Crop Revenue Insurance: Problems With New Plans Need to Be Addressed
(Chapter Report, 04/29/98, GAO/RCED-98-111).

Pursuant to a congressional request, GAO reviewed various issues
pertaining to the Department of Agriculture's new crop revenue insurance
plans.

GAO noted that: (1) the three government-subsidized revenue insurance
plans differ in the revenue guarantees they provide to farmers and in
their relative cost to the government; (2) two of the plans, Revenue
Assurance and Income Protection, set the revenue level that is to be
protected at the time that crops are being planted, while the third,
Crop Revenue Coverage, determines the protected revenue at either
planting or at harvest, depending on when crop prices are higher; (3) in
terms of potential government costs, Crop Revenue Coverage is likely to
cost the government significantly more than the other two plans because
of its higher reimbursement for administrative expenses and because of
potentially higher total underwriting losses; (4) furthermore, the
plan's promise base the revenue guarantee on the price at planting or
the price at harvest, whichever is higher, exposes the government to
higher claims payments in the years when widespread crop losses are
coupled with rapidly increasing prices; (5) in their first two years of
availability to farmers, the crop revenue insurance plans, especially
Crop Revenue Coverage, achieved a significant share of the crop
insurance market, accounting for about one-third of the total crop
insurance sales in the areas where they were offered; (6) in terms of
the claims payments for 1997, all types of crop insurance experienced
much lower than average levels of claims as a result of favorable
growing conditions in most of the country; (7) morever, primarily
because revenue insurance plans were often marketed in lower-risk areas,
they experienced lower levels of claims payments than did multiple-peril
crop insurance; (8) GAO identified shortcomings in each revenue
insurance plan's approach to establishing premium rates; (9) Crop
Revenue Coverage is especially problematic because its rate structure
does not take into account the interrelationship between crop prices and
yields--an essential component of actuarially sound rate settings; (10)
while good weather and stable crop prices generated very favorable
claims experience over the first 2 years of the plans' availability, GAO
has doubts about whether the rates established for each plan are
actuarially sound over the long term and are appropriate to the risk
each farmer presents; and (11) furthermore, while the plans were
initially approved on a limited basis only, the Federal Crop Insurance
Corporation, acting within its authority, approved the substantial
expansion of one of these plans--Crop Revenue Coverage--before initial
results were available.

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

 REPORTNUM:  RCED-98-111
     TITLE:  Crop Revenue Insurance: Problems With New Plans Need to Be 
             Addressed
      DATE:  04/29/98
   SUBJECT:  Agricultural programs
             Insurance premiums
             Insurance claims
             Grain and grain products
             Farm income stabilization programs
             Insurance cost control
             Agricultural production
IDENTIFIER:  USDA Revenue Assurance Plan
             USDA Income Protection Plan
             USDA Crop Revenue Coverage
             
******************************************************************
** 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 Ranking Minority Member, Committee on Agriculture,
House of Representatives

April 1998

CROP REVENUE INSURANCE - PROBLEMS
WITH NEW PLANS NEED TO BE
ADDRESSED

GAO/RCED-98-111

Problems With New Crop Revenue Insurance Plans

(150429)


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

  APH - actual production history
  CRC - Crop Revenue Coverage
  FCIC - Federal Crop Insurance Corporation
  IP - Income Protection
  MPCI - multiple-peril crop insurance
  NASS - National Agricultural Statistics Service
  RA - Revenue Assurance
  USDA - U.S.  Department of Agriculture

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


B-279365

April 29, 1998

The Honorable Charles W.  Stenholm
Ranking Minority Member
Committee on Agriculture
House of Representatives

Dear Mr.  Stenholm: 

This report responds to your request that we study various issues
pertaining to the U.S.  Department of Agriculture's new crop revenue
insurance plans.  The report contains a recommendation to the
Secretary of Agriculture to correct deficiencies in the methods used
to establish premium rates for these plans. 

We are sending copies of this report to appropriate House and Senate
committees; interested Members of Congress; the Secretary of
Agriculture; the Administrator of the Risk Management Agency; the
Director, Office of Management and Budget; and other interested
parties.  We will also make copies available to others upon request. 

If you or your staff have any questions, I can be reached on (202)
512-5138.  Major contributors to this report are listed in appendix
VII. 

Sincerely yours,

Robert A.  Robinson
Director, Food and
 Agriculture Issues


EXECUTIVE SUMMARY
============================================================ Chapter 0


   PURPOSE
---------------------------------------------------------- Chapter 0:1

Farming is an inherently risky enterprise.  In conducting their
operations, farmers are exposed to both production and price risks. 
Over the years, the federal government has played an active role in
helping to mitigate the effects of these risks on farm income.  A new
tool, crop revenue insurance, has been introduced to help farmers
manage their risks.  Three federally subsidized crop revenue
insurance plans--Crop Revenue Coverage, Revenue Assurance, and Income
Protection--are now being sold to farmers in various parts of the
country.  The plans protect farmers from the effect of declines in
either crop prices or yields by guaranteeing an agreed-upon level of
revenue. 

In light of the rapid expansion of the new crop revenue insurance
plans and the government's significant financial participation in
them, the Ranking Minority Member of the House Committee on
Agriculture asked GAO to (1) identify the differences between the
three new revenue insurance plans, (2) report on the plans' sales and
claims experience, and (3) analyze the methodologies used to set the
plans' premium rates. 


   BACKGROUND
---------------------------------------------------------- Chapter 0:2

Farm production levels can vary significantly from year to year,
primarily because farmers operate at the mercy of nature and
frequently are subjected to weather-related and other natural
disasters.  Farmers can also experience wide swings in the prices
they receive for the commodities they grow, depending on domestic and
international production levels and demand. 

Prior to 1996, the U.S.  Department of Agriculture (USDA)
administered programs known as deficiency payment programs for
several major crops--wheat, feed grains, cotton, and rice.  These
programs protected farmers' income against declines in prices through
a complicated array of pricing mechanisms.  The government's role in
agricultural production changed with the passage of the Federal
Agriculture Improvement and Reform Act in 1996.  Under the 1996 act,
farmers are encouraged to produce in response to market forces,
rather than to the expectation of federal payments.  As part of this
new direction in policy, the act replaced the income support programs
with "production flexibility contracts"--agreements between the
federal government and participating farmers that provide for fixed
but declining 7-year annual payments that are not tied to market
prices.  USDA estimates that the production flexibility contracts
will cost a total of $35.6 billion over the 7-year period. 

Since the 1930s, federally subsidized multiple-peril crop insurance
has been farmers' principal means of managing the risk associated
with crop losses.  The Federal Crop Insurance Corporation, a wholly
owned government corporation under the management of USDA's Risk
Management Agency, administers the federal crop insurance program. 
Between 1980 and 1998, USDA expanded the availability of crop
insurance from 30 to 67 crops and from about one-half of the nation's
counties to virtually all areas of the country.  The federal
government's crop insurance costs totaled about $8.9 billion from
1990 through 1997. 

To manage the risk to their incomes resulting from price
fluctuations, many farmers use crop insurance in combination with
nongovernmental strategies such as forward contracting or hedging on
national commodity exchanges.  As an alternative to using crop
insurance and forward contracting or hedging separately, new
government-supported revenue insurance plans allow farmers to buy a
single policy that protects against both production and price risks. 
Crop Revenue Coverage and Revenue Assurance were developed by private
insurance companies that requested and received federal support for
the plans, whereas the Federal Crop Insurance Corporation developed
Income Protection.  The new revenue insurance plans are also
administered by the Federal Crop Insurance Corporation as part of its
overall responsibility for the crop insurance program. 

As it does for traditional multiple-peril crop insurance, USDA
supports the revenue insurance plans by (1) subsidizing the premiums
farmers pay, (2) paying private insurance companies to sell the
insurance and process claims, and (3) agreeing to pay a large portion
of any underwriting losses that occur if claims exceed premiums. 


   RESULTS IN BRIEF
---------------------------------------------------------- Chapter 0:3

The three government-subsidized revenue insurance plans differ in the
revenue guarantees they provide to farmers and in their relative cost
to the government.  Two of the plans, Revenue Assurance and Income
Protection, set the revenue level that is to be protected at the time
that crops are being planted, while the third, Crop Revenue Coverage,
determines the protected revenue at either planting or at harvest,
depending on when crop prices are higher.  In terms of potential
government costs, Crop Revenue Coverage is likely to cost the
government significantly more than the other two plans because of its
higher reimbursements for administrative expenses and because of
potentially higher total underwriting losses (the excess of claims
payments over total premiums).  Furthermore, the plan's promise to
base the revenue guarantee on the price at planting or the price at
harvest, whichever is higher, exposes the government to higher claims
payments in the years when widespread crop losses are coupled with
rapidly increasing prices. 

In their first 2 years of availability to farmers, the crop revenue
insurance plans, especially Crop Revenue Coverage, achieved a
significant share of the crop insurance market, accounting for about
one-third of the total crop insurance sales in the areas where they
were offered.  In terms of the claims payments for 1997, all types of
crop insurance experienced much lower than average levels of claims
as a result of favorable growing conditions in most of the country. 
Moreover, primarily because revenue insurance plans were often
marketed in lower-risk areas, they experienced lower levels of claims
payments than did multiple-peril crop insurance. 

GAO identified shortcomings in each revenue insurance plan's approach
to establishing premium rates.  Crop Revenue Coverage is especially
problematic because its rate structure does not take into account the
interrelationship between crop prices and yields--an essential
component of actuarially sound rate setting.  While good weather and
stable crop prices generated very favorable claims experience over
the first 2 years of the plans' availability, GAO has doubts about
whether the rates established for each plan are actuarially sound
over the long term and are appropriate to the risk each farmer
presents.  Furthermore, while the plans were initially approved on a
limited basis only, the Federal Crop Insurance Corporation, acting
within its authority, approved the substantial expansion of one of
these plans--Crop Revenue Coverage--before initial results were
available. 


   PRINCIPAL FINDINGS
---------------------------------------------------------- Chapter 0:4


      THREE PLANS' APPROACHES TO
      INSURING REVENUE RESULT IN
      DIFFERENT LEVELS OF
      PROTECTION AND GOVERNMENT
      COSTS
-------------------------------------------------------- Chapter 0:4.1

For both Revenue Assurance and Income Protection, the farmer's
revenue guarantee is established when crops are planted by
multiplying the farmer's historical average production per acre by
the prevailing futures market price.  If the farmer's revenue at
harvest is below the guaranteed level, the farmer receives an
insurance payment.  A farmer whose revenue is at or above the
guaranteed level does not receive a payment.  The farmer's total
revenue is the determining factor; no payment would be made if a
price decline is offset by an increase in production or a loss in
production is offset by an increase in prices.  In contrast, the
calculation of the amount of revenue guaranteed under Crop Revenue
Coverage is more complicated.  At planting, Crop Revenue Coverage
guarantees a minimum revenue that is determined by multiplying the
prevailing futures market price at planting by the farmer's
historical production per acre.  At harvest, the revenue guarantee
calculation is revisited, and the final guarantee is determined by
multiplying the farmer's historical production by the higher of the
price at planting or the price at harvest.  If the price has
increased in the period between planting and harvest, the farmer
receives a payment for any lost production at the higher harvest
price.  This upward price protection feature assures the farmer that
any lost production will be replaced at the prevailing market price,
thus facilitating forward contracting by the farmer.  If, however,
the harvest price is lower, and production was lost, the original
guarantee is in force. 

The revenue insurance plans also differ in several operational
features.  Although futures prices form the basis for all the revenue
guarantees, the plans adjust these prices differently to account for
variations between local and national prices.  In addition, the
methods used for establishing which fields will be covered for
insurance purposes vary from plan to plan.  Finally, only Crop
Revenue Coverage is available in most areas of the country. 

Crop Revenue Coverage is likely to be more costly to the government
than multiple-peril crop insurance and the other revenue insurance
plans because of its higher reimbursements for administrative
expenses and higher potential underwriting losses.  First, with
respect to administrative expenses, the reimbursements for Crop
Revenue Coverage are likely to be higher than those for
multiple-peril crop insurance.  This is because the premiums per acre
are much higher and the administrative expense reimbursements are
based on a percentage of these premiums.  While the reimbursement
rate paid by the Federal Crop Insurance Corporation is lower for Crop
Revenue Coverage than the rate paid for multiple-peril crop
insurance, it is not low enough to offset Crop Revenue Coverage's
much higher premium levels.  Second, with respect to underwriting
losses, because both plans are expected to generate such losses at a
fixed percentage of premiums paid over time, the higher volume of
premiums for Crop Revenue Coverage is likely to result in higher
losses than for multiple-peril crop insurance.  Furthermore, the
claims experience with Crop Revenue Coverage is likely to have a more
exaggerated, or magnified, impact during any given year because of
the plan's unique upward price protection feature.  For example, if
Crop Revenue Coverage had been available for winter wheat in 1996,
when widespread wheat losses were coupled with significant increases
in commodity prices, the Federal Crop Insurance Corporation would
have had to pay an additional 43 percent, or $172 million more, in
claims than it actually paid under traditional multiple-peril crop
insurance.  However, part of the potential underwriting loss would
have been offset by the higher premiums paid for this plan. 
Nevertheless, the government's exposure to loss would have been
substantially increased.  Alternatively, because the price increases
that occurred in 1996 more than offset the average production loss,
the provisions of Income Protection or Revenue Assurance would have
resulted in claims payments that were about $200 million less than
the claims actually paid under multiple-peril crop insurance. 


      NEW INSURANCE PLANS
      ACHIEVING SIGNIFICANT SHARE
      OF CROP INSURANCE MARKET
-------------------------------------------------------- Chapter 0:4.2

Crop revenue insurance plans as a group had strong sales, obtaining a
significant portion of the total crop insurance sales in 1997, the
first year that all three plans were available.  Crop Revenue
Coverage, the most widely available of the three revenue insurance
plans, took away a considerable amount of business from
multiple-peril crop insurance--obtaining a 32-percent share of the
market--in the areas where it was sold.  In contrast, neither Revenue
Assurance nor Income Protection attracted many purchasers--obtaining
6-percent and 3-percent shares, respectively--in the areas where they
were sold. 

All types of crop insurance experienced relatively low levels of
claims in 1997.  The crop insurance industry discusses the extent of
losses in terms of the claims paid per premium dollar collected.  For
1981 through 1996, traditional multiple-peril crop insurance paid an
average of $1.26 in claims per $1 of premium.  However, in 1997,
because of relatively favorable growing conditions in the country
overall, the crop insurance program experienced a much lower level of
claims--$0.49 per $1 of premium.  Moreover, the revenue insurance
plans experienced even lower levels of claims payments than did
multiple-peril crop insurance--ranging from $0.06 to $0.36 per $1 of
premium.  According to the Risk Management Agency, the lower claims
experience could have occurred for several reasons, such as a
concentration of sales in lower-risk areas, stable crop prices, or a
combination of these and other factors. 

Crop Revenue Coverage policies written in 1997 insured higher acreage
levels and were associated with operations having lower production
variability over time.  Crop insurance research has shown that
policies with these characteristics tend, on average, to have a lower
incidence of claims payments.  This lower level of risk may have
occurred because the initial marketing efforts were targeted to
operators of larger farms in the most consistently productive areas. 
As such, the differences in risk may diminish over time as marketing
expands into the general farming community.  With respect to Income
Protection and Revenue Assurance, GAO could not analyze their risk
characteristics because of their small sales volume. 


      APPROACHES USED TO ESTABLISH
      PREMIUM RATES MAY NOT
      ADEQUATELY PROTECT THE
      GOVERNMENT FROM FINANCIAL
      LOSSES
-------------------------------------------------------- Chapter 0:4.3

GAO identified shortcomings in the way premium rates are established
for each of the revenue insurance plans.  Appropriate methods for
setting rates for these plans are critical to ensuring the financial
soundness of the crop insurance program over time.  GAO found that
the Crop Revenue Coverage plan does not base its rate structure upon
the interrelationship between crop prices and farm-level yields--an
essential component of actuarially sound rate setting.  For example,
a decline in yields is often accompanied by an increase in prices,
which mitigates the impact of the decline in yields on a farmer's
revenue.  Because this plan does not recognize this
interrelationship, the premium adjustments may not be sufficient over
the long term to cover claims payments and may not be appropriate to
the risk each farmer presents.  GAO is not able to determine whether
premium rates for this plan are too high or too low.  In contrast,
the rate-setting approaches for Revenue Assurance and Income
Protection are based on a likely distribution of revenues that
reflects the interrelationship between crop prices and yields. 
However, the plans have several shortcomings that are not as serious
as the problem GAO identified for Crop Revenue Coverage.  For
example, in constructing its revenue distribution, Revenue Assurance
uses only 10 years of yield data (1985-94), which is not a sufficient
historical record to capture the fluctuations in yield over time. 
Furthermore, 3 of these 10 years had abnormal yields:  1988 and 1993
had abnormally low yields, and 1994 had abnormally high yields. 
Additionally, Income Protection bases its estimate of future price
increases or decreases on the way that prices moved in the past. 
This approach could be a problem because price movements in the past
occurred in the context of past government programs.  In the absence
of these government programs, the price movements may be considerably
more pronounced.  While favorable weather and stable crop prices
generated very favorable claims experience over the first 2 years
that the plans were available to farmers, these shortcomings raise
questions about whether the rates established for each plan will be
actuarially sound and are fair--that is, appropriate to the risk each
farmer presents over the long term. 

