U.S. Agriculture: Grain Fungus Creates Financial Distress for North
Dakota Barley Producers (Letter Report, 03/22/99, GAO/RCED-99-59).

Pursuant to a congressional request, GAO reviewed the effect of scab and
vomitoxin on North Dakota barley crops, focusing on: (1) the financial
impact from scab and vomitoxin on barley farmers; (2) the performance of
vomitoxin test methods; and (3) short- and long-term actions that could
help reduce the impact of scab and vomitoxin on North Dakota barley
farmers.

GAO noted that: (1) North Dakota barley farmers have experienced
extensive revenue losses from scab and vomitoxin damage; (2) from 1993
through 1997, these farmers suffered estimated cumulative losses of
about $200 million from scab and vomitoxin--equal to about 17 percent of
the $1.2 billion in total barley revenues they received during this
period; (3) while most of the revenue losses resulted from decreases in
barley production, losses also resulted from severe price discounts; (4)
maltsters and brewers, the traditional buyers of North Dakota's malting
barley, have reacted to the scab and vomitoxin damage by purchasing less
barley from North Dakota farmers and more from Canadian and other
western U.S. sources; (5) three tests are generally used to measure
vomitoxin concentrations in barley produced in North Dakota; (6) one is
a field kit, called Veratox, which is commonly used by grain elevators
and commercial testing facilities and is the test that most directly
affects the prices farmers receive for their barley; (7) the Veratox
test can produce results that vary at concentrations critical to pricing
decisions; (8) testing experts attribute variations in test results to
several sources, including the skill of the technician conducting the
test; (9) they stress the importance of quality assurance measures and
training to help reduce this variation; (10) the other two
tests--high-pressure liquid chromatography and gas chromatography--are
reference methods that are used primarily in research laboratories for
such purposes as checking the performance of the Veratox kit; (11)
according to analytical chemists and other testing experts, these tests
provide accurate and consistent test results; (12) however, because of
the complexity and the cost of the equipment for these two tests, they
are not practical for use at commercial testing facilities and other
locations that serve barley farmers; (13) short-term actions, such as
rotating crops and spraying with fungicides, may help reduce scab and
vomitoxin's impact under conditions of light infestation; (14) however,
according to North Dakota agriculture experts, the benefits of these
actions are negligible during periods of moderate to severe infestation;
(15) from 1993 through 1997, several counties in the Red River Valley of
North Dakota experienced moderate or severe scab and vomitoxin
infestation; and (16) the longer-term action of developing more
scab-resistant barley may also help reduce the disease's impact under
conditions of light infestation.

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

 REPORTNUM:  RCED-99-59
     TITLE:  U.S. Agriculture: Grain Fungus Creates Financial Distress 
             for North Dakota Barley Producers
      DATE:  03/22/99
   SUBJECT:  Economic analysis
             Agricultural production
             Grain and grain products
             Plant diseases
             Agricultural chemicals
             Quality control
             Grain inspection
             Testing
             Prices and pricing
IDENTIFIER:  North Dakota
             
******************************************************************
** 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 Honorable
Byron L.  Dorgan, U.S.  Senate

March 1999

U.S.  AGRICULTURE - GRAIN FUNGUS
CREATES FINANCIAL DISTRESS FOR
NORTH DAKOTA BARLEY PRODUCERS

GAO/RCED-99-59

Grain Fungus

(150651)


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

  ARS - Agricultural Research Service
  CRD - Crop Reporting District
  ERS - Economic Research Service
  FDA - Food and Drug Administration
  FSA - Farm Service Agency
  GC - gas chromatography
  GIPSA - Grain Inspection, Packers and Stockyards Administration
  HPLC - high pressure liquid chromatography
  NASS - National Agricultural Statistics Service
  QA - Quality Assurance
  USDA - United States Department of Agriculture

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


B-281798

March 22, 1999

The Honorable Byron L.  Dorgan
United States Senate

Dear Senator Dorgan: 

Barley has traditionally been a major source of farm income in North
Dakota, second only to wheat in total acreage planted and crop
income.  However, since 1993, substantial portions of North Dakota's
barley crop have been damaged by a fungal disease--known as
scab--that frequently produces a toxin called vomitoxin.  The scab
and vomitoxin epidemic has reduced the amount of barley produced and
sold in North Dakota and the prices paid for barley.  Specifically,
scab causes barley kernels to become discolored and shriveled,
reduces crop yield, depresses grain weight, and ultimately forces
farmers to sell fewer bushels of barley at reduced prices. 
Furthermore, barley (in the form of barley malt) is a key ingredient
in beer, and vomitoxin in barley can cause beer to produce too much
foam, either during the fermentation process, thereby reducing the
amount of beer produced, or when beer cans or bottles are opened,
thus creating a less desirable product.  In response, the malting and
brewing industries will pay only a discounted price for barley that
tests positive for vomitoxin; conversely, the industries offer top,
or premium, prices for barley that is vomitoxin-free.  Barley that is
not of malting quality is sold primarily for animal feed and commands
a much lower price.  Discounted prices for vomitoxin-contaminated
barley cover a narrow range of concentrations, beginning at 0.6 parts
per million (ppm)--the approximate level at which field tests can
begin quantifying the amount of vomitoxin--to about 3 ppm.\1 Beyond 3
ppm, barley is usually sold as animal feed. 

Concerned about the effect of these losses on North Dakota barley
farmers, you asked us to (1) determine the financial impact from scab
and vomitoxin on these farmers,\2 (2) assess the performance of
vomitoxin test methods, and (3) identify short- and long-term actions
that could help reduce the impact of scab and vomitoxin on North
Dakota barley farmers. 


--------------------
\1 To put these concentrations in context, 1 ppm is approximately
equivalent to 1 kernel of contaminated barley in almost two bushels
of barley. 

\2 To estimate losses because of scab and vomitoxin, we first
estimated what barley revenues would have been for 1993 through 1997
had there been no scab and vomitoxin outbreak.  We then compared our
estimate of barley revenues with actual barley revenues for these
years to determine losses.  In estimating losses, we controlled for
other variables, such as weather, that can affect barley production. 


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

North Dakota barley farmers have experienced extensive revenue losses
from scab and vomitoxin damage.  From 1993 through 1997, these
farmers suffered estimated cumulative losses of about $200 million
from scab and vomitoxin--equal to about 17 percent of the $1.2
billion in total barley revenues they received during this period.\3
While most of the revenue losses resulted from decreases in barley
production, losses also resulted from severe price discounts. 
Maltsters and brewers, the traditional buyers of North Dakota's
malting barley, have reacted to the scab and vomitoxin damage by
purchasing less barley from North Dakota farmers and more from
Canadian and other western U.S.  sources. 

Three tests are generally used to measure vomitoxin concentrations in
barley produced in North Dakota.  One is a field kit, called Veratox,
which is commonly used by grain elevators and commercial testing
facilities and is the test that most directly affects the prices
farmers receive for their barley.  The Veratox test can produce
results that vary at concentrations critical to pricing decisions. 
Testing experts attribute variations in test results to several
sources, including the skill of the technician conducting the test. 
They stress the importance of quality assurance measures and training
to help reduce this variation.  The other two tests--high pressure
liquid chromatography and gas chromatography--are reference methods
that are used primarily in research laboratories for such purposes as
checking the performance of the Veratox kit.  According to analytical
chemists and other testing experts, these tests provide accurate and
consistent test results.  However, because of the complexity and the
cost of the equipment for these two tests, they are not practical for
use at commercial testing facilities and other locations that serve
barley farmers. 

Short-term actions, such as rotating crops and spraying with
fungicides, may help reduce scab and vomitoxin's impact under
conditions of light infestation.  However, according to North Dakota
agriculture experts, the benefits of these actions are negligible
during periods of moderate to severe infestation.  From 1993 through
1997, several counties in the Red River Valley of North Dakota
experienced moderate or severe scab and vomitoxin infestation. 
Furthermore, many of these actions have tradeoffs, such as causing
environmental problems (like soil erosion), that barley farmers must
take into account.  The longer-term action of developing more
scab-resistant barley may also help reduce the disease's impact under
conditions of light infestation.  But many scientists say that more
resistant barley may not be commercially available for at least 6
years.  They also emphasize that developing a variety that is
100-percent resistant to scab is unlikely anytime soon. 


--------------------
\3 All measures of farm revenues and losses are stated in 1997
constant dollars.  Crop insurance payments for scab and
vomitoxin-damaged barley covered only a very small portion (less than
2 percent) of cumulative revenue losses from the epidemic. 


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

Vomitoxin, a toxin associated with a fungal disease called scab, only
occurs when scab is present.  Since 1993, scab and vomitoxin have
affected wheat and barley crops in the Northern Great Plains, which
includes North Dakota; Minnesota; South Dakota; and Manitoba, Canada. 
Crops in the Red River Valley region (the eastern part of North
Dakota, the western part of Minnesota, and a corner of northeast
South Dakota) have been the most severely affected.  The mold that
produces vomitoxin grows primarily on grains, particularly on wheat
and barley, and can cause vomiting in farm animals that ingest
vomitoxin-contaminated feed grains.  The Food and Drug Administration
(FDA), which is responsible for ensuring food safety in certain
foods--including grains--has not issued any guidance on vomitoxin in
barley or barley products.  However, it has issued advisory levels\4
for vomitoxin in wheat and wheat products and feed grains for
animals. 

The scab and vomitoxin epidemic has added to the financial stress of
farmers in North Dakota and the rest of the Northern Great Plains. 
North Dakota suffered a drought in 1988 and floods in 1993 and 1997. 
The U.S.  Department of Agriculture's (USDA) Farm Service Agency
(FSA) estimates that in the barley-producing regions of North Dakota
most affected by scab and vomitoxin, 768 (or about 14 percent) of the
farmers stopped farming between 1996 and 1998.  Although this figure
includes farms that failed because of flood, drought, and other
reasons, FSA officials stated that scab and vomitoxin were the
primary reasons for leaving farming. 

