SOVIET CLIMATE CHANGE: IMPLICATIONS FOR GRAIN PRODUCTION

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP97R00694R000400720001-7
• Release Decision: RIPPUB
• Original Classification: S
• Document Page Count: 25
• Document Creation Date: December 22, 2016
• Document Release Date: October 7, 2010
• Sequence Number: 1
• Publication Date: May 1, 1985
• Content Type: REPORT

Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Directorate of 1)r-V11 Secret
N
Soviet Climate Change:
Implications for
Grain Production
A Research Paper
Secret
GI 85-10128
May 1985
COPY 3 2 2
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Directorate of Secret
Intelligence
Soviet Climate Change:
Implications for
Grain Production[
Office of Global Issues. Comments and queries are
welcome and may be directed to the Chief,
This paper was prepared by
Strategic Resources Division, OGI,
Secret
GI 85-10128
May 1985
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
25X1
25X1
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Soviet Climate Change:
Implications for
Grain ProductionF_
Summary During the past several decades, grain production in the USSR has
Information available benefited from a general improvement in climate-increased precipitation
as of 23 April 1985 and slightly higher temperature-and massive investments in agrotech-
was used in this report.
nology. Of these two factors, climate and agrotechnology, the former has
been the most important. Aside from its direct impact on crop growth,
climate indirectly determines in large measure the effectiveness of technol-
ogy such as fertilizer applications. Other determinants of grain output such
as political decisions, the quality of management, and worker incentives,
while important, have had much less impact
While the long-term climate trend has been favorable, there has been a
slight drop in precipitation in the 1980s. However, it is too early to assume
a permanent change in the long-term pattern. Our analysis indicates that
precipitation probably will remain near its present level during the rest of
the decade. At the same time, we expect temperatures to continue their rise
in the grain area because of worldwide increases -in atmospheric carbon
dioxide. Temperature increases will lengthen the growing season in the
north-providing opportunities for increased production of hardier crops
such as rye. Increased temperatures, however, will exacerbate the dry
conditions in the southern Urals, lower Volga, and Kazakhstan-areas that
account for 20 percent of Soviet grain production
Long-term weather patterns and trends in fertilizer deliveries to agricul-
ture suggest that Soviet grain production during the 1986-90 period most
likely will average 195 million metric tons annually-about 60 million tons
below target. Thus, at this level of production, Moscow will remain
dependent on foreign sources for grains if the leadership intends to achieve
projected levels of livestock production and fulfill promises to improve the
diet of Soviet citizens. Given the uncertainties of climate prediction, the
grain-growing environment in the USSR could be somewhat better or
worse than this most likely estimate:
? If a more favorable climate prevails, one with precipitation equal to the
1976-80 period-the best five-year average of the last 65 years-and if
fertilizer deliveries reach planned levels, the Soviets could produce an
average of 221 million tons per year.
? With bad weather similar to the 1961-65 period-the lowest five-year
average precipitation of the last 25 years-and fertilizer deliveries
increasing at only the average rate of the last 10 years, production could
average as low as 165 million tons annually
iii Secret
GI 85-10128
May 1985
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret LOA-1
Given the scenarios, Moscow will not be able to make rapid progress on two
key goals of the Food Program-improving food supplies while reducing
dependence on Western farm products. Indeed, the Soviets would need to
import an average of 15 to 65 million tons of grain annually during the
1986-90 period to meet domestic requirements. Although imports at the
upper end of this range are logistically and financially feasible, they would
strain the transportation system and could force reduction in other hard
currency impor
At least two options could enable the Soviets to boost grain output
substantially above our most likely estimate. For example, grain yields
could be raised significantly by importing more and better agrochemicals
and improving application techniques. We believe the Soviets could also
increase grain production by changing the crop mix. Specifically, we
estimate that a substitution of corn for wheat and other grains on irrigated
land could result in a net increase of as much as 12 to 14 million tons by
1990. Despite the benefits associated with such options, the Soviets have al-
ways been slow to change agricultural polic$
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Climate and Technology
Implications
8
10
10
A Simple Regression Model for Estimating Grain Yields of the USSR 15
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Soviet Climate Change:
Implications for
Grain Production
Introduction
Soviet efforts to put more and better meat on the
table-a principal measure by which Soviet citizens
judge their standard of living-have resulted in mas-
sive investments in agriculture and in grain produc-
tion in particular. Over the past two decades, Moscow
has committed billions of rubles to land reclamation
and irrigation, the production of agrochemicals, farm
machinery and equipment, and to a wide variety of
construction. These measures, coupled with generally
favorable weather, have caused grain production to
increase impressively. Nonetheless, the demands of
the steadily increasing livestock herds for grain still
far exceed the amount farms have been reliably able
to produce. As a result, the Soviets have been forced
to expend sizable amounts of hard currency for grain
imports.
Numerous factors have prevented the Soviets from
achieving their grain production goals, including poor
management and lack of incentives for farm workers.