Furthermore, while the plans were initially approved only on a
limited basis, the Federal Crop Insurance Corporation authorized the
substantial expansion of Crop Revenue Coverage before the initial
results of claims experience were available.  In doing so, the
Corporation was acting within its authority to approve privately
developed crop insurance plans and in response to strong demand from
farmers.  USDA's Office of General Counsel advised against the
expansion, noting that an expansion without any data to determine
whether the plans or rates are sound might expose the Corporation to
excessive risk.  While Crop Revenue Coverage was expanded rapidly,
Revenue Assurance and Income Protection essentially remain pilot
plans with no nationwide availability. 


   RECOMMENDATION
---------------------------------------------------------- Chapter 0:5

To be more certain that the revenue insurance plans are actuarially
sound over the long term and are appropriate to the risk each farmer
presents, GAO recommends that the Secretary of Agriculture direct the
Administrator of the Risk Management Agency to address the
shortcomings in the methods used to set premiums.  Specifically, with
respect to all three plans, the Secretary should direct the Risk
Management Agency to reevaluate the methods and data used to set
premium rates to ensure that each plan is based on the most
actuarially sound foundation.  With respect to Crop Revenue Coverage,
which does not incorporate the interrelationship between crop prices
and farm-level yields, the Risk Management Agency should base premium
rates on a revenue distribution or another appropriate statistical
technique that recognizes this interrelationship. 


   AGENCY COMMENTS AND OUR
   EVALUATION
---------------------------------------------------------- Chapter 0:6

In commenting on a draft of this report, the U.S.  Department of
Agriculture expressed concern with GAO's recommendation that the
agency reevaluate the data and methods used to set premiums for the
three revenue insurance plans.  Specifically, the Department noted
that while it does not necessarily endorse or feel fully comfortable
with all aspects of the rating models, it does not believe GAO's
report provides evidence that there are "fatal flaws" in the rating
methods for the revenue insurance plans.  Therefore, the Department
believes that the plans' continued use of these rating methods is
appropriate. 

GAO believes that its recommendation is appropriate.  While GAO does
not state in this report, nor does it believe, that the plans contain
"fatal flaws," GAO does believe that the shortcomings identified in
all three revenue insurance plans are serious enough to warrant a
reevaluation of the methods and data used to set premium rates to
ensure that each plan is based on the most actuarially sound
foundation.  This is especially the case for Crop Revenue Coverage,
which does not base its rate structure upon the interrelationship
between crop prices and farm-level yields. 

The Department also provided clarifying comments to the report that
have been incorporated where appropriate.  The Department's comments
and GAO's responses are presented in detail in appendix VI. 


INTRODUCTION
============================================================ Chapter 1

Farming is inherently risky because farmers are exposed to both
production and price risks.  Farm production levels can vary
significantly from year to year, primarily because farmers operate at
the mercy of nature and frequently are subjected to weather-related
and other natural disasters.  Farm operators can also experience wide
swings in the prices they receive for the commodities they grow,
depending on total domestic and international production and demand. 

Over the years, the federal government has played an active role in
helping to mitigate the effects of risk on farm income.  On the
production side, the government has subsidized the federal
multiple-peril crop insurance program, allowing covered farmers to
receive an indemnity payment when production falls below a certain
level.  To help mitigate price risk, the government administered
price and income support programs for farmers of major field crops
such as wheat, feed grains, cotton, and rice.  However, the Federal
Agriculture Improvement and Reform Act of 1996, commonly known as the
1996 farm bill, terminated the previous income support programs and
replaced them with fixed but declining 7-year annual payments. 
Because these payments are not tied to market prices, farmers now
have to take greater responsibility for managing their risk. 

To help farmers manage their risk, the U.S.  Department of
Agriculture (USDA), has introduced a new risk management tool,
revenue insurance.  Unlike the traditional multiple-peril crop
insurance program, which insures against losses in the level of crop
production, revenue insurance plans insure against losses in revenue. 
The plans protect the farmer from the effects of either declines in
crop prices or declines in crop yields.  The guarantees are based on
market prices and on the historical yields associated with the
insured acreage.  As it does for traditional crop insurance, USDA
shares in the cost of these plans by (1) subsidizing the premiums
farmers pay, (2) paying private insurance companies to sell the
insurance and process claims, and (3) paying a large portion of the
plans' underwriting losses (the difference between premiums and
claims). 


   FEDERALLY SUBSIDIZED
   MULTIPLE-PERIL CROP INSURANCE
   PROTECTS FARMERS FROM
   PRODUCTION LOSSES
---------------------------------------------------------- Chapter 1:1

Since the 1930s, federally subsidized multiple-peril crop insurance
has been a principal means of managing the risk associated with crop
losses.  The Federal Crop Insurance Corporation (FCIC) administers
the crop insurance program.\1 Over time, this program has grown from
covering a few crops and areas to covering most crops and areas.  In
addition, the Congress has periodically appropriated funds for
disaster assistance to farmers when farming areas have suffered
widespread crop losses because of weather conditions, such as drought
or flooding. 

Between 1980 and 1998, USDA expanded the availability of crop
insurance from 30 to 67 crops and from about one-half of the nation's
counties to virtually all areas of the country.  Participation,
measured in terms of the percent of eligible acres insured, rose from
about 10 percent in 1980 to about 40 percent in the early 1990s. 
Under the Federal Crop Insurance Reform and Department of Agriculture
Reorganization Act of 1994, the Congress required farmers wishing to
participate in other USDA farm programs to purchase a minimum amount
of crop insurance.  This requirement helped increase participation to
over 70 percent of eligible acres. 

As the crop insurance program was expanded, federal costs (in
constant 1997 dollars) averaged over $1.1 billion annually during the
1990s.  As shown in table 1.1, the government's costs for crop
insurance totaled about $8.9 billion from 1990 through 1997. 



                                        Table 1.1
                         
                            Government's Cost of Federal Crop
                                        Insurance

                           (Constant 1997 dollars in millions)

               Claims paid in
                    excess of                Administrative        FCIC's
                 premiums and       Premium         expense     operating    Government's
Fiscal year      other income       subsidy  reimbursements         costs      total cost
-------------  --------------  ------------  --------------  ------------  --------------
1990                     $279          $255            $325          $104            $963
1991                      285           226             282            97             890
1992                      261           221             277            99             858
1993                      822           217             274           115           1,428
1994                    (136)           264             312            83             524
1995                      196           807             389           109           1,501
1996                       90           996             499            65           1,650
1997                    (373)           945             450            74           1,096
 (estimated)
=========================================================================================
Total                  $1,423        $3,931          $2,808          $747          $8,909
-----------------------------------------------------------------------------------------
Note:  Totals may not add because of rounding. 

Source:  USDA. 

Several types of government costs are associated with the traditional
crop insurance program.  For every dollar of premium established, the
government pays an average of 40 cents and the farmer pays 60 cents. 
The government's portion of the premiums totaled $3.9 billion from
1990 through 1997.  In addition, for every dollar of premium, the
government pays the participating insurance companies another 27
cents for the administrative costs of selling and servicing the
policies.\2 These administrative expense reimbursements to the
private insurance companies totaled $2.8 billion from 1990 through
1997.  Furthermore, the government paid a portion of program losses
(the difference between premiums and claims).  Over the years, the
established premiums have not been sufficient to pay the claims on
the policies.  Under the 1994 reform act, USDA is required to achieve
a loss ratio of 1.10--that is, for every dollar in premiums taken in,
the claims paid would be expected to average no more than $1.10.\3
For 1981 through 1996, the claims paid have averaged $1.26 per $1 of
premium, but the increases in premium rates in recent years by the
Risk Management Agency are now expected to lower the loss ratio to
about 1.10.  Under the government's standard reinsurance agreements
with the companies, the companies share a limited portion of any
program losses, but the government absorbs the vast majority of them,
totaling $1.4 billion over the period.  Finally, the government paid
$747 million for FCIC's own operating costs. 

In 1993, we reported the high costs associated with crop insurance
through the years, and we pointed out that the insurability problems
faced by the program hindered its actuarial soundness.\4 Unlike
insurers in other insurance industries, such as property and
casualty, crop insurers cannot minimize their risk of loss by pooling
participants with different levels of risk in their insurance
program.  In these other industries, the losses for one insured are
independent of the losses for another insured.  For the agriculture
sector, however, losses are not generally independent of each other. 
For example, weather conditions, such as widespread drought, can
cause production losses for many of the farmers in the same insurance
pool. 

Furthermore, as we pointed out in the 1993 report, the crop insurance
program is subject to conditions known as adverse selection and moral
hazard.  Because FCIC does not have sufficient farm-level information
to differentiate among farmers' risks, it may charge similar premiums
to both high-risk and low-risk farmers.  Consequently, high-risk
farmers are more likely to find premiums attractive and therefore
participate in the program in greater numbers than do low-risk
farmers--a situation referred to as adverse selection.  The report
also noted that FCIC lacks sufficient information about individual
farmers to detect moral hazard--when an insured farmer's actions
increase the chance for or the extent of loss.  For example, when
insurance payments seem to offer a better financial return than
marketing a partial crop, a farmer may reduce inputs, such as
fertilizer or pesticides, thereby increasing the risk of a production
loss. 


--------------------
\1 FCIC is a wholly owned government corporation.  It is governed by
a board of directors composed of USDA officials, an insurance
industry representative, and farming industry representatives.  It is
under the management of USDA's Risk Management Agency. 

\2 The reimbursement rate declined from 34 cents per $1 of premiums
for 1988 through 1991, to 33 cents in 1992, 32.5 cents in 1993, 31
cents in 1994 through 1996, 29 cents in 1997, and 27 cents in 1998. 
In 1997, we reported that FCIC could further lower the reimbursement
rate in the range of 24 cents per $1 of premiums and still adequately
compensate participating companies for the reasonable costs
associated with selling and servicing multiple-peril crop insurance. 
See Crop Insurance:  Opportunities Exist to Reduce Government Costs
for Private-Sector Delivery (GAO/RCED-97-70, Apr.  17, 1997). 

\3 The target reduces to $1.075 per $1 of premium after Sept.  30,
1998. 

\4 Crop Insurance:  Federal Program Faces Insurability and Design
Problems (GAO/RCED-93-98, May 24, 1993). 


   INCOME AND PRICE SUPPORT
   PROGRAMS MITIGATED PRICE RISKS
---------------------------------------------------------- Chapter 1:2

The federal government also used income and price support programs in
an effort to protect farmers' incomes.  Prior to 1996, USDA
administered programs known as deficiency payment programs for
several major crops--wheat, feed grains, cotton, and rice.  These
programs were designed to protect farmers' incomes against declines
in prices through a complicated array of pricing mechanisms.  In
return for participating in these programs, farmers agreed to limits
on the number of acres they placed into production.  Unlike the
deficiency payment programs, which were not reauthorized by the 1996
farm bill, a number of price support programs, such as the marketing
loan program, are still in place.  Marketing loan programs are
designed, among other things, to help farmers in periods of severely
low prices. 

Under the 1996 farm act, farmers' were encouraged to produce in
response to market forces alone, rather than to the expectation of
federal payments.  As part of this new direction in policy, the 1996
act replaced the previous income support programs with "production
flexibility contracts"--agreements between the federal government and
participating farmers that provide for fixed but declining 7-year
annual payments through 2002.  These annual payments are not tied to
market prices.  Farmers who signed these agreements are not
restricted to the type or amount of any crop they plant.  USDA
estimates that the production flexibility contracts will cost a total
of $35.6 billion over the 7-year period. 


   OTHER WAYS FARMERS MANAGE RISK
   OF PRICE FLUCTUATION
---------------------------------------------------------- Chapter 1:3

Many farmers also use crop insurance in combination with
nongovernmental strategies to manage the risk to their income
resulting from price fluctuations.  A common strategy is forward
contracting.  With this technique, farmers contract to sell the crop,
well before it is actually harvested, and thus are able to establish
a pre-harvest selling price and guarantee an outlet for the crop. 
Additionally, some farmers use hedging--a process whereby the farmer
directly uses the commodity futures markets to establish a
pre-harvest price for the crop.  The farmers using these techniques
to manage their price risk generally continue to use traditional
multiple-peril crop insurance to manage the risk of crop loss. 


   NEW FEDERALLY SUBSIDIZED CROP
   REVENUE INSURANCE PLANS PROTECT
   FARMERS FROM BOTH PRICE AND
   YIELD DECLINES
---------------------------------------------------------- Chapter 1:4

As an alternative to buying crop insurance and separately forward
contracting or hedging, three new government-supported revenue
insurance plans--Crop Revenue Coverage, Revenue Assurance, and Income
Protection--provide farmers with a single policy that protects
against both production and price risk.  Crop Revenue Coverage and
Revenue Assurance were developed by private insurance companies that
requested and received federal reinsurance for the plans,\5 whereas
FCIC developed Income Protection as a pilot project under the terms
of the 1994 crop insurance reform act, which called for a risk
protection plan based on the cost of production. 

Income Protection and Revenue Assurance are similar in that each plan
pays indemnities when the income from crop production is less than
the revenue guaranteed at planting.  Crop Revenue Coverage adds an
additional dimension that allows the farmer to receive a larger
payment if market prices have increased in the intervening period. 
For all three plans, market prices are tied to the futures prices on
the commodity exchanges, such as the Chicago Board of Trade. 
Premiums for Crop Revenue Coverage are established as surcharges to
the traditional multiple-peril crop insurance rates, whereas Income
Protection and Revenue Assurance use methods to establish new rates
that are independent of the traditional rate. 

USDA shares in the cost of these new plans in a manner similar to the
method used to support traditional multiple-peril crop insurance. 
First, just as with traditional multiple-peril crop insurance, USDA
subsidizes the premiums farmers pay.  The subsidy, which averages 40
percent of premiums for multiple-peril crop insurance, is limited, in
the case of the new revenue plans, to the same dollar amount that
would apply to the comparable multiple-peril insurance policy. 
Second, just as with traditional multiple-peril crop insurance, USDA
pays private insurance companies a reimbursement for administrative
expenses to sell the revenue insurance policies and process claims. 
This administrative reimbursement is a preestablished percentage of
the premiums paid by the farmers.  In 1998, USDA will pay the
companies 27 percent of premiums to sell and service the
multiple-peril, Income Protection, and Revenue Assurance policies. 
Because the premiums are significantly higher for Crop Revenue
Coverage policies, USDA has limited the administrative payment on
these policies to 23.25 percent of premiums.  Finally, just as with
multiple-peril crop insurance, USDA pays a large portion of any
underwriting losses that may result if premiums are not high enough
to pay all claims arising under the revenue policies.  For 1998, USDA
increased the portion of these losses that the companies must absorb,
but the government continues to absorb most of the losses. 
Conversely, if underwriting gains occur--when premiums are higher
than claims--the insurance companies and the federal government share
in the gains. 


--------------------
\5 Under reinsurance, FCIC agrees to subsidize the purchasers'
premiums, pay the companies an administrative fee, and limit the
amount of loss that the companies could suffer under the policies. 
Reinsurance for privately developed plans is authorized under the
1990 farm act.  Crop Revenue Coverage was developed by Redland
Insurance Company, and Revenue Assurance was developed by Farm Bureau
Mutual Insurance Company.  Once a new plan is approved for
reinsurance, it may be sold by any participating company. 


   OBJECTIVES, SCOPE, AND
   METHODOLOGY
---------------------------------------------------------- Chapter 1:5

In light of the rapid expansion of, and the government's significant
financial participation in, the new crop revenue plans, the Ranking
Minority Member of the House Committee on Agriculture asked us to (1)
identify the differences between the three new revenue insurance
plans, (2) report on the plans' sales and claims experience, and (3)
analyze the methodologies used to set the plans' premium rates. 

We identified the differences in the various revenue insurance plans
by reviewing USDA's documentation for each plan as provided by the
plans' developers and comparing the plans' features and protection
levels.  We confirmed our understanding of the various features of
each plan by interviewing the Administrator of USDA's Risk Management
Agency at USDA's headquarters in Washington, D.C., and the Senior
Actuary at the Risk Management Agency's main field office in Kansas
City, Missouri; and by interviewing the developers of the revenue
plans at Kansas State University, Iowa State University, and Montana
State University. 