Barley is economically important to North Dakota agriculture. 
Traditionally, it is second only to wheat in acreage planted and
total crop income.  For example, in 1992, the last year before the
scab and vomitoxin epidemic, North Dakota's farm income from all
crops totaled $2.2 billion, of which $1.2 billion (about 54 percent)
was from wheat and $237 million (about 11 percent) was from barley. 
Furthermore, for the last 50 years, North Dakota has been the leading
barley producer in the United States; in 1997, it accounted for 27
percent of the nation's total barley production.\5

Most farmers sell their barley to grain dealers, who then resell it
to maltsters and brewers.  To determine the price they will offer
farmers for their barley, including the need for discounts, grain
dealers have the barley tested for vomitoxin.  Most testers of
vomitoxin in North Dakota use a test kit called Veratox because it is
relatively quick, inexpensive, and practical for commercial use.  The
high pressure liquid chromatography (HPLC) and gas chromatography
(GC) tests, which are used by researchers for purposes such as
advancing research on vomitoxin, are also used by maltsters and
USDA's Grain Inspection, Packers and Stockyards Administration
(GIPSA) to check Veratox test results.  These two reference methods
are generally not used by commercial testing facilities and grain
dealers because they are more costly, time-consuming, and complex to
operate. 

GIPSA is the USDA agency that oversees federal grain inspections and
has several key associated responsibilities.  It authorizes certain
commercial testing facilities to perform tests following its official
procedures and standards.  It also approves various testing methods,
such as the Veratox kit, for use by these authorized facilities. 
Approved test methods, for which GIPSA provides training, must meet
the agency's performance criteria.  GIPSA also monitors the
consistency of test results across its authorized facilities.  For
example, GIPSA conducts quarterly reviews of the test results from
its authorized testing facilities.  For these reviews, GIPSA uses the
HPLC test method as a reference for, or check on, test results from
these facilities.  GIPSA considers the scab and vomitoxin epidemic to
be a serious problem and has taken actions to address vomitoxin
testing issues, such as conducting a study in 1998 to assess the
extent to which sampling methods can affect vomitoxin test results. 

However, GIPSA oversees only a portion of commercial grain testing
nationwide.\6 Commercial testing facilities unaffiliated with GIPSA
and large grain elevators where in-house testing with the Veratox kit
is cost-effective also perform vomitoxin testing.  The North Dakota
Barley Council estimates that 40 percent of commercial vomitoxin
testing in North Dakota occurs at GIPSA's authorized facilities; the
remaining 60 percent occurs at either the unaffiliated testing
facilities or large grain elevators.  GIPSA has no oversight
responsibility for vomitoxin tests performed by these other entities. 
Currently, GIPSA has four authorized agents in North Dakota that
operate six commercial testing facilities.  In addition, North Dakota
has about nine commercial testing facilities that are not affiliated
with GIPSA and between 12 and 20 grain elevators that test for
vomitoxin. 


--------------------
\4 Advisory levels are FDA's initial guidance on the amount of toxin
allowed in food before public health is threatened. 

\5 In 1997, other leading U.S.  barley producers were Montana, Idaho,
Washington State, and Minnesota. 

\6 Only exports of U.S.  grains are required to have a GIPSA grain
inspection.  However, unless requested, these inspections do not
include vomitoxin testing.  In addition, domestic grain elevators can
request GIPSA to inspect grain, including testing for vomitoxin. 


   NORTH DAKOTA BARLEY FARMERS
   HAVE EXPERIENCED LARGE REVENUE,
   YIELD, AND ACREAGE LOSSES
   BECAUSE OF SCAB AND VOMITOXIN
------------------------------------------------------------ Letter :3

From 1993 through 1997, we estimate that North Dakota barley farmers
suffered cumulative revenue losses from scab and vomitoxin of about
$200 million (in 1997 dollars)\7 --equal to almost 17 percent of the
$1.2 billion in total barley revenues farmers received during this
period.\8 The losses from these diseases varied significantly, both
over the years and across the regions of the state, with the Red
River Valley suffering the greatest losses.  However, crop insurance
payments for scab- and vomitoxin-damaged barley covered only a very
small portion, less than 2 percent, of these cumulative losses.  U.S. 
maltsters and brewers, the traditional buyers of North Dakota's
malting barley, have reacted to scab and vomitoxin by expanding their
imports of malting barley from Canada by about 380 percent. 


--------------------
\7 Technically, using a range of assumptions, we estimated losses of
$177 million to $224 million.  For the purpose of this report, we are
stating this as approximately $200 million in revenue losses. 

\8 See app.  I for a detailed description of our data sources,
methodology, and the results of our analysis. 


      LARGE REVENUE LOSSES
      RESULTING FROM SCAB AND
      VOMITOXIN ARE CAUSED BY
      PRODUCTION DECLINES AND
      PRICE DISCOUNTS
---------------------------------------------------------- Letter :3.1

From 1993 through 1997, we estimate that North Dakota farmers lost
about $200 million (in 1997 dollars) in revenues as a result of
declines in both production and price discounts.  These losses were
equal to almost 17 percent of the $1.2 billion in total revenues
barley farmers received during these years.  About 70 percent of
these losses, or $139 million, were from reduced barley yields (in
bushels per acre) and from farmers' leaving more barley unharvested. 
For example, between 1992 and 1997, average North Dakota barley
yields dropped from a pre-disease level of 65 bushels an acre to 45
bushels an acre.  Also, as shown in figure 1, from 1993 through 1997
(the years of the epidemic), the number of acres planted with barley
fell from 2.9 million to 2.4 million and the number of harvested
acres of barley fell from 2.4 million to 2.25 million.\9 Differences
between the amount of acres planted and actually harvested were the
largest in 1993 and 1996.  For instance, in 1993, North Dakota
farmers harvested about 500,000 fewer barley acres than they had
planted. 

   Figure 1:  North Dakota Barley
   Acreage--Gap Between Planted
   Acres and Harvested Acres,
   1992-97

   (See figure in printed
   edition.)

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

Price discounts for barley contaminated with vomitoxin also played a
key role in reducing farmers' revenues.  From 1993 through 1997,
price discounts because of vomitoxin accounted for about 30 percent,
or $61 million, of total revenue losses.  The relationship between
vomitoxin and price discounts is complex.  Discounting in the
marketplace stems from the U.S.  brewing industry's desire to use
little or no vomitoxin-contaminated barley.  In general, U.S. 
brewers send price signals that reflect their specific quality and
quantity requirements to merchandisers and maltsters.  These price
signals are subsequently incorporated into price discount schedules
that reflect buyers' reluctance to purchase barley with vomitoxin
unless they receive a highly discounted price.  Grain elevators use
these schedules, in conjunction with other quality premium or
discount factors,\10 to determine an overall price quote to farmers. 
Price discount schedules for barley vomitoxin can change over time,
sometimes on a daily basis, depending on market conditions.  Price
signals for malting barley come largely from the four large firms
that dominate the U.S.  brewing industry.  One of these firms
represents nearly half of the market.\11 A limited number of buyers
in a given industry, such as the brewing industry, can influence the
market price for a given commodity. 

According to industry experts, although vomitoxin can cause excessive
foaming during the malting process and in finished beer products,
brewers require discounts for malting barley primarily because they
are concerned about the potential for a negative public perception of
beer containing vomitoxin.  The industry is concerned that consumers
will switch brands or purchase other alcoholic beverages if it is
reported that beer contains vomitoxin.  As a result, brewers are
willing to pay top prices for vomitoxin-free barley, but only highly
discounted prices for barley contaminated with vomitoxin. 

Table 1 shows an example of a price schedule for barley,
incorporating discounts for different levels of vomitoxin.  Although
discounting strategies vary, grain dealers generally begin
discounting the price of vomitoxin- contaminated barley at 0.6 ppm. 
This first discount, usually the largest of several, ranges from
about 40 cents to 60 cents a bushel.  As shown in table 1, this first
discount would result in a price of about $2 per bushel.  Grain
dealers apply subsequent discounts of about 5 cents to 15 cents for
concentrations of vomitoxin that range from 1.1 ppm to 3.0 ppm.  At
vomitoxin concentrations above 3.0 ppm, dealers generally purchase
barley as feed grain, which receives the lowest price, about $1.75
per bushel.  The American Malting Barley Council reported that for
1997 only 9 percent of all midwestern malting barley had a vomitoxin
level that fell into the premium price category of 0.5 ppm or less. 



                                Table 1
                
                Example of a Price Schedule for Barley,
                Discounted for Different Concentrations
                              of Vomitoxin

Concentration of vomitoxin                Price per bushel
----------------------------------------  ----------------------------
0-0.5 ppm                                 $2.55 (premium malting
                                          barley price--no discount )

0.6-1.0 ppm                               $2.00

1.1-2.0 ppm                               $1.85

2.1-3.0 ppm                               $1.80

3.1-3.5 ppm                               $1.75 (feed grain price)
----------------------------------------------------------------------
Source:  GAO's analysis of May 1998 barley price schedules from two
grain dealers in Bottineau County, North Dakota. 

Along with steep price discounts, vomitoxin has had the effect of
shifting the amount of malting versus feed grain barley produced in
North Dakota.  In the years before the scab and vomitoxin epidemic,
the largest part of the state's barley production, and hence barley
revenues, came from premium-priced malting barley--60 to 70 percent
of all North Dakota barley sales.  However, since the scab and
vomitoxin epidemic, this trend has changed.  Specifically, in several
years during 1993 through 1997, many regions of North Dakota sold
over 50 percent of the barley produced to the lower-valued feed grain
market.\12


--------------------
\9 In 1996, the amount of barley planted and harvested temporarily
increased.  Planted acres increased because of farmers' response to
record high barley prices in 1995; harvested acres increased because
of favorable weather conditions that were less conducive to the
development of scab and vomitoxin. 

