simulation Results

                                     Account    Percent of
                                  balance in       no-loan
                                     year 35       balance
------------------------------  ------------  ------------
No loan                             $952,977

Maintain contributions to pension account during loan re
----------------------------------------------------------
6.3 percent loan                    $892,209  93.6 percent
7.0 percent loan                    $900,892  94.5 percent
8.0 percent loan                    $913,526  95.9 percent
9.5 percent loan                    $932,968  97.9 percent

Suspend contributions to pension account during loan re
----------------------------------------------------------
6.3 percent loan                    $687,863  72.2 percent
7.0 percent loan                    $696,546  73.1 percent
8.0 percent loan                    $709,180  74.4 percent
9.5 percent loan                    $728,622  76.5 percent
----------------------------------------------------------

*** End of document. ***