We believe that the most important factor in the year-
to-year variation in Soviet grain production has been
weather. Indeed, the climate,' although gradually
getting warmer and wetter, is generally unfavorable
for grain cultivation. Because climate changes slowly
over many years, while weather varies widely from
year to year, a long weather record is necessary to
analyze climate trends properly. We developed a
climate data base, covering 1920-84, to analyze his-
torical weather records and project weather for the
next five years. Our research also indicated that
fertilizer deliveries to agriculture, an important factor
in Soviet attempts to increase grain yields, can serve
as a surrogate for other kinds of technological im-
provements in regression analysis. This study shows
that past changes in Soviet climate and technology-
particularly the precipitation component of climate
' Climate is weather over a longer period. For example, daily mean
temperature is used to describe weather, while mean temperature
for 10 years or longer is used to characterize climate. Both are
and the fertilizer component of technology-
correspond well with historical variations in grain
production and should provide a key to future Soviet
performance. 25X1
The Role of Climate
Grain is grown primarily in a zone extending from the
borders of Eastern Europe to western Siberia-nearly
4,500 kilometers (km) west to east-and from the dry
steppes of Central Asia to the tundra regions-some
1,800 km south to north. For the most part, soils in
the zone are comparable or somewhat inferior to those
of the northern plains of the United States. Soil
deficiencies aside, our analysis indicates that the low
precipitation in the relatively more important south-
ern regions (Ukraine, Volga Valley, and the Caucasus)
has been the key limitation to grain production in the
USSR. Primarily because of yearly variation in pre-
cipitation, total grain production during the 1971-80
period ranged from a low of 140 million metric tons in
1975 to an alltime high of about 237 million tons in
1978. Only about 2 percent of the grain area in the
USSR is irrigated, and it accounts for only ab25X1
of
The timing of rainfall can be as important as the
annual volume. In the Soviet Union, most grain crops
are grown with less reliable precipitation than in the
United States. Moreover, in most grain areas a
smaller percentage of annual precipitation is concen-
trated during the growing season than in the United 25X1
States. This is the case, for example, throughout most
2 For previous studies on climate change in the USSR, see
joint Economic Com-
mittee, Congress of the United States, Soviet Economy in the
1980s: Problems and Prospects, Part II, Selected Papers, Decem-
ber 1982, pp. 10-12, "Climate and Grain Production in the Soviet
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
of the Ukraine and northern Caucasus the most
important Soviet grain-growing region-which is gen-
erally comparable in climate to eastern Nebraska and
southern Minnesota. The Nebraska/ Minnesota areas
generally receive over 610 millimeters (mm) of precip-
itation per year, of which 75 percent occurs during
April to September (when the grain is growing) and
thus can be used most effectively by the plants. In
contrast, the Ukraine/Caucasus region receives about
510 mm, with only 55 percent falling during the
growing season
Although precipitation is the principal determinant of
grain yields in the most important grain regions of the
USSR, agricultural decisions by the Soviet leadership
can have a lesser but still important impact on
production. For example, the record-high five-year
average production of 205 million tons during 1976-
80 resulted not only from three consecutive years of
good weather (1976-78)-highly unusual-but also
from a decision to substantially increase planted area.
Total harvested area during the three-year period
averaged almost 129 million hectares (ha). In contrast,
during the 1981-84 period, the weather was relatively
poorer and the average harvested area dropped to an
estimated 122 million ha, mainly because of a deci-
sion to increase fallow, a technique used in the USSR
primarily to build up soil moisture and dampen year-
to-year fluctuations in production. Largely because of
these factors, production during 1981-84 declined to
an estimated average of 180 million tons
25X1
Climate Change
Our analysis of the two most common measures of
climate, average annual precipitation and tempera-
ture, shows a slow change in the climate throughout
the Soviet grain-growing region. Both temperature
and precipitation are increasing. While precipitation
and temperature are related, our analysis indicates
that precipitation is normally the more important
factor in determining Soviet grain yields
Long-Term Trend. Since the 1930s there has been a
general trend of increasing precipitation in the Soviet
grain-growing region (figure 2). Although precipita-
tion has varied greatly from year to year, on average
it has increased about 20 mm per decade since the
The precipitation and temperature regimes of the
major grain-growing regions of the Soviet Union were
compiled from data recorded at 66 Soviet climatolog-
ical stations.- The stations are distributed nearly
evenly across the grain-growing regions of the USSR
(figure 1). Of the 66 stations, 21 provided data from
1920 to 1949, all provided data from 1950 to 1974,
and 36 provided data from 1975 to 1984F__1
Because of the good correspondence between annual
averages obtained from the sets of 21, 36, and 66
stations for the period 1950 to 1974, we were able to
use the data from only the 21 stations for 1920-49
and the 36 stations for 1975-84 with confidence. The
grain region's annual temperature and precipitation
averages were obtained by weighting each station's
average by the fraction of total grain area within a
surrounding polygon.b The annual precipitation aver-
ages of the 21 and 36 station sets were in most cases
within 2 to 3 percent of the annual averages of the 66
stations, and the five-year averages of the 21 and 36
station sets were within 1 to 1.5 percent of the five-
year averages of the 66 stations (table 1). Even better
correspondence was obtained in the temperature com-
parison
., Sources of information for this data base are "World Weather
Records, "published by the old US Weather Bureau, and "Month-
ly Climatic Data for the World, "published by the National
Oceanographic and Atmospheric Administration (NOAH)
n A standard technique called the Thiessen polygon met ho was
used. The technique assumes that the precipitation at any station
can be applied halfway to the next station in any direction. The
polygons are formed by the perpendicular bisectors of the line
joining nearby stations. The grain area in each polygon is used to
weight the precipitation nt (or temperature) of the station in
the center of the polygon
1940s. Precipitation averaged about 476 mm in the
1970s-25 mm more than the 1960s and 71 mm
(almost 3 inches) more than the dry 1930s. The latest
five-year average (1980-84) shows a slight decrease to
470 mm-still considerably above the long-term
(1920-84) average of 435 mmF_~
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Figure 1
Selected Climatic Stations
1920-84 1951-84
1920-74 1951-74
I Major grain-producing area
Economic region boundary
Union republic (SSR) boundary
Baltic Sea n -'
..ig
l uop ip: d "I
Kalinin ra
R.S.f S R.
