AGRICULTURAL

ECONOMICS
EL~EVIER Agncultural Economic; 10 (1994) 107-123

Sustaining growth in agriculture:

a quantitative review of agricultural research investments
Jock R. Anderson *·a, Philip G. Pardey b,c, Johannes Roseboom b

"Agncultural Polzctes Du•l5ton, Agnculture and Natural Resource Department, World Bank,
1818 H Street N.W, Waslungton, DC 20433, USA
h lntematwnal SerVice for Natwnal Agncultura/ Research (ISNAR), The Hague, The Netherlands
' Department of Agncultural and Applwd Economtcs, Unu erslly of Mumesota, Mmneapohs, MN, USA
(Accepted 21 June 1993)

Abstract

Growth in agriculture depends on many things but one of the most important is investment in agricultural
research. Decision making in the agricultural research policy area can only be aided by access to better information.
This article overviews a recent endeavor to move policy dialogue beyond merely qualitative impressions towards a
process that is underpinned with new and cogent data. The data used have been assembled at ISNAR in a manner
designed to make comparisons both over time and between countries more valid than has been the case in the past.
The comparisons thus possible reveal considerable diversity both between countries and between broad regional
aggregations. Also illuminated here are issues related to the commodity orientation, capital and labor intensity, and
size and scope of particular national programs.

Policy-makers are having to contend with un-

precedently rapid changes in the market for agri-
* Corresponding author. cultural science and technology services. In the
The research data reported here are largely from a study less-developed countries, the rapid expansion of
undertaken by ISNAR with assistance from the then Agncul- agricultural research capacity experienced during
ture and Rural Development Department of the World Bank the 1960s and 1970s slowed considerably in the
(Pardey, Roseboom and Anderson, 1991). The monetary val- 1980s. Particularly in debt-ridden regions, such as
ues throughout this paper have been expressed in constant
1980 PPP dollars. PPP stands for purchasing power parity and
sub-Saharan Africa, and Latin America and
represents a synthetic exchange rate that attempts to reflect Caribbean, investment in agricultural research
the purchasmg power of a currency. Monetary values m stagnated or even declined. In addition, the level
current local currency were first deflated to base-year 1980 of development aid to less-developed countries
with a local (GoP) deflator and then converted into constant stalled during the 1980s (OECD, 1989) while the
1980 dollars using 1980 PPP exchange rates taken from Sum-
mers and Heston (1988) and, for China, Summers and Heston
small but significant share that was channeled to
(1991). The authors thank Hugh Quigley for hts assistance in agricultural research is threatened by other de-
preparing this paper. mands. A reversal of these trends seems unlikely

0169-5150j94j$07.00 © 1994 Elsevier Science B.V. All nghts reserved

SSDJ 0169-5150(93)E0035-L
108 J.R. Anderson et al. /Agricultural Economics 10 (1994) 107-123

in the near future and therefore resources for upon to broaden their research agendas and to
public agricultural research in less-developed give greater attention to concerns of environmen-
countries could well tighten even further in the tal degradation and resource management. The
coming years. In the more-developed countries, international system is also restructuring its re-
public support for agricultural research is under search portfolio with regard to forestry, fisheries
close review and there is a strong tendency to and vegetable research in addition to its tradi-
have those who most directly benefit from re- tional emphasis on basic food crops and livestock.
search pick up at least part of the bill. Moreover, These changes raise major policy issues about the
agricultural surpluses, declining agricultural appropriate division of labor and problem focus
prices, and a continuing decline in farm numbers between the national, regional and international
in many of the more-developed countries have centers that are yet to be resolved.
led to populist calls for a moratorium on further There are large variations across countries and
public investment in agricultural research. over time in the level of investment in agricul-
Against this backdrop of fiscal stringencies, tural research. As a country's per-capita income
the demands being placed on national, and in- grows, its support for agricultural research - as
deed international, research systems are intensi- indexed by an agricultural research intensity
fying. In addition to the traditional emphasis on (ARI) ratio that expresses research expenditures
stimulating productivity growth within agricul- relative to agricultural output - tends also to
ture, many of these systems are also being called grow. But there are offsetting tendencies, includ-

Table 1
Annual agricultural research personnel and expenditures, regional totals
Region 1961-65 1971-75 1981-85 Growth rate(%) a

Agricultural research personnel (full-time equivalents per year)

Sub-Saharan Africa (43) b 1323 2416 4941 6.8
China 7469 11781 36335 8.2
Asia and Pacific, excl. China (28) 6641 12439 22576 6.3
Latin America and Caribbean (38) 2666 5840 9000 6.3
West Asia and North America (20) 2157 4 746 8995 7.4

Less-developed countries (130) 20256 37221 81848 7.2

More-developed countries (22) 40395 48123 56376 1.7

Total (152) 60651 85344 138224 4.2

Agricultural research expenditures (millions 1980 PPP dollars per year)

Sub-Saharan Africa (43) b 149.5 276.9 372.3 4.7
China 486.7 874.8 1712.7 6.5
Asia and Pacific, excl. China (28) 316.7 651.5 1 159.6 6.7
Latin America and Caribbean (38) 229.1 486.6 708.8 5.8
West Asia and North America (20) 126.9 300.7 455.4 6.6

Less-developed countries (130) 1308.9 2 590.5 4408.7 6.3

More-developed countries (22) 2190.7 3 726.3 4812.9 4.0
Total (152) 3499.6 6 316.8 9 221.6 5.0
Source: Most of this table is drawn from Pardey, Roseboom and Anderson (1991, p. 200), as are most of the data reported in this
paper. The China data are from Fan and Pardey (1992).
Totals may not add up exactly because of rounding.
a Compound annual average growth rate between 1961-65 and 1981-85.
b Bracketed figures indicate the number of countries in the regional totals. The appendix to Pardey, Roseboom and Anderson
(1991) indicates which specific countries were included in the regional aggregates.
 l.R. Anderson et al. j Agricultural Economics 10 (1994) 107-123 109

ing one whereby agricultural research expendi- 1980s, most noticeably in sub-Saharan Africa, and
tures rise less than proportionately with agricul- Latin America and Caribbean, both of which
tural output, due possibly to economies of size or have been struggling to contain soaring interna-
economies of scope in research. In this paper we tional debts. In fact, the 1976-80 to 1981-85
present a quantitative review of the global pat- annual rate of growth in real research expendi-
tern of investment in agricultural research using a tures was only 0. 7% for sub-Saharan Africa and
new compilation of data that serves to completely 0.9% for Latin America and Caribbean compared
revise and update the data series reported by with a more-and less-developed country average
Evenson and Kislev (1975), Oram and Bindlish of 2.9% and 4.3%, respectively. Although spend-
(1981) and Judd, Boyce and Evenson (1986). Our ing on agricultural research increased faster in
intention is to illustrate what is actually happen- less-developed than in more-developed countries
ing in the world of agricultural research and to during the past two decades, the less-developed
help move the policy dialogue beyond merely countries' share of total expenditure on research
qualitative impressions toward a process that is rose only to 48% from its 37% in 1961-65.
underpinned with a set of basic data and quanti- Of the less-developed regions, only in Asia and
tative indicators. Pacific did annual growth in research expendi-
tures exceed the annual increase in researchers.
In more-developed countries, on the other hand,
1. Public investments in national agricultural spending on research increased twice as fast as
research the number of researchers.

