Previous Page Table of Contents Next Page


Chapter 4 - The economic contributions of agricultural extension to agricultural and rural development

Robert Evenson

Robert Evenson is a Professor at Yale University, and serves as Director of The International and Development Economics Program.

Investment indicators: Agricultural research and extension
The conceptual foundation for extension impact
A note on statistical methods and issues for economic evaluation
Estimates of economic impacts: A summary
Lessons
Notes
References

Agricultural extension programmes are quite diverse from an international perspective. Most are managed as public sector agencies, usually located in the ministry of agriculture, but some are located in other ministries such as education or rural development. Many are managed by nongovernmental organizations (NGOs). Many private firms and private organizations (for example, coffee-growers' associations) conduct extension programmes. Even within the most typical organizational structure, where extension is part of the government's ministry of agriculture, there is great variation in the degree of decentralization of management of extension services. In some countries, extension is decentralized, as in India, where it is a state subject. In most developing countries, however, governmental services are highly centralized, with varying forms of regional and subregional units designed to serve local areas.

Further, there is great variation in the skill level and agricultural competence of field staff. In some systems, field staff have little formal technical training in the agricultural sciences. In some cases, this is dictated by a village worker philosophy, in others by local language demands. But, in most cases, it simply is the result of the decisions to expand agricultural extension programmes rapidly during the 1950s and 1960s, when few highly trained agriculturalists were available (see Bindlish & Evenson, 1993 and Bindlish, Gbetibouo, & Evenson, 1993 for African studies; and Swanson & Claar, 1984 for a general history).

Finally, this diversity of skills, management systems, and objectives has changed over time in many countries. Perhaps the major changes in the management and design of agricultural extension systems over the past four decades is associated with the training and visit (T&V) system introduced in the 1970s by Benor, Harrison, and Baxter (1984) and implemented in many countries with World Bank lending support.

Given this diversity, broad generalizations about the economic contribution of agricultural extension to agricultural development are not feasible. Many situation-specific factors impinge on the effectiveness of extension programmes. The fact that substantial reform and redesign of many extension programmes has taken place indicates that some of them were perceived by their supporters to have been less than fully effective. However, we now have a substantial body of economic studies of extension services in a number of countries; 75 studies of economic impacts of extension systems have been published to date. My task in this chapter is to review the findings of 57 of these studies and to draw out some of the lessons they have to offer.

I begin this review with a brief summary of investment patterns for both agricultural research and extension. This is designed to provide historical perspective and to call attention to some of the economic and institutional diversity in which extension systems must function. In the second section, I review the conceptual foundation for measuring the economic impact. Statistical procedures are reviewed in the third section, and in the fourth I summarize the findings of the studies under review and attempt to relate these to some of the differences in economic and institutional settings. In the final part, I summarize policy lessons.

Investment indicators: Agricultural research and extension

Several relevant indicators of investment in "technological infrastructure" are reported in Table 1. Agricultural extension, agricultural research, and human capital investments are included. The country groups are formed to reflect diversity in levels of "technology infrastructure." These country group categories will be used throughout this paper as a means of recognizing that extension programmes are conducted in different settings, and that their design, management, and effectiveness are conditioned by these settings. It is thus important to describe these country groups in some detail (for more detail, see Evenson & Westphal, 1994).

There are three Type 1 categories and three Type 2 categories for developing countries. The Type 1 categories cover countries which have not yet mastered production of a full range of goods and services using modem technology.

Type 1a includes approximately 20 countries (including Yemen, Laos, Surinam, Zaire) that lack basic infrastructure of all types. Governments have limited influence, and little manufacturing capacity exists in these countries.

Type 1b includes approximately 30 countries (including Nepal, Papua New Guinea, Haiti, Ethiopia, Burkina Faso) with rudimentary technological infrastructure. Some direct foreign investment has taken place, and this pro vides some access to foreign technology. Agricultural policy in these countries is often dominated by parastatal organizations.

Table 1. International Technology Investment Indicators: Investment Intensities (R&D/GDP, Extension/GDP, Expenditures/staff, 1990)

Indicators

Technological Infrastructure Type

Type 1 Developing Countries

Type 2 Developing Countries

Industrialized Countries

1a

1b

1c

2a

2b

2c

Traditional Technology

First Emergence

Islands of Modernization

Mastery of Conventional Technology

Transition to NIC-Hood

NIC-Hood

Recently

OECD

R&D/GDP

Agriculture


Public-NARs

.002

.004

.005

.006

.007

.010

.010

.015


Public-IARCs

.0005

.0005

.0005

.0005

.0005

0

0

0


Private

0

0

.0002

.002

.003

.005

.005

.015

Industry


Public

0

.0001

.0001

.0002

.0005

.001

.003

.003


Private

0

0

.0002

.005

.007

.010

.015

.023

Extension/GDP

Agriculture


Public

.005

.005

.010

.010

.010

.010

.010

.010


Private

0

0

0

0

.001

.002

.005

.010

Expenditures/staff (1980 000 dollars)


Research

47

47

47

40

45

50

70

95


Extension

2

2

4

4

10

15

35

35

Some higher education is provided. Most graduates are employed by government agencies, including agricultural extension services.

Type 1c includes 25 countries that have achieved partial modernization (including Sri Lanka, Tunisia, Kenya, Ivory Coast, Bangladesh). Modem agricultural practices have been introduced in most of these countries, and most have well-developed agricultural research and extension services. Most workers are literate. Universities have begun to train scientists and engineers. Graduates have begun to work outside the government. No significant private sector R&D capacity has yet been built in these countries.

The 20 Type 2 developing countries, on the other hand, have made sufficient investments in technology infrastructure to realize "followers" or "catch-up" growth. In other words, these countries are catching up to the developed countries.