~yy! M
~h rp n
^#!Yr~~Md4C a
MOLDAVi...
S.S.R.
Kishin
Sbvi t Union
R. S. 1F,
Caspian J
Sea +~
tdlov.
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Table I
USSR: Precipitation and Temperature Averages
for the Grain Area
21-Station 36-Station 66-Station 21-Station 36-Station 66-Station
Average Average Average Average 'Average Average
1920 375 5.3
1921 335 4.6
1922 443 4.5
1923 413 4.9
1924 402 5.4
1920-24 394 4.9,
1925 388 6.0
1926 470 4.6
1927 429 5.0
1928 491 3.5
1929 389 3.8
1925-29 433 4.6
1930 431 5.1
1931 412 4.1
1932 417 5.2
1933 459 4.2
1934 405 4.1
1930-34 425 4.5
1935 403 5.0
1936 369 4.8
1937 433 5.4
1938 359 5.7
1939 363 5.1
1935-39 385 5.2
1940 415 4.3
1941 472 4.0
1942 429 2.7
1943 383 4.7
1944 406 5.8
1940-44 421 ' 4.3-
1945 404 3.5
1946 393 5.0
1947 419 3.8
1948 422 5.7
1949 396 5.2
1945-49 407 4.6-
1950 442 449 448 4.3 4.7 4.7
1951 351 376 377 4.9 4.9 4.9
1952 385 400 412 4.9 4.8 4.8
1953 446 452 455 4.8 4.8 4.8
1954 405 390 400 3.2 3.6 3.5
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Table 1 (continued)
Precipitation (millimeters)
21-Station
21-Station
Average
36-Station
Average
66-Station
Average
21-Station
Average
36-Station
Average
66-Station
Average
1950-54
406a
413-
419a
4.5a
4.6
4.5
1955
442
443
438
5.2
5.1
5.1
1956
478
471
466
3.3
3.7
3.7
1957
426
420
414
5.3
5.1
5.1
1958
480
481
487
4.8
4.7
4.7
1959
397
393
399
5.3
5.2
5.2
1955-59
445-
442 a
441
4.8a
4.8 a
4.8
1960
437
456
457
3.6
3.9
3.9
1961
479
462
461
5.4
5.4
5.4
1962
459
430
439
5.8
5.6
5.7
1963
418
407
417
4.8
4.8
4.8
1964
429
459
469
4.7
4.3
4.4
1960-64
444 a
443 a
449-
4.9 a
4.8 a
4.8
1965
402
395
402
5.3
4.9
5.1
1966
505
540
516
5.6
5.4
5.5
1967
454
459
451
4.7
4.5
4.8
1968
453
458
444
5.3
5.2
5.3
1969
454
463
450
2.7
2.9
2.7
1965-69
454 a
463 a
453 a
4.7-
4.6 a
4.7-
1970
606
585
547
4.9
4.8
5.0
1971
478
462
468
4.7
4.6
4.6
1972
438
425
441
5.1
4.9
4.7
1973
453
474
465
5.4
5.2
5.0
1974
459
448
447
5.1
5.0
5.1
1970-74
487 a
478-
474 a
5.0 a
4.9
4.9 a
1975
401
6.4
1976
462
3.6
1977
490
4.2
1978
545
4.7
1979
494
4.6
1975-79
478 a
4.7-
1980
498
4.1
1981
470
5.8
1982
489
5.2
1983
464
6.5
1984
432
5.3
1980-84
470 a
.4 a
5.4-
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Figure 2
Annual Precipitation in Major Grain Area; 1920-84
MiIIimetcrs
600
III'.,. I I Ii I I I IIIII~I.L!
0 1920 30 40
tutu) is liu Oclohcr through Septemher
\,erages li>r 21 alation,, 66 stations, and 36 stations, respectivcly, were
used liar the periods 19211-4'1, 19 0-74, and 1975-84.
i iL 1I I 11. i ii J. 1
60 70
Although precipitation in the grain-producing zone
has increased overall, analysis of the weather data
shows a significant change in the geographical distri-
bution of precipitation during the last 10 years (1975-
84) compared with the 1950-74 period (figure 3). Most
of the grain area experienced an increase in precipita-
tion-as much as 75 mm in parts of European
RSFSR and eastern Ukraine. Significant decreases
occurred, however, in some important grain-
producing areas of the southern Urals and western
and eastern Kazakhstan in some cases a decline of
about 25 mm or more occurred in these regions
As for temperature, our analysis also shows a gradual
increase overall in the grain-growing region, from a
10-year average of about 4.5?C in the 1940s to about
5.0?C for the 1975-84 period (table I and figure 4).