\10 Other quality factors that can affect malting barley premiums and
discounts include moisture, color, test weight, protein, foreign
material, and damage. 

\11 These top four firms are Anheuser-Busch, Inc., the Miller Brewing
Company, the Coors Brewing Company, and the Stroh Brewery Company. 
In 1998, these firms constituted 45 percent, 22 percent, 10 percent,
and 9 percent, respectively, of the U.S.  beer market. 

\12 Other than its use in beer production, barley produced for human
consumption represents a very small part, less than 5 percent, of
North Dakota barley production. 


      REVENUE LOSSES BECAUSE OF
      SCAB AND VOMITOXIN VARIED BY
      LOCATION AND YEAR
---------------------------------------------------------- Letter :3.2

While scab and vomitoxin have reduced North Dakota barley farmers'
revenues, the amount of loss has varied by region and year.  As seen
in table 2, the most severely affected area in North Dakota in terms
of total revenue losses has been in the upper Red River Valley--the
East Central and Northeast regions of the state--while the Southeast
region has been least affected.  Barley farmers suffered their
greatest losses overall from vomitoxin in 1993 and in 1997, with
losses of $62 million and $68 million, respectively.  However, as the
table shows, some regions--because they were less affected by
vomitoxin and thus had more premium quality malting barley to
sell--had a small increase in revenues in certain years. 



                          Table 2
          
          Changes in North Dakota Barley Farmers'
              Revenues as a Result of Scab and
               Vomitoxin, by Region, 1993-97

                    (Dollar in millions)

           Changes in barley revenues by crop reporting
                            districts
        --------------------------------------------------
         North                      East
        Centra  Northeas  Centra  Centra  Southe  Total by
Year         l         t       l       l     ast      year
------  ------  --------  ------  ------  ------  --------
1993     ($11)     ($21)    ($8)   ($16)    ($6)     ($62)
1994        $4     ($14)    ($1)   ($13)    ($1)     ($26)
1995        $3     ($10)    ($2)   ($10)    ($1)     ($20)
1996      ($3)     ($21)      $1    ($3)      $1     ($25)
1997     ($12)     ($36)    ($7)   ($11)    ($2)     ($68)
==========================================================
Total    ($19)    ($102)   ($17)   ($53)   ($10)    ($201)
 by
 regio
 n
----------------------------------------------------------
Note:  Dollars are stated in 1997 constant dollars.  Losses are in
parenthesis and represent revenue declines from barley contaminated
by scab and vomitoxin.  For example, farmers in the North Central
District received $11 million less for their barley in 1993 because
of this contamination.  Losses were determined by comparing actual
barley revenues with predicted revenues in the absence of vomitoxin
for each year.  Predicted revenues were developed using historical
data from the pre-vomitoxin years of 1959 through 1992.  (See app.  I
for details on our methodology.)

Source:  GAO's analysis of data from various organizations. 


      IMPORTS OF MALTING BARLEY
      FROM CANADA INCREASED AS A
      RESULT OF SCAB AND VOMITOXIN
---------------------------------------------------------- Letter :3.3

Because the scab and vomitoxin outbreak has reduced the supply of
high- quality malting barley in the Northern Great Plains, the
traditional purchasers of North Dakota malting barley--U.S.  brewers
and maltsters--have increasingly turned to Canadian and other U.S. 
sources.  Figure 2 illustrates the increase in Canadian barley
production and the increase in exports of malting barley to the
United States.  During the years of the scab and vomitoxin epidemic,
average annual Canadian exports of malting barley to the United
States increased by about 380 percent.\13 From 1993 through 1997,
average annual barley exports from Canada reached 705,000 metric
tons, compared to 147,125 metric tons from the pre-epidemic years of
1985 through 1992.  In addition, to meet the increased U.S.  demand
for premium quality malting barley, Canadian production of malting
barley grew from 1 million metric tons in 1993 to 2.2 million metric
tons in 1997.  Agriculture Canada\14 reported in 1997 that the United
States has been Canada's largest market for malting barley over the
past 4 years because of the shortage of quality U.S.  malting barley. 
And, in 1997, malting barley imports from Canada represented over 25
percent of all malting barley consumed by the U.S.  brewing industry. 
In comparison, from 1988 through 1992, malting barley imports from
Canada represented about 5 percent of all malting barley consumed by
the industry. 

   Figure 2:  Canadian Malting
   Barley Production and Exports
   to the United States, 1985-97

   (See figure in printed
   edition.)

Note:  Canadian production of malting barley is estimated by
Agriculture Canada as including that country's domestic barley
consumption and its exports of barley and barley malt. 

Source:  GAO's analysis of Agriculture Canada's data. 

Although scab and vomitoxin have decreased North Dakota barley
farmers' revenues, Canadian imports have somewhat moderated the
blight's impact on U.S.  brewers and maltsters.  Shortages of malting
barley in the United States as a result of these diseases would
normally tend to increase U.S.  malting premiums and prices, but
these increases have been tempered by the large imports of Canadian
malting barley.  That is, even with a smaller domestic supply of
malting quality barley, larger Canadian imports produce competitive
pressures to keep prices below the levels they would be if imports
were not part of the U.S.  malting barley market. 


--------------------
\13 While we did not consider them central to discussing Canadian
barley imports, we recognize that other factors influence these
imports, such as exchange rates and the North American Free Trade
Agreement. 

\14 Agriculture Canada is the Canadian government's department of
agriculture. 


   VERATOX TEST RESULTS HAVE
   INCREASED VARIABILTY AT LEVELS
   CRITICAL TO PRICING DECISIONS
------------------------------------------------------------ Letter :4

According to testing experts,\15 while the Veratox test kit serves
the market's need for a relatively fast and cost-effective method for
measuring vomitoxin in barley, it can produce test results that vary,
particularly at concentrations critical to pricing decisions. 
Testing experts state that this variability can be reduced to some
extent through quality assurance measures and training.  Testing
experts believe the HPLC and GC tests produce more accurate and
consistent results, in part because they are conducted under
controlled laboratory conditions.  However, because of their
complexity and cost, these tests are not practical for commercial
use. 


--------------------
\15 Testing experts we interviewed included officials from the
Association of Official Analytical Chemists.  See the scope and
methodology section of this report for a listing of the organizations
of the testing experts that we contacted. 


      BARLEY PRICES ARE SET AT
      LEVELS WHERE VERATOX TEST
      RESULTS VARY THE MOST
---------------------------------------------------------- Letter :4.1

Testing experts state that all tests for vomitoxin, including
Veratox, experience variability in test results, particularly at the
upper and lower limits of the test's ability to measure vomitoxin. 
This variability at the Veratox kit's lower limits of measuring
vomitoxin can affect whether barley farmers receive a price discount. 
According to the manufacturer of the Veratox kit, the kit's lower
limit of measurement is 0.5 ppm.  At this concentration, Veratox test
results can range from 0 ppm (where barley receives no discount) to
1.1 ppm (where barley would incur a substantial price discount).  The
market, therefore, is making crucial pricing decisions at
concentration levels where the Veratox kit has substantial
variability. 

Our analysis of selected test data results supports expert opinion
regarding Veratox's variability.  To conduct this analysis, we
compared 1,068 Veratox test results between (1) Neogen (the
manufacturer of Veratox) and a North Dakota commercial grain testing
facility and (2) these two facilities and GIPSA's in-house HPLC
reference method.\16 We found that Veratox test results on the same
samples of barley varied between Neogen and the commercial grain
testing facility.  Specifically, assuming that the HPLC test results
represent the true concentration of vomitoxin, we found that at
concentrations between 0.7 and 4 ppm, Neogen's estimate of vomitoxin
levels was, on average, higher than the testing facility's. 
Consequently, farmers could have received different prices from each
testing location, had the test results been the basis for a
commercial sale.  For example, we found instances in which, at a HPLC
vomitoxin concentration of 1 ppm, the manufacturer's test results
measured 1.1 ppm vomitoxin or greater, while the testing facility's
measured 0.4 ppm.  Had these results been the basis for a sale, a
producer would have received $1.85 per bushel or less from the
manufacturer but $2.55 per bushel from the testing facility.  (See
table 1 for an example of a price schedule for barley.)

Testing experts said that test methods have two types of
variability--inherent and systemic.  Inherent variability exists in
all vomitoxin test methods and increases at the higher and lower
limits of a given test method's ability to measure vomitoxin levels. 
Experts state that this variability cannot be controlled, which is
why it is called inherent.  The inherent variability of the Veratox
test may affect barley buyers and farmers differently.  According to
GIPSA officials, grain elevators, which purchase barley from farmers
and sell it to maltsters and brewers, may be less affected because
they handle larger volumes of barley, with a correspondingly greater
number of test results.  Thus, the prices based on test results that
were too high or too low--because of the inherent variability
associated with the test kit--could counterbalance each other.  As a
result, grain elevators may be less affected by variable test results
than barley farmers, who receive prices based on fewer test results. 
Because farmers may be more affected, some testing experts believe
that if price discounts were started at 1 ppm, rather than at 0.5 ppm
(the lower testing limit of the kit), farmers could receive more
equitable test results.  Some cereal scientists told us that no
appreciable increases in beer production problems occur when brewing
with barley having vomitoxin concentrations of 1 ppm versus 0.5 ppm. 
However, U.S.  brewers and maltsters we talked to had varying
opinions on whether beer production problems would increase at
concentrations of 1 ppm. 

Systemic variability, which refers to differences in how testers
obtain and process grain samples and conduct tests, can also affect
test results--for Veratox as well as for other testing methods.  For
example, a Veratox test involves many actions--selecting and
processing the grain sample, extracting the vomitoxin from the grain,
and measuring the vomitoxin.  Furthermore, the test equipment must be
maintained and cleaned in order to achieve optimal results.  Experts
said that, because the potential exists for mistakes at each stage of
the process, the accuracy of the kit's results is affected by the
skill of the technician using it. 