To determine the sales and claims experience of the three revenue
insurance plans and traditional multiple-peril crop insurance, we
obtained USDA's computer files for crop years 1996 and 1997--the
first years in which revenue insurance policies were sold.  We
identified national sales and claims information for each plan and
analyzed this information, controlling for the differences in
availability because of location, crop, and level of protection.  We
also examined the characteristics of Crop Revenue Coverage policies
by measuring average acres insured, variability of year-to-year crop
yields, and average yields per insured policy unit and comparing this
information with the characteristics of multiple-peril crop insurance
policies.  Because Income Protection's and Revenue Assurance's sales
were limited, we could not analyze their risk characteristics. 

To analyze the methods used to set premium rates and to identify
uncertainties pertaining to premium rates, we reviewed academic
literature on setting insurance rates and agricultural economics
literature on crop revenue insurance and other issues such as the
correlation between local crop yields and national prices.  We also
interviewed officials at USDA's Economic Research Service, Office of
the Chief Economist, and Risk Management Agency; the academic
consultants on the plans at Kansas State University, Iowa State
University, and Montana State University; and agricultural economists
at several other universities who have performed research on crop
and/or revenue insurance issues.  In order to examine each revenue
insurance plan, we interviewed the developers of the plans and
reviewed the documentation they had provided to USDA as well as
additional information they provided to us.  We also evaluated each
plan in light of our economic analysis, our discussions with the
experts in these fields, and our review of the pertinent insurance
and agricultural economics literature.  We discussed our analysis
with the developers of the plans and several independent reviewers. 

We conducted our review from July 1997 through March 1998 in
accordance with generally accepted government auditing standards.  We
used the same files USDA uses to manage the crop insurance program. 
These files provide the most comprehensive information on farmers who
have purchased crop revenue insurance. 


THREE PLANS' APPROACHES TO
INSURING REVENUE RESULT IN
DIFFERENT LEVELS OF PROTECTION AND
GOVERNMENT COSTS
============================================================ Chapter 2

The three government-subsidized revenue insurance plans--Income
Protection, Revenue Assurance, and Crop Revenue Coverage--differ in
the revenue guarantees they provide to the farmer and in their
relative cost to the government.  Two of the plans, Income Protection
and Revenue Assurance, set the revenue level that is to be protected
at the time that crops are being planted, while the third, Crop
Revenue Coverage, determines the protected revenue at either planting
or at harvest, depending on when prevailing crop prices are higher. 
In terms of potential government costs, Crop Revenue Coverage is
likely to cost the government significantly more than the other two
plans over time. 


   PLANS DIFFER IN THE REVENUE
   PROTECTION THEY OFFER
---------------------------------------------------------- Chapter 2:1

The three government-subsidized revenue insurance plans--Income
Protection, Revenue Assurance, and Crop Revenue Coverage--establish a
revenue target, or guarantee, for farmers.  But they differ in how
that guarantee is determined. 

For both Income Protection and Revenue Assurance, the farmer's
revenue guarantee is established when crops are planted.  To
determine that guarantee, the insurer multiplies the farmer's
expected production by a price established at planting.  If the
farmer's revenue at harvest is below that expected preseason income,
the farmer receives an insurance payment.  Farmers whose revenue is
at or above the guaranteed level do not receive a payment.  Total
revenue from the crop is the determining characteristic, not the
level of production or the price alone.  No payment would be made if
a price decline is sufficiently offset by an increase in production
or if a loss in production is offset by a sufficient increase in
price. 

In contrast, the calculation of the amount of revenue guaranteed
under Crop Revenue Coverage is more complicated.  Crop Revenue
Coverage guarantees a minimum revenue at planting that is determined
by multiplying the prevailing futures market price at planting by the
farmer's historical production per acre.  At harvest, the revenue
guarantee is revisited, and the final guarantee is determined by
multiplying the farmer's historical production by the price at
planting or at harvest, whichever price is higher.  If the price has
increased in the period between planting and harvest, the farmer
receives a payment for any lost production at the higher harvest
price.  This upward price protection feature assures the farmer that
any lost production will be replaced at the prevailing market price,
thus facilitating forward contracting by the farmer.  If, however,
the harvest price is lower, the original guarantee is in force.  (For
additional information on how the revenue payment is calculated, see
app.  I.)


   OPERATIONAL FEATURES ARE
   SOMEWHAT DIFFERENT IN THE THREE
   REVENUE INSURANCE PLANS
---------------------------------------------------------- Chapter 2:2

The revenue insurance plans also differ in several operational
features.  Although futures prices form the basis for the payments
under all three plans, the plans adjust these prices differently to
account for variations between local and national prices.  In
addition, the methods used to establish which parcels of land will be
covered for insurance purposes vary from plan to plan.  Finally, only
one of the plans is available across the country. 


      PRICE USED IN REVENUE
      CALCULATION VARIES BY TYPE
      OF INSURANCE PLAN
-------------------------------------------------------- Chapter 2:2.1

The insurance payment for all three plans is determined by
subtracting the revenue realized at harvest from the revenue
guarantee.  The starting point for determining revenue is the futures
prices for a particular commodity on its commodity exchange. 
However, each plan adjusts those prices somewhat differently.  The
differences center around how the national price prevailing on a
commodity exchange is adjusted for local conditions.  Generally,
prices in local markets are a few cents per bushel less than the
national price on the board of trade.  These local differences are
generally greater in the areas more distant from major market
centers, and the differences decline nearer to the market centers. 
Income Protection and Crop Revenue Coverage do not adjust for this
factor, while Revenue Assurance makes a county-by-county adjustment. 
Table 2.1 shows the revenue guarantee features of the three plans. 



                                        Table 2.1
                         
                           Revenue Guarantee Features for Three
                             Risk Management Plans, for Corn

                                                                     Crop Revenue
Plan feature             Income Protection     Revenue Assurance     Coverage
-----------------------  --------------------  --------------------  --------------------
Revenue guarantee        Historical            Historical            Historical
calculation              production            production            production
                         multiplied by 100%    multiplied by 100%    multiplied by the
                         of the Chicago Board  of a projected        higher of (1) 95% or
                         of Trade's February   county price (the     100% of Chicago
                         price for the         Chicago Board of      Board of Trade's
                         December contract     Trade's February      February price for
                                               price for the         December delivery or
                                               December contract,    (2) 95% or 100% of
                                               adjusted by a county  the Chicago Board of
                                               factor)               Trade's price in
                                                                     November for the
                                                                     December contract

Actual harvest revenue   Actual production     Actual production     Actual production
calculation              multiplied by 100%    multiplied by USDA's  multiplied by 95% or
                         of the November       posted county price   100% of the November
                         price for the                               price of the
                         December contract                           December contract
-----------------------------------------------------------------------------------------
Source:  USDA. 

As the table shows, Income Protection makes no adjustment for the
difference in prices that occur from county to county.  Instead, the
plan uses one national price for all policies in all counties.  For
corn, the price used to determine the revenue guarantee for all
policies is the Chicago Board of Trade's average corn futures price
in February for the December contract.  Similarly, Income Protection
determines actual revenue for all policyholders by multiplying the
farmer's actual production by the Chicago Board of Trade's corn
futures average price in November for the December contract. 

In contrast, Revenue Assurance determines the revenue guarantee for
each farmer using the Chicago Board of Trade's February prices for
the December corn contract, adjusted by a county-specific factor. 
Revenue Assurance establishes this adjustment on the basis of the
historical relationship of local harvest prices in each county to the
Chicago Board of Trade's prices in the harvest month.  To determine
the value of the harvested crop, Revenue Assurance departs from the
Chicago Board of Trade's prices.  Instead, it uses a price USDA
establishes for other purposes in each county--referred to as the
posted county price. 

During 1996 and 1997, Crop Revenue Coverage calculated each farmer's
revenue guarantee using the higher of (1) 95 percent of the average
corn futures price on the Chicago Board of Trade in February for the
December contract, or (2) 95 percent of the average corn price on the
Chicago Board of Trade in November for the December contract.  To
determine the crop's harvested value, Crop Revenue Coverage used 95
percent of the average corn price on the Chicago Board of Trade in
November for the December contract.  For 1998, farmers may choose to
insure at either 95 or 100 percent of the futures price. 


      LAND COVERED VARIES BY PLAN
-------------------------------------------------------- Chapter 2:2.2

The three risk management plans differ in the choices they offer the
farmer to combine the various individual fields on their farm or
farms for insurance purposes.  These differences in the way farmers
can insure the land they farm are important because revenue payments
differ depending on the actual configuration.  Four land
configuration arrangements are available to farmers:  (1) whole farm
(combining coverage on all fields for all combinations of covered
crops in the county in which the farmer has a share in the crops
produced); (2) enterprise unit (combining each of the fields in which
the farmer owns or has a share of the crop produced in the county,
regardless of ownership arrangement); (3) basic unit (combining each
of the fields of a crop under a single type of ownership
arrangement); and (4) optional unit (essentially, insuring on a
field-by-field basis).  In general, the more a farmer's land is
consolidated, the less likely it is that the farmer will have a loss
large enough to trigger an insurance payment.  This is because a
farmer's production, for insurance purposes, is averaged across all
the insured fields. 

Income Protection is available only on the basis of the enterprise
unit.  In contrast, for Revenue Assurance, farmers can choose to
configure their farm with any type of units.  Initially, Revenue
Assurance establishes the premium rate for those choosing the basic
unit.  If the farmer wants to further divide the basic unit into
optional units, the policy imposes a surcharge.  However, if the
farmer elects to consolidate coverage on the basis of an enterprise
unit, the policy offers a discount from the initial basic unit rate. 
The policy provides an additional discount for the farmer who chooses
whole farm coverage.  Finally, Crop Revenue Coverage allowed basic
and optional coverage in 1996 and 1997 and received approval from
FCIC to add enterprise coverage for 1998.  In 1997, 61 percent of
Crop Revenue Coverage policies were based on optional units.\1


--------------------
\1 In comparison, 54 percent of multiple-peril policies were based on
this configuration. 


      THE THREE PLANS VARY IN THE
      CROPS COVERED AND AREAS OF
      AVAILABILITY
-------------------------------------------------------- Chapter 2:2.3

As table 2.2 shows, the three plans are not available for all crops
in all areas, although Crop Revenue Coverage is rapidly expanding to
cover more crops in more states.  All three plans are relatively new,
which accounts for their limited availability in some areas of the
nation. 



                                        Table 2.2
                         
                          Differences in Crops Covered and Areas
                              of Availability for Three Risk
                                     Management Plans

Income Protection Plan         Revenue Assurance Plan        Crop Revenue Coverage Plan
-----------------------------  ----------------------------  ----------------------------
1996: Corn farmers in 14       1996: Not available           1996: Corn and soybean
pilot counties, cotton                                       farmers in Iowa and
farmers in 8 pilot counties,   1997: Corn and soybean        Nebraska
and spring wheat farmers in 7  farmers in Iowa
pilot counties                                               1997: Continued 1996
                               1998: No change               coverage and added corn
1997: Continued 1996 coverage                                farmers in 11 more states,
and added farmers of grain                                   wheat farmers in 8 states,
sorghum in 25 counties,                                      cotton farmers in 4 states,
soybeans in 56 counties, and                                 grain sorghum farmers in 6
winter wheat in 18 counties                                  states, and soybean farmers
                                                             in 10 additional states
1998: Continued 1996 and 1997
coverage and added winter                                    1998: Continued 1996 and
wheat farmers in 12                                          1997 coverage and added
additional counties                                          wheat farmers in 25
                                                             additional states, soybean
                                                             farmers in 12 additional
                                                             states, and cotton farmers
                                                             in 13 additional states
-----------------------------------------------------------------------------------------
Source:  USDA. 

Crop Revenue Coverage, since its introduction in 1996, has rapidly
expanded to all major crops in the major growing areas.  Income
Protection, developed by USDA in 1996, has been expanded slightly but
is only available in scattered counties covering certain crops around
the nation.  Finally, Revenue Assurance, which became available in
1997, only covers corn and soybeans in Iowa. 


   HYPOTHETICAL EXAMPLE OF HOW
   DIFFERENT CROP INSURANCE PLANS
   WOULD COVER DIFFERENT
   PRODUCTION AND PRICE SITUATIONS
---------------------------------------------------------- Chapter 2:3

To illustrate the differences between traditional multiple-peril crop
insurance and the three revenue insurance plans, we examined the
premiums and insurance claim payments for a hypothetical Iowa corn
farmer.  For this illustration, we assumed the farmer purchased crop
insurance at the 75-percent coverage level and established a record
of normal production of 120 bushels per acre.  We also used 1997
prices under various combinations of 30-percent price and production
increases and declines.  Of course, payment amounts at other
combinations of production and prices would be different.  As shown
in table 2.3, premiums for Crop Revenue Coverage would be higher than
for traditional multiple-peril crop insurance because Crop Revenue
Coverage provides additional benefits.  In contrast, for this
example, premiums for Income Protection and Revenue Assurance would
be lower than for traditional crop insurance.  The table also shows
that, in the event of normal production combined with 30-percent
decline in prices, no payment would be due under the traditional
multiple-peril crop insurance policy, but each of the revenue
policies would provide payments.  In the event of 30-percent declines
in both production and price, each type of policy would pay, but the
amounts paid would vary.  However, in the event of a 30-percent
decline in production combined with a 30-percent increase in price,
the traditional policy and Crop Revenue Coverage would result in
claims payments, but no claim payment would result under the terms of
Income Protection and Revenue Assurance. 



                               Table 2.3
                
                 Hypothetical Illustration of Insurance
                Premium and Insurance Claim Payment per
                  Acre for a Corn Farmer at Different
                    Levels of Production and Prices

                         Multiple-        Crop
                        peril crop     Revenue      Income     Revenue
                         insurance    Coverage  Protection   Assurance
----------------------  ----------  ----------  ----------  ----------
Premiums                    $11.20      $16.50       $5.90       $8.40

Payments
----------------------------------------------------------------------
Normal production and
 30% price increase
                                 0           0           0           0
Normal production with
 30% price decline
                                 0       15.90       16.50       13.80
Production decrease by
 30% with 30% price
 decline\a                   14.70       81.06       85.26       73.92
Production decrease by
 30% with 30% price
 increase                    14.70       20.22           0           0
----------------------------------------------------------------------
\a This situation is unlikely to occur in the aggregate, but it may
occur for individual farmers. 

Source:  GAO's analysis of USDA's data. 

Appendix I describes how the premiums and payments shown in table 2.3
were calculated. 


   CROP REVENUE COVERAGE IS LIKELY
   TO COST THE GOVERNMENT
   SIGNIFICANTLY MORE THAN OTHER
   PLANS
---------------------------------------------------------- Chapter 2:4

Crop Revenue Coverage is likely to be more costly to the government
than the other insurance plans because of its higher reimbursements
for administrative expenses to participating companies and because of
potentially higher total underwriting losses (the excess of claims
payments over total premiums).  Furthermore, the plan's promise to
base the revenue guarantee on the price at planting or at harvest,
whichever is higher, exposes the government to higher claims payments
in years when widespread crop losses are coupled with rapidly
increased prices. 


      ADMINISTRATIVE COSTS ARE
      LIKELY TO BE HIGHER FOR CROP
      REVENUE COVERAGE THAN FOR
      OTHER PLANS
-------------------------------------------------------- Chapter 2:4.1

The government pays insurance companies a smaller fee per premium
dollar to sell and service Crop Revenue Coverage than the other
revenue plans or multiple-peril crop insurance.  However, the total
cost of administrative reimbursements for Crop Revenue Coverage is
greater because the reimbursement rate is not low enough to offset
the much higher premiums under this plan.  That is, the government
reimburses the companies at a rate of 23.25 cents for every dollar of
premium sold, which is less than the rate of 27 cents per dollar of
premium for the other plans; but because Crop Revenue Coverage's
premiums per acre average about 30-percent higher than the premiums
for the other crop insurance plans, the effective cost to the
government is actually higher for Crop Revenue Coverage than for the
other insurance plans. 

For example, for insurance sales that generate $1 million of
premiums, the government's costs for reimbursing administrative
expenses under Income Protection and Revenue Assurance is $270,000
(27 percent in administrative costs multiplied by $1 million in
premiums).  In contrast, the government's equivalent cost for Crop
Revenue Coverage for the same number of insured acres is $302,250
(23.25 percent in administrative costs multiplied by the higher
premiums--$1.3 million), an increase of $32,250. 

Assuming Crop Revenue Coverage premiums of $300 million for crop year
1998,\2 the government's administrative reimbursement cost for this
plan will be over $7 million higher than for either of the other two
revenue insurance plans or traditional multiple-peril crop insurance. 
The participating companies receive higher reimbursements but also
incur some additional expenses, including higher processing and
training costs, and higher loss adjustment costs in the years when
Crop Revenue Coverage makes payments while multiple-peril crop
insurance does not. 


--------------------
\2 In 1997, Crop Revenue Coverage premiums totaled $281 million. 