Part of this long-term temperature increase may
reflect urbanization (increased pollution and city
heat-island effects). The rest may represent an in-
crease in air temperature worldwide that US scientists
generally attribute to a rise in atmospheric carbon
dioxide.
' Changing Climate, National Academy of Science, National
Academy Press, l98
25X1 25X1
An analysis of changes in temperatures by geographic
area shows regional changes in annual temperature
during the last 10 years (1975-84) compared with the
1950-74 period (figure 5). Temperature increases of
about 0.5? to 1.0?C are evident over most of the
north, central, and eastern regions of the grain area.
A climatic increase in temperature usually causes a
lengthening of the growing period, which in the future
may permit additional areas in Siberia and northern
European RSFSR to come under cultivation, especial-
ly with the hardier rye varieties that are already
showing success. On the other hand, we believe future
temperature increases in the southern Urals, lower
Volga, and Kazakhstan would further exacerbate the
already dry climate there
Future Trends. Our analysis of climate indicates that
temperature will continue to increase and precipita-
tion is likely to remain above the long-term average.
Although average annual precipitation during the
1980-84 period was slightly less than during the
1970s-470 mm compared with 476 mm--it is still
too early to conclude that a downward trend-or
leveling off-has set in. Indeed, recent precipitation
levels are still well above the pre-1970 averages
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Figure 3
Change in Mean Annual Precipitation
for 1975-84 Compared With 1950-74
-25 0 +25 50 75
Major grain-producing limit
Economic region boundary
Union republic (SSR) boundary
75
The Un+Yed States Government h** host
recongnrsad the inoorporhon of Wopip,
Lativia, and Lithuania into the SOY at Un,on.
Boundary rapveaetttation is not naceraaatily
authoritative, TheMoldovbn SSRis not
part of any aoanomic roveon.
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
-Mfg
Baltic
Kiev.(
L hr
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Figure 4
Average Annual Temperature for the Soviet Grain Area," 1920-84
I)egrccs centigridU
IlllilIIIII~IIIlilillllllll
U 192 )) 3)) 40 s() 60
rr,,a i s I u r Oclnher I hrI'ugh Sc hte III he r
' el.l4e, IOU 21 sLItluni, 06 staII?ns, an(1 36 suit ions, ICspecti?e I), were
used liu Inc tperiods 1920-49, 19N0-'4, u1(1 197>-84.
We postulate that, for the 1986-90 period, average
rainfall should not depart greatly from the 1980-84
average of 470 mm even though year-to-year precipi-
tation amounts will continue to vary widely. Statisti-
cal analysis of the change in precipitation between
sequential five-year intervals during the 1920-84 peri-
od showed about a 50-percent probability that precipi-
tation in the 1986-90 period will average 476 mm or
above, and about a 15-percent probability that it will
25X1 be above 500 mm or below 440 mm.
As for temperature, we expect an increase that will
generally follow the long-term trend, as a result of a
continued increase in atmospheric carbon dioxide.
The temperature increase, however, may not be as
great as that experienced from 1975-79 to 1980-84
(4.7? to 5.4?C), since the latter period was considera-
bly above trend. Continuation of the trend of five-year
averages from the 1940s to 1990 would place the
average 1985-90 temperature at about 5.0? to 5.2?C.
I I I I I I I I 1 I I I I L ?_ili
70 80
25X1
Climate and Technology
Climate directly affects crop growth and technology
inputs, namely fertilizer. Together climate and tech-
nology are key determinants of grain yields. Further-
more, other technological improvements-such as
new seed varieties and the application of improved
herbicides and pesticides-are not totally effective
without good weather. For example, should precipita-
tion in the USSR return to the low levels of the 1930s
and 1940s, the benefits of most of the new technology
would be greatly reduced.' If the climate continues to
improve for grain production or remains about the
same, more likely in our view, Soviet success in raising
production will increasingly rely on technology. Our
analysis indicates that chemical fertilizer has been an
' Under such circumstances, lack of moisture would be the key
yield-limiting factor. For example, fertilizer needs moisture to be
used effectively by crops. Furthermore, additional technological
investments such as new seed varieties, herbicides, or farm tillage
and irrigation equipment would have little value without timely and
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Figure 5
Change in Mean Annual Temperature
for 1975-84 Compared With 1950-74
Celsius (?C)
Major grain-producing limit
Economic region boundary
Union republic (SSR) boundary
-25 0 [phone]
Baltic Sea Lenin ad,)
/' -,Riga N t I
Baltic
Kara
Sea
S,bviE
R. S. F R.
Caspian
^ Sea
West
Iberia
75
The Un+t d Slaf. ~arirh#iR nc
racahgnitod thu ncor ni ki
Lativwh#AdLithuania Nit! Union.