The Type 2a countries (including India, Colombia, Mexico, Argentina, Turkey) have achieved modern engineering capabilities. Significant private sector R&D is undertaken. Universities are advanced. Most have not achieved the industrial, trade, technology, and macro-economic policy regimes to realize rapid growth.

The Type 2b countries (including Indonesia, Thailand, Malaysia, Chile, China) have achieved transition to newly industrialized country (NIC) status and the rapid growth associated with it. These countries have advanced technological capabilities and effective policy environments. Some of these countries (for example, Indonesia) have managed to move through the Type 2a stage in a relatively short period of time.1

The Type 2c countries (including Hong Kong, South Korea, Singapore, Taiwan) are well-established NICs.

Table 1 reports two sets of indicators of investment intensities (investment/GDP) and one set of "price" indicators for these types of countries.

Agricultural research indicators show that public sector research capacity exists in most countries; perhaps 10 or so of the Type la countries would not have such capacity. In contrast, private sector R&D relevant to agriculture or to industry is not important in Type 1 countries. Such R&D becomes increasingly important as countries become more advanced (that is, for Type 2a to Type 2c).

Research capacity for the industrial sector clearly lags behind research capacity for agriculture in all except the most advanced developing countries. Only the Type 2 countries have significant industrial R&D capacity.

Agricultural extension programmes, by contrast, serve almost all countries. For the Type la and 1b countries, these programmes represent the only modernizing investments of significance. Extension spending intensity exceeds research spending intensities until NIC-hood is reached.2

The expenditures-staff data show that the real costs of supporting a research scientist are roughly constant across developing country groups and are roughly half the level prevailing in industrialized countries. For extension, however, the real cost of supporting extension field staff is very low in the poorer countries, in fact too low for efficiency in many cases. This ratio explains why many poor countries have very large extension staffs and large staff-farmer ratios. Extension staff are perceived to be low-cost producers of economic growth relative to researchers, and to some degree they are.3

The conceptual foundation for extension impact

Two conceptual themes are relevant to extension impact. The first is the awareness-knowledge-adoption-productivity (AKAP) sequence. The second is the "growth gap" interrelationship between extension, schooling, and research

The AKAP Sequence

It is convenient to visualize extension as achieving its ultimate economic impact by providing information and educational or training services to induce the following sequence:

A: Farmer awareness
K: Farmer knowledge, through testing and experimenting
A: Farmer adoption of technology or practices
P: Changes in farmers' productivity

Changes in farmer behaviour will be reflected in quantities of goods produced, the quantities of inputs used, and in their prices. These, in turn, can be measured as "economic surplus," which is the added value of goods produced from a given set of inputs made possible by the extension activities.

Studies of extension impacts have measured farmer awareness (and sources of awareness), knowledge (and testing of practices), adoption, and productivity. Not all studies have examined all parts of the sequence. Most have shown a statistical relationship between the quantity of extension services made available to farmers and increases in awareness, knowledge, adoption, and productivity.

While the AKAP sequence has a natural ordering, it is clear that real resources in the form of skills and activities by both extension staff and farmers are required to move along the sequence. Awareness is not knowledge. Knowledge requires awareness, experience, observation, and the critical ability to evaluate data and evidence. Knowledge leads to adoption, but adoption is not productivity. Productivity depends not only on the adoption of technically efficient practices, but of allocatively efficient practices as well. Productivity also depends on the infrastructure of the community and on market institutions.

Extension services affect each part of the sequence. They can be seen as both substitutes for and complements to the acquired skills of their clientele farmers. Empirical evidence indicates that they are, on balance, net substitutes for farmers' skills as reflected in farmers' schooling. For example, extension services are typically not the only sources of information (awareness). Skilled farmers can seek information on their own. Farmers with few skills may not do so. Extension information then may have a higher impact on farmers with less schooling. It appears, however, that the awareness-knowledge part of the sequence is where extension services are strong substitutes for farmer schooling. Through organized frequent contact, they "teach" farmers, and this is more than simply informing farmers.4

The teaching versus informing distinction is also relevant to the "newness" of the information (that is, of the recommended practice or other technology) and of the nature of the practice or new technology. When technology is new (as for example with a recently released variety of rice) and is also "simple" to evaluate and adopt (where it is a matter of using new seed without altering other practices), information-awareness is relatively easily converted to knowledge and adoption. Farmers with few skills usually adopt such technology with a time lag. When the technological practice is more complex and requires substantial changes in activities and sometimes capital investment, teaching is required. Repeated messages clearly stated, followed up by field staff and often community organization, are required to proceed through the AKAP sequence in this case.

Productivity Gaps and Extension

The AKAP sequencing is, as noted above, related to the flow of new technical information and to the existing state of unadopted technology. We can see this interrelationship more clearly in the context of productivity "gaps." Figure 1 portrays what is meant by gaps and relates these to the technological infrastructure types used in Table 1.5

Figure 1 depicts crop yields, adjusted for fertilizer and other inputs, for five technology infrastructure types. Four yield levels are depicted for each type:

A: Actual yields
BP: Best practice yields
BPBI: Best practice, best infrastructure yields
BPBIRP: Best practice, best infrastructure, research potential yields

These yield levels in turn define three "gaps":

G(P): A practices gap between the best practice (BP) yield and actual (A) farmers' yields

G(I): An infrastructure-institutions gap between the best institutions, best practice (BPBI) yield and best practice (BP) yield

G(R): A research gap between the research potential yield (BPBIRP) and the best practice, best institutions (BPBI) yield

These gaps provide a way to classify the contribution of extension activities and to show how research and extension are linked. A stylized sequence across technology types is depicted. This could also be visualized as a time