1.1. Investment trends

1.2. Expenditures per researcher
Differences in patterns of agricultural growth
are in large part the result of national differences Average spending per researcher in less-devel-
in factor and environmental endowments and in oped countries has been falling since the early
the policies adopted towards agriculture and, 1970s. In 1981-85, it was actually lower in real
more specifically, agricultural research. For two terms than in 1961-65. In more-developed coun-
decades, up to the mid-1980s, global investment tries meanwhile, spending per researcher has been
in public agricultural research increased substan- rising steadily and the emphasis has consistently
tially. 1 Between 1961-65 and 1981-85, the total been towards greater investment in human capi-
number of public-sector agricultural researchers tal within the NARSs.
grew at an average annual rate of 4.2%. The One reason for the relatively lower spending
number of researchers in less-developed coun- per researcher in less-developed countries is that,
tries increased by 7.2% a year, just over four of late, the rapidly expanded university systems in
times the annual rate in more-developed coun- these countries have produced many more gradu-
tries (Table 1). ates than previously. Many governments in less-
In the period 1981-85, the less-developed developed countries insist that public bodies, in-
countries employed 59% of the world's agricul- cluding research systems, employ graduates, but
tural researchers, compared with 33% in 1961-65. then fail to provide adequate matching funds.
Annual growth rates in research investment in In the Asia and Pacific region, expenditure per
less-developed countries have slowed during the researcher has always been lower than in other
less-developed regions. This is partly because rel-
ative prices for labor are lower, which induces a
substitution of labor for other inputs in the sys-
1 The global comparisons presented in this paper do not

include the former USSR, Eastern Europe, Cuba, Vietnam,

tem. But it is also because they are dominated by
Cambodia, the Republic of South Africa, and a number of larger research systems that may well be able to
very small countries for which no data were available. realize economies of scale and scope.
110 J.R. Anderson eta/. I Agricultural Economics 10 (1994) 107-123

In sub-Saharan Africa, expenditure per re- Table 2

searcher has for a long time been higher than in Nationality and degree status of agricultural researchers,
1981-85 average (%)
other regions. During the 1960s, the NARSs in
this region continued to be heavily staffed by Regionjcountry Expatriates Share of
postgraduates a
relatively expensive expatriates from the former
colonial powers. The poor quality of Africa's in- Sub-Saharan Africa a 29 45
Asia and Pacific b 11 53
frastructure and the need to import nearly all Latin America and Caribbean 2 51
equipment also drive up the research costs in this West Asia and North Africa 18 27
region. Although still higher than in most other Less-developed countries 12 48
less-developed regions, spending per researcher Australia na 57
in sub-Saharan Africa is falling, in part a reflec- New Zealand na 78
tion of the fact that expatriate researchers are United States na 93
increasingly being replaced by less expensive na- a Measures the proportion of national researchers holding a
tional research staff, and in part because of the PhD or MSc degree or equivalent. Figures for Australia and
above-mentioned budgetary crises. New Zealand are for 1981, and for the United States for 1980
only.
b Does not include China and India, the two major NARSs in
1.3. Human capital
the region.

One of the fundamental strengths (or, too of-

ten, weaknesses) of NARSs, and a major factor in roughly one-half of agricultural researchers, in-
determining the success of agricultural research, cluding expatriates, in less-developed countries
lies in the quality, composition and deployment held a postgraduate research degree (Table 2). If
of their research staff. expatriate researchers are excluded from the cal-
Developing meaningful measures of this hu- culation, no less-developed region has a share of
man capital component is challenging both con- researchers with a postgraduate qualification
ceptually and practically. Indicators such as uni- greater than 60%. For a significant number of
versity qualifications and years of research expe- countries it is even lower than 40%. Somewhat
rience may explain much of the difference in surprisingly, poorer less-developed regions have a
quality between research systems, but they are relatively high proportion of qualified staff, al-
not the only factors. The composition of the though that is much lower if expatriate re-
research staff will depend, among other things, searchers are excluded from the calculation.
on the NARS's size and the type of research it is In many less-developed countries, the early
conducting. These influences vary greatly be- agricultural research institutes were established
tween regions and will change within a system by European colonial powers and, during the first
over time. For instance, a smaller NARS whose half of this century, these institutes were staffed
activities are focused more on capturing research with expatriate researchers. On independence,
spillovers and adapting them to local circum- most former colonies moved to replace expatri-
stances is unlikely to need a cadre of researchers ates with national researchers. In some countries,
similar to that required by a large NARS that is this change took place gradually but in others it
likely to confront an altogether different scale was a more abrupt process and caused major
and set of research problems. Similarly, while a disruptions in agricultural research programs. At
system dominated by researchers holding PhDs present, only the smaller countries of sub-Saharan
and 20 years of experience may be considered Africa, the Caribbean, the Pacific and the oil-rich
highly qualified, it is not necessarily the most countries of West Asia have relatively large pro-
appropriate labor force to confront the applied portions of expatriates on their research staffs.
and site-specific problems that face many na- The share of expatriates is declining rapidly, how-
tional research systems today. ever. In sub-Saharan Africa, for example, the
Data for the period 1981-85 indicate that share of expatriates in NARSs was about 90% in
 J.R. Anderson eta!. j Agricultural Economics 10 (1994) 107-123 Ill

1960 but had declined to some 30% in the early ticularly when such training may be of question-
1980s. Making the plausible assumption that the able relevance.
numbers of expatriate researchers working within
the Chinese and Indian systems are negligible,
the percentage of expatriate researchers working 1.4. Commodity orientation
throughout the less-developed world in 1981-85
is estimated to be around 3%. In less-developed countries, agricultural re-
Although economic development can be ex- search is directed predominantly at crops. Based
pected to increase the supply of university gradu- on a sample of 83 countries, roughly two-thirds of
ates, research systems in some of the wealthier all agricultural researchers are engaged in work
less-developed countries appear to have difficulty related to crops. For the remainder, 19% are
recruiting or retaining qualified staff. In part, this engaged in livestock research, 7% in forestry re-
is because salaries and conditions in public agri- search, and 6% in fisheries research.
cultural research institutes are not competitive There are some limited regional disparities in
with other employment opportunities. In a num- the share of resources devoted to a particular
ber of countries, for example, universities employ commodity orientation (Table 3). While such dis-
large numbers of PhDs in the agricultural sci- parities are inevitable, given regional variations in
ences, while the national public agricultural re- the pattern of production, it has been argued by
search institutes employ few or none. some that less research is devoted to fisheries and
On the other hand, as argued earlier, a large forestry than their reported economic importance
proportion of PhDs on the research staff does not warrants (see Mergen et a!., 1988, on forestry
necessarily indicate a successful or a mature re- research). In fact, the data - as shown in Table 3
search program. The contemporary systems of - do not generally support this proposition. Re-
Australia and New Zealand, for instance, have search into forestry attracts more resources than
apparently achieved significant successes with a its congruent share in agricultural output in all
high proportion of staff trained only to the BSc or regions. In Asia and Pacific, and West Asia and
MSc level, as did the U.S. system in earlier years. North Africa, this is also true of fisheries.
While certainly not discounting the value of train- Nevertheless, the actual facilities for research
ing researchers to the PhD level, these observa- into forestry and fisheries are limited, primarily
tions would suggest that greater attention should because NARSs in less-developed countries are
be given to the research orientation and training generally small. The majority (73%) of them em-
within BSc and especially MSc programs at local ploy fewer than 200 researchers in total, while
universities rather than simply seeking a high only small percentages of these are engaged in
proportion of PhDs through training abroad, par- research into fisheries or forestry.