9oundaq rspeyteatatF 't- #~1/pi
?uthorBaCaw of
Liatfiwiin i. h~
part of an 'c
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
important input in enhancing grain yields. We also
found that, in regression analysis, fertilizer can serve
as a surrogate for the other kinds of technological
improvements gradually introduced during the last 25
year 25X1
Looking Ahead
To evaluate the impact of climate change on future
grain output in the Soviet Union, we selected three
alternative weather scenarios-favorable, most likely,
and unfavorable. Because of the important impact of
technology on grain production, we also developed
three corollary scenarios for fertilizer deliveries to
agriculture. The combinations of weather and fertiliz-
er scenarios were used in a newly developed regression
model to estimate future grain yields. We believe that
the nominal-or most likely- combination of weather
and fertilizer scenarios provides the best indication of
Soviet grain production for the rest of the I980s.`F-
Weather Scenarios. We estimate with confidence
from weather trends (figures 2 and 4 and table I) that
during 1986-90 the average precipitation in the grain
area will most likely range between 450 and 490 mm
and that temperature will average between 5.0? to
5.2?C. The general upward trend in temperature and
precipitation is consistent with the findings of the
National Academy of Science,' which projects that
mean global temperature and precipitation will in-
crease because of increases in atmospheric carbon
dioxide. Nevertheless, the Academy cannot predict
the magnitude or location of such increases. The
climate probably would not suddenly revert to the
lower precipitation levels of the 1940s and 1950s,
although sudden shifts in precipitation-as i25X1
The three weather scenarios we selected to estimate
the range of Soviet production in the 1986-90 period
are:
? The most likely, derived from the precipitation and
temperature regimes of the 1970-84 period with
annual averages of 474 mm and 5.0?C.
Average grain yields for the 1986-90 period were estimated using
a simple regression model (see the appendix). To derive these
estimates, we examined various factors which influence grain
production. Statistical analysis showed that precipitation, tempera-
ture, and the level of fertilizer deliveries to agriculture adequately
rapture the variability in Soviet grain yields
('hanging Climate, National Academy Press, 1983.
25X1
? The favorable, based on the 1976-80 period, which
shows the highest five-year precipitation average
(498 mm) of our 65-year record.
? The unfavorable and least likely, based on the five-
year period 1961-65, which averaged 438 mm, the
lowest of the last 25 years.
Fertilizer Delivery Scenarios. Following a four-year
lull that began in the mid-1970s, fertilizer deliveries
to agriculture regained their upward momentum after
1979, growing at an average rate of 1.1 million tons
per year to a record 23.1 million tons in 1984. Such a
continued rate of growth (almost 6 percent per year)
in fertilizer deliveries during the next six years would
fulfill Soviet plans to deliver 30-32 million tons of
fertilizer for crops in 1990.1
We developed three fertilizer delivery scenarios:
? The high or best case, which projects an annual 6-
percent increase in fertilizer delivery, or an average
of about 1.5 million tons per year. Although the 6-
percent rate of growth approximates the 1979-84
average, we doubt that the Soviets will be able to
maintain this rate because of expected lags in the
commissioning of new facilities for the production of
fertilizers, poor management, and the underuse of
existing facilities. The 1984 rate of growth was in
fact less than I percent.
? The medium, or most likely case, which projects
that deliveries will increase by about 1.0 million
tons per year, or a 4-percent growth, yielding a total
delivery to agriculture of 29 million tons by 1990.
We judge this scenario the most likely because we
expect the Soviets to fall 1-2 million tons short of
plan in 1985 and be unable to produce enough
fertilizer in 1986-90 to make up the 1981-85 short-
falls and meet 1986-90 goals as well.
A statistical analysis of the change in precipitation between five-
year intervals during 1920-84 results in the following approximate
probabilities of occurrence for the three precipitation scenarios
chosen for the 1986-90 period:
15-20 percent probability that precipitation will be 498 mm or
above.
45-50 percent probability that precipitation will be 474 mm or
above.
5-10 percent probability that precipitation will be 438 mm or
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97ROO694ROO0400720001-7
,secret
? The low case, which projects a 2-percent-per-annum
growth rate. This rate was derived from a model
using the last 10 years' deliveries of fertilizer to
agriculture. The model results project a total deliv-
ery of 26 million tons by 1990, for an increase of
only about 0.5 million tons per yea
The projected fertilizer deliveries to agriculture for
the entire USSR for the three scenarios described
above were translated into fertilizer delivery rates in
kilograms per hectare (kg/ha) for each republic by
dividing by agricultural area. In all cases, we assume
total harvested area will approximate 124 million ha,
roughly equal to the annual average hectarage for
1979-83. This relatively low hectarage figure assumes
that the Soviets will maintain current levels of fallow.