For all testing methods, a number of actions--including training and
quality assurance\17 efforts--can be used to reduce systemic
variability.  First, test results from grain samples known to have
vomitoxin can be compared across various testing facilities.  This
method, often referred to as a "check-sample program," helps ensure
that testing facilities will achieve consistent test results. 
GIPSA's offices and its authorized testing facilities use this
approach.  Specifically, GIPSA sends samples of barley or wheat with
known concentrations of vomitoxin to its authorized facilities for
testing.  GIPSA then compares the test results to determine if all
the facilities are measuring about the same amount of vomitoxin. 
GIPSA officials believe that their check-sample program helps keep
vomitoxin test results consistent among its testing facilities. 

Second, testing experts stress the importance of using "quality
assurance (QA) pools" to reduce systemic variability.  QA pools
consist of samples of naturally contaminated barley that a testing
facility has tested many times in order to identify the true amount
of vomitoxin in the sample.  Testing facilities that practice quality
assurance using QA pools will run tests on a pool in conjunction with
daily vomitoxin tests.  If a test on the QA pool detects an amount of
vomitoxin that differs significantly from the known amount of
vomitoxin in the pool, technicians are alerted that the tests on
other samples also may be incorrect. 

Finally, testing experts said that the training of the technicians
who conduct the tests is critical for obtaining optimal test results. 
GIPSA, for instance, provides Veratox training to all personnel who
work at GIPSA-authorized testing facilities in North Dakota.\18
However, GIPSA does not oversee the training given to other
commercial grain testing facilities.  Neogen, the Veratox kit's
manufacturer, also provides training to new customers.\19


--------------------
\16 The test data were drawn from GIPSA's 1998 Sampling Variability
Study.  The purpose of the study was to determine how sample size
affects variability in vomitoxin test results, but was not intended
to represent all vomitoxin sampling efforts across North Dakota. 
Similarly, our analysis is not intended to represent all vomitoxin
testing that occurs in North Dakota, but rather to address testing
results from the two study participants that use the Veratox kit--the
manufacturer and a North Dakota commercial grain testing facility. 
See app.  II for a technical discussion of our analysis. 

\17 Quality assurance efforts are defined as any policy, procedure,
or program whose purpose is to identify and address errors and
inconsistencies in test results. 

\18 GIPSA facilitates 1 day of training with the Veratox kit in
coordination with the manufacturer--4 hours of lecture on the nature
of mycotoxins and how the kit works and 4 hours of demonstration and
practice led by a Neogen representative. 

\19 Neogen's training consists of hands-on practice with the test
kit.  A Neogen official said that training normally lasts from 4 to 8
hours, but Neogen trainers will work with customers as long as
necessary to ensure that they can use the kit. 


      REFERENCE METHODS PRODUCE
      ACCURATE RESULTS, BUT ARE
      NOT PRACTICAL FOR COMMERCIAL
      USE
---------------------------------------------------------- Letter :4.2

According to testing experts, the HPLC and GC testing methods are
widely accepted among analytical chemists for providing accurate and
consistent results.  For example, the Association of Official
Analytical Chemists\20 has approved a GC method and reviewed an HPLC
method; and the American Society of Brewing Chemists\21 has approved
a GC method for industry use.  In addition, the HPLC and GC methods
are sometimes used to assess the performance of commercial test kits,
including Veratox, because these chromatographic methods, according
to testing experts, have less variability in their test results.  For
instance, GIPSA evaluates the performance of any new commercial test
kit against its HPLC reference method before permitting its use by
GIPSA employees and GIPSA-authorized testing facilities. 
Furthermore, GIPSA uses the HPLC method in its check-sample program. 

While these reference methods have less variability than Veratox,
they are not practical for use at commercial testing facilities and
grain elevators for several reasons, according to experts we spoke
with.  First, the procedures for preparing and testing the vomitoxin
samples for these methods take several hours to complete.  However,
during the barley harvest, farmers typically deliver their barley to
grain elevators by trucks that must unload and return to the fields
for other loads.  Because of the need for quick turnaround, the
farmers, elevators, and truck drivers cannot wait several hours for a
vomitoxin test to be conducted.  In comparison, the Veratox test
takes about 30 minutes to conduct.  Second, the HPLC and GC methods
require thousands of dollars in equipment investments.  For example,
HPLC and GC test equipment cost between $40,000 to $60,000 to
purchase, while the Veratox test equipment costs about $3,200. 


--------------------
\20 The Association of Official Analytical Chemists (International)
is an independent association devoted to promoting methods validation
and quality of measurements in the analytical sciences through, among
other things, studying, validating, and approving methods of
analysis. 

\21 The American Society of Brewing Chemists was founded to improve
and bring uniformity to the brewing industry on a technical level by
resolving technical problems on an industrywide basis, keeping
current on the technical needs of the brewing industry, and
anticipating the industry's future concerns. 


   UNDER CONDITIONS OF LIGHT
   INFESTATION, SHORT- AND
   LONGER-TERM ACTIONS MAY REDUCE
   THE IMPACT OF SCAB AND
   VOMITOXIN
------------------------------------------------------------ Letter :5

In barley, scab, and the vomitoxin resulting from scab, can be
reduced somewhat through the use of fungicides and certain farming
practices, such as crop rotation and deeper tillage of the soil.\22
However, costs and other factors limit the usefulness of these
actions, and their impact is minimal when the infestation is
severe.\23 In addition, varieties of barley that are more resistant
to scab and vomitoxin will not be commercially available for at least
6 years.  According to cereal scientists, improved barley varieties
combined with short-term actions may eventually help some farmers to
better manage scab and vomitoxin infestations, thereby reducing
farmers' financial losses.  However, it is unlikely that vomitoxin
will be completely eliminated in the foreseeable future. 


--------------------
\22 As mentioned previously, vomitoxin is caused by the scab disease. 
Thus, all efforts to reduce vomitoxin must begin with reducing the
occurrence of scab. 

\23 Although a formal definition does not exist, one North Dakota
extension agent described a year of light infestation as one in which
5 to 10 percent of the barley crop acres in one county are
contaminated with scab and vomitoxin; a year of moderate infestation
as one in which about 50 percent of the barley acres in one county
are contaminated; and a year of severe infestation as one in which 75
percent or more of the barley acres in one county are contaminated. 


      SHORT-TERM ACTIONS MAY BE
      HELPFUL, BUT HAVE
      LIMITATIONS
---------------------------------------------------------- Letter :5.1

According to North Dakota extension agents\24 and cereal scientists,
a number of short-term actions can help farmers reduce scab, and thus
vomitoxin concentrations, in barley.  First, crop rotation--changing
the type of crop planted each growing season--enriches the nutrients
in the soil and decreases the incidence of crop disease.  Although
most farmers rotate crops routinely, the inclusion of more broadleaf
crops in a rotation is likely to help decrease the levels of scab in
the soil.  Broadleaf crops, such as sunflowers, canola, and
sugarbeets, are not as susceptible to scab as cereal grains, such as
barley and wheat.  However, even if rotation initially helps reduce
scab levels, infestation could occur from airborne spores from other
locations.  Furthermore, other problems could discourage the use of
crop rotations:  (1) some broadleaf crops (such as sugarbeets)
require costly equipment and costly contractual agreements and (2)
many broadleaf crops cannot be grown in certain parts of North
Dakota, thereby limiting the number of crops that can be included in
rotations.  For example, some farmers in north central North Dakota
cannot easily grow beans because the climate is generally too cold
and the growing season is too short.  As a result, these farmers have
shorter rotation cycles and are forced to more quickly return to
crops (such as barley and wheat) that are highly susceptible to scab. 

Second, deep tilling to completely overturn the soil--which does not
occur with conventional tilling--could reduce scab levels.  Since
scab stays through the winter in infected crop stubble, tilling
deeper into the soil buries any infected residue and can help prevent
scab from spreading to the next year's crop.  However, deep-till
practices result in less moisture in the soil, causing farmland to
become more prone to wind and water erosion, and are therefore not
practical for farmers in the drier portions of North Dakota (such as
the western portion of the state).  Deep tilling also requires
farmers to purchase more expensive tilling equipment.  Furthermore,
as with crop rotation, infestation can occur from airborne spores if
even one scab-infected farm in an area does not use deep tilling. 
Thus, for optimal effectiveness, deep tilling has to be conducted
across many farms. 

Third, applying fungicides can help reduce vomitoxin.  However,
fungicides are not always reliable because of weather conditions and
the difficulties associated with applying them.  North Dakota farmers
primarily use two types of fungicides, protectant and systemic. 
Protectant fungicides (which cover the plant externally) have been
used for a number of years and are easily washed off by rain and
degraded by sunlight.  Systemic fungicides, which are newer, get
absorbed into the barley plant within 4 to 8 hours of application and
are not affected by sunlight or water.  However, the timing of the
application of both systemic and protectant fungicides is critical. 
They must be applied immediately after the barley flower blossoms
because a new flower can become infected with airborne scab spores
within 3 to 4 days.  Once the barley flower is infected with the scab
fungus, the fungus has the potential to produce vomitoxin.  In
addition, a farmer can expect to spend between about $90,000 and
$138,000 to spray a 3,000-acre barley crop with a fungicide.\25 Thus,
in deciding whether to use fungicides, farmers must compare the costs
they will incur in applying them with the higher price they could
receive if their barley is less contaminated with vomitoxin. 

North Dakota extension agents told us that using the deep-till and
rotation farming practices with fungicides increases the overall
effectiveness of these short-term actions in reducing scab and
lowering vomitoxin levels.  However, they also noted that if airborne
scab spores are widespread and weather conditions are favorable to
fungal growth, barley crops would still become contaminated.  Thus,
they believe that these short-term actions will be effective only in
years of light infestation. 