Table 3
Regional congruence between agricultural GDP and research personnel, 1981-85 (%)
Region Crops and Livestock Forestry Fisheries
AgGDP Research AgGDP Research AgGDP Research
Sub-Saharan Africa (22) a 88.6 87.3 4.7 7.3 6.6 5.4
Asia and Pacific, excl. China (10) 89.7 81.1 5.2 9.4 5.0 9.6
Latin America and Caribbean (20) 94.2 92.8 2.9 5.4 2.8 1.8
West Asia and North Africa (7) 95.9 91.6 2.4 5.7 1.7 2.7
Less-developed countries (59) 90.7 87.0 4.6 7.3 4.6 5.7
Data may not add up exactly because of rounding.
a Bracketed figures represent number of countries included in the regional samples on which the AgGDP breakdown is based. The
research breakdown is based on regional samples that include a somewhat larger number of countries.
112 J.R. Anderson et al. j Agricultural Economics 10 (1994) 107-123

1.5. Factor shares costs in less-developed country NARSs is that

capital items are often relatively more expensive
A major challenge for managers of research in less-developed countries and they often lack
systems is to make the most effective use of adequate repair and maintenance facilities, lead-
available resources. The best mix of spending on ing to an early write-off of equipment. Factor
capital equipment, personnel and operating costs substitution, where less expensive inputs are sub-
will depend to a large degree on the relative stituted for more expensive, may counterbalance
availability and cost of research inputs, their qual- this effect somewhat, but it is not likely that it
ity, and the type of research being conducted. will outweigh it.
Since the most effective combination will differ Whereas salaries and operating costs in agri-
between regions and change within a system over cultural research expenditures represent service
time, it is unrealistic to propose standards for flows, capital expenditures represent additions to
determining the 'optimal' mix of inputs in re- a stock. Thus, the high share of capital in annual
search. The data presented in this section should spending may also exaggerate the actual share of
be regarded as indications of magnitude based on capital in the services used to perform research.
a sample of countries, not as economic optima Capital equipment can last for many years, and a
necessarily to be targeted. measure of service flow, rather than of expendi-
The available data suggest that, in 1981-85, ture, would probably reduce the share of capital
NARSs in less-developed countries on average actually used by a research system in any given
devoted 19% of annual expenditures to capital year, particularly if such a system expands rapidly.
investment, compared with 8% in the U.S. The The recurrent costs of NARSs can be divided
higher share of spending on capital equipment by between salaries and operating costs. In less-de-
NARSs in less-developed countries supports the veloped countries, salaries tend to be lower and
conclusion, also evidenced by their rapid growth, operating costs higher than in more-developed
that most are in an expansionary phase. During countries. In 1981-85, a sample of 43 less-devel-
this phase, not only must capital stock be re- oped countries spent an average of $48,100 in
placed but additional capital stock must be ac- constant 1980 dollars per researcher, compared
quired. The pattern of expenditures in the U.S., with $91,200 per researcher in the U.S. The con-
on the other hand, reflects that of a mature temporary level of spending per researcher in
system, most of whose capital spending entails less-developed countries again appears to reflect
replacement of existing equipment. The emphasis the situation in the early years of the U.S. experi-
in the U.S. system has, moreover, been consis- ment stations. In fact, in the U.S., total spending
tently towards greater investment in human capi- per researcher fell steadily in the early years as
tal rather than physical capital in recent years. the recruitment of researchers outpaced rises in
Today the system performs with around 14 cents research expenditures. In the first 30 years, real
of physical capital for each dollar of human capi- operating expenditures per researcher fell to
tal, compared with about 27 cents 50 years ago roughly one-third of original levels and did not
(Pardey, Eveleens and Hallaway, forthcoming). recover until the 1970s - some 60 years later!
The contemporary pattern of expenditure in One of the major difficulties in making plausi-
less-developed countries mirrors that of U.S. ex- ble cross-country comparisons of factor shares is
periment stations in their early years at the turn that cross-country differences in price levels,
of this century, when the share of capital in total which are not consistent across different expendi-
expenditures rose steadily to peak at 29% in ture items, act to confound the comparisons.
1912, before steadily falling. Capital (i.e., land, Thus, if spending on operating costs and salaries
buildings, equipment, etc.) has claimed the same is adjusted to take account of price differences
share of overall spending in U.S. research sta- between countries, research in less-developed
tions (about 8%) for the past three decades. countries is seen to be more labor-intensive rela-
A second factor in the higher share of capital tive to the U.S. Looked at this way, the share of
 J.R. Anderson eta!. I Agricultural Economics 10 (1994) 107-123 113