25X1
Projected Yields and Production
Grain yields and production to 1990 were calculated
with the regression model using the three fertilizer
scenarios and the actual weather variables for 1961-
65, 1976-80, and 1970-84, periods typical of unfavor-
able, favorable, and most likely weather conditions
(table 2). 25X1
The model forecasts that, given what we consider the
most likely weather and fertilizer scenario, the
USSR's average grain yield during 1986-90 will be
15.7 centners per hectare (ce/ha). Using a harvested
area of about 124 million ha, this equates to an
average annual production of 195 million tons. Given
this scenario, the model projects that there is a 95-
percent probability 10 that Soviet grain production
during 1986-90 will average between 180 million tons
and 210 million tons. 25X1
With afavorable weather scenario similar to 1976-80
and the high fertilizer delivery levels that the Soviets
are striving to achieve, Moscow could average 17.8
cc/ha or 221 million tons, with a 95-percent probabil-
ity that the average will be more than 206 million tons
but less than 236 million tons.l 25X1
" Roughly 60 percent of the 75-million-ton Soviet grain production
increase from the 1961-65 period to the 1976-80 period (130 million25X1
tons versus 205 million tons) was caused by an increase in average
precipitation (438 mm versus 498 mm). The remaining 40 percent
was caused by improvements in agrotechnology
" This assumes that neither the mix of feed nor current levels of
The 95-percent probability range is approximately defined by the
model's estimate ? two standard errors of estimate, or within 15
million tons of the projected average of 195 million tons. One
standard error of estimate was calculated to be 7.5 million tons
An unfavorable weather scenario typical of 1961-65
(the least likely of the three scenarios) and low
fertilizer delivery growth rates could plunge average
grain production to 165 million tons, with less than a
5-percent probability that it would be above 180
million tons 25X1
Analysis of the regression model results, using the
three weather and three fertilizer scenarios and as-
suming a harvested area of 124 million ha, shows that,
at a constant fertilizer level, every 10-mm increase in
average annual precipitation will result in a 7.5-
million-ton increase in average grain production. Cor-
respondingly, at a constant precipitation level, every
million-ton increase in fertilizer deliveries to agricul-
ture will produce about an additional 2.5 million tons
of grain.' 25X1
Implications
The three grain production scenarios for the 1986-90
period suggest that the USSR will not progress
rapidly on two key goals of the Food Program-
improving food supplies while reducing dependence on
Western farm products. Indeed, even if grain produc-
tion averages 221 million tons, Moscow would still
need to import at least 15 million tons of grain
annually to maintain current levels of seed, food, and
industrial use, as well as to achieve planned output
levels for meat, milk, and eggs." Given the most likely
scenario of 195 million tons, grain imports would have
to exceed 35 million tons annually
In the unlikely event that production falls to 165
million tons, the Soviets would require an average of
roughly 65 million tons of grain imports annually.
This would be an enormous amount, but probably not
one beyond the USSR's improved logistic capability.
It would be financially difficult, but possible, so long
as grain prices remain relatively low. If Moscow
animal productivity chang~
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97ROO694ROO0400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Table 2
USSR: Projections of All-Grain Average
Yields and Production, 1986-90
Weather Scenario
Increase in
Fertilizer Deliveries
to Agriculture
Yield (centners
per hectare)
Average Production b
(million metric tons)
95-Percent Probability
Range of Production
(million metric tons)
Unfavorable weather
Low
13.3
165
150-180
Medium
13.6
169
154-184
High
14.2
177
162-192
Favorable weather
Low
16.8
209
194-224
Medium
17.2
214
199-229
High
17.8
221
206-236
Most likely weather
Low
15.3
190
175-205
Medium
15.7
195
180-210
High
16.2
202
187-217
Low, medium, and high increases in fertilizer deliveries to
agriculture correspond to approximately 2-, 4-, and 6-percent
increases per year.
I Production is estimated by assuming an average grain area of 124
million hectares, similar to that of the 1979-83 period.
The 95-percent probability range is approximately defined by the
average ? 2 standard errors of estimate.
chooses not to test the transportation system and/or
not to reduce other hard currency imports, however,
the need for grain could be reduced in several ways.
Planners could save a few million tons by reducing the
quantity of grain used for food, industrial purposes,
and export, as they have done in the past. They could
cut grain demand by reducing livestock inventories-
a tactic strenuously avoided since 1975. This would
increase meat supplies temporarily but would proba-
bly slow subsequent growth in meat production. A
third alternative is to curtail quantities of grain fed
per animal. But the reduction would have to be offset
by other feeds or animal productivity, and thus meat,
milk, and egg production would suffer. Consumers
would be faced with diets of somewhat lesser variety
and quality, but the extensive special food distribution
systems put in place during 1979-81 to cope with the
widespread food shortages probably would help offset
the effects 25X1
Soviet Policy Options
In our view, the Soviets have at least two policy
options that may enable them to boost output substan-
tially above the levels indicated in our most likely
weather and fertilizer scenario by 1990:
? Grain yields could be raised significantly if a deci-
sion were made to purchase more and better agro-
chemicals, that is, pesticides, herbicides, fungicides,
plant-disease protective agents, etc., from foreign
suppliers, and if steps were taken to improve appli-
cation at the farm level.
? We also believe the Soviets could increase overall
grain production by changing the crop mix." For
example, by substituting corn for wheat and other
grains on irrigated lands, Moscow could boost out-
put by as much as 12-14 million tons by 1990.
" The potential to increase grain production in the USSR by
changing crop mixes is the topic of a forthcoming CIA research
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
LYV IVI
cantly affect production by 1990.