      TOTAL LOSSES ARE LIKELY TO
      BE HIGHER
-------------------------------------------------------- Chapter 2:4.2

Crop Revenue Coverage's higher volume of premiums also results in
higher costs to the government than multiple-peril crop insurance,
given equal levels of underwriting losses.  Under current law, both
multiple-peril crop insurance and the new revenue insurance plans are
expected to operate over time with an underwriting loss of $1.10 paid
in claims for every $1 in premium.  In other words, the government
expects, over time, to pay claims averaging $1.10 for every $1 in
premium.  Therefore, by applying the same loss rate to Crop Revenue
Coverage's larger volume of premiums, the absolute dollar value of
the loss will be higher. 


      EXPOSURE TO HIGH PAYOUTS IN
      A SINGLE YEAR ARE MORE
      LIKELY
-------------------------------------------------------- Chapter 2:4.3

In addition to generating potentially higher total losses, the claims
experience with Crop Revenue Coverage is likely to have a more
exaggerated, or magnified, impact during any given year because of
the plan's unique upward price protection feature.  This feature,
which gives the farmer an increased revenue guarantee when market
prices rise between the time the farmer plants and harvests the crop,
significantly raises the government's exposure to large claims
payments in years when widespread crop losses are coupled with
rapidly increased prices.  The two other plans reduce the
government's exposure during such years. 

For example, in 1996, adverse weather conditions destroyed winter
wheat in sections of the Great Plains and Midwest, contributing to an
increase in prices from $3.65 a bushel at planting to $5.47 per
bushel at harvest.  If Crop Revenue Coverage had been available for
winter wheat in 1996, FCIC would have had to pay an additional 43
percent, or $172 million more, in claims than it actually paid under
traditional multiple-peril crop insurance.  As shown in table 2.4,
assuming Crop Revenue Coverage had been available and protected 50
percent of the acres insured in 1996, FCIC would have paid an
estimated $569.8 million in wheat claims instead of the $397.7
million it actually paid.  Alternatively, because the price increase
that occurred in 1996 more than offset the average production loss,
the provisions of the Income Protection or Revenue Assurance plans
would have resulted in claims payments that were much less than those
actually paid under multiple-peril crop insurance--an estimated
$198.8 million. 



                               Table 2.4
                
                 Estimated Winter Wheat Claims Payments
                 Assuming Crop Revenue Coverage, Income
                 Protection, or Revenue Assurance on 50
                    Percent of Acres Insured, 1996\

                         (Dollars in millions)

                                                             Estimated
                         Actual claims                          claims
                        payments under       Estimated  payments under
                           traditional          claims          Income
                             multiple-  payments under   Protection or
                            peril crop    Crop Revenue         Revenue
State                        insurance      Coverage\a     Assurance\a
----------------------  --------------  --------------  --------------
Kansas                          $137.6          $200.5           $68.8
Oklahoma                          59.8            84.1            29.9
Texas                             48.8            69.2            24.4
Colorado                          26.9            38.6            13.4
Illinois                          20.5            32.9            10.3
South Dakota                      18.7            24.1             9.3
Nebraska                          14.4            19.0             7.2
All other states                  71.0           101.4            35.5
======================================================================
Total                           $397.7          $569.8          $198.8
----------------------------------------------------------------------
Note:  This analysis may reflect some spring wheat losses because the
records in USDA's databases do not always distinguish between the
types of wheat that were insured. 

\a Assumes revenue insurance protection on 50 percent of the acres
insured and traditional multiple-peril crop insurance on the other 50
percent.  This is a hypothetical example:  The plans were not
available for this crop in these states. 

Source:  GAO's analysis of USDA's data. 

This example involves one crop.  Widespread droughts often affect a
number of crops, in which case the government's financial exposure
under Crop Revenue Coverage would increase more.  However, part of
the potential underwriting loss for Crop Revenue Coverage is reduced
by the increased premiums in effect.  Additionally, under reinsurance
agreements, underwriting losses are borne in part by the
participating companies, but the majority of the losses are paid by
USDA.  Furthermore, during years with favorable claims experience,
Crop Revenue Coverage would generate higher underwriting gains than
either multiple-peril crop insurance or the other two revenue
insurance plans. 


NEW INSURANCE PLANS ACHIEVING
SIGNIFICANT SHARE OF CROP
INSURANCE MARKET
============================================================ Chapter 3

In their first 2 years, the crop revenue insurance plans, especially
Crop Revenue Coverage, have already achieved a significant share of
the crop insurance market, accounting for about one-third of crop
insurance premiums in the areas where they were offered.  In the
initial years, the new plans' claims payment experience was similar
to the experience of traditional multiple-peril crop insurance. 

With respect to the characteristics of the farming operations covered
by the plans, Crop Revenue Coverage policies written in 1997 insured
higher acreage levels and were associated with operations having
lower production variability, over time, than traditional
multiple-peril crop insurance.  Therefore, the Crop Revenue Coverage
policies, on average, appear to be less risky.  This lower level of
risk may have occurred because initial marketing efforts were
targeted to larger farmers in the most consistently productive farm
areas.  As such, the differences in risk may diminish over time as
the marketing expands into the general farming community. 


   CROP REVENUE INSURANCE SALES
   STRONG IN 1997, WHILE CLAIMS
   WERE LOW
---------------------------------------------------------- Chapter 3:1

Crop revenue insurance plans, as a group, had strong sales, claiming
a significant portion of crop insurance sales in 1997, the first year
that all three plans were available.  Crop Revenue Coverage, the most
widely available of the three revenue insurance plans, took away a
considerable amount of business from multiple-peril crop
insurance--obtaining a 32-percent share of the market--in the areas
where it was sold.  In contrast, neither Revenue Assurance nor Income
Protection were able to attract many purchasers--obtaining 6-percent
and 3-percent shares, respectively--in the areas where they were
sold. 


      CROP REVENUE COVERAGE
      CAPTURED ABOUT ONE-THIRD OF
      MARKET WHERE AVAILABLE
-------------------------------------------------------- Chapter 3:1.1

By its second year, Crop Revenue Coverage had captured a significant
portion of the crop insurance business from traditional
multiple-peril crop insurance in areas where both were available.  As
shown in table 3.1, Crop Revenue Coverage in 1997 accounted for 32
percent of the premiums, 29 percent of the acres insured, and 25
percent of the policies in the areas where it was sold.  According to
a senior Risk Management Agency official, this plan has attracted
many purchasers in part because the premiums for the plan, on a
cost-per-acre basis, were relatively low in the areas where the plan
was introduced, and in these locations, the premiums appeared
reasonable for the potential additional benefits they provide. 



                               Table 3.1
                
                 Market Share of Crop Revenue Coverage
                and Multiple-Peril Crop Insurance Where
                     Both Plans Were Offered, 1997

                                     Measures of market share
                            ------------------------------------------
                                Percent of      Percent of  Percent of
Insurance plan              total premiums   insured acres    policies
--------------------------  --------------  --------------  ----------
Crop Revenue Coverage                   32              29          25
Multiple-peril crop                     68              71          75
 insurance
----------------------------------------------------------------------
Source:  GAO's analysis of USDA's data. 


      INCOME PROTECTION HAS NOT
      CAPTURED SIGNIFICANT MARKET
      SHARE
-------------------------------------------------------- Chapter 3:1.2

In the counties where Income Protection is available for purchase,
few farmers have opted to buy it.  As shown in table 3.2, Income
Protection obtained from 3 to 5 percent of the total crop insurance
market, depending on the measure used.  In the 41 counties where both
Income Protection and Crop Revenue Coverage were offered in 1997, the
sales achieved by Income Protection appear to come at the expense of
Crop Revenue Coverage rather than multiple-peril crop insurance. 



                               Table 3.2
                
                Market Share of Revenue Insurance Plans
                  Where Income Protection Was Offered,
                                  1997

                                     Measures of market share
                            ------------------------------------------
                                Percent of      Percent of  Percent of
Insurance plan               total premium   insured acres    policies
--------------------------  --------------  --------------  ----------
Income Protection                        3               4           5
Crop Revenue Coverage                   25              22          18
Multiple-peril crop                     72              74          76
 insurance
----------------------------------------------------------------------
Source:  GAO's analysis of USDA's data. 


      REVENUE ASSURANCE HAS NOT
      CAPTURED LARGE MARKET SHARE
      IN IOWA
-------------------------------------------------------- Chapter 3:1.3

In the one state where it was sold--Iowa--Revenue Assurance met with
only moderate success.  For crop year 1997, Iowa was the only state
where farmers were able to choose between traditional multiple-peril
crop insurance and all three revenue insurance plans.  As shown in
table 3.3., in terms of total premiums, Revenue Assurance achieved a
6-percent share of the Iowa corn insurance market and an 8-percent
share of the Iowa soybean insurance market.  In contrast, Crop
Revenue Coverage achieved higher market penetration in Iowa--52
percent of the corn and 49 percent of the soybean market--than it did
nationally.  Income Protection--available in six counties in
Iowa--achieved less than 1 percent of the sales for both corn and
soybeans. 



                               Table 3.3
                
                Market Share of Revenue Insurance Plans,
                               Iowa, 1997

                                     Measures of market share
                            ------------------------------------------
Crop and                        Percent of      Percent of  Percent of
insurance plan              total premiums   insured acres    policies
--------------------------  --------------  --------------  ----------
Corn
Revenue Assurance                        6               9           7
Income Protection                      0\a             0\a         0\a
Crop Revenue Coverage                   52              41          37
Multiple-peril crop                     42              50          55
 insurance

Soybeans
----------------------------------------------------------------------
Revenue Assurance                        8               9           7
Income Protection                      0\a             0\a         0\a
Crop Revenue Coverage                   49              40          35
Multiple-peril crop                     43              52          58
 insurance
----------------------------------------------------------------------
\a Rounds to less than 1 percent. 

Source:  GAO's analysis of USDA's data. 


      MULTIPLE-PERIL CROP
      INSURANCE AND REVENUE
      INSURANCE PLANS HAD
      RELATIVELY LOW LEVELS OF
      CLAIMS IN 1997
-------------------------------------------------------- Chapter 3:1.4

All types of crop insurance had relatively low levels of claims in
1997.  The crop insurance industry discusses the extent of losses in
terms of the claims paid per premium dollar collected.  For 1981
through 1996, traditional multiple-peril crop insurance paid an
average of $1.26 in claims per $1 of premium (including the
government's subsidy).  However, in 1997, because of the relatively
favorable growing conditions in the nation, the crop insurance
program had a much lower level of claims--$0.49 per $1 of premium. 
Moreover, the revenue insurance plans had lower levels of claims
payments than did multiple-peril crop insurance--ranging from $0.06
to $0.36 per $1 of premium, as shown in table 3.4.  According to the
Risk Management Agency, the lower claims experience could have
occurred for several reasons, such as a concentration of sales in
lower-risk areas, stable crop prices, and/or a combination of these
and other factors. 



                               Table 3.4
                
                  Claims Experience by Insurance Plan,
                                  1997

                         (Dollars in millions)

                                 Total          Claims   Claims per $1
Insurance plan                premiums        payments      of premium
--------------------------  ----------  --------------  --------------
Multiple-peril crop           $1,331.1          $689.9           $0.52
 insurance
Crop Revenue Coverage            280.7           100.4            0.36
Income Protection                  2.7             0.6            0.23
Revenue Assurance                  8.1             0.5            0.06
======================================================================
Total                         $1,622.5          $791.4           $0.49
----------------------------------------------------------------------
Source:  USDA. 

The generally low level of claims experienced for the revenue
insurance plans also may be attributed in part to the fact that the
new insurance products were generally purchased by larger, slightly
lower-risk farmers.  See appendix II for detailed sales and claims
data by state and insurance plan. 


   CROP REVENUE COVERAGE POLICIES
   INSURED MORE ACRES AND HAD
   LOWER RISK CHARACTERISTICS THAN
   TRADITIONAL CROP INSURANCE
---------------------------------------------------------- Chapter 3:2

Crop Revenue Coverage policies written in 1997 insured a higher
number of acres and were associated with operations having lower
production variability over time, thus appearing to be less risky, on
average, than traditional multiple-peril crop insurance.  Crop
insurance research has shown that policies with these characteristics
tend, on average, to have a lower incidence of claims payments.  The
differences between Crop Revenue Coverage and traditional
multiple-peril crop insurance may have occurred because initial
marketing efforts were targeted to larger farmers in the most
consistently productive farm areas.  As such, the differences may
diminish over time as marketing expands into the general farming
community.  While the two plans differ in these respects, we found
that they were similar in other respects, such as the average yield
per acre.  Because Income Protection's and Revenue Assurance's sales
were limited, we could not analyze their risk characteristics. 


      CROP REVENUE COVERAGE
      INSURED HIGHER-THAN-AVERAGE
      ACREAGE LEVELS
-------------------------------------------------------- Chapter 3:2.1

In 1997, Crop Revenue Coverage insured more acres, on average, than
did traditional multiple-peril crop insurance.  Specifically, the
policies for traditional multiple-peril crop insurance, insured, on
average, about 132 acres per policy in 1997,\1 while Crop Revenue
Coverage policies insured about 160 acres, or 21 percent more. 
According to a senior Risk Management Agency official, these
differences may have occurred because crop insurance agents' initial
marketing efforts may have targeted larger farming operations, and
this difference may decline over time as marketing expands into the
general farming community. 


--------------------
\1 Generally, a farmer has one policy for each crop insured.  Thus, a
farmer who insures both corn and soybeans would have two policies. 


      CROP REVENUE COVERAGE
      POLICIES INSURED OPERATIONS
      WITH LOWER PRODUCTION
      VARIABILITY
-------------------------------------------------------- Chapter 3:2.2

In 1997, Crop Revenue Coverage policies insured farming operations
with slightly less variation in their production history over time,
on average, than traditional crop insurance.  Specifically, these
policies had an average variation of 22 percent, compared with an
average variation of 25 percent for traditional multiple-peril crop
insurance.  These percentages represent the average deviation of each
insured unit's actual yield per acre each year from the unit's
average yield over the period for which production history was
provided.  As we noted in 1993, farmers having a high variation in
their production are more likely to experience a loss than farmers
having low variation in their production and, thus, are riskier to
insure.\2 With a variation in production that is 3 percentage points
lower, the holders of Crop Revenue Coverage policies are less likely
to experience a loss. 


--------------------
\2 Crop Insurance:  Federal Program Faces Insurability and Design
Problems (GAO/RCED-93-98, May 24, 1993). 


      CROP REVENUE COVERAGE AND
      TRADITIONAL MULTIPLE-PERIL
      CROP INSURANCE SHARE SOME
      CHARACTERISTICS ASSOCIATED
      WITH RISK
-------------------------------------------------------- Chapter 3:2.3

While Crop Revenue Coverage and traditional multiple-peril crop
insurance differ in some respects, they are similar in others.  For
example, multiple-peril crop insurance and Crop Revenue Coverage
policies generally were based on similar years of production history
provided by the policyholders--an average of 7.3 years for insured
units under multiple-peril insurance compared with 7.4 years under
Crop Revenue Coverage.  Similarly, multiple-peril crop insurance
policies and Crop Revenue Coverage policies had levels of insured
production per acre that exceeded the average yield for all farmers
in the particular county by about the same percentage.  The
multiple-peril crop insurance policy units insured yields per bushel
that were 115 percent of the average yield per bushel for all farmers
in the particular county, while Crop Revenue Coverage policy units
had insured yields that were 116 percent of their county's average
yield per bushel. 


APPROACHES USED TO ESTABLISH
PREMIUM RATES MAY NOT ADEQUATELY
PROTECT THE GOVERNMENT FROM
FINANCIAL LOSSES
============================================================ Chapter 4

We identified shortcomings in the way premium rates are established
for each of the revenue insurance plans.  While favorable weather and
stable crop prices generated a very favorable claims experience over
the first 2 years that the plans were available to farmers, these
shortcomings raise questions about whether the rates established for
each plan are actuarially sound over the long term and are
appropriate to the risk each farmer presents.  Furthermore, while the
plans were initially approved on a limited basis only, FCIC approved
the substantial expansion of one of these plans--Crop Revenue
Coverage--before the initial results of claims experience were
available.  Since this initial expansion, FCIC has made and proposed
a number of changes to provide safeguards in its process for
approving new plans. 


   PROBLEMS WITH METHODS USED TO
   ESTABLISH PREMIUMS
---------------------------------------------------------- Chapter 4:1

According to insurance principles, insurance companies need
information on likely future losses in order to establish premium
rates that would cover those losses.  For crop revenue insurance,
reliably projecting future losses requires an accurate depiction of
the revenues that insured farmers are likely to generate.  Premium
rates can then be established on the basis of the probability that
actual revenues will diverge from insured revenues in a given year. 
Such a depiction of revenues for farmers as a whole is commonly
referred to as a revenue distribution. 