operating costs in recurrent expenditures in cal specificity. For example, certain activities can
1981-85 averaged 15% in less-developed coun- relate to improving crops or to developing new
tries. breeds of plants suitable for specific agroecologi-
After adjusting for cross-country price differ- cal zones within a country. Alternatively, research
entials, operating expenditures per researcher are can be devoted to developing improved crop and
also much smaller in less-developed regions than soil management practices that will allow farmers
in the U.S. Agricultural researchers in sub- across a range of agroecological zones to increase
Saharan Africa, Asia and Pacific, Latin America yields or improve pest and disease control.
and Caribbean, and West Asia and North Africa Even in the absence of size economies with
work with only 50%, 43%, 74% and 22%, respec- regard to any particular line of research (e.g., a
tively, of the operating resources provided to a particular commodity research program), a sys-
U.S. researcher. However, the salary component tem may well be able to generate economies of
of recurrent expenditures (including the salaries scope through a judicious choice in its portfolio
of both scientific and support staff) in the less-de- of research activities. Such scope economies arise
veloped regions is much closer to the U.S. level. when a system can undertake a whole range of
And in sub-Saharan Africa it is even higher. This research endeavors more cheaply than if these
may be accounted for by the relatively high num- endeavors were undertaken by separate research
ber of expatriates still working in African NARSs entities. These economies can be achieved, for
and the fact that the employment policies of example, by sharing staff, equipment, informa-
many governments in less-developed countries re- tion, or know-how between different lines of re-
sult in NARSs employing large numbers of sup- search.
port staff. An important implication is that, when a sys-
tem can create sufficiently strong economies of
1.6. Size, scope, and spillovers scope, these can, in turn, lead to economies of
size across the whole range of its activities, even
Since 1961-65, the average size of NARSs has if such economies of size do not arise for some
more than doubled, from around 400 to 910 re- individual research programs (Baumol, Panzar
searchers, as has average expenditure per system. and Willig, 1988). Of course, there can also arise
In less-developed countries the average size of diseconomies of scope, particularly among small
NARSs has increased from 155 to 630 full-time systems that spread their limited resources across
equivalent researchers. Nevertheless, 95 of the numerous research areas. Thus, small NARSs
considered 130 NARSs in less-developed coun- will be unable to conduct research in all areas
tries still employ fewer than 200 researchers, while that may warrant attention in the agricultural
39 systems employ fewer than 25 researchers. systems they serve. They will have to make choices
Only 14 employ more than 1000, illustrating that between areas of study and this, in turn, requires
the growth and development of NARSs in the some specialization and flexibility in response to
past two decades has diverged significantly. opportunities as circumstances change.
When analyzing the cost structures and effec- The efficiency of a research system can also be
tiveness of a NARS, one needs to consider both increased by adapting research conducted else-
the overall size and diversity of its operations and where to local circumstances. The ability to cap-
the agricultural system it serves. ture research spillovers is particularly important
The evidence on whether or not research oper- for small NARSs with the capacity to do little
ations are subject to economies of size is limited more than search and screening for suitable tech-
and far from definitive. In the case of a NARS, nologies. The best source of spillover would seem
considerations of economies of size are con- to be neighboring countries with similar agricul-
founded by the fact that these systems generate a tural systems and agroecological features. This
wide diversity of products and services that vary strategy would require a policy of hiring staff for
in their commodity, technology, and agroecologi- their ability to adapt research to local situations
114 J.R. Anderson eta!./ Agricultural Economics 10 (1994) 107-123

rather than necessarily to carry out original re- still have an effect in increasing output for as
search. It also requires flexibility in the research long as 30 years after the research was initiated.
system to identify and act upon opportunities To consider gains in agricultural gross domestic
arising from developments elsewhere. product (AgGDP) as a measure of the impact of
There is some disturbing evidence that many contemporaneous research expenditure could
smaller NARSs are unable to take up informa- thus be misleading. Differences in the quality of
tion quickly enough and that the knowledge they land and labor, and in the intensity of use of
work with is increasingly out-of-date. In a world other inputs such as fertilizers and machinery,
of growing international interconnectedness, will influence output and so distort international
adapting spillovers from other research systems is comparisons of output that are measured only in
most effective if they can be adapted to local terms of research.
circumstances in a timely manner. Using out-of- In fact, all the more- and less-developed re-
date information only undermines a country's rel- gions steadily increased their research intensities
ative technological capacity and therefore its during that period (Table 4). In 1981-85, less-de-
competitiveness. veloped countries spent an average of nearly $4
per agricultural worker on research, 2.5 times the
I. 7. Research and productivity amount spent two decades before. In more-devel-
oped countries, spending on research increased
Research is best seen as an investment activity. 4.4 times over the same period, to $214 for every
The research process itself takes time, and a agricultural worker.
further period elapses before the results of re- A final factor urging caution in assessing the
search are taken up. Recent evidence suggests, benefits of research in terms of increasing AgGDP
furthermore, that the benefits of research can is that a large share of agricultural research may

Table 4
Agricultural research factor-intensity ratios
Region 1961-65 1971-75 1981-85
Agricultural research expenditures per economically active person in agriculture
Sub-Saharan (37) a 1.7 2.7 3.1
China 1.7 2.5 4.1
Asia and Pacific, excl. China (15) 1.2 2.2 3.4
Latin America and Caribbean (26) 6.5 12.8 17.7
West Asia and North Africa (13) 4.5 10.5 14.3
Less-developed countries (92) 1.8 3.2 4.6
More-developed countries (18) 48.6 119.1 213.5
Total (110) 4.7 7.5 9.5

Agricultural research expenditures per hectare of agricultural/and

Sub-Saharan (37) 0.2 0.4 0.6
China 1.2 2.1 4.1
Asia and Pacific, excl. China (15) 1.0 2.1 3.6
Latin America and Caribbean (26) 0.4 0.7 1.0
West Asia and North Africa (13) 0.4 1.0 1.4
Less-developed countries (92) 0.6 1.1 1.8
More-developed countries (18) 1.8 3.1 4.0
Total (110) 1.0 1.8 2.5
All expenditures are in constant 1980 PPP dollars.
a Bracketed figures indicate the number of countries in regional totals.
 I.R. Anderson et al. /Agricultural Economics 10 (1994) 107-123 115

be directed towards maintaining gains from ear- research, 4.3% if related to spending by and for
lier research rather than enhancing output levels less-developed countries. Its budget rose, in nom-
per se. Recent estimates suggest that, in the U.S., inal terms, from $20 million contributed by 20
around one-third of research expenditures are donors in 1971, to $280 million from 40 donors in
spent on maintenance, and it is probable that 1990. If corrected for inflation, the CG expendi-
many less-developed countries devote at least as tures show clearly different phases of growth:
much proportional effort to such work (Adusei rapid expansion during the 1970s, slower growth
and Norton, 1990). during the 1980s, and apparent stagnation or
even decline in the past few years.
The U.S. was the largest single donor to the
2. International investments in agricultural CG, although both Europe and Japan increased
research their share of contributions during the 1980s. The
World Bank acted as a balancing "donor of last
Contrary to the situation in many areas of resort," allocating its funds after other donor
scientific research, there has always been an im- intentions were known. It lately has contributed
portant element of international cooperation in around 15% of the system's total budget each
agricultural research. Much of this is due to the year.
legacy of the colonial relationships that existed at During the 1980s, although the CG was estab-
the time institutionalized agricultural research lished partly in response to the high levels of
was developing. poverty and hunger in Asia, the emphasis in the
One of the leading international organizations allocation of funds has shifted to sub-Saharan
in the field is the Consultative Group on Interna- Africa. Between 1986 and 1988, sub-Saharan
tional Agricultural Research (CGIAR), an um- Africa accounted for 39% of the CG's core ex-
brella body of around 40 donor countries and penditures, compared with 26% to Asia, 21% to
international agencies that foster the activities of Latin America and Caribbean, and 14% to West
some dozen supranational research centers. (Re- Asia and North Africa.
cently the number was enlarged to 18 but our The "appropriate" regional allocation of funds
discussion here refers to the original 13.) Ten of is just one of the policy issues facing the CG.
these centers have their headquarters in less-de- Although much of the increase in funding to
veloped countries. Most are engaged in research sub-Saharan Africa has been for special projects,
into either food commodities or agricultural pro- it is arguable that the concentration of resources
duction problems in a particular tropical region, has swung too much towards that region at the
and three undertake worldwide research into spe- ultimate expense of Asia, which has several-fold
cific commodities. more poor than sub-Saharan Africa.
Established in 1971, the stated objective of the The commodity orientation of the system has
CGIAR (CG for short), was to assist efforts to been subject to change over time. As the system
increase food production in the less-developed expanded, the share allocated to cereals research
world. The goals were extended in 1990, in recog- declined steadily to about 40%, of which rice
nition of agriculture's broader role in economic research still accounts for the largest share at
development, to helping less-developed countries 17% of the system's total. Food crops, such as
achieve self-reliance in food. Self-reliance is taken potatoes, other roots and tubers, and legumes,
to mean the capacity of a country to provide account for 24% of the total, while livestock
sufficient food for its population, either directly research accounts for around 20%. The remain-
through local food production or indirectly by ing resources are allocated to noncommodity pro-
generating agricultural exports that will allow food grams, including farming systems, food policy,
to be imported. genetic resources, and NARS capacity building.
In 1981-85, the CG accounted for only 1.8% The recent expansion of the system has broad-
of global public-sector spending on agricultural ened the commodity coverage to include fish-
116 l.R. Anderson eta!./ Agricultural Economics 10 (1994) 107-123