--------------------
\24 University-based extension specialists interact with scientists
and relay scientific and other knowledge to farmers and other
research customers. 

\25 The range of costs for fungicides is due primarily to whether a
crop is sprayed from the ground or from the air.  If a farmer chooses
to spray a field from the ground, a large capital investment in
equipment is required. 


      MORE RESISTANT BARLEY
      VARIETIES ARE EXPECTED TO BE
      COMMERCIALLY AVAILABLE IN
      SEVERAL YEARS
---------------------------------------------------------- Letter :5.2

North Dakota State University, the University of Minnesota, South
Dakota State University, and Busch Agricultural Resource, Inc.,\26
began a cooperative breeding effort to develop more scab-resistant
barley in 1994.  The four institutions exchange and test potential
new varieties of barley.  They also share information about new
barley varieties that show resistance to scab and vomitoxin. 

In March 1997, a U.S.  Wheat and Barley Scab Initiative was formed by
scientists, members of the wheat and barley industries, commodity
groups, and others to call national attention to the scab problem and
to set national priorities for scab research.  In fiscal year 1998,
the Congress appropriated $500,000 to USDA to fund the scab research
plan established by the leaders of the initiative; in fiscal year
1999, an additional $3 million was appropriated for the effort. 
Several of the research areas focus on developing more resistant
varieties and assessing the effectiveness of fungicides in combating
scab.  Although USDA's Agricultural Research Service (ARS) is funding
the initiative, scientists at state land grant universities,
including North Dakota State University, will perform most of the
research. 

According to barley breeders and farming experts, because of many
scientific and commercial requirements, it takes about 8 to 10 years
to breed, test, and release a new variety of barley.  The breeding
process includes several steps.  First, a breeder must identify the
genetic characteristics that could make the barley more resistant to
vomitoxin.  Second, these characteristics need to be combined and
strengthened through successive new generations of barley varieties. 
Third, new varieties must be tested under multiple environmental
conditions to ensure that they are truly resistant.  During the
breeding process, new varieties may sometimes appear to be resistant
to scab when, in fact, they are not.  For example, if a greenhouse
containing a new variety being tested for resistance is kept cool and
limited moisture is allowed to accumulate on the barley, little scab
will grow.  This may lead the breeder to believe that the variety is
scab-resistant, while, in fact, the greenhouse environment suppressed
scab growth. 

Fourth, after a breeder is confident that new varieties are truly
resistant, they must be tested and screened for necessary malting and
brewing qualities.  For example, a new variety of barley must be
uniform in size and have plump kernels (necessary for successful beer
brewing) or maltsters and brewers will not be interested in buying
it.\27 According to scientists, while some more resistant barley
varieties are currently undergoing commercial trials by maltsters and
brewers, none contain all of the characteristics that the industry
requires. 

Lastly, new barley varieties must be tested for commercial viability. 
Any new variety of barley that meets the malting and brewing
industry's requirements would also have to be high-yielding in order
for it to be commercially attractive to farmers.  Scientists estimate
that a commercially acceptable, more scab-resistant barley variety is
at least 6 years away. 

Breeders expect that, over time, new, more resistant barley, combined
with short-term actions may help farmers to better manage scab and
vomitoxin infestations and reduce their financial losses.  However,
these experts state that a more resistant barley variety will not
completely eliminate the incidence of scab and vomitoxin,
particularly during periods of moderate or severe infestation. 


--------------------
\26 Busch Agricultural Resource, Inc., is an agricultural research
and operations subsidiary providing brewing raw materials to
Anheuser-Busch, Inc. 

\27 A University of Minnesota plant pathologist estimates that
breeders must consider at least 30 different barley traits in order
to develop barley that meets the brewing industry's needs. 


   AGENCY COMMENTS
------------------------------------------------------------ Letter :6

We provided a draft of this report to USDA for its review and
comment.  We met with the Deputy Administrator, Grain Inspection,
Packers and Stockyards Administration, and with other officials from
that organization and USDA's Agricultural Research Service.  The
officials generally agreed with the information presented in the
report and provided several technical changes and clarifications.  We
have incorporated these changes as appropriate. 


   SCOPE AND METHODOLOGY
------------------------------------------------------------ Letter :7

You asked us to (1) determine the financial impact of scab and
vomitoxin on North Dakota barley farmers, (2) assess the performance
of vomitoxin test methods, and (3) identify short- and long-term
actions that could help reduce the impact of scab and vomitoxin on
North Dakota barley farmers. 

To address the first question, we collected and developed historical
data on North Dakota barley prices and production for 1959 through
1992--the period before the scab and vomitoxin epidemic--and on key
weather factors affecting production for both that period and the
blighted years.  We used these data to estimate (1) what barley
prices and production would have been in 1993 through 1997 in the
absence of scab and vomitoxin and (2) what revenues would have been
in the absence of scab and vomitoxin.  We then compared this estimate
of revenues with actual barley revenues to determine farmers' losses
by year and by crop reporting district.  We also developed
information on how prices are transmitted from the maltsters and
brewers down to the farmers, and collected data on Canadian
production and exports of malting barley to the United States during
this time period.\28 To conduct these tasks, we used data from the
North Dakota State University, GIPSA, the North Dakota Department of
Agriculture, USDA's National Agricultural Statistics Service and its
Economic Research Service, the North Dakota Barley Council, and
Agriculture Canada.  We also conducted interviews with officials from
these organizations and with North Dakota grain dealers.  (See app. 
I for a detailed description of our data sources, methodology and the
results of our analysis.)

To address the second question, we reviewed GIPSA, industry, and
academic studies on the test methods; interviewed testing experts;
and analyzed Veratox test data on vomitoxin from GIPSA's 1998
Sampling Variability Study.  Using data from the study, we assessed
the performance of vomitoxin test results on the basis of the
variability of test results between testing facilities.  Testing
experts we spoke with included officials at GIPSA, FDA, and major
U.S.  malting and brewing companies; academic researchers; and
representatives of the Association of Official Analytical Chemists,
the American Society of Brewing Chemists, the American Malting Barley
Association, the North Dakota Barley Council, and the North Dakota
Grain Dealers Association.  (See app.  II for a detailed description
of our methodology and the results of our analysis.)

To address the third question, we (1) obtained information on
academic, public, and private research on actions to reduce the
impact of scab and vomitoxin and on progress in developing more
scab-resistant barley and (2) interviewed scientists at North Dakota
State University and the University of Minnesota and officials at
USDA's Economic Research Service and Agricultural Research Service. 

Finally, we had a draft of this report reviewed for accuracy and
objectivity by several economists and agricultural experts from
academia. 

We did not independently verify the data obtained from our sources. 
Our work was conducted from April 1998 through February 1999 in
accordance with generally accepted government auditing standards. 


--------------------
\28 To convert the revenue losses to 1997 constant dollars, we used
the Department of Commerce's chain-type price index for gross
domestic product. 


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

As arranged with your office, unless you publicly announce its
contents earlier, we plan no further distribution of this report
until 30 days from the date of this letter.  At that time, we will
provide copies of this report to Chairman Richard Lugar and Ranking
Minority Member Tom Harkin of the Senate Committee on Agriculture,
Nutrition, and Forestry; Chairman Larry Combest and Ranking Minority
Member Charles Stenholm of the House Committee on Agriculture; other
interested congressional committees; and the Honorable Dan Glickman,
the Secretary of Agriculture.  We will also make copies available to
others upon request. 

If you or your staff have any questions about this report, please
contact me at (202) 512-5138.  Major contributors to this report are
listed in appendix III. 

Sincerely yours,

Robert E.  Robertson
Associate Director, Food and
 Agriculture Issues


ESTIMATION OF REVENUE LOSSES TO
NORTH DAKOTA BARLEY FARMERS AS A
RESULT OF SCAB AND VOMITOXIN
=========================================================== Appendix I

This appendix explains the methods and data we used to estimate the
revenue losses for North Dakota barley as a result of the scab and
vomitoxin epidemic for 1993 through 1997.\29 To develop this
estimate, we first estimated what barley revenues would have been in
the absence of the vomitoxin epidemic in North Dakota.\30 This
required estimating what production levels and prices would have been
in each district in each year.  In turn, estimating production levels
required estimating both yields and the ratios of
harvested-to-planted acres in the absence of the disease.  We then
compared estimated barley revenues without the disease to actual
barley revenues received, which we calculated from price and
production data, to obtain estimated losses.  We then totaled all the
crop reporting districts and all the years to obtain an estimate of
total losses during this period. 


--------------------
\29 The method used in this analysis is adapted from a report by D. 
Demcey Johnson, George K.  Flaskerud, Richard D.  Taylor, and
Vidyashankara Satyanarayana, "Economic Impacts of Fusarium Head
Blight in Wheat," Agricultural Economics Report No.  396, June 1988,
Department of Agricultural Economics, North Dakota State University,
Fargo, North Dakota. 

\30 North Dakota has nine major crop reporting districts (CRD).  For
this analysis, we focused on the five regions where substantial scab
and vomitoxin outbreaks have occurred--the North Central, Northeast,
Central, East Central, and Southeast CRDs of the state. 


   ESTIMATING REVENUE LOSSES IN
   THE ABSENCE OF SCAB AND
   VOMITOXIN
--------------------------------------------------------- Appendix I:1

To estimate losses resulting from scab and vomitoxin, we first
estimated what barley revenues would have been during this time
period if the epidemic had not occurred, but all other relevant
factors (such as weather) had been unchanged.  We estimated both
production levels and prices and multiplied them to obtain estimated
revenues. 