Data on individual farmers' actual revenues are not available. 
However, a reasonable approximation of these revenues can be obtained
by multiplying a farmer's yields by crop prices.  In this way, a
simulated revenue distribution can be developed that provides a
reasonable basis for establishing premium rates. 

Crop Revenue Coverage is problematic because it uses neither a
revenue distribution nor another appropriate statistical technique
that takes into account the relationship between prices and yields as
a basis for estimating premiums and future claims payments.  Instead,
rate setting for this plan begins with the premium rate structure for
traditional multiple-peril crop insurance and increases rates by
introducing an additional charge to cover the risk of a price
increase and another charge to cover the risk of revenue that is less
than the guarantee.  By not recognizing the interrelationship between
prices and yields, the premium adjustments may not be actuarially
sound over the long term or appropriate to the risk each farmer
presents.  Thus, we are not able to determine whether premium rates
for this plan are too high or too low. 

In contrast, the rate-setting approaches for Revenue Assurance and
Income Protection are much less problematic because they are based on
revenue distributions, although they use different approaches to
develop these distributions.  We also identified several shortcomings
in these two plans.  However, these shortcomings are less serious
than Crop Revenue Coverage's lack of a revenue distribution or other
statistical technique that takes into account the interrelationship
between prices and yields. 

Revenue Assurance has shortcomings in two respects.  First, in
constructing its revenue distribution, the plan uses only 10 years of
yield data (1985-94), which is not a sufficient historical record to
capture the fluctuations in yield over time.  Furthermore, 3 of the
10 years had abnormal yields:  1988 and 1993 had abnormally low
yields, and 1994 had abnormally high yields.  Second, Revenue
Assurance assumes that the interrelationship between crop prices and
yields is the same in all production areas.  This is not the case. 
That is, the link between yield declines and price increases or yield
increases and price declines is much stronger in some areas than
others.  By using the same estimate of the interrelationship for all
areas, the resulting estimate of claims may be too high in some areas
and too low in others.  As a result, there is no assurance that the
plan's premiums are appropriate for all farmers and will actually
cover all claims over time. 

With respect to Income Protection, the plan's major shortcoming is
that it bases its estimate of future price increases or decreases on
the way that prices moved in the past.\1

This method of developing estimates could be a problem because past
price movements occurred in the context of past government programs,
and in the absence of the government programs, the price movements
may be considerably more pronounced, according to some analysts. 
Instead, price volatility estimates based on commodity futures prices
are more appropriate for forecasting expected claims payments because
they reflect current expectations of the extent to which prices may
increase or decrease between planting and harvest. 

The methods used in the three plans to set premium rates are
described and evaluated in greater detail in appendixes III, IV, and
V. 


--------------------
\1 This shortcoming applies to Crop Revenue Coverage as well. 


   FCIC EXPANDED CROP REVENUE
   COVERAGE PLAN BEFORE INITIAL
   RESULTS WERE AVAILABLE
---------------------------------------------------------- Chapter 4:2

Crop Revenue Coverage was initially approved for sale in December
1995 for two crops--corn and soybeans--in two states--Iowa and
Nebraska.  Given FCIC's lack of experience with revenue insurance and
the uncertainty surrounding the soundness of the premiums charged,
restricting the initial sales to a limited area was prudent. 
However, in July 1996, 7 months after it initially approved Crop
Revenue Coverage and, before it knew the claims experience in these
areas, FCIC's board of directors approved the expansion of Crop
Revenue Coverage to include wheat farmers in Kansas, Michigan,
Nebraska, South Dakota, Texas, Washington State, and 19 counties in
Montana.  This expansion occurred under the board's authority to
approve privately developed insurance products.  The board required
that the companies add a 10-percent surcharge, referred to as a
catastrophic load factor, to the rates initially established.  This
surcharge was not based on the initial experiences in the original
states but was a judgmental adjustment added in response to the
concerns about the adequacy of premium rates expressed by USDA and
university economists. 

In January 1997, the board, acting again within its authority,
expanded Crop Revenue Coverage to essentially cover all major crops
in the major states where the crops are grown.  It was clear at this
time that Crop Revenue Coverage was more popular than had been
initially expected.  National producer organizations expressed strong
interest in expanding the program to additional geographical areas
and to additional crops.  The board expanded Crop Revenue Coverage,
although it was cautioned by USDA officials, USDA's Office of General
Counsel, and USDA's Office of Inspector General about problems with
the continued expansion of the plan.  Specifically, the Administrator
for the Risk Management Agency informed the board that no
underwriting experience was available to evaluate Crop Revenue
Coverage.  He also noted that the amount of liability under the plan
can increase between planting and harvest, thereby increasing crop
insurance liability in a loss situation and potentially having a
major impact on FCIC's overall loss ratio.  However, the
Administrator also pointed out that an expanded program would have
the advantage of giving farmers in most states an additional risk
management tool.  Furthermore, USDA's Office of General Counsel
advised the board to reject expansion because widespread expansion
might expose FCIC to excessive risk in the absence of any data that
could be used to determine whether the rates were actuarially
appropriate.  Finally, USDA's Office of Inspector General cautioned
FCIC several times that expansion was occurring without adequate
controls in place. 

Income Protection and Revenue Assurance have not been significantly
expanded since their introduction. 


   FCIC HAS INITIATED A NUMBER OF
   CHANGES TO BETTER SAFEGUARD THE
   GOVERNMENT'S INTEREST
---------------------------------------------------------- Chapter 4:3

To avoid problems with the introduction of future revenue insurance
plans, USDA is developing new regulations that would require any new
plan to undergo a preapproval review before it could be sold
nationwide that is much more rigorous than the review undertaken for
Crop Revenue Coverage, Revenue Assurance, and Income Protection.  The
draft regulations require that a company proposing a new plan include
a detailed description of the rating method used, simulations of the
performance of the premiums under various scenarios, and the results
of a review by a peer review panel or accredited actuary.  The
regulations also require that the requester provide detailed
information concerning plans for future expansion of the plan. 

Additionally, FCIC has made changes to the gain- and loss-sharing
portions of the reinsurance arrangements with the companies that
better protect the government's interest with respect to the revenue
insurance plans.  For 1998, FCIC decreased the companies' share of
underwriting gains and increased the companies' share of underwriting
losses. 


CONCLUSIONS AND RECOMMENDATION
============================================================ Chapter 5


   CONCLUSIONS
---------------------------------------------------------- Chapter 5:1

With the government's phasing out of income support for farmers, risk
management tools are increasingly important.  Of the available risk
management tools, farmers are increasingly turning to the revenue
insurance plans.  Accordingly, it is important that the premium
structures for the revenue policies be set in a fashion that will be
appropriate to the risk each farmer presents and will protect the
government from undue exposure to loss.  Despite very positive early
underwriting experiences, our analysis indicates that the premium
structures for the three revenue insurance plans have weaknesses in
their underlying assumptions and methods that could result in their
being actuarially unsound.  Crop Revenue Coverage, the plan that has
become the most popular, is the most problematic.  While we
identified some problems in the methods used to set premiums for all
three plans, we found the most serious deficiencies in Crop Revenue
Coverage, which did not base its rates on a revenue distribution or
other appropriate statistical technique that takes into account the
interrelationship between crop prices and yields. 

Apart from its rate-setting deficiencies, Crop Revenue Coverage is
also more costly to the government than the other plans.  Because
Crop Revenue Coverage's premiums are higher, the federal government
pays higher reimbursement costs for administrative expenses and has
higher underwriting losses over time. 


   RECOMMENDATION TO THE SECRETARY
   OF AGRICULTURE
---------------------------------------------------------- Chapter 5:2

To be more certain that the revenue insurance plans are actuarially
sound over the long term and are appropriate to the risk each farmer
presents, we recommend that the Secretary of Agriculture direct the
Administrator of the Risk Management Agency to address the
shortcomings in the methods used to set premiums.  Specifically, with
respect to all three plans, the Secretary should direct the Risk
Management Agency to reevaluate the methods and data used to set
premium rates to ensure that each is based on the most actuarially
sound foundation.  With respect to Crop Revenue Coverage, the Risk
Management Agency should base premium rates on a revenue distribution
or another appropriate statistical technique that recognizes the
interrelationship between farm-level yields and expected prices. 


   AGENCY COMMENTS AND OUR
   EVALUATION
---------------------------------------------------------- Chapter 5:3

In commenting on a draft of this report, USDA expressed concern with
our recommendation that it reevaluate the data and methods used to
set premiums for the three revenue insurance plans.  Specifically,
USDA noted that while it does not necessarily endorse or feel fully
comfortable with all aspects of the rating models, the agency does
not believe our report provides evidence that there are "fatal flaws"
in the plans' rating methods.  Therefore, the Department believes
that the plans' continued use of these rating methods is appropriate. 

We disagree.  While we do not state in this report, nor do we
believe, that the plans contain "fatal flaws," we believe that the
shortcomings we identified in all three revenue insurance plans are
serious enough to warrant a reevaluation of the methods and data used
to set premium rates to ensure that each plan is based on the most
actuarially sound foundation.  This is especially the case for Crop
Revenue Coverage, which does not base its rate structure upon a
distribution of likely revenues from farming operations.  Without a
distribution of likely revenues or other appropriate statistical
technique, the plan does not take into account the interrelationship
between crop prices and yields, and many crop insurance experts agree
that such an interrelationship must be considered.  Thus, we stand by
our recommendation that the Risk Management Agency needs to address
the shortcomings in the rating methods. 

USDA also provided clarifying comments to the report that have been
incorporated where appropriate.  USDA's comments and our responses
are presented in detail in appendix VI. 


METHODOLOGY USED TO ESTIMATE
PREMIUMS AND CLAIMS PAYMENTS TO
FARMERS
=========================================================== Appendix I

This appendix explains the methodology we used to calculate the
premiums and payments for a hypothetical Iowa farmer under
multiple-peril crop insurance, Crop Revenue Coverage, Income
Protection, and Revenue Assurance.  We assumed that the farmer would
plant nonirrigated corn, have a production history of 120 bushels per
acre, and would choose to buy insurance at the 75-percent coverage
level.  This farmer is located in Adair County, Iowa--a county in
which all three revenue insurance policies were available in 1997. 
The prices used in the example are those that were established by the
Federal Crop Insurance Corporation (FCIC) for each plan for 1997. 
The examples of claims payments assume various combinations of
30-percent increases and decreases in prices and production levels. 
We chose these percentages to illustrate the operation of the various
insurance plans.  Other combinations of changes in prices and/or
production levels would produce different results. 


   ESTIMATING PREMIUMS
--------------------------------------------------------- Appendix I:1

To purchase traditional multiple-peril crop insurance, our
hypothetical Iowa corn farmer chose basic unit coverage and insured
at 100 percent of the crop price available for 1997 ($2.45).  Given
our assumptions, the farmer would have paid $11.20 per acre for
traditional multiple-peril crop insurance.  For Crop Revenue
Coverage, our hypothetical farmer also selected basic unit coverage. 
The projected crop price for Crop Revenue Coverage in 1997 was $2.59
per bushel for corn.  On the basis of our assumptions, we determined
that the farmer choosing Crop Revenue Coverage would have paid $16.50
per acre in 1997.  For Revenue Assurance, with a projected price of
$2.38 per bushel for corn, this same farmer would have paid $8.40 per
acre.  The Income Protection price we used for our estimate was $2.73
per bushel for corn, and we determined that the farmer would have
paid premiums of $5.90 per acre in 1997. 


   ESTIMATING PAYMENTS IN THE
   EVENT OF NORMAL PRODUCTION IN
   COMBINATION WITH DECLINING
   PRICES
--------------------------------------------------------- Appendix I:2

In the event of normal production combined with a 30-percent decline
in price, no payment would be due under the traditional
multiple-peril policy, but each of the revenue insurance policies
would provide payments.  No payment would be due under the
traditional multiple-peril policy because, by definition, it only
pays when the farmer's production falls below the guarantee, which in
the case of the 75-percent coverage level, would be 75 percent of 120
bushels, or 90 bushels. 

If the farmer purchased Crop Revenue Coverage, the revenue guarantee
would be the 75-percent coverage level multiplied by the normal
production of 120 bushels, and the resulting production multiplied by
the higher of the projected price ($2.59 per bushel in 1997) or the
harvest price ($1.81 if the price declined 30 percent).  The
guarantee under these conditions would be $233.10 (.75 x 120 x $2.59
= $233.10).  The guarantee is then compared with the value of the
farmer's harvested crop, determined by multiplying the actual
production by the harvest price (120 x $1.81 = $217.20).  Thus, in
the case of normal production combined with a 30-percent decline in
price, the farmer who obtained Crop Revenue Coverage would receive a
payment of $15.90 per acre ($233.10 - $217.20 = $15.90). 

If, instead, the farmer had purchased an Income Protection policy,
the revenue guarantee would be determined by multiplying the coverage
level (.75) by the normal production (120 bushels), and multiplying
the resulting production by the projected price ($2.73 per bushel in
1997).  The per-acre guarantee under these conditions would be
$245.70 (.75 x 120 x $2.73 = $245.70).  The policy bases the payment
on the difference between this guarantee and the $229.20 per-acre
value of the farmer's crop--determined by multiplying the actual
production (120 bushels per acre) by the harvest price ($1.91 if the
price declined 30 percent).  Thus, in the case of normal production
combined with a 30-percent decline in the price, the per-acre payment
for the farmer who purchased Income Protection would be $16.50
($245.70 - $229.20 = $16.50). 

If the farmer had purchased a Revenue Assurance policy instead, the
revenue guarantee would be determined by multiplying the coverage
level (.75) by the normal production (120 bushels), and multiplying
the resulting production by the projected county price.  The price
varies by county, depending on the extent to which the price in the
county has tended to be higher or lower than the price on the
national commodity market.  For Adair County in 1997 for corn, the
projected county price was $2.38 ($2.73 per bushel national price in
1997 minus $0.35 county adjustment = $2.38).  The per-acre guarantee
under these conditions would be $214.20 (.75 x 120 x $2.38 =
$214.20).  The policy bases the payment on the difference between
this guarantee and the per-acre value of the farmer's crop
($200.40)--determined by multiplying the actual production (120
bushels per acre) by the harvest price ($1.67 if the price declined
30 percent).  Thus, in the case of normal production combined with a
30-percent decline in price, the per-acre payment for the farmer who
purchased Revenue Assurance would be $13.80 ($214.20 - $200.40 =
$13.80). 


   ESTIMATING PAYMENTS IN THE
   EVENT OF REDUCED PRODUCTION IN
   COMBINATION WITH DECLINING
   PRICES
--------------------------------------------------------- Appendix I:3

In the event of both a 30-percent decline in production and a
30-percent decline in price, each type of policy would pay, but the
amounts paid would vary.  The traditional policy pays on the basis of
a decline in production, while the revenue policies pay on the basis
of a decline in gross revenue. 

The traditional multiple-peril crop insurance policy pays when the
farmer's production falls below the guarantee, which in the case of
the 75-percent coverage level, would be 75 percent of 120 bushels, or
90 bushels.  If the farmer purchasing this policy experienced a
30-percent reduction in production, production would average 84
bushels per acre (70 percent of 120).  Thus, the farmer would be paid
for a reduction of 6 bushels per acre (90 - 84 = 6).  The actual
price prevailing at harvest does not affect the payment under the
traditional policy.  Assuming the farmer had selected the 100-percent
price option, the payment would be made at $2.45 per bushel (the
price election announced by the U.S.  Department of Agriculture prior
to the 1997 crop insurance sales period), although national prices
had declined to $1.72 in this example.  Thus, in the case of a
30-percent reduction in production combined with a 30-percent decline
in price, the farmer who obtained traditional multiple-peril crop
insurance would receive a payment of $14.70 per acre (6 bushels x
$2.45 = $14.70). 

If the same farmer had purchased a Crop Revenue Coverage policy
instead, the revenue guarantee would be the 75-percent coverage level
multiplied by the normal production of 120 bushels, and the resulting
production multiplied by the higher of the projected price ($2.59 per
bushel in 1997) or the harvest price ($1.81 if prices declined 30
percent).  The guarantee under these conditions would be $233.10 (.75
x 120 x $2.59 = $233.10).  The guarantee is then compared with the
value of the farmer's harvested crop, determined by multiplying the
actual production by the harvest price (84 bushels x $1.81 =
$152.04).  Thus, in the case of a 30-percent reduction in production
combined with a 30-percent decline in price, the farmer who obtained
Crop Revenue Coverage would receive a payment of $81.06 per acre
($233.10 - $152.04 = $81.06). 

If the same farmer had purchased an Income Protection policy instead,
the revenue guarantee would be determined by multiplying the coverage
level (.75) by the normal production (120 bushels), and multiplying
the resulting production by the projected price ($2.73 per bushel in
1997).  The per-acre guarantee under these conditions would be
$245.70 (.75 x 120 x $2.73 = $245.70).  The policy bases the payment
on the difference between this guarantee and the per-acre value of
the farmer's crop ($160.44)--determined by multiplying the actual
production (84 bushels per acre) by the harvest price ($1.91 if the
price declined 30 percent).  Thus, in the case of 30-percent
reduction in production combined with a 30-percent decline in price,
the per-acre payment for the farmer who purchased Income Protection
would be $85.26 ($245.70 - $160.44 = $85.26). 