eries, agroforestry and forestry, and bananas and and resource management concerns into its re-
plantains. It may also eventually include horticul- search program are currently being implemented.
tural commodities. This is being done by expanding the system and
The recent reorientation of CG objectives to redesigning its approach in order to incorporate
emphasize self-reliance instead of self-sufficiency institutional entities that specifically address re-
in food was a recognition of the fact that increas- search concerns within an agroecological perspec-
ing food production is not, in itself, a solution to tive. Aware of the fact that socioeconomic - not
the hunger problem. Future policies must reflect just natural - conditions constrain the effective-
the role of agricultural growth as a means of ness and spillover potential of the system's re-
generating additional on-farm and off-farm in- search, this agroecological aspect is being over-
come and employment, and the need to sustain laid on a geopolitical or regional dimension to
the natural-resource base on which continued generate a so-called "ecoregional" perspective
gains in agricultural productivity depend. (TAC/ Center Directors Working Group, 1993).
The CG's initial efforts were largely targeted
toward more favorable production environments.
Technology packages were developed that in- 3. Private investments in agricultural research
volved higher rates of fertilizer application, im-
proved water management and cultural practices, Any formulation of future public-sector re-
along with new crop cultivars that were particu- search policy must take into account the level of
larly responsive to more intensified production activity and changing role of the private sector in
regimes. While the dramatic contribution of these agricultural research. As farmers use more pur-
technology packages to increasing global food chased inputs and the value-added in agriculture
supplies is unquestioned (Anderson, Herdt and increasingly moves off-farm to the marketing and
Scobie, 1988), by the mid-1970s the CG had also processing sectors, it is likely that the incentives
begun to address production constraints in the for private-sector investment in research will
more marginal environments of the semi-arid and grow. While there is a general perception that the
(sub-)humid tropics. private-sector has increased its participation in
At present, about 30% of CG funding is tar- and funding of agricultural research, there are no
geted towards technology for marginal lands, available data to give an accurate quantitative or
which is roughly equal to the percentage of the even qualitative perspective of these develop-
poor population in less-developed countries that ments at the global, regional, and, in many in-
live in these areas. The issue of the relative stances, even country-level.
merits of seeking to improve or maintain produc- There are various reasons why these data are
tivity of favorable versus marginal lands will con- not readily available. Firms may feel their com-
tinue to be an important one for the CG, particu- petitive interests are not well served by a full and
larly with respect to the potential opportunity frank disclosure of their R & D activities and so
costs (in terms of productivity gains foregone) of may be less than forthcoming in this regard. Even
diverting scarce research resources away from when such data are reported, there are genuine
more responsive areas towards the more marginal difficulties in identifying the R & D component
ones. Analyses of the type reported by Byerlee that relates specifically to agriculture. This is
and Morris (1993) will be needed for guiding particularly a problem for multiproduct firms in
future investment policy in this regard. the chemical, pharmaceutical, biological and me-
Research on resource management will be- chanical industries that pursue economies of
come more important as the need for continued scope by sharing research resources across a
increases in food production places an ever number of lines of research. Apportioning these
greater strain on the world's natural resources. R & D expenditures to a particular country in any
The CG has taken the first step in this direction. meaningful way is also problematic, especially
Strategies on how best to include agroforestry when dealing with multinational firms that cen-
 J.R. Anderson eta/. I Agricultural Economics 10 (1994) 107-123 117

tralize various aspects of their global or regional 1991, the private sector accounted for 37% of
R&D operations. total agricultural R&D expenditures in Colombia
Recent estimates by Pray and Neumeyer (1989) (compared with 22% in 1970) and in Ecuador the
for the U.S. and Thirtle et al. (1991) for the UK private-sector share is now 27% (up from 19% in
suggest that private-sector investments in agricul- 1986). To the extent these developments are rep-
tural and food (i.e., largely post-harvest) R&D resentative of the situation in other Latin Ameri-
are substantial and at least as great as the public can (and perhaps some Asian) countries, they call
effort. Reliable global estimates of private expen- for a radical rethink of the roles of the public-sec-
ditures on agricultural R&D are simply unavail- tor research agencies in these regions.
able. Persley (1990, p. 48) reports that about 540 Having said this, however, there are still many
million dollars were spent worldwide by the pri- countries, especially in Africa, where the low
vate sector on agricultural biotechnology research level of purchased inputs in agriculture limits the
in 1985, accounting for roughly 60% of the 900 size of the derived market for privately produced
million dollars spent on modern biotechnology agricultural technologies. This situation is likely
research for agriculture in that year. This to continue for some time to come. Nevertheless,
private-sector R&D figure is significantly larger governments have a number of policy instruments
than the 36 million dollars Pray and Echeverria with which to influence private R&D. Public-sec-
(1991) estimate was spent annually by multina- tor research can foster private-sector research by
tional companies on all types of agricultural (in- providing (or selling) research results and by
cluding post-harvest) R&D in less-developed training the personnel needed by private compa-
countries during the latter half of the 1980s. nies to conduct research. Patents and plant-variety
Taken together these figures suggest that an protection laws, if they are well designed and
overwhelmingly large share of private-sector enforced, can create the necessary incentives for
spending on agricultural R&D occurs in the private companies to invest in R&D. Technology
more-developed countries. imports can stimulate local R&D, so more liber-
The data reported by Pray and Echeverria alized technological trade could also increase lo-
(1991) support this view. During the mid-1980s, cal private-sector R & D activities.
spending by U.S. firms on R&D in the food and Innovative institutional arrangements can also
agricultural sectors was around 2.4 billion dollars help foster those complementarities that exist
per annum (with about 58% of this total going to when the generally more upstream or "pretech-
agriculture). Comparable figures for the UK and nology" types of research best suited to the pub-
France are 530 million and 270 million dollars, lic domain are married with the more applied,
respectively. Much of the privately sourced funds technology-generating types of research best
for agricultural R&D in the less-developed re- suited to the private domain. For instance, joint-
gions of the world comes from Latin America and venture research endeavors, where both public
Asia and, according to Pray and Echeverria and private agencies jointly undertake and/ or
(1991), it is concentrated in a few large countries cofinance a program of research are becoming
such as Brazil, Mexico, Argentina and India. more frequent. Fee-for-service or contract re-
Our understanding of the scale and scope of search is also increasingly being used to privatize
these private-sector efforts is, unfortunately, woe- the financing of research being performed by
fully inadequate. The recent and careful efforts public-sector research institutions.
by Falconi (1992, 1993) to compile time-series Private, for-profit research should not be seen
data on private-sector, agricultural R&D expen- as something intrinsically detrimental to the pub-
ditures in Ecuador and Colombia are quite re- lic good, but neither is it likely that an unfettered
vealing in this respect. These new data show that, private sector has the incentives to invest suffi-
in both countries, private-sector spending on agri- ciently in researching those problems that will
cultural R & D grew much more rapidly than pub- optimize social welfare. Public-policy formulators
licly sourced expenditures during the 1980s. By will need to become increasingly sensitive to a
118 J.R. Anderson et al. I Agricultural Economics 10 (1994) 107-123