      ESTIMATING PRODUCTION IN THE
      ABSENCE OF SCAB AND
      VOMITOXIN
------------------------------------------------------- Appendix I:1.1

As a first step in estimating production, we used a regression
analysis to estimate barley yields from 1959 through 1992 (before the
scab and vomitoxin epidemic) for region i in time period t as a
function of weather events and a time trend: 

(1)



(See figure in printed edition.)

where

yit = harvested yield in region i in year t

Pit = the difference between average total precipitation and total
precipitation during the growing season divided by the standard
deviation of total rainfall for region i and year t

Pit\2 = the squared value of Pit , the precipitation deviation
variable

Tit = the difference between historical average temperature during
the growing season and average temperature during the growing season
divided by the standard deviation of average temperature for region
i, year t

t = a time trend variable, t=1,...,34

In this regression, we transformed both average growing season
temperature and total rainfall to measures of deviations by
subtracting their historical average levels from their actual levels
and dividing by their standard deviations.\31 As a result, these
variables measure how close a particular year's average temperature
or total rainfall is to its historical average.  For example, values
greater than +1 are associated with hot weather or wet months; values
less than -1 are associated with dry or cool months; and values
between +1 and -1 are near the average.  We used these transformed
weather variables in the regression rather than the actual values
because they were more significantly related to yield and contained
less multicollinearity.  In addition, because there is an optimum
level of precipitation, beyond which yields may decrease, we included
a squared precipitation term in our equation.  Other agricultural
economists analyzing yield have also used squared precipitation
terms.  Finally, we inserted an annual time trend to represent yield
changes because of changes in such things as technology, input use,
or farm size. 

Table I.1 displays our estimates of the parameters of these
regression equations for each CRD analyzed.  Except for Pit \2 in
CRDs 3 and 5, all independent variables were significant at the 0.05
level and above and displayed the expected signs. 



                                        Table I.1
                         
                             Barley Yield Equation Parameter
                           Estimates by Crop Reporting District

                                       Crop reporting district
             ----------------------------------------------------------------------------
                    CRD 2
                    North           CRD 3           CRD 5           CRD 6           CRD 9
                Central\a       Northeast         Central    East Central       Southeast
-----------  ------------  --------------  --------------  --------------  --------------
Intercept         25.87\b         24.43\b         21.28\b         27.20\b         26.43\b
                   (9.06)          (8.48)          (7.73)         (10.01)          (9.66)
Precipitati        4.10\b          3.26\b          5.37\b          2.96\b          5.49\b
 on                (3.72)          (2.72)          (4.25)          (2.33)          (3.56)
 deviation
 (Pit )
Precipitati       -2.65\b           -2.31           -1.27         -2.27\b         -3.60\b
 on               (-2.47)         (-1.65)         (-1.34)         (-2.14)         (-3.83)
 deviation
 squared
 (Pit\2)
                  -3.89\b         -3.41\b         -4.56\b         -3.57\b         -2.87\b
 Temperature      (-3.31)         (-2.47)         (-3.44)         (-2.67)         (-2.07)
 deviation
 (Tit)
                   0.72\b          1.15\b          0.93\b          1.17\b          0.92\b
 Time trend        (5.64)          (9.00)          (7.02)          (9.09)          (7.01)
 (t)
R\2                  0.71            0.75            0.69            0.76            0.70
Adjusted             0.68            0.72            0.64            0.72            0.66
 R\2
DW                   1.74            1.79            1.62            1.91            1.79
Number of              34              34              34              34              34
 Observatio
 ns
-----------------------------------------------------------------------------------------
Note:  Numbers in the parentheses are t-values. 

\a Indicates error structure corrected for first order
autocorrelation. 

\b Indicates parameter is statistically significant at the 0.05 level
or higher

We also performed a Chow test to determine whether barley yields were
homogeneous across CRDs and, thus, if we could pool all of our data
into one regression equation.  This hypothesis, however, was rejected
at the 0.05 level, and we therefore used our yield estimates from the
regressions of the separate CRDs in our analysis. 

In equation 2, we calculated yield in the absence of scab and
vomitoxin as a weighted average of predicted yield from equation 1
and actual yield. 

(2)



(See figure in printed edition.)

In equation 2, ynit denotes yield in the absence of scab and
vomitoxin, yfit the predicted yield from equation 1, and ysit the
actual yield in a scab-infected year.  The fraction of yield
shortfall attributable to scab and vomitoxin is denoted it. 
If vomitoxin were the only factor accounting for a shortfall during
the scab-infected years, then  = 1 and ynit = yfit ; that
is, the yield that would have occurred in the absence of the disease
equals the predicted yield from equation 1. 

Since scab and vomitoxin occur simultaneously with other crop
diseases or weather problems, such as flooding, in these regions, we
needed to estimate the fraction of yield shortfall resulting only
from the disease.  In order to obtain these estimates, we first asked
experts at the North Dakota State Extension Service as well as the
North Dakota State Plant Pathology Department what yield losses (in
percentages) were due only to scab and vomitoxin for each year and
each district.  We then multiplied these percentages for each CRD and
each year by our predicted yields, or the yields in the absence of
vomitoxin, to determine yield loss in bushels per acre.  In order to
estimate the percent of yield loss resulting only from the disease,
we then divided these yield losses by the difference between
predicted and actual yields (or total yield shortfall) in order to
estimate the fraction of yield shortfall resulting only from the
disease.  These calculated percentages are shown in table I.2: 


--------------------
\31 Lloyd D.  Teigen and Milton Thomas, Jr., Weather and Yield,
1950-1994:  Relationships, Distributions, and Data, Economic Research
Service Staff Paper, Commercial Agriculture Division, Number 9527. 


      VOMITOXIN
------------------------------------------------------- Appendix I:1.2



                                    Table I.2
                     
                      Fraction of Yield Shortfall Resulting
                     from the Presence of Scab and Vomitoxin
                                 by CRD, 1993-97

                              Crop reporting district
    ----------------------------------------------------------------------------
           CRD 2
Ye         North           CRD 3           CRD 5           CRD 6           CRD 9
ar       Central       Northeast         Central    East Central       Southeast
--  ------------  --------------  --------------  --------------  --------------
19          1.00            0.24      0.26 -0.43      0.17 -0.28      0.19 -0.31
 93
19          1.00      0.32 -0.65            1.00      0.18 -0.30      0.19 -0.32
 94
19    0.26 -0.52            0.34      0.16 -0.32      0.15 -0.31      0.15 -0.30
 95
19    0.84 -1.00      0.69 -0.93            1.00            0.66      0.75 -1.00
 96
19    0.43 -0.87      0.40 -0.53      0.45 -0.68      0.11 -0.26      0.19 -0.38
 97
--------------------------------------------------------------------------------
In addition to estimating yield in the absence of scab and vomitoxin,
we needed to calculate the ratio of harvested-to-planted acres to
estimate barley production.  During the years of the epidemic, many
acres that were planted to barley actually went unharvested.  Because
the ratio of actual harvested-to-planted acreage during the
scab-infected years might have differed from the predicted ratio for
reasons other than scab and vomitoxin, we again used a weighted
average of the predicted and actual ratios to estimate the ratio in
the absence of the disease.  We used past values of the ratio of
harvested-to-planted acreage as the predicted values, but we used the
same  values to measure the fraction of yield shortfall
resulting from scab and vomitoxin.  Specifically, in equation 3, we
calculated the ratio of harvested-to-planted acres to account for
acreage that was left abandoned because of scab and vomitoxin for
each region for each time period as:\32

(3)



(See figure in printed edition.)

where

ahit = actual harvested acres in time period t in CRD i

apit = actual planted acres in time period t in CRD i

Ri = the average\33 of the ratio of harvested-to-planted acres,
1983-92

Rnit = the ratio of planted-to-harvested acres, in the absence of
vomitoxin

it = the same adjustment factor used to calculate yield
without vomitoxin

Finally, we combined our estimates of yield and the ratio of
harvested-to-planted acreage in the absence of vomitoxin to estimate
production in the absence of vomitoxin, qnit: 

(4)



(See figure in printed edition.)

In order to estimate production in the absence of scab and vomitoxin,
without overestimating losses, we used the maximum of estimated yield
in the absence of vomitoxin and actual yield and the maximum of the
calculated ratio of harvested-to-planted acres without vomitoxin and
the actual ratio.  For example, if the estimated yield falls below
actual yield in a scab year, actual yield would be used instead of
the estimated yield (without scab/vomitoxin) to estimate production. 
The product of the second term and acres planted, apit, equals
harvested acres in a year without the presence of scab and vomitoxin. 


--------------------
\32 While this calculation is an adjustment for reductions in
harvested-to-planted barley acres, we did not make an adjustment for
the decrease in planted acres.  Overall, however, we assumed that
farmers who reduced planting barley acreage increased their acreage
of other crops.  Therefore, while revenue went down because of
reduced barley acres, revenue for the whole farm may not have
decreased because of the increased plantings of other crops. 

\33 Here, we took an olympic average of the ratio of
harvested-to-planted acres for the 10 years from 1983 through 1992. 
An olympic average omits the maximum and minimum values contained in
a given sample.  It is used when the sample is small and may have
observations that are unrepresentative (such as a drought year). 


      ESTIMATING BARLEY MALTING
      PREMIUMS AND FEED GRAIN
      PRICES IN THE ABSENCE OF
      VOMITOXIN
------------------------------------------------------- Appendix I:1.3

As the next step in determining barley revenue in the absence of
vomitoxin, we estimated both malting barley premiums and feed grain
prices for 1993 through 1997 had there been no disease.  To do this,
we used regression analysis and historical data on price and
production from 1959 through 1992 to estimate price equations for
both malting barley premiums and feed grain prices.  First, we
explain malting premium price movements by total barley production or
its relationship to the larger national barley market.  Since the
proportion of malting barley in the entire crop was fairly stable in
the years prior to the vomitoxin epidemic, increases in total barley
production translate into increases in the quantities of malting
barley.  Moreover, while there are differences in premiums from
region to region, prices are generally transmitted from the malting
and brewing industries at a more aggregate market level.  Therefore,
in equation 5 we specify the historical association between malting
premiums, Pi \m and total U.S.  barley production, QT, for each CRD
analyzed, i: 

(5)



(See figure in printed edition.)