If the same farmer had purchased a Revenue Assurance policy instead,
the revenue guarantee would be determined by multiplying the coverage
level (.75) by the normal production (120 bushels), and multiplying
the resulting production by the projected county price.  The price
varies by county, depending on the extent to which prices in the
county have tended to be higher or lower than the prices on the
national commodity market.  For Adair County in 1997 for corn, the
projected county price was $2.38 ($2.73 per bushel national price in
1997 minus $0.35 county adjustment = $2.38).  The per-acre guarantee
under these conditions would be $214.20 (.75 x 120 x $2.38 =
$214.20).  The policy bases the payment on the difference between
this guarantee and the $140.28 per-acre value of the farmer's
crop--determined by multiplying the actual production (84 bushels per
acre) by the harvest price ($1.67 if prices declined 30 percent). 
Thus, in the case of a 30-percent reduction in production combined
with a 30-percent decline in price, the per-acre payment for the
farmer who purchased Revenue Assurance would be $73.92 ($214.20 -
$140.28 = $73.92). 


   ESTIMATING PAYMENTS IN THE
   EVENT OF REDUCED PRODUCTION IN
   COMBINATION WITH INCREASING
   PRICES
--------------------------------------------------------- Appendix I:4

In the event of a decline in production combined with an increase in
price, the traditional policy and the Crop Revenue Coverage policy
would result in payments, but no payment would result under the terms
of the Income Protection and Revenue Assurance policies. 

Because the harvest price has no effect on the payment under the
traditional crop insurance policy, the claim payment for a farmer
with a 30-percent decline in production in combination with a
30-percent increase in price would be the same as the payment under
constant or decreasing prices ($14.70 as calculated in the previous
section). 

If the same farmer had purchased Crop Revenue Coverage instead, the
revenue guarantee would be the 75-percent coverage level multiplied
by the normal production of 120 bushels, and the resulting production
would be multiplied by the higher of the projected price ($2.59 per
bushel in 1997) or the harvest price ($3.37 if the price increased 30
percent).  The guarantee under these conditions would be $303.30 (.75
x 120 x $3.37 = $303.30).  The guarantee is then compared with the
value of the farmer's harvested crop, determined by multiplying the
actual production by the harvest price (84 bushels x $3.37 =
$283.08).  Thus, in the case of a 30-percent reduction in production
combined with a 30-percent increase in price, the farmer who obtained
Crop Revenue Coverage would receive a per-acre payment of $20.22
($303.30 - $283.08 = $20.22). 

If the same farmer had purchased an Income Protection policy instead,
no payment would be due because the value of the harvested crop would
exceed the revenue guarantee.  The revenue guarantee would be
determined by multiplying the coverage level (.75) by the normal
production (120 bushels), and multiplying the resulting production by
the projected price ($2.73 per bushel in 1997).  The per-acre
guarantee under these conditions would be $245.70 (.75 x 120 x $2.73
= $245.70).  No payment would be due because this guarantee is less
than the per-acre value of the farmer's crop ($298.20)--determined by
multiplying the actual production (84 bushels per acre) by the
harvest price ($3.55 if prices increased 30 percent).  Thus, in the
case of a 30-percent reduction in production combined with a
30-percent increase in price, no insurance payment would be made to
the farmer who purchased Income Protection ($245.70 - $298.20 =
-$52.50--thus, no payment is due). 

Similarly, if the same farmer had instead purchased a Revenue
Assurance policy, no insurance payment would be due because the value
of the harvested crop would exceed the revenue guarantee.  The
revenue guarantee would be determined by multiplying the coverage
level (.75) by the normal production (120 bushels), and multiplying
the resulting production by the projected county price.  The price
will vary by county, depending on the extent to which the price in
the county has tended to be higher or lower than the prices on the
national commodity market.  For Adair County in 1997 for corn, the
projected county price was $2.38 ($2.73 per bushel national price in
1997 minus $0.35 county adjustment = $2.38).  The per-acre guarantee
under these conditions would be $214.20 (.75 x 120 x $2.38 =
$214.20).  No payment would be required because the guarantee is less
than the $259.56 per-acre value of the farmer's crop--determined by
multiplying the actual production (84 bushels per acre) by the
harvest price ($3.09 if prices increased 30 percent).  Thus, in the
case of a 30-percent reduction in production combined with a
30-percent increase in price, no insurance payment would be made to
the farmer who purchased Revenue Assurance ($214.20 - $259.56 =
-$45.36--thus, no payment is due). 


CROP INSURANCE EXPERIENCE, 1997
========================================================== Appendix ii

The tables in this appendix show crop insurance results for 1997 for
traditional multiple-peril crop insurance (MPCI), Insurance
Protection (IP), Crop Revenue Coverage (CRC), and Revenue Assurance
(RA).  Table II.1 shows various sales and claims payments experience,
by state and by insurance plan.  Table II.2 combines all states to
show sales and claims payments experience by insurance plan only. 



                                        Table II.1
                         
                          Crop Insurance Experience by State and
                                 by Insurance Plan, 1997

                          (Policies in force, acres insured, and
                                  dollars in thousands)

                                                           Government
                 Insuranc  Policies     Acres       Total     premium    Claims      Loss
State            e plan\a  in force   insured    premiums     subsidy  payments     ratio
---------------  --------  --------  --------  ----------  ----------  --------  --------
Alabama          MPCI             6       828     $14,524      $7,374   $36,753      2.53
                 IP             0\b         3          99          42         0      0.00
Alaska           MPCI           0\b         8          21          21        40      1.90
Arizona          MPCI             1       428       5,677       3,698     3,633      0.64
                 CRC            0\b         5         238          48        83      0.35
Arkansas         MPCI            17     3,686      32,248      27,576     8,163      0.25
                 IP             0\b       116         996         552       344      0.35
California       MPCI            20     2,884      82,478      54,540    17,869      0.22
Colorado         MPCI            16     3,357      27,801      14,960    17,569      0.63
                 CRC              1       242       3,885       1,141       589      0.15
Connecticut      MPCI           0\b        22         790         478       768      0.97
Delaware         MPCI             1       207         987         659     1,078      1.09
Florida          MPCI             1       206       6,148       3,353     5,552      0.90
Georgia          MPCI            11     1,950      33,491      21,205    13,185      0.39
                 CRC            0\b        32       1,520         515     1,490      0.98
                 IP             0\b         6         320          80         0      0.00
Hawaii           MPCI           0\b        13         108          81         0      0.00
Idaho            MPCI             6     1,102      11,086       5,603     5,834      0.53
Illinois         MPCI           110    11,209      67,926      34,466    12,683      0.19
                 CRC             10     1,252      16,384       4,004     4,167      0.25
                 IP               2       197         978         339        66      0.07
Indiana          MPCI            34     4,382      30,659      13,469    18,939      0.62
                 CRC              4       679       9,153       1,898     3,663      0.40
                 IP             0\b        18          67          28        16      0.23
Iowa             MPCI           101    11,068      63,304      29,003     4,791      0.08
                 CRC             43     5,534      64,474      16,192     6,876      0.11
                 IP             0\b         4          18           6         1      0.04
                 RA               9     1,185       8,061       2,779       494      0.06
Kansas           MPCI           113    11,175      70,040      34,989    13,044      0.19
                 CRC             17     2,929      31,104       9,597     7,707      0.25
                 IP             0\b         4           6           2         0      0.00
Kentucky         MPCI             6       963       7,197       4,478     3,465      0.48
Louisiana        MPCI            11     2,575      27,021      22,423     6,949      0.26
Maine            MPCI             1        83       2,402       1,505       885      0.37
Maryland         MPCI             3       479       3,107       1,885     6,389      2.06
Massachusetts    MPCI             1        29       1,606         951       255      0.16
Michigan         MPCI            17     2,180      20,042      13,803     4,271      0.21
                 CRC              2       296       4,401       1,397     1,686      0.38
Minnesota        MPCI            81    12,123     106,790      52,208    49,526      0.46
                 CRC             14     2,839      34,481      10,369    12,179      0.35
                 IP             0\b        33         179          66       163      0.91
Mississippi      MPCI             9     3,287      28,941      22,353     9,586      0.33
Missouri         MPCI            47     4,640      38,282      25,507     6,983      0.18
                 CRC              4       536       8,778       2,901     2,697      0.31
Montana          MPCI            19     6,137      36,427      16,032    12,913      0.35
                 CRC              1       301       2,380         774       642      0.27
Nebraska         MPCI            73     7,659      54,414      25,968    22,324      0.41
                 CRC             32     4,271      48,232      13,957    17,558      0.36
Nevada           MPCI           0\b        17         135          91         0      0.00
New Hampshire    MPCI           0\b         9         140          94         0      0.00
New Jersey       MPCI             1        96       1,144       1,058       480      0.42
New Mexico       MPCI             2       580       4,752       3,204       880      0.19
New York         MPCI             4       474       3,818       3,197       998      0.26
North Carolina   MPCI            15     2,161      20,427      13,539    10,622      0.52
North Dakota     MPCI            82    17,386     122,807      58,140   148,928      1.21
                 CRC              2       710       4,727       1,856     5,748      1.22
                 IP             0\b       0\b           2           1         4      1.72
Ohio             MPCI            29     3,069      16,899       8,940     6,791      0.40
                 CRC              3       418       5,466       1,341     1,874      0.34
Oklahoma         MPCI            26     4,837      28,457      15,066    15,494      0.54
                 CRC            0\b        67         978         329       343      0.35
Oregon           MPCI             3       715       4,000       2,180       842      0.21
Pennsylvania     MPCI             6       553       5,077       3,241     8,455      1.67
Rhode Island     MPCI           0\b         1          22          13        22      1.02
South Carolina   MPCI             5       872       9,133       7,055     3,510      0.38
South Dakota     MPCI            59     9,160      68,558      33,734    68,584      1.00
                 CRC             12     2,369      26,375       8,503    20,102      0.76
Tennessee        MPCI             4       810       7,454       5,932     2,524      0.34
Texas            MPCI            90    13,154     205,498     107,559   114,110      0.56
                 CRC              5     1,135      17,080       5,884    12,827      0.75
                 IP             0\b         3          11           4        30      2.70
Utah             MPCI             1       108         764         433       379      0.50
Vermont          MPCI           0\b        44         239         216       204      0.85
Virginia         MPCI             5       687       7,403       3,882    11,696      1.58
Washington       MPCI            11     2,175      18,205      11,117     4,885      0.27
                 CRC            0\b       125       1,046         215       148      0.14
West Virginia    MPCI             1        52         744         457       769      1.03
Wisconsin        MPCI            30     3,032      29,042      16,806     4,892      0.17
Wyoming          MPCI             3       351       2,821       1,200     1,383      0.49
=========================================================================================
Total\c                       1,239   178,334  $1,622,492    $820,561  $791,390      0.49
-----------------------------------------------------------------------------------------
\a MPCI includes the group risk plan. 

\b Rounds to less than 1. 

\c Total excludes special plans that cover peanuts, tobacco, fruit
trees, and various minor crops. 

Source:  GAO's analysis of the U.S.  Department of Agriculture's
(USDA) data. 



                                        Table II.2
                         
                            U.S. Crop Insurance Experience by
                                   Insurance Plan, 1997

                          (Policies in force, acres insured, and
                                  dollars in thousands)

                                                         Government
           Insuranc    Policies       Acres       Total     premium      Claims      Loss
U.S.       e plan\a    in force     insured    premiums     subsidy    payments     ratio
---------  --------  ----------  ----------  ----------  ----------  ----------  --------
           MPCI           1,080     153,023  $1,331,053    $735,743    $689,893      0.52
           CRC              148      23,743     280,701      80,919     100,379      0.36
           IP                 3         383       2,677       1,119         624      0.23
           RA                 9       1,185       8,061       2,779         494      0.06
=========================================================================================
Total\b                   1,239     178,334  $1,622,492    $820,561    $791,390      0.49
-----------------------------------------------------------------------------------------
\a MPCI includes the group risk plan. 

\b Total excludes special plans that cover peanuts, tobacco, fruit
trees, and various minor crops. 

Source:  GAO's analysis of USDA's data. 


METHODOLOGY USED TO SET PREMIUM
RATES FOR CROP REVENUE COVERAGE
AND OUR ANALYSIS
========================================================= Appendix III

The Crop Revenue Coverage plan was developed by a private insurance
company in the early 1990s.  The plan is designed to guarantee
farmers (1) a certain level of income and (2) the replacement value
of the difference between insured yields and actual yields if actual
yields are below the insured level.  Crop Revenue Coverage's premiums
are based on three components:  "yield risk," "upward price risk,"
and "revenue risk." Premiums calculated for each of the components
are added together to generate the total premium for each policy. 

This appendix defines each component and explains how it is
developed.  The first section describes the calculation of the yield
risk component of the premium, which is based on the multiple-peril
crop insurance program.  The second section describes the calculation
of the upward price risk component, which refers to the expected
payout by the insurer as a result of a yield loss and a price
increase between planting (insurance sales period) and harvest.  The
third section shows how the revenue risk component is developed,
which is the risk that, if prices are lower at harvest than at
planting, actual revenue is less than guaranteed revenue.  The fourth
section demonstrates how the three components are summed to form a
base premium.  In these calculations, yields and prices are treated
as if they are independent of one another.  Finally, we present our
analysis of the method used to set premiums for Crop Revenue
Coverage. 


   CALCULATING THE YIELD RISK
   COMPONENT
------------------------------------------------------- Appendix III:1

For Crop Revenue Coverage, yield risk relates to situations in which
the actual yield is lower than the insured yield and the price at
harvest is not higher than the price guaranteed at planting.  Through
yield risk coverage, the insured farmer is eligible for a payment
equivalent to the difference between the insured yield and the actual
yield, multiplied by the planting price.  The portion of the premium
related to yield risk is derived from the premium rate schedules for
multiple-peril crop insurance.  The yield risk accounts for
two-thirds of the expected payout by the insurer. 

The yield risk premium is the product of the multiple-peril crop
insurance base rate, the farmer's actual production history (APH),
the coverage level, and the planting price or: 

Yield risk premium = MPCI Base Rate x APH x Coverage Level x Planting
Price

Equation 1 estimates the portion of the premium that relates to yield
risk: 

(1)



(See figure in printed edition.)

where PRy is the calculated premium, R is the multiple-peril crop
insurance base rate, Yg is the insured yield, Pp is the planting
price, and EL is expected yield loss.  The premium is not exactly
equivalent to the product of the planting price and expected losses
because expected losses for each farm can only be approximated. 

Multiple-peril crop insurance base rates are derived from historical
losses relative to historical premiums for various yield and coverage
levels.\1 The relevant market price is equal to 95 percent of the
average closing price of the harvest period's futures contract price
during the planting period.  The expected yield loss equals the
multiple-peril crop insurance base rate multiplied by the yield
guarantee to the farmer. 


--------------------
\1 The multiple-peril crop insurance base premium rate is derived
from the ratio of historical claims payments to historical
liabilities (loss-cost method) at the midpoint of FCIC's nine premium
rate levels--called "R-spans"--for the 65-percent coverage level. 
From that midpoint, or fifth rate span--called R05--the spans are a
series of levels of rates that decrease as yield levels increase and
increase as yield levels decrease. 


   CALCULATING THE UPWARD PRICE
   RISK COMPONENT
------------------------------------------------------- Appendix III:2

Upward price risk is a component developed especially for Crop
Revenue Coverage.  It refers to the risk of a higher price at harvest
than at planting when the actual yield is lower than the insured
yield.  Under the upward price risk component, the insured farmer is
eligible for a payment equal to the difference between the insured
and actual yields multiplied by the harvest price. 

The total upward price risk equals the product of the multiple-peril
crop insurance base rate, the farmer's APH, the coverage level, and
the upward price factor (which is the product of the upward price
rate times the maximum liability for that crop) or: 

Upward Price Risk = MPCI Base Rate x APH x Coverage Level x Upward
Price Factor

Equations 2 through 10 are used to estimate the upward price factor,
that is, the risk of prices increasing between planting and harvest,
when the farmer has a loss in yield. 

Equation 2 estimates a premium rate for a yield loss by dividing
expected crop losses by the yield guarantee: 

(2)



(See figure in printed edition.)

where R is the insurance premium rate, EL is expected loss, and Yg is
the yield guarantee. 