rapidly changing technological and institutional Table 5

environment in order to take advantage of the Agricultural research-intensity ratios by region and income
group, total weighted average percentages(%)
opportunities that exist to mobilize both private
and public research resources in a socially desir- Region/income group • 1961-65 1971-75 1981-85
able manner. Sub-Saharan Africa (37) b 0.26 0.42 0.49
China 0.42 0.40 0.41
Asia and Pacific, excl. China (15) 0.14 0.22 0.32
Latin America and Caribbean (26) 0.30 0.46 0.58
West Asia and North Africa (13) 0.28 0.50 0.52
4. Political and financial support for agricultural
research Less-developed countries (92) 0.26 0.34 0.41
More-developed countries (18) 0.96 1.41 2.03

A fundamental task facing NARSs is to win Low (30) 0.25 0.30 0.37
Lower-middle (28) 0.24 0.35 0.40
public support for research and translate it to Middle (18) 0.25 0.46 0.57
financial support. This must be done in the con- Upper-middle (18) 0.27 0.44 0.55
text of a public sector subject to competing claims High (16) 1.08 1.57 2.23
on its scarce resources from various interest Total sample (110) 0.48 0.63 0.71
groups in society, be they producers, consumers,
• Countries assigned to income classes based on 1971-75
or taxpayers.
per-capita GOP averages where low is < $600; lower middle is
From this perspective, governments give differ- $600-1499; middle is $1500-2999; upper middle is $3000-
ential preference to various programs both within 5999; and high is > $6000.
and between sectors of the economy in response b Bracketed figures represent number of countries in each

to such pressures. Thus, agricultural expenditures region or income class.

are committed to such programs as rural infras-
tructure, education and credit as well as to the
generation and dissemination of new agricultural research in the context of the overall level of
technologies. In addition, many poor countries support for agriculture. 2
implement distortionary pricing and marketing A traditional measure of the level of support
policies in the (short-run) pursuit of cheap-food for agricultural research is the agricultural re-
policies and the like, that ultimately tax agricul- search-intensity ratio that expresses levels of re-
ture and accelerate the transfer of resources from search spending as a percentage of agricultural
the sector. These same policies, in part by under- gross domestic product (AgooP ). Although the
valuing the sector-specific assets in agriculture majority of the less-developed countries spent
(e.g., land, irrigation facilities, etc.), can also lead well above 0.5% of AgooP on agricultural re-
to an underinvestment in agricultural research search in 1981-85, the weighted average was only
and the level of effort invested by farmers in 0.4% (Table 5). This apparent difference between
searching for, evaluating, and adapting new agri- simple and weighted average is caused by the fact
cultural technologies and practices. that the smaller less-developed countries tend to
To gain a full understanding of the observed have substantially higher agricultural research-in-
disparities in the nature and level of support for tensity ratios than the large less-developed coun-
agricultural research (and the ultimate productiv- tries. The weighted average of the more-devel-
ity effects that flow therefrom) would require oped countries barely reached 2% in 1981-85.
detailed consideration of these "political econ- The Southern European countries lagged signifi-
omy" forces - an exercise that would take us well cantly behind the other more-developed coun-
beyond our present brief (Gardner, 1990; Roe tries. When calculated by income group, a (not so
and Pardey, 1993; Alston and Pardey, 1994).
Rather, the aim here is to present some compara-
tive evidence on the level of support for public 2 For an earlier version of these data, see Pardey, Kang and
agricultural research, and to place publicly funded Elliott (1989).
 J.R. Anderson et al. /Agricultural Economics 10 (1994) 107-123 119

surprising) strong correlation appears between because the efficacy of a country's research en-
per-capita income and the agricultural research- deavor differs between regions and over time. It
intensity ratio. could, therefore, be more helpful, instead of set-
Although agricultural research-intensity ratios ting arbitrary targets for research investment, to
approximately doubled in both more- and less-de- fix a desired rate of return from the investment
veloped countries between 1961-65 and 1981-85, made - to set targets that would push rates of
they declined in the latter half of that period in return to below 20%, for example.
37% of the less-developed countries, one-half of The data presented in Table 6 show that low-
which were in sub-Saharan Africa. income countries spend a considerably greater
Research investment has traditionally pro- share of overall public expenditures on agricul-
duced high levels of return compared with invest- ture and agricultural research than high-income
ments in other areas, up to and exceeding 35% in countries, around 11% on agriculture and 0.7%
some instances (Echeverria, 1990). This fact, and on agricultural research, compared with 2.7%
the gap in investment compared with more-devel- and 0.2%, respectively, in high-income countries.
oped countries, has led some authorities to con- Moreover, the share of public expenditures on
clude that many less-developed countries under- agriculture directed specifically to research re-
invest in agricultural research. It has also led to mains surprisingly constant, at around 8% in
calls from the World Bank, for example, to set a 1981-85, for both poor and rich countries alike.
research investment target of 2% of AgGDP To understand why this is so would involve, at
(World Bank, 1981). a minimum, a detailed consideration of the deci-
Research-intensity ratios are useful to policy- sion-making processes whereby public research
makers because they indicate the importance investments, pricing policies and the like are
other countries attach to agricultural research. jointly determined. Particular attention would
But they may be an unreliable indicator of the need to be given to the relative incidence of
appropriateness of a nation's research investment research benefits and costs (across producers,

Table 6
Agricultural and agricultural research shares in public-sector expenditures
Income group a 1971-75 1976-80 1981-85
Percentage of agricultural expenditures in total government expenditures
Low (13) b 10.5 11.7 11.2
Lower-middle (18) 7.5 8.1 9.3
Middle (12) 6.5 5.7 5.2
Upper-middle (12) 6.7 4.7 4.3
High (15) 3.0 2.7 2.5
Total sample (70) 7.1 6.9 6.8

Percentage of agricultural research expenditures in total government expenditures

Low (13) 0.8 0.7 0.7
Lower-middle (18) 0.7 0.5 0.6
Middle (12) 0.5 0.4 0.4
Upper-middle (12) 0.2 0.2 0.2
High (12) 0.3 0.2 0.2