Table I.3 shows the results of this analysis. 



                                        Table I.3
                         
                             Malting Barley Premium Parameter
                           Estimates by Crop Reporting District

                                         Crop reporting district
                 ------------------------------------------------------------------------
                        CRD 2
Independent             North         CRD 3         CRD 5           CRD 6           CRD 9
variable              Central     Northeast       Central    East Central       Southeast
---------------  ------------  ------------  ------------  --------------  --------------
Intercept              0.88\a        1.42\a        1.07\a          2.05\a          1.07\a
                       (3.68)        (6.16)        (4.48)          (6.85)          (4.23)
Total               -0.0015\a     -0.0026\a     -0.0018\a       -0.0039\a       -0.0018\a
 production           (-2.78)       (-5.29)       (-3.54)         (-6.07)         (-3.18)
 (QT)
Reg R\2                  0.20          0.47          0.29            0.54            0.25
DW                       1.66          1.77          1.81            1.84            2.00
Observations               34            34            34              34              34
-----------------------------------------------------------------------------------------
Note:  Numbers in the parentheses are t-values. 

\a Indicates the parameter is statistically significant at the 0.05
level or higher. 

As table I.3 shows, we found a negative and highly significant
association between malting premiums and total barley production at
the national level for all CRDs.  We also tried other variations of
this regression model, including ones using combinations of stocks as
well as barley yields for independent variables.  However, these
variables did not perform as well as the total barley production
variable.  Because of the presence of positive serial correlation in
all CRDs, we used the Yule-Walker regression technique\34 to derive
our estimates.  In general, serial correlation causes standard errors
to be biased downward, thus indicating that parameter estimates are
more precise than they actually are.  Therefore, correcting for this
problem leads to more efficient parameter estimates. 

In the feed grain market, corn is the primary feed grain product,
accounting for more than 80 percent of total feed grain consumption. 
Because barley feed grain prices, Pi\f , are driven primarily by corn
prices, in equation 6, we specify the historical association between
feed grain barley prices, the price of corn, PC , and total U.S. 
barley production, QT , as: 

(6)



(See figure in printed edition.)

To correct for first-order serial correlation, as in the malting
premium regression models, we used the Yule-Walker regression
technique for the feed grain regressions.  As table I.4 indicates,
the total barley production variable displayed a negative sign and
was significant at the 0.10 percent level and above in all CRDs
except 6.  In all CRDs, the price of corn was positively related to
barley feed grain prices and highly statistically significant. 



                                        Table I.4
                         
                         Feed Grain Barley Parameter Estimates by
                                 Crop Reporting District

                                         Crop Reporting District
                 ------------------------------------------------------------------------
                        CRD 2
Independent             North         CRD 3         CRD 5           CRD 6           CRD 9
variable              Central     Northeast       Central    East Central       Southeast
---------------  ------------  ------------  ------------  --------------  --------------
Intercept                0.24          0.28          0.21            0.22            0.21
                       (1.19)        (1.48)        (1.19)          (1.13)          (1.04)
Corn                   0.78\a        0.75\a        0.77\a          0.75\a          0.78\a
 price (PC)           (17.75)       (18.18)       (19.81)         (17.42)         (17.49)
Total               -0.0009\a     -0.0008\a     -0.0007\b         -0.0006       -0.0007\b
 production           (-2.07)       (-2.10)       (-2.00)         (-1.39)         (-1.76)
 (QT)
Reg R\2                  0.91          0.92          0.93            0.91            0.91
DW                       1.94          1.93          1.91            1.87            1.91
Observations               34            34            34              34              34
-----------------------------------------------------------------------------------------
Note:  Numbers in parentheses are t-values. 

\a Indicates parameter is statistically significant at the 0.05 level
or higher. 

\b Indicates parameter is statistically significant at the 0.10
level. 

Substituting in actual values of barley production and corn prices
for years 1993 through 1997, we used these regression parameters to
predict what malting barley and feed grain barley prices would have
been in the absence of the vomitoxin epidemic for these years.  We
assume that malting barley prices are the sum of estimated feed grain
prices plus estimated malting premiums. 


--------------------
\34 The Yule-Walker regression technique starts by forming the
ordinary least-squares estimate of the parameters.  Next, given the
vector of autoregressive parameters (using the Yule-Walker equations)
and the variance matrix of the error vector, efficient estimates of
the regression parameters are computed using generalized least
squares. 


      ESTIMATION OF BARLEY REVENUE
      IN THE ABSENCE OF SCAB AND
      VOMITOXIN
------------------------------------------------------- Appendix I:1.4

As the final step in estimating barley revenue in the absence of
vomitoxin, we combined our previously obtained estimates of
production and prices without the disease to obtain revenue as the
product of production and price.  However, since barley production
data are only for total production, and are not separated out for the
malting and the feed grain markets, we first needed to allocate total
production to these markets.  We derived the proportion of the crop
sold as malting barley and feed grain barley by using actual data on
the prices of malting barley, PM , feed grain barley, PF, and the
total average barley price, PB .  Because the overall price of
barley, PB , is a weighted average of the malting and feed grain
price, using\35 equations 7 and 8, we can obtain the proportion of
barley that is sold to the malting market, nbari\m , and the
proportion sold to the feed grain market (1 - nbari\m ) as equation 7
shows: 

(7)



(See figure in printed edition.)

Rearranging terms, we can express the proportion of barley sold to
the malting market as a function of observed prices as equation 8
shows: 

(8)



(See figure in printed edition.)

Using historical malting and feed grain prices from 1959 through
1992, we obtained proportions for each year and took their average to
derive malting barley and feed grain weights.  These weights
represent the proportions of malting and feed grain barley in the
market in a year without the disease.  Table I.5 shows, for each CRD,
these estimated average weights: 



                                        Table I.5
                         
                         Estimated Average Malting and Feed Grain
                           Weights by Crop Reporting District,
                                         1959-92

                        CRD 2
                        North         CRD 3         CRD 5           CRD 6           CRD 9
Barley market         Central     Northeast       Central    East Central       Southeast
---------------  ------------  ------------  ------------  --------------  --------------
Malting                  0.71          0.68          0.62            0.79            0.60
 (nbar\mi)
Feed grain (1-           0.29          0.32          0.38            0.21            0.40
 nbar\mi)
-----------------------------------------------------------------------------------------
To estimate the amount of production that would have gone to the
malting barley and feed grain markets for each district in each year
from 1993 through 1997 in the absence of vomitoxin, we multiplied
these weights by our estimate of total barley production (without
vomitoxin).  For instance, in order to account for the amount of
barley that typically went into the malting side of the market for
CRD i in year t, we multiplied nbari\m , the average proportion of
malting barley for that CRD, by the estimated production in that
district in year t in the absence of the disease, qnit (from equation
4).  Finally, to estimate malting barley revenue for years 1993
through 1997 in the absence of vomitoxin, we multiplied the estimated
malting barley production for each district and year by the predicted
malting barley price (in the absence of vomitoxin), nPi\m , for that
district and year.  We used the same procedure to estimate the
revenue for all CRDs for the feed barley market.  Equation 9
summarizes how we estimated total barley revenue in a particular
district and year in the absence of vomitoxin (NREVit): 

(9)



(See figure in printed edition.)

where

nbari\m = proportion of malting barley production without vomitoxin
for CRD i

nbari\f = proportion of feed barley production without vomitoxin for
CRD i

qnit = total quantity of barley production for CRD i in year t, (from
eq.  4)

nPit\m = predicted malting barley price, without vomitoxin, for CRD
i, time t

nPit\f = predicted feed grain barley price, without vomitoxin, CRD i,
time t

We used the chain-type price index for gross domestic product to
express all revenues in 1997 dollars and then totaled over the years
1993 through 1997 to obtain an estimate for each district of what
barley revenues would have been during this period in the absence of
vomitoxin. 


--------------------
\35 In order to calculate average barley prices, the National
Agricultural Statistics Service (NASS) surveys farm elevators for
malt and feed grain prices as well as quantities sold in North Dakota
counties.  Typically, this covers about one-third of the barley
farmers in a county.  According to the Associate Administrator of
NASS, three caveats are associated with this method.  First, there
may be some error because of sampling expansion factors; second, the
measurement is taken at the elevator, not where the grain actually
comes from; and third, there may be some discrepancy because of
carryover stocks from one year to another. 


      CALCULATING ACTUAL REVENUE
      FROM MALTING BARLEY AND FEED
      GRAIN BARLEY
------------------------------------------------------- Appendix I:1.5

Using equation 10, we calculate the actual amount of revenue from
barley production, AREVit , for CRD i, in time period t as: 

(10)



(See figure in printed edition.)

The actual amount of production in each CRD in each year is denoted
qait , while the actual market prices of malting barley and feed
barley are represented by aPit\m and aPit\f , respectively.  For each
year between 1993 and 1997 and for each CRD, we calculated the
proportion of barley sold to the malting market, abarit\m and the
proportion sold as feed, abarit\f , using the same method as we did
in equation 8.  Table I.6 displays these weights for each CRD. 