Other options, such as the purchase of turnkey agro-
chemical plants, are possible but would not signifi- 25X1
Despite the potential benefits associated with these
options, history has shown that the Soviets are slow to
change their agricultural policy, particularly for
wheat production. We judge the possibility of the
Soviets deciding to import larger amounts of agro-
chemicals and agrochemical technology to be some-
what greater than the likelihood of the crop mix being
tion before 1990
1985 with imported agrochemicals. Purchases of large
quantities of agrochemicals from the United States
and other Western nations could help boost grain
production above trend in the near term, but the time
required to install turnkey production facilities would
preclude domestic chemical output from reaching a
high enough level to significantly affect grain produc-
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
secret
Appendix
A Simple Regression Model
for Estimating Grain Yields
of the USSR
A grain-yield regression model enables us to estimate
grain yields during the 1986-90 period under different
weather and technology growth scenarios. The model,
therefore, has to be a function of variables that
measure the contribution of weather and technology
to grain yields 25X1
Figure 6 illustrates the historical all-grain yields, total
precipitation in the grain area during the growing
period (October-July), and the average amount of
fertilizer (kg/ha) delivered to agriculture in the
USSR. The graph shows a considerable increase in
yields from the mid-1960s to the late 1970s, with
simultaneous increases in fertilizer delivery and levels
of precipitation. With a few exceptions, there is a
general correspondence between high and low points
of precipitation and yield. Thus, precipitation and
fertilizer delivery rates are likely candidates for de-
scribing grain yields by means of a ---;ion
25X1
Because of the paucity of published Soviet grain data
since 1975, our grain-yield equations were derived for
large areas covering one or more republics and having
sufficient climatic stations to adequately describe
weather parameters. For example, from 1975 through
1980, only republic grain yields were published by the
Soviets; after 1980 practically no rain-yield informa-
tion was published 25X1
We used the RSQUARE procedure of the Statistical
Analysis System (SAS) computer software package to
narrow down the selection of variables for the predic-
tive model. The RSQUARE procedure performs all
possible regressions for a dependent variable (grain
yield, in this instance) and a collection of independent
variables, and gives the r-square value for each model.
With the selected parameters, we then derived the
yield equations using the General Linear Model
(GLM) procedure of SA$
Table 3 lists the variables tested by the RSQUARE
routine and the equations finally adopted.14 An inter-
esting result of the selection process was that fertilizer
application variable (FERTH) produced higher
r-squares than the variable YEAR, a term traditional-
ly used as a surrogate for technology." Fertilizer
application rates to grain area would be an even better
parameter to use in the regression, but these data are
not generally available at the republic level. We found
no improvement in estimating Soviet all-grain yields
by using separate winter and spring grain yield equa-
tions. We therefore elected to use the all-grain yield
equations for the combinations of republics shown in
table 3, which also gave better results than a single
equation derived for the entire Soviet Union.
The major assumptions inherent in the use of the
regression model for forecasting grain production in
the 1986-90 period are:
? That projected increases in fertilizer deliveries to
agriculture represent the major contribution of tech-
nology to grain-yield increases.
" We tested three variables for describing the technology contribu-
tion to yield: YEAR, total fertilizer deliveries to agriculture
(FERTD), and average fertilizer deliveries per hectare of agricul-
tural land (FERTH) from Soviet published data. We also tested
cross terms such as FERTH*PREC to detect any interaction
between fertilizer response and precipitation amounts, and nonlin-
ear terms such as log(FERTH) to describe diminishing yield returns
at high fertilizer delivery levels. In all instances, except one, we
found no significant increase in r-square when crossterms or other
nonlinear terms were added to the candidate models. We believe
that this occurred because the geographic areas covered by the
model's equations were too large to adequately capture the interac-
tion between FERTH and PREC. Only in Belorussia and in the
Baltic, where fertilizer delivery levels are among the highest in the
country, did we find that the use of a log(FERTH) term produced
" Fertilizer delivered per hectare of agricultural land (FERTH) has
increased nearly linearly with time (FERTH and YEAR show a
correlation coefficient of 0.98). Therefore, FERTH, in addition to
being directly related to grain-yield increases, is also a surrogate for
other technological improvements that have gradually been intro-
duced during the last 25 years and have also been responsible for
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Figure 6
Growing Period Precipitation, All-Grain Yields, and Average Fertilizer
Per Hectare," 1960-84
NoI .cJIe chanuc~
\lillimctcr. t ;nln.]
>Up `1
f )i i,~brrJuh ["'1 ilaliun fnr the '-'ICI 2111111 arc,i_ Avcrapc fcrtil
d.ln.rc~l i. per I Tiara nI iericullur.iI Ianll.
25X1 urc (-I \ cUima~c,.
? That any changes in the mix of grains planted or in
other agricultural practices, such as the amount of
cropland under irrigation, will take place gradually
and therefore will be included in the model variable
representing the delivery of fertilizer nPr hectare
(FERTH). 25X1
? That the mean square error of our regression model
adequately describes the errors of the modelF---]
Figure 7 and table 4 show how the model's estimated
yields compare with the actual yields for 1960-80, the
period used to derive the model. Also plotted on figure
7 are the model estimates for 1981-84 compared with
CIA estimates. The model fits the observations with
an average error of 1.1 ce/ha and a mean square error
of 1.4 ce/ha for individual years and 0.6 ce/ha for a
five-year period." The model is able to explain 80
percent of the variation in the all-grain yieldsF_
'? The mean square error for a five-year period is 1.4/ V 5 = 0.6.