Total sample (70) 0.5 0.4 0.4

All data represent simple averages across all countries in each income class.
a Countries assigned to income classes based on 1971-75 per-capita GDP averages where low is < $600; lower middle is $600-1499;
middle is $1500-2999; upper-middle is $3000-5999; and high is ~ $6000.
b Bracketed figures represent number of countries in each income class.
120 J.R. Anderson eta!. I Agricultural Economics 10 (1994) 107-123

consumers and taxpayers) in relation to alterna- per-capita incidence of "costs" associated with
tive policy instruments, be they investing in rural such programs. If .this were the case, the willing-
public goods such as agricultural research versus ness of rural-based coalitions to lobby govern-
taxes, subsidies and production quotas (Alston ments in support of agricultural research (and
and Pardey, 1994). In the absence of available other forms of interventions that transfer re-
case-by-case data, the macro-level figures in Table sources to agriculture rather than the nonagricul-
7 are suggestive of some of the political economy tural sector) may, in turn, be positively associated
forces at work here. with per-capita income. Modelling and quantify-
While total government spending on agricul- ing governments' incentives to invest in rural pub-
ture, indexed over the agricultural population, lic goods such as research is necessary but far
increases dramatically by a factor of 85 times, from sufficient to develop policies that help sus-
from around $21 per capita in the low-income tain support to public-sector agricultural re-
countries to $1800 per capita in the high-income search.
countries, there is only a corresponding 8-fold
increase in agricultural spending indexed over the Donor support. While funding for agricultural re-
total population. Per-capita spending on agricul- search is only a small part of international devel-
tural research follows a similar pattern. Thus, as opment aid programs, it constitutes a significant
one moves from low- to high-income countries, contribution to the financing of numerous less-
the level of per-capita "benefits" or transfers developed NARSs. Aid from donor countries or
accruing to rural-based coalitions may well in- organizations is particularly vital for countries
crease at a disproportionately larger rate than the where high levels of international debt and an

Table 7
Public spending per capita on agriculture and agricultural research
Income group a Government expenditure on agriculture Agricultural research expenditures
1971-75 1976-80 1981-85 1971-75 1976-90 1981-85
Per head of agricultural population
Low (13) b 14.0 18.9 21.1 0.9 1.1 1.3
Lower-middle (18) 44.0 69.5 102.1 3.7 4.0 5.3
Middle (12) 77.8 94.8 119.2 5.5 6.1 7.6
Upper-middle (12) 218.8 358.7 552.3 12.4 19.8 26.5
High (15) 1338.2 1423.1 1801.0 91.8 113.2 140.6

Total 362.4 404.1 531.2 23.9 29.9 37.6

Per head of total population

Low (13) 10.0 13.4 14.1 0.7 0.8 0.9
Lower-middle (18) 20.9 29.6 38.7 1.5 1.8 2.3
Middle (12) 31.6 35.3 38.1 2.4 2.3 2.6
Upper-middle (12) 66.0 62.1 93.0 2.2 2.5 2.7
High (15) 111.5 112.4 115.0 7.3 8.1 8.5
Total (70) 47.9 50.9 56.3 2.9 3.2 3.5
All data represent simple averages across all countries in each income class and are expressed in constant 1980 PPP dollars.
a Countries assigned to income classes based on 1971-75 per-capita GDP average where low is < $600; lower-middle is $600-1499;
middle is $1500-2999; upper-middle is $3000-5999; and high is ~ $6000.
b Bracketed figures represent number of countries in each income class.
 l.R. Anderson eta/./ Agricultural Economics 10 (1994) 107-123 121

inadequate tax base make it virtually impossible and North Africa were much lower, 7% and 11%,
for the national government to adequately sup- respectively.
port a viable agricultural research program.
There is a serious lack of data available on
precise levels of donor support to NARSs. What 5. Concluding observations
available data there are, are difficult to stand-
ardize given the disparate reporting methods used While the past contributions of agricultural
by NARSs as well as donors. Figures from donor research to productivity gains and the improve-
countries and organizations for the period 1981- ments in living standards that followed have been
85 put contributions to agricultural research at an impressive, the challenges that lie ahead are con-
average of $658 million a year, which amounts to siderable indeed. There will be unprecedented
a very modest 1.9% of total official development increases in the demand for additional food and
assistance to less-developed countries. Based on fibre production while the threats to even achiev-
data from the NARSs, donor contributions in ing, let alone sustaining, such levels of output in
that period amounted to only about $355 million the face of a degrading natural resource base for
a year. This discrepancy can probably be ex- agriculture loom large. Such threats appear as
plained by the fact that NARSs commonly under- real for the more-favored, intensively cultivated
estimate the full extent of contributions they re- production environments as they are for the more
ceive. In estimating donor support, NARSs, quite marginal areas (Pingali, 1994).
understandably, often exclude payments in kind, There are unlikely to be any quick technologi-
such as the salaries and expenses of expatriates cal "fixes" to addressing these concerns. In fact,
working for them, which can be a substantial for the more immediate term at least, maintain-
element of aid contributions. It is also difficult to ing as well as enhancing past productivity in-
compile accurate figures on the amount of aid to creases is likely to come from the incremental
a research system when it is given as part of a gains arising from a whole array of new technolo-
wider package of aid that is distributed through gies and management practices (Byerlee, 1994).
the country's national government. While individually less "newsworthy" than the
The available data suggest that, in real terms, Green-Revolution technologies of the past, these
donor aid to NARSs has fallen since 1980 as sources of growth, when taken as a group, will
overall levels of development aid have stood still. nevertheless be just as real.
The World Bank accounts for around 25% of But to realize these output gains in a manner
donor funds applied to agricultural research in that preserves the environment will require a
the less-developed world, but the Bank's support sustained commitment to national and interna-
for individual national research endeavors de- tional research endeavors. While many countries
clined during the 1980s. Moreover, this support is experienced a substantial growth in their research
concentrated in just a few NARSs. Of the $817 capacity in the 1960s and 1970s, a considerable
million it allocated to strengthen less-developed number saw an erosion of their public-sector re-
NARSs during the period 1981-87, two-thirds search capacity in the 1980s. Although privately
went to only six projects. sponsored research endeavors are sure to grow in
The levels of external funding to national sys- the future, the corollary is not necessarily to cut
tems vary enormously, from none in Venezuela back on public-sector investments. In fact, the
and South Korea to 85% in Tuvalu. Sub-Saharan substantial growth in privately sponsored re-
Africa had the highest rate of donor funding, search in the U.S. over the past several decades
receiving on average 35% of its expenditures from occurred in conjunction with a continued, albeit
donors. NARSs in the Asia and Pacific region slower, growth in public-sector research invest-
received an average of 26% of their funding from ments.
donors. The levels of donor support to NARSs in To fully harness the potential complementari-
Latin America and Caribbean, and West Asia ties and synergy between public and private re-
122 J.R. Anderson et al. /Agricultural Economics 10 (1994) 107-123