Equation 3 integrates the price distribution above the planting price
in order to estimate expected loss from an upward price change: 

(3)



(See figure in printed edition.)

where EL is the expected loss in dollars, Pp is the planting price,
Pa is the actual price, and f (P) is the probability density function
for price changes.  The function is constrained by the maximum price
difference reimbursable for each insured crop.\2

In order to facilitate the estimation of expected losses, Crop
Revenue Coverage uses the polynomial function for the integration of
a normally distributed probability distribution from Abramowitz and
Stegun\3 (Equations 4 and 5) along with a procedure developed by
Botts and Boles\4 (Equations 6 through 9).  The Botts and Boles
procedure estimates the mean of a truncated normal distribution, one
in which a portion is cut off and isolated for analysis.  The
truncated distribution is bounded by the maximum compensated price
change and the mean of price changes for the entire normal
distribution.  This is the portion of the price distribution that
reflects prices above the planting price. 

Equation 4 estimates the probability of a loss (or in this case the
probability of an upward price change) using the polynomial function
for integration of a normal distribution: 

(4)



(See figure in printed edition.)

where P is the probability that the insurer would be required to pay
insured farmers under the "upward price risk" provision of Crop
Revenue Coverage.  Moreover, a1, a2, and a3 are constants from
Abramowitz and Stegun.  Variables, Z and T are estimated in Equations
5 and 6. 

Equation 5 estimates the value of T, which measures the area under a
normal curve: 

(5)



(See figure in printed edition.)

where b is a parameter of the price distribution from Abramowitz and
Stegun. 

Equation 6 estimates Z, which is the height (measured parallel to the
Y axis) of the ordinate of the truncated distribution: 

(6)



(See figure in printed edition.)

where EP is the expected or mean price change of the entire
distribution, Pp is the planting price, and SD is the standard
deviation of price changes for the entire normal distribution. 

Equation 7 estimates M, the mean of the truncated normal
distribution, in this case the mean of the distribution of upward
price changes: 

(7)



(See figure in printed edition.)

where EP is the mean price for the entire normal distribution
(untruncated), Z is as defined above, P is the probability of a price
change above the guaranteed price, and SD is the standard deviation
of price changes for the entire distribution. 

Equation 8 estimates expected losses: 

(8)



(See figure in printed edition.)

where M is the mean of price changes for a truncated normal
distribution, P is the probability of a loss and Pp is the planting
price. 

Substituting Equation 7 into Equation 8 gives Equation 9, which
expresses expected loss per bushel: 

(9)



(See figure in printed edition.)

Equation 10 (as in Equation 2 for a yield loss) expresses the premium
rate per bushel for an upward price change as the result of dividing
expected losses per bushel by the planting price: 

(10)



(See figure in printed edition.)

The premium rate calculated above, however, must be adjusted to
reflect Crop Revenue Coverage regulations, which require payment for
price increases under conditions of actual yield losses only.  In
order to account for this feature of the program, a conditional
probability, that is, the probability of a price increase given a
yield loss must be calculated.  In order to calculate a premium rate
for this factor (R adjusted for the probability of a price increase,
given a yield loss), the unadjusted R (as in Equation 10) is
multiplied by the multiple-peril crop insurance base rate for yield
loss. 


--------------------
\2 For example, the maximum price increase that is insured for corn
is $1.50 per bushel. 

\3 Abramowitz, Milton, and Irene Stegun, Handbook of Mathematical
Functions (Washington, D.C.:  Government Printing Office, 1968). 

\4 Botts, Ralph R., and James N.  Boles, "Use of Normal-Curve Theory
in Crop Insurance Rate Making," Journal of Farm Economics 39 (1957): 
733-40.  The Botts and Boles method is a general approach that
assumes it is possible to estimate expected insurers' indemnity
payments once the values of the mean and standard deviation of a
normal distribution of the insured factor are known.  Botts and Boles
apply the method to yields in their article.  For Crop Revenue
Coverage, the method is applied to prices and yields. 


   CALCULATING THE REVENUE RISK
   COMPONENT
------------------------------------------------------- Appendix III:3

Revenue risk refers to the risk of harvest revenue that is lower than
the revenue guaranteed at planting.  Guaranteed revenue is the
product of the insured yield and the planting price.  Under the
revenue risk component, as long as harvest revenue is lower than
guaranteed revenue, the insured farmer is eligible for a payment. 
When the harvest price is lower than the planting price, harvest
revenue can be lower than guaranteed revenue when yield is at or
above the insured yield or when yield is lower than the insured
level.\5

The revenue risk factor is the product of the revenue rate,\6 the
farmer's actual production history, the coverage level, and the
downward price factor (which is the downward price rate times the
maximum liability for that crop) or: 

Revenue Risk = Revenue Rate x APH x Coverage Level x Downward Price
Factor. 

In order to calculate the revenue risk, Crop Revenue Coverage
estimates two factors:  the downward price factor and the revenue
rate.  The downward price factor is calculated using the same method
as the upward price factor, but here the risk evaluated is that
prices will be lower at harvest than at planting.  The revenue rate
is derived from the area under the yield curve below the yield
guarantee, given a price decline.  The revenue rate must cover the
risk, when price declines, of harvest revenue that is less than the
planting revenue guarantee.  The revenue rate does not cover the risk
of a yield loss, because the yield risk factor compensates for that
by paying the insured farmer the product of the yield loss and the
planting price.  However, the revenue rate must cover the risk of the
guaranteed revenue being higher than the sum of market revenue and
payments under the yield component.  For a given price decline, the
largest such payout under the revenue rate would occur at the yield
guarantee, when no payments are made under the yield risk component. 
Alternatively, the greatest payment under the yield risk component
would occur at zero production, when no payment is made under the
revenue risk component. 

The Crop Revenue Coverage base rate is calculated in six steps. 
First, a mean yield and standard deviation are calculated by county,
by crop, and by farming practice using data on APH and multiple-peril
crop insurance base rates.  Second, using these data, a yield curve
is generated.  Third, using the polynomial function for the
integration of a normally distributed density function (Abramowitz
and Stegun), the area under the curve below the yield guarantee is
calculated to obtain the probability of collecting indemnities, given
a price decline.  Fourth, the expected yield loss is calculated using
the Botts and Boles method.  Fifth, this expected loss is subtracted
from the yield guarantee because this part of the yield loss is
already covered by the multiple-peril crop insurance or "yield risk"
portion of the Crop Revenue Coverage premium.  Sixth, the expected
yield is divided by the mean yield and multiplied by the probability
of collecting indemnities in any given year, given a price decline. 

In steps 1 and 2 above, the yield curves are generated by using the
mean and standard deviations of yield that are derived from the Risk
Management Agency's published APH and base rate data. 

In the third step, Equations 11, 12, and 13 calculate the area
underneath the yield curve between 0 and the yield guarantee, or the
probability, P, of an indemnity being paid, given a price decline: 

(11)



(See figure in printed edition.)

(12)



(See figure in printed edition.)

(13)



(See figure in printed edition.)

where a1 , a2, a3, and b are constants, P is the probability of
collecting indemnities in any given year, T measures the area
underneath the normal curve, Z measures the ordinate between the
x-axis and the normal curve, y is the mean yield of the
distribution, yg is the guaranteed yield, and SD is the standard
deviation of yields. 

In the fourth step, Equation 14, the expected yield loss, EL, is
calculated: 

(14)



(See figure in printed edition.)

In the fifth step, Equation 15, the expected yield, EY, is calculated
by subtracting the expected loss, EL, from the yield guarantee:\7

(15)



(See figure in printed edition.)

In the sixth step, Equation 16, the revenue rate is obtained by
multiplying the ratio of the expected yield to the mean yield by the
probability, P, of the farmer collecting an indemnity from a price
decline: 

(16)



(See figure in printed edition.)

There is no provision in this rate for the possibility that yields
could be above the mean while prices are declining, triggering an
indemnity. 


--------------------
\5 Crop Revenue Coverage does not guarantee price.  If prices fall
but yields do not decline, in most cases there will be no indemnity
payments.  For example, the 1997 wheat price fell by nearly 50 cents
per bushel during the insurance period, but most Kansas growers
received no Crop Revenue Coverage indemnity payments because they had
sufficient yields to offset the revenue loss caused by lower prices. 

\6 The revenue rate is also called the Crop Revenue Coverage base
rate. 

\7 The expected yield loss is subtracted because it is already
covered under the yield risk component. 


   MAKING THE FINAL CALCULATION
   FOR THE CROP REVENUE COVERAGE
   PREMIUM
------------------------------------------------------- Appendix III:4

The calculation of the total Crop Revenue Coverage base premium,
before subsidy, is the sum of the following three products: 

  -- Yield risk premium = MPCI Base Rate x APH x Coverage Level x
     Planting-Period Price

  -- Upward price risk premium = MPCI Base Rate x APH x Coverage
     Level x Upward Price Factor, and

  -- Revenue risk premium = Revenue Rate x APH x Coverage Level x
     Downward Price Factor. 


   ANALYSIS OF CROP REVENUE
   COVERAGE
------------------------------------------------------- Appendix III:5

Crop Revenue Coverage differs significantly in its rate-setting
method from the two other insurance plans.  Unlike the methods used
for Income Protection and Revenue Assurance, the method used to
establish premiums for Crop Revenue Coverage is not based on a
revenue distribution or another appropriate statistical technique. 
Instead, Crop Revenue Coverage establishes rates by adding together
yield, upward price, and revenue risk factors.  The yield risk
component is based on rates established under traditional
multiple-peril crop insurance.  The upward price risk component is
used to estimate losses to the insurer in the case of a price
increase, given a yield loss.  The revenue risk component is used to
estimate losses to the insurer of harvest revenue that is lower than
the revenue guaranteed in the planting period.  Using this additive
procedure, the private insurance company developer assumed that price
and yield are independent of each other and derived them separately. 

However, the price-yield correlation is needed to help establish
premium rates that are not too high to discourage participation or
too low to cover losses.  This correlation would be greatest in
concentrated production areas, such as the midwestern cornbelt, where
the price-yield correlation is highest, and decline as the distance
from these areas increases because the price-yield correlation
decreases the further production for corn is from the central area. 

Analysts disagree about the impact of omitting the correlation
between price and yield.  Some have suggested that omitting this
correlation may not be as serious a shortcoming as might be expected. 
Although the price-yield relationship is an important component of
revenue distributions, especially for major crop production areas,
Crop Revenue Coverage premiums, on average, may still be appropriate
to cover losses over time, according to these analysts.  This is
because, although the rate for price increases (upward price risk)
may be too low and the rate for price decreases (revenue rate) may be
too high, they may offset each other.  However, other analysts point
out that there is no evidence that the failure to incorporate the
price-yield correlation has a neutral effect on premiums.  They say
that government outlays in years of very low yields could be
extensive because the plan understates the probability of a yield
loss when prices increase. 


METHODOLOGY USED TO SET PREMIUM
RATES FOR REVENUE ASSURANCE AND
OUR ANALYSIS
========================================================== Appendix IV

In response to the Iowa Farm Bureau's proposal that federal
deficiency payments be replaced with a federally subsidized insurance
product, Revenue Assurance was developed to provide a payment to
insured farmers when farm revenues fall below a predetermined trigger
level.  The payment is the difference between the trigger, or
guaranteed, revenue and the actual revenue. 

In order to develop premiums that will likely cover future losses
over time, insurers need to accurately depict a revenue distribution,
or use another appropriate statistical technique, to reflect receipts
at the farm level.  Three primary steps are essential to determining
the revenue distribution--developing the price distribution,
developing the yield distribution, and estimating the price-yield
correlation. 

The first section of this appendix describes how the price
distribution, using futures prices adjusted for local differentials,
is calculated for Revenue Assurance.  The second section describes
how the yield distribution is estimated.  Certain parameters are
imposed on the price distribution and on the yield distribution.  The
third section shows how the price and yield distributions are
combined to form a revenue distribution that incorporates a
price-yield relationship.  The fourth section shows how expected
losses are used to calculate premiums.  Finally, we present our
analysis of the methodology used to set premium rates for Revenue
Assurance. 


   DEVELOPING THE PRICE
   DISTRIBUTION
-------------------------------------------------------- Appendix IV:1

Current prices, which have the advantage of reflecting current market
conditions, are used for developing price distributions for Revenue
Assurance.  The premiums are based on the prices set during planting
for futures prices during the harvest period, adjusted for local
conditions.  Following an analysis of the responsiveness of cash
prices to changes in futures prices, the difference between futures
and cash prices for each county was found to be constant over time. 

Equation 1 uses current futures price and price volatility to
estimate a lognormal price distribution, F(P): 

(1)



(See figure in printed edition.)

where  and  are the parameters of the lognormal
price distribution.  The current price used is the average of the
planting period price of the harvest period futures contract.  The
price volatility used is calculated by applying the Black options
pricing formula to the price of the planting period put option on the
harvest period futures contract. 


   YIELD DISTRIBUTION DEVELOPMENT
-------------------------------------------------------- Appendix IV:2

Revenue Assurance assumes that crop yields follow a beta
distribution.  The beta distribution exhibits three major
characteristics:  First, it can exhibit negative or positive
skewness; second, it has finite minimum and maximum values; and
third, it can take on a wide variety of shapes. 

Equation 2 describes the beta distribution of yields, y, as: 

(2)



(See figure in printed edition.)



(See figure in printed edition.)

where p, q, ymax ,and ymin are the four parameters and
(p+q), (p), and (q) refer to the gamma
function of (p+q), p, and q, respectively, which is directly related
to the beta distribution. 

Equations 3 and 4 estimate the values of p and q using the method of
moments technique:\1

(3)



(See figure in printed edition.)

(4)



(See figure in printed edition.)

where  is the mean of yield and  is the standard
deviation of yield for each county, ymax is the maximum, and ymin is
the minimum yield.  P is from equation 3.  The mean yield, ,
is derived from a discrete range of the farmers expected yields.  The
maximum and minimum yields determine the degree and direction of
skewness and of kurtosis.\2


--------------------
\1 The method of moments is a technique that uses a moment generating
function, (t), to generate all of the moments, such as the
mean or the first moment, of a random variable X and its probability
distribution.  Using (t), all of the moments of X can be
obtained by successively differentiating (t). 

\2 A distribution that lacks symmetry with respect to a vertical axis
is said to be skewed or to have skewness.  Kurtosis has to do with
the degree of peakedness that the distribution exhibits:  how steeply
the curve rises and falls. 


   DEVELOPING THE REVENUE
   DISTRIBUTION
-------------------------------------------------------- Appendix IV:3

Using the Johnson and Tenenbein approach,\3 Revenue Assurance
estimates a revenue distribution by joining the lognormal price
distribution and the beta yield distribution.  A continuous bivariate
revenue distribution is constructed by taking random draws of
variables from the specified marginal distributions for price and
yield.  The variables already reflect the dependence measure,
, Spearman's rank correlation coefficient, to account for
the yield-price correlation. 

The needed variables to form a revenue distribution, price and yield,
x and y, are generated through the following procedure.  Capital
letters represent random variables and lower case letters represent
drawn values of these random variables. 

In Equation 5 , A and B are assumed to have a common standard normal
density function with mean 0 and standard deviation of 1. 

(5)



(See figure in printed edition.)

Equation 6 defines r as a, the value of the drawn variable from a
standard normal distribution: 

(6)



(See figure in printed edition.)

Equation 7 defines s, the linear combination of the values of a and b
weighted by c, which reflects the yield-price correlation, : 

(7)



(See figure in printed edition.)

where a and b are identically and independently distributed random
variables with a common density function and c is a weight reflecting
the relationship between the two random variables, in this case price
and yield. 

Equations 8 and 9 define w and z, the cumulative density functions of
R and S, respectively. 

(8)



(See figure in printed edition.)

(9)



(See figure in printed edition.)

where (.) is the cumulative density function for a standard
normal variate. 

Finally, Equations 10 and 11 result in the variables price, x, and
yield, y: 

(10)



(See figure in printed edition.)

(11)



(See figure in printed edition.)

where FX(.) and FY(.) are the known marginal cumulative density
functions for price and for yield, and FX\-1 (.) and FY\-1 (.) are
the corresponding inverse functions. 

After the correlated price and yield observations are drawn from the
inverse marginal distributions, they are multiplied together to
generate thousands of revenue observations.  In this way, a revenue
distribution is generated. 


--------------------
\3 Johnson, Mark E.  and Aaron Tenenbein, "A Bivariate Distribution
Family With Specified Marginals," Journal of the American Statistical
Association.  Vol.  76 (Mar.  1981):  198-201.  The basis for the
Johnson and Tenenbein approach is that a linear combination of two
independent deviates creates dependence. 


   CALCULATING EXPECTED LOSSES AND
   PREMIUM RATES
-------------------------------------------------------- Appendix IV:4

Revenue Assurance premiums are derived from an average of expected
losses.  The expected loss for a hypothetical policy is derived by
taking the difference between the guaranteed revenue and market
revenue as reflected in the revenue distribution developed above.  If
the guaranteed level is higher than the revenue realized, the
difference is the amount of indemnity owed.  Potential indemnities
associated with each guaranteed revenue are totaled.  The losses are
averaged across all policies to develop premium rates. 