Three Nears (1971, 1973, and 1976) show particularly large model
errors of the order of 2 to 3 cc/ha. We will investigate the causes of
~ "Id
Precipitation
The model's errors may be caused by a combination
of factors. The first is the gross nature of the model
itself. Because of a paucity of data, the model must
use meteorological variables averaged for relatively
long periods (four to 10 months) and for very large
areas (as large as the RSFSR). Second, although the
years used in the model (1960-80) are the most
relevant in terms of describing recent Soviet agricul-
tural and climate changes, they may not be sufficient
to capture the range of errors inherent in the model.
Third, the variables in the model may be related to
yield in a more complex, nonlinear, and interactive
way than can be represented by our simple linear
model. Finally, there are certainly other variables
such as short-term weather events, the quality of
management, work incentives, and political decisions,
that influence yield but could not be included in the
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Table 3
USSR: An All-Grain Yield Regression Model
Variables Tested by the RSQUARE Procedure a
PREC (4-9)
PREC (10-3)
PREC (io-9)
PREC (10-8)
PREC (4-7)
TEMP (4-9)
FERTD
FERTH
YEAR
(FERTH) X (PREC)
TEMP (10-3)
LOG(FERTD)
LOG(FERTH)
YEAR'
TEMP ((o-9)
SQRT(FERTD)
SQRT(FERTH)
TEMP ((0-8)
TEMP (4-7)
Equations Selected for Estimating All-Grain Yields b R2
For the RSFSR: YIELD, = -3.97 + 0.0875 PREC (4-7) + 0.0141 FERTH 0.80
For Kazakhstan: YIELDk = 3.52 + 0.0472 PREC ((0-8) - 0.5367 TEMP (4-7) + 0.1277 FERTH 0.73
For Ukraine + Moldavia: YIELD? = 25.44 + 0.0313 PREC (4-9) + 1.334 TEMP (10-3) - 1.156 TEMP (4-7) + 0.0544 FERTH 0.81
For Belorussia + Baltic: YIELDb = -15.069 - 1.1584 TEMP (4-7) + 9.519 LOG(FERTH) 0.83
a PREC-average region precipitation in millimeters weighted by b Letter subscripts r, k, u, and b refer respectively to RSFSR,
grain area. Kazakhstan, Ukraine plus Moldavia, Belorussia plus Baltic; p
TEMP-average region temperature in ?C weighted by grain refers to all these areas combined, representing about 96 percent of
area. total Soviet grain area.
FERTD-total fertilizer delivered to agriculture in million metric
tons.
FERTH-average fertilizer delivered per hectare of agricultural
land in kilograms.
YIELD average region grain yield of major grain area in
centners per hectare.
Number subscripts refer to first and last months of period
averaged for temperature (TEMP), or totaled for precipitation
(PREC). For example, PREC (10-3) refers to total precipitation
during October-March.
For all areas combined: YIELDP = (A, YIELD, + Ak YIELDk + A. YIELD? + Ab YIELDb)/Ap,
where A? Ak, A?, Ab are the grain areas, and Ap = A, + Ak + A? + Ab
For the USSR: YIELD = -1.472 + 1.104 YIELDP
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Figure 7
Comparison of Observed All-Grain Yields and Model's Yields, 1960-84
C11111Crti pCr 11c'Imc
'()
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Table 4
USSR: All-Grain Yields and Production, 1960-84
Actual Yield
(centners per hectare)
Actual Production
(million metric tons)
Model Yield
(centners per hectare)
Model Production
(million metric tons)
1960
10.9
125.5
11.9
137.5
1961
10.7
130.8
11.6
141.8
1962
10.9
140.2
10.7
137.7
1963
8.3
107.5
8.5
110.4
1964
11.4
152.1
11.2
149.3
1965
9.5
121.1
10.1
129.3
1961-65
130.3
133.7
1966
13.7
171.2
14.4
179.7
1967
12.1
147.9
12.2
149.0
1968
14.0
169.5
14.8
179.8
1969
13.2
162.4
12.6
154.6
1970
15.7
186.8
15.4
183.7
1966-70
167.5
169.4
1971
15.4
181.2
13.5
159.2
1972
14.0
168.2
12.9
155.1
1973
17.6
222.5
14.2
180.0
1974
15.4
195.7
14.7
187.0
1975
11.0
140.1
12.0
153.5
1971-75
181.6-
167.0
1976
17.5
223.8
15.5
198.1
1977
15.0
195.7
15.7
204.6
1978
18.5
237.4
18.5
237.7
1979
14.2
179.2
15.6
197.1
1980
14.9
189.1
16.3
206.3
1976-80
205.5-
208.8
1981
NA
13.0
163.3
1982
NA
15.9
195.5
1983
NA
16.3
196.9
1984
NA
14.5
173.3
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
 Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7
Secret
Secret
Sanitized Copy Approved for Release 2010/10/07: CIA-RDP97R00694R000400720001-7