search endeavors will require that more attention vice for National Agricultural Research, The Hague, The
be given to each sector's comparative research Netherlands, 46 pp.
Falconi, C.A. (1993) Public and private-sector interactions in
advantage. In particular, the gains to researching agricultural research in less-developed countries: the case
improved agricultural management and produc- of Colombia. ISNAR Disc. Paper 93-14, International
tion practices - those that will play a large role in Service for National Agricultural Research, The Hague,
realizing sustainable improvements in agricultural The Netherlands, 53 pp.
output - are generally difficult to appropriate Fan, S. and Pardey, P.G. (1992) Agricultural research in
China: its institutional development and impact. Interna-
and likely to remain the domain of the public tional Service for National Agricultural Research, The
sector. So too are the more basic, pretechnology Hague, The Netherlands, 96 pp.
types of research that, in turn, lay the founda- Gardner, B.L. (1990) Price supports and optimal spending on
tions for the privately-sponsored applied and agricultural research. Seminar Paper 89-02, Centre for
adaptive research programs of tomorrow. Failure International Economic Studies, University of Adelaide,
S.A., 37 pp.
to support and nurture today's research endeav- Hazell, P.B.R. and Ramasamy, C. (1991) The Green Revolu-
ors may well reap many unfortunate and unde- tion Reconsidered: The Impact of the High-Yielding Rice
sired consequences in the not-too-distant future. Varieties in South India. Johns Hopkins University Press,
Baltimore, MD, 286 pp.
Judd, M.A., Boyce, J.K. and Evenson, R.E. (1986) Investing in
agricultural supply: the determinants of agricultural re-
References search and extension investment. Econ. Dev. Cult. Change,
35: 77-113.
Adusei, E.O. and Norton, G.W. (1996) The magnitude of Lipton, M.L. with Longhurst, R. (1989) New Seeds and Poor
agricultural maintenance research in the USA. J. Prod. People. Johns Hopkins University Press, Baltimore, MD,
Agric. 3: 1-6. 289 pp.
Alston, J.M. and Pardey, P.G. (1994) Distortion in prices and Mergen, F., Evenson, R.E., Judd, M.A. and Putnam, J. (1988)
agricultural research investments. Chapter 4 in: J.R. An- Forestry research: a provisional global inventory. Econ.
derson (Editor), Agricultural Technology Policy: Issues for Dev. Cult. Change, 37: 149-171.
the International Community. CAB International, OECD (1989) Development Cooperation in the 1990s - Ef-
Wallingford, pp. 57-82. forts and Policies of the Members of the Development
Anderson, J.R., Herdt, R.W. and Scobie, G.M. (1988) Science Assistance Committee. Organization for Economic Coop-
and Food: The CGIAR and Its Partners. World Bank, eration and Development, Paris, 287 pp.
Washington, DC, 134 pp. Oram, P.A. and Bindlish, V. (1981) Resource allocation to
Baumol, W.J., Panzar, J.C. and Willig, R.D. (1988) Con- national agricultural research: trends in the 1970s - a
testable Markets and the Theory of Industry Structure review of third world systems. International Service for
(Revised Edition). Harcourt, San Diego, CA, 538 pp. National Agricultural Research, The Hague, The Nether-
Byerlee, D. (1993) Technology transfer systems for improved lands, 104 pp.
crop management: lessons for the future. Chapter 12 in: Pardey, P.G., Kang, M.S. and Elliott, H. (1989) Structure of
J.R. Anderson (Editor), Agricultural Technology Policy: public support for national agricultural research systems: a
Issues for the International Community. CAB Interna- political economy perspective. Agric. Econ., 3: 261-278.
tional, Wallingford, pp. 206-228. Pardey, P.G., Roseboom, J. and Anderson, J.R., Editors (1991)
Byerlee, D. and Morris, M. (1993) Research for marginal Agricultural Research Policy: International Quantitative
environments: Are we underinvested? Food Policy, 18: Perspectives. Cambridge University Press, Cambridge, 462
381-393. pp.
Echeverria, R.G. (1990) Assessing the impact of agricultural Pardey, P.G., Eveleens, W.M. and Hallaway, M.L. (forthcom-
research. In: R.G. Echeverria (Editor), Methods for Diag- ing). A Statistical History of US Agricultural Research:
nosing Research System Constraints and Assessing the 1889 to 1991. Center for International Food and Agricul-
Impact of Agricultural Research, II. International Service tural Policy, University of Minnesota, St. Paul.
for National Agricultural Research, The Hague, The Persley, G.J. (1990) Beyond Mendel's Garden: Biotechnology
Netherlands, pp. 1-31. in the Service of World Agriculture. CAB International,
Evenson, R.E. and Kislev, Y. (1975) Investment in agricultural Wallingford, 155 pp.
research and extension: a survey of international data. Pingali, P.L. (1994) Technological prospects for reversing the
Econ. Dev. Cult. Change, 23: 507-521. declining trend in Asia's rice productivity. Chapter 21 in:
Falconi, C.A. (1992) Public and private-sector interactions in J.R. Anderson (Editor), Agricultural Technology Policy:
agricultural research in less-developed countries: the case Issues for the International Community. CAB Interna-
of Ecuador. ISNAR Disc. Paper 92-13, International Ser- tional, Wallingford, pp. 380-397.
 J.R. Anderson et al. j Agricultural Economics 10 (1994) 107-123 123

Pray, C.E. and Echeverria, R.G. (1991) Private-sector agricul- comparisons of real product and price level estimates for
tural research in less-developed countries. In: P.G. Pardey, 130 countries, 1950-1985. Rev. Income Wealth, 34: 1-25.
J. Roseboom and J.R. Anderson (Editors), Agricultural Summers, R. and Heston, A. (1991) The Penn World Table
Research Policy: International Quantitative Perspectives. (Mark 5): an expanded set of international comparisons,
Cambridge University Press, Cambridge, pp. 343-364. 1950-1988. Q. J. Econ., 106: 327-368.
Pray, C.E. and Neumeyer, C.F. (1989) Impact of technology TACjCenter Directors Working Group (1993) The ecore-
policy on research in the agricultural input industries. gional approach to research in the CGIAR. TAC Secre-
Paper presented at the American Agricultural Economics tariat, Food and Agriculture Organization, Rome, 23 pp.
Association Meetings, Baton Rouge, LA. Thirtle, C.G., Beck, H.S., Palladino, P., Upton, M. and Wise,
Roe, T.L. and Pardey, P.G. (1991) Economic policy and W.S. (1991) Agriculture and Food. In: R. Nicholson, C.
investment in rural public goods: a political economy per- Cunningham and P. Gummett (Editors), Science and
spective. In: P.G. Pardey, J. Roseboom and J.R. Anderson Technology in the United Kingdom. Longmans, London,
(Editors), Agricultural Research Policy: International pp. 116-146.
Quantitative Perspectives. Cambridge University Press, World Bank (1981) Agricultural research - Sector policy
Cambridge, pp. 7-49. paper. World Bank, Washington, DC, 110 pp.
Summers, R. and Heston, A. (1988) A new set of international