To develop premium rate tables for similar production levels, every
permutation of a discrete range of prices, yields, and coverage
levels corresponding to average expected losses are simulated.  These
data are used to estimate premium rates by developing a translog
equation that links expected losses with (1) expected farm and county
yield, (2) yield variability, (3) price volatility, (4) coverage
levels, and (5) the cross-products and squares of these variables. 


   ANALYSIS OF REVENUE ASSURANCE
-------------------------------------------------------- Appendix IV:5

Although the Revenue Assurance model has the advantage of being based
on current prices and a revenue distribution that incorporates the
price-yield interrelationship, assumptions about relevant
distributions and application of a key statistical technique raise
questions about the adequacy of the plan's premium rates. 

The Revenue Assurance method uses prices that may be more appropriate
than Income Protection's or Crop Revenue Coverage's for calculating
future revenues.  That is, Revenue Assurance uses prices at the
pre-planting period for harvest period futures contracts as the
expected prices and derives the variance of prices from current
options contracts on the relevant futures contract.  These current
prices and variances are more likely to reflect future market
conditions than historical prices because they reflect traders'
expectations of prices in the future. 

However, the developers use too few years of yield data to estimate
yield variability.  Furthermore, yields in 3 separate years during
the period 1985 through 1994 reflect events that are likely to occur
much less frequently than every 10 years.  Exceedingly low yields
were observed in 1988 and 1993, and very high yields were observed in
1994.  By limiting the basis for yield analysis to the 1985-94
period, the model would forecast these unusual yields more frequently
than historical yields would indicate. 

Revenue Assurance uses a parametric statistical method that requires
that the underlying distribution function be normal or some other
specified form.  If properly applied, this method generates efficient
estimates that have smaller variances than those of the nonparametric
method. 

However, the assumptions Revenue Assurance makes about yield
distributions may not reflect actual yield data at the farm level. 
According to several analysts, there is no consensus about the
correct functional form of yield distributions.  Furthermore,
estimates of the yield distribution are very sensitive to the assumed
minimum and maximum values for yield. 

In addition, the Johnson and Tenenbein statistical technique imposes
a constraint that is not appropriate.  Specifically, a constant value
for the price-yield correlation for all farmers in all years, which
is required for the proper application of the technique, does not
reflect actual experience.  In the areas further from the heaviest
concentration of production, the interrelationship between prices and
yields is weaker than in the heart of the production area. 
Furthermore, in catastrophic years, the correlation between prices
and yield is usually stronger than the average value over time. 
Because it is not appropriate to assume a constant price-yield
correlation, it is difficult to have confidence that rates based on
such a revenue distribution would be actuarially sound over the long
term and appropriate to the risk each farmer presents. 


METHODOLOGY USED TO SET PREMIUM
RATES FOR INCOME PROTECTION AND
OUR ANALYSIS
=========================================================== Appendix V

In response to a mandate under the Crop Insurance Reform Act of 1994,
USDA developed Income Protection, an insurance plan designed to
guarantee a certain level of income from crop production.  Premiums
for Income Protection are based on revenue distributions that show
expected losses and payouts at different levels of guaranteed income. 

Three primary steps occur in developing the Income Protection rating
methodology--the construction of yield distributions, the
construction of price distributions, and the construction and
simulation of the revenue distributions on the basis of the results
of the first two constructions. 

The first section of this appendix describes how the components of
the simulated yield distributions are calculated using regional,
county, and farm-level yield data.  The second section describes how
the components of the price distribution are calculated by estimating
an equation relating prices to the yields already estimated.  The
third section shows how the price and yield observations developed
from the distributions are combined to construct revenue
distributions.  No statistical restrictions are imposed on the yield,
price, or revenue distributions.  The fourth section shows how
average indemnities and thus rates are calculated.  Finally, we
present our analysis of the methodology used to set premium rates for
Income Protection. 


   CONSTRUCTING THE YIELD
   DISTRIBUTION
--------------------------------------------------------- Appendix V:1

The yield distributions for Income Protection are derived from data
on three major sources of yield variability--trends over time,
regional events, and individual farm production characteristics. 
Trends over time are represented by 50 years of regional yield data. 
Yield data for years when actual yields were vastly different from
expected yields are included and weighted relative to the 50 years of
data used. 

Regional events are represented by regional yield data adjusted for
differences in county yield.  Regional data are also used to capture
price-yield interactions, or correlations.  For information on the
yield on individual farms, APH records are used for farms for which
actual yield data are available for 6 or more years.  Additional
yield data provided by farmers supplement historic records. 

The regional data are the acre-weighted averages of county yields
provided by USDA's National Agricultural Statistics Service (NASS)
for all counties that the Federal Crop Insurance Corporation (FCIC)
has specified as risk-rating regions.  The county yields are NASS
county yields per planted acre.  The pooled farm data consist of the
most recent APH data reported by farmers and recorded in FCIC's files
on yield history.  For estimating rates, data are used from farms
that report 6 or more years of actual yields. 

To determine the Income Protection premium for a farmer, the
predicted yield for the farmer's county is adjusted by the difference
between the farmer's yield as reflected in the yield data provided by
the farmer and the county average yield. 


      REGIONAL YIELD TREND
      EQUATION
------------------------------------------------------- Appendix V:1.1

Equation 1 estimates regional yields: 

(1)



(See figure in printed edition.)

where R is the regional yield, t is time, a\R is the region's yield
intercept, g(t) is the region's estimated yield trend over time, and
et\R is the regional residual yield variation.  The same yield trend
is imposed on all counties in a risk-rating region.  The errors or
remaining variability in yields, after the trend has been accounted
for, are used to construct the revenue equation. 

Equation 2 shows the method used to test for heteroskedasticity, that
is, whether the variability of regional yields has changed over time: 

(2)



(See figure in printed edition.)

The results indicated that the variability had changed over time and
a scaling process was applied to the errors to correct for the
heteroskedasticity. 

Equation 3 shows one method used to correct for heteroskedasticity. 
Here, the predicted values of the absolute yield errors are used to
scale the original yield errors from the regional equation to 1997
units.  The estimated values of these errors make up the yield
distribution from which observations are drawn to develop revenue
functions: 

(3)



(See figure in printed edition.)


      CONSTRUCTING THE
      COUNTY-ADJUSTED REGIONAL
      YIELDS
------------------------------------------------------- Appendix V:1.2

After the regional trend is estimated, g(t), a county-specific
intercept, a1\C is estimated to account for county-specific
differences in productivity.  The intercepts are calculated as the
simple averages of the differences in yields between each county and
the average yield for the region.  All farms in a county are used to
calculate yield variations if at least 6 years of yield data are
provided by 50 or more farms in the county; if fewer than 50 farms
provided yield data for at least 6 years, yields for all farms in the
region are used. 

Equation 4 is used if the yield trend, g(t), is linear: 

(4)



(See figure in printed edition.)

where Ct is the county yield and ut\C is the error term. 

Equation 5 is used if the function of the yield trend is not linear,
with the county intercept, a1\C : 

(5)



(See figure in printed edition.)

where Ct is county yield and Tc is the number of years in the data
set and g(t) from the regional equation detrends the county data. 

Equation 6 is used to construct a county-adjusted regional yield
series for each county in a risk-rating region to maintain a
consistent rating process across regions: 

(6)



(See figure in printed edition.)

where Rt\c is the unknown county-adjusted regional yield, a1\C is the
county-specific intercept, g(t) is the regional trend function, and
et\R is the regional residual as estimated above.  The intercept,
trend value, and error term are summed to construct the
county-adjusted regional yield. 


      CALCULATING DIFFERENCES
      BETWEEN FARM-LEVEL AND
      COUNTY-ADJUSTED REGIONAL
      YIELDS
------------------------------------------------------- Appendix V:1.3

In order to determine yield variability attributable to the farm
yield only, it is necessary to isolate yield variability at the
county-adjusted regional level.  (These two sources of variation are
reconstituted during the premium estimation process.) Isolating the
variability in this manner allows the county-adjusted regional data
set, which is longer than the farm data set, to be used to estimate
the severity and frequency of large regional events. 

Equation 7 shows the construction of the yield variability
attributable to the farm level only: 

(7)



(See figure in printed edition.)

where dt\f is the deviation from the county-adjusted regional yield
for each farm in time t, yt\f is the farm yield, and Rt\C is the
county-adjusted regional yield in time t. 

Equation 8 shows the construction of the farm's average yield
variability attributable to the farm level only: 

(8)



(See figure in printed edition.)

where d\f (bar)\ is the average deviation from the county-adjusted
regional yield for each farm.  The deviation is calculated by
subtracting the average county-adjusted regional yield, R\C (bar),
from the average farm yield, y\f (bar): 

Equation 9 shows the remaining variability after accounting for
variability at the farm and county-adjusted regional levels: 

(9)



(See figure in printed edition.)

The variability, or statistical errors, remaining is expressed as a
function of the difference of the farm's deviation from its average
yield and the county-adjusted regional's deviation from its average
yield for the same period of time.  If a given county has 50 or more
farms with 6 years or more of data, the residuals from the county's
farms are used.  However, if there are fewer than 50 farms with 6
years or more years of data, residuals from all farms in the
risk-rating region are used. 


   CONSTRUCTING PRICE
   DISTRIBUTIONS
--------------------------------------------------------- Appendix V:2

For the major field crops, price distributions are based on monthly
average prices from planting to harvest over a 37-year period. 
Prices for commodity futures at planting and harvest are used to
develop price distributions for the major field crops.  Monthly
averages of the futures price contracts for the 1960-96 period are
constructed for each insured crop.  The planting period price, Pt \0
, is defined as the average of a 30-day period ending 2 weeks before
the crop insurance sign-up for that crop and location, while the
harvest period price, Pt\1 , is an average 30-day price for the month
prior to the close of the harvest futures contract.  Since
proportional prices are used, there is no need to deflate prices. 

Equation 10 estimates seasonal price relationships in the futures
market as a function of historical yield deviations: 

(10)



(See figure in printed edition.)

where Pt\1 is the harvest time futures price of the crop and Pt\0 is
the planting time futures price (or forecast) of the harvest crop,
a1\p is the intercept, a2\p measures the relationship between price
and yield, Rt\C is the county-adjusted regional yield, and Rt\C (hat)
is the forecasted county-adjusted regional yield for year t.  The
term inside the parenthesis adjusts for the lower number of price
observations relative to yield observations in the calculation of
revenue under Income Protection.  This term is constructed in order
to generate a zero mean set of proportional regional yield deviations
for the subset of yield data used in the expression. 

The equation is also used to estimate the price-yield correlation and
the remaining statistical errors, which were not accounted for by the
variation in the county prices.  The error term from this equation is
used in a later step to obtain a consistent estimate of revenue. 


   CONSTRUCTING AND SIMULATING THE
   REVENUE DISTRIBUTION
--------------------------------------------------------- Appendix V:3

In order to construct the revenue distribution, errors from three
estimated equations are drawn randomly: 

  -- et\R from the yield trend Equation (1),

  -- et\f from the remaining farm variability Equation (9), and

  -- et\p from the price-yield equation (10). 

Equation 11 represents the construction of a simulated
county-adjusted regional yield: 

(11)



(See figure in printed edition.)

Equation 12 represents the construction of a simulated farm yield: 

(12)



(See figure in printed edition.)

Equation 13 represents the construction of a simulated price
realization: 

(13)



(See figure in printed edition.)

Equation 14 represents the construction of a simulated revenue
realization: 

(14)



(See figure in printed edition.)

If REVS , actual revenue, is less than the guaranteed revenue, a
payment or indemnity of the difference is assumed to be made and the
amount recorded.  The above process is repeated 10,000 times, and a
running total of the payouts is recorded for each of the possible
indemnity levels.  The average indemnity (total indemnities divided
by 10,000) is used as an estimate of the actuarially neutral
premiums. 

Using this method, rates were developed for discrete combinations of
farm and regional average yields for use in insurance rate tables. 


   ANALYSIS OF INCOME PROTECTION
--------------------------------------------------------- Appendix V:4

The premium rates offered in Income Protection are developed through
a nonparametric statistical model that constructs a revenue
distribution on the basis of actual price and yield data.  This model
does not make any assumptions about the shape of the actual revenue
distribution or about price and yield distributions; such assumptions
could bias the estimates of expected losses. 

As part of the rate-setting method, this model takes into account all
of the variability in price and yield data, as well as changes in the
price-yield correlation; therefore, it is not necessary to estimate
these factors separately.  The advantage of this approach is that the
shape of the revenue distribution is generated by the actual crop
data.  Therefore, the error of incorrectly imposing a shape on the
revenue distribution is avoided.  However, if the underlying
distribution is known, a nonparametric method may have an inherent
disadvantage of producing less efficient estimates than a parametric
method. 

While Income Protection appropriately relies on an integrated
statistical model to estimate probable losses, it does not consider
how revenue may change in response to the new farm policy.  That is,
Income Protection relies on historical crop prices to estimate rates,
which, as previously discussed, may not reliably predict future crop
prices.  Developers of the Income Protection plan believe that
because of the effect of previous farm programs, historical data sets
may underestimate the variances in future farm prices.  This would
mean that premium rates would be too low to accommodate future price
fluctuations and therefore future losses.  In order to account for
this effect, a 20-percent loading factor was added to the premium. 




(See figure in printed edition.)Appendix VI
COMMENTS FROM THE U.S.  DEPARTMENT
OF AGRICULTURE
=========================================================== Appendix V



(See figure in printed edition.)



(See figure in printed edition.)


The following are GAO's comments on the U.S.  Department of
Agriculture's letter dated April 6, 1998. 

GAO'S COMMENTS

1.  We disagree.  While we do not state in this report, nor do we
believe, that the plans contain "fatal flaws," we believe that the
shortcomings we identified in all three revenue insurance plans are
serious enough to warrant a reevaluation of the methods and data used
to set premium rates to ensure that each plan is based on the most
actuarially sound foundation.  In particular, as we reported, the
rating method for Crop Revenue Coverage is especially problematic
because it does not take into account the relationship between crop
prices and yields. 

2.  Contrary to the agency's assertion, we do not assume that the
premium rates for revenue coverage (without replacement coverage) are
always lower than the premium rates charged for yield coverage.  We
agree with the agency that the rates for these revenue plans can be
higher than those for yield coverage when both yields and prices
decline.  For this reason, throughout the report, we say that a
decline in yield is "often" accompanied by an increase in prices. 

3.  In using the term "actuarial soundness," we mean that the
premiums established for each plan are sufficient over the long term
to cover the indemnities paid, and that individual premiums are
appropriate to the risk each farmer presents.  We have revised the
report to clarify our use of this term. 

4.  We have removed the word "mechanical" from the report. 

5.  We disagree.  Estimates of future price volatility based on
historical prices and estimates of price volatility based on current
market expectations are not equally appropriate.  Crop Revenue
Coverage and Income Protection base their estimates of future price
increases or decreases on the way that prices moved in the past, when
certain farm programs were in place that set a price floor.  This
situation has changed.  Under current policy, when prices are tied to
market conditions, we continue to believe that the market's
expectation of price volatility is the best barometer of intra-year
price changes. 

6.  In the executive summary, we have modified the language to
reflect the partially offsetting effects of Crop Revenue Coverage's
higher premiums.  Our discussion in chapter 2 already reflected this
point.  Nevertheless, when crop prices are higher at harvest than at
planting, claims payments for Crop Revenue Coverage will exceed those
paid for multiple-peril crop insurance. 

7.  We have modified our report to reflect the fact that in recent
years the agency has improved its expected loss ratio for traditional
multiple-peril crop insurance to achieve the current legislatively
mandated 1.10 loss ratio. 

8.  See comment 1.  In addition, we believe that as shortcomings in
the methods used to establish premium rates are identified, the Risk
Management Agency should take action to correct the deficiencies to
the extent possible. 

9.  We agree that the agency must continually evaluate all
rate-making methodologies.  However, when this evaluation reveals
shortcomings, as we point out in this report, then evaluations should
be translated into actions to ensure that each plan is based on the
most actuarially sound foundation. 

10.  We have modified our report to reflect the agency's authority to
approve expansion of Crop Revenue Coverage. 

11.  The Risk Management Agency's senior actuary informed us that the
premium rates for Crop Revenue Coverage average about 30 percent
higher than comparable premium rates for traditional multiple-peril
crop insurance.  Because administrative expense reimbursements are
based on fixed percentage of premiums, higher premiums for Crop
Revenue Coverage will result in higher administrative costs to the
government.  A judgment on whether the reimbursement is adequate to
cover expenses was beyond the scope of our work. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================= Appendix VII

Robert C.  Summers, Assistant Director
Thomas M.  Cook, Evaluator-in-Charge
Barbara J.  El Osta
Donald L.  Ficklin
Mary C.  Kenney
Robert R.  Seely, Jr.
Carol Herrnstadt Shulman


*** End of document. ***