                                    Table I.6
                     
                     Estimated Malting Barley and Feed Grain
                        Barley Weights, by Crop Reporting
                                District, 1993-97

                           CRD 2                               CRD 6
                           North       CRD 3       CRD 5        East       CRD 9
Market          Year     Central   Northeast     Central     Central   Southeast
------  ------------  ----------  ----------  ----------  ----------  ----------
Maltin          1993        0.22        0.28        0.61        0.42        0.18
 g
 barle
 y
                1994        0.78        0.53        0.67        0.32        0.53
                1995        0.85        0.72        0.70        0.55        0.67
                1996        0.58        0.43        0.91        0.77        0.43
                1997        0.58        0.20        0.68        0.28        0.84
Barley          1993        0.78        0.72        0.39        0.58        0.82
 feed
 grain
                1994        0.22        0.47        0.33        0.68        0.48
                1995        0.15        0.28         0.3        0.45        0.33
                1996        0.42        0.57        0.09        0.23        0.57
                1997        0.42         0.8        0.32        0.72        0.16
--------------------------------------------------------------------------------

      CALCULATING CHANGES IN FARM
      REVENUE RESULTING FROM SCAB
      AND VOMITOXIN
------------------------------------------------------- Appendix I:1.6

Using equation 11, we calculated total changes in revenue from barley
production for North Dakota due to scab and vomitoxin, it,
as: 

(11)



(See figure in printed edition.)

This total represents the sum of the differences between the actual
revenue, AREVit, and the predicted revenue in each year, in the
absence of vomitoxin, NREVit , for each CRD i in each year, 1993
through 1997. 


   DATA USED TO ESTIMATE LOSSES IN
   FARM REVENUES FROM SCAB AND
   VOMITOXIN
--------------------------------------------------------- Appendix I:2

We gathered data from several sources for our calculation of the
revenue losses for North Dakota barley as a result of scab and
vomitoxin.  Our main source of data, the North Dakota Agricultural
Statistics Service, provided information by CRD on planted and
harvested barley acres, total barley production, malting barley
prices, feed grain barley prices, and average barley yields for 1959
through 1997.  We used weather data, average temperature and total
precipitation by CRD from 1950 to 1997, supplied by USDA's Economic
Research Service as well as by North Dakota State University.  North
Dakota State University area crop extensionists and plant
pathologists familiar with vomitoxin provided estimates of the
fraction of yield shortfall attributed to vomitoxin for 1993 through
1997.  Finally, we used data on U.S.  barley production and corn
prices from 1959 to 1997 from NASS in our estimation of malting
barley premiums and barley feed prices. 


ANALYSIS OF THE VERATOX TEST KIT'S
PERFORMACE
========================================================== Appendix II

According to our analysis of the data from the Grain Inspection,
Packers and Stockyards Administration's (GIPSA) 1998 Sampling
Variability Study, the Veratox test results from Neogen (the kit's
manufacturer) and Grand Forks Grain Inspection, Inc.  (a
GIPSA-authorized testing facility) differed significantly from each
other and from GIPSA's high pressure liquid chromatography (HPLC)
results.  Our analysis is not projectable to all Veratox test results
in North Dakota because the data from GIPSA's sampling study are not
representative of all Veratox testing and barley sampling throughout
the state. 


   GIPSA'S 1998 DATA FROM ITS
   SAMPLING VARIABILITY STUDY
-------------------------------------------------------- Appendix II:1

GIPSA's sampling study was designed to determine how sampling size
and method affect variability in vomitoxin test results.  In this
study, GIPSA (1) obtained six bulk barley samples from various
elevators to study the effect of sampling size on variability and (2)
sampled 10 trucks using different sampling methods to determine the
effect of sampling method on variability.  All samples were cleaned,
ground, and subdivided into portions for testing by the Grand Forks
Grain Inspection's laboratory.  Additional portions were provided to
Neogen and Romer (another test kit manufacturer) for testing in their
respective laboratories, and portions of the truck samples were
tested by GIPSA in its Kansas City, Missouri, laboratory using the
HPLC method.  Neogen and Grand Forks Grain Inspection tested each
subsample using Neogen's Veratox test kit.\36 Neogen performed two
tests on each subsample it received, and Grand Forks Grain Inspection
performed one test on each subsample it received. 

GIPSA did not intend to have the results from its barley sampling
study represent the variability that exists with all barley sampling
in North Dakota.  It selected its test lots to ensure that vomitoxin
concentration levels in the samples would fall within the Veratox
test kit's range of measurement ability--that is, from 0.5 parts per
million (ppm) to 5 ppm.  In addition, test data were from samples
that differed in size and method of collection because GIPSA's
purpose was to assess the effect of these variables (size and
sampling method) on vomitoxin test results.  However, because GIPSA
found that sample size and sampling method did not significantly
alter the variability of test results, we concluded that the lack of
uniformity in sample size and sampling method is not a significant
limitation to our analysis. 

We analyzed 376 Veratox tests performed by Grand Forks Grain
Inspection and 692 tests performed by Neogen.\37 According to GIPSA
officials, greater variability occurs when results from multiple test
facilities are analyzed.  Thus, since our analysis is based on data
from only two testing facilities, our results may not be
representative of the true amount of variability in vomitoxin test
results conducted in North Dakota. 


--------------------
\36 The samples were also tested by North Dakota State University
using gas chromatography and Romer Labs, Incorporated, using its
FluoroQuant test kit and the HPLC method. 

\37 We originally obtained 1,287 Veratox tests (which had
corresponding HPLC test results of less than 5.1 ppm) from GIPSA's
sampling study, but excluded 132 Veratox tests from our analysis
because they either lacked comparative data or exceeded the
calibration range of the Veratox method.  We then excluded the
remaining 87 records in the concentration category 4.1-5 parts per
million (ppm) because the removal of records from the tests that
exceeded the calibration range of the Veratox method greatly reduced
the number of records within this category. 


   GAO'S ANALYSIS
-------------------------------------------------------- Appendix II:2

According to GIPSA officials, the variability of test results differs
depending on the concentration of vomitoxin in the barley sample.  At
their recommendation, we used GIPSA's HPLC test result to represent
the true concentration of vomitoxin in a sample and grouped the
Veratox test results into four ranges.  The first range contains
results from barley samples with relatively low concentrations of
vomitoxin--those with HPLC results of 0.7 ppm to 1 ppm.  The last
category contains results from samples with the highest
concentrations of vomitoxin--those with HPLC results of 3.1 ppm to 4
ppm. 

Our analysis of Veratox test results from Neogen and the Grand Forks
testing facility showed differences in the amount of vomitoxin
measured at each location (see table II.1).  That is, testing
identical samples of barley at the testing facility and at the
manufacturer resulted in different measurements of vomitoxin. 
Specifically, using the HPLC test results to represent the true
concentration of vomitoxin, we found that at concentrations between
0.7 and 4 ppm, Neogen's estimation of vomitoxin was, on average,
higher than the testing facility's.  Given these differences, and the
fact that small differences in the amount of vomitoxin measured can
affect barley prices, we concluded that producers could have received
different prices from each testing location if the test results had
been the basis for a commercial sale. 



                               Table II.1
                
                  Difference in Veratox Results at Two
                           Testing Facilities

                         (In parts per million)

                                              We estimate that average
                                           test results at Neogen will
                                           exceed those at Grand Forks
At an HPLC result of                                              by\a
----------------------------------------  ----------------------------
0.7 --1.0                                                    0.2 --0.6
1.1 --2.0                                                    0.2 --0.6
2.1 --3.0                                                    0.6 --1.0
3.1 --4.0                                                    0.7 --1.2
----------------------------------------------------------------------
\a Our estimates represent the 95-percent confidence intervals for
the differences between the mean test results for Neogen and Grand
Forks Grain Inspection. 

Source:  GAO's analysis of data from GIPSA's 1998 Sampling
Variability Study. 

We also found in some cases that the results from Neogen and Grand
Forks Grain Inspection differed, on average, from GIPSA's HPLC
reference method (see table II.2).  Specifically, test results from
the manufacturer were higher than test results from the HPLC
reference method at three of the four concentration ranges we
reviewed.  For example, when HPLC results ranged from 0.7 to 1.0 ppm,
we estimated that the average Neogen's results would be between 1.3
to 1.5 ppm, which is higher than the average HPLC results.  In
addition, average Veratox results from the Grand Forks facility were
lower than the reference method at two of four concentration ranges. 
For instance, when HPLC results ranged from 2.1 to 3.0 ppm, we
estimated that the average test result from the testing facility
would be between 1.7 to 2.0 ppm, which is lower than the average for
the reference method.  The fact that in one case the manufacturer's
test results were higher, on average, than the reference method's
results, while the testing facility's results were lower, further
demonstrates that variability can occur among testing facilities
using the Veratox test kit. 



                               Table II.2
                
                 Comparison of Average Veratox and HPLC
                              Test Results

                         (In parts per million)

                         We estimate the average Veratox result from
                             repeated tests will fall between:\a
                        ----------------------------------------------
At an HPLC result of:   Neogen                  Grand Forks
----------------------  ----------------------  ----------------------
0.7 --1.0               1.3 --1.5\b             0.9 --1.2

1.1 --2.0               1.8 --2.0\b             1.4 --1.7\

2.1 --3.0               2.5 --2.8               1.7 --2.0\b

3.1 --4.0               3.5 --3.8\b             2.5 --3.0\b
----------------------------------------------------------------------
\a Our estimates represent the 95-percent confidence intervals for
the average Veratox result. 

\b Indicates a statistically significant difference between the
average HPLC and Veratox result. 

Source:  GAO's analysis of data from GIPSA's 1998 Sampling
Variability Study. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================= Appendix III

Jerilynn B.  Hoy, Assistant Director
Kurt W.  Kershow, Project Manager
Patricia M.  Crown
Barbara J.  El-Osta
Leanne M.  Flama
Jay R.  Cherlow
Karen E.  Bracey
Carol Herrnstadt Shulman


RELATED GAO PRODUCTS
============================================================ Chapter 0

Wheat Pricing:  Information on Transition to New Tests for Protein
(GAO/RCED-95-28, Dec.  8, 1994). 

Midwest Grain Quality (GAO/RCED-94-66R, Nov.  1, 1993). 


*** End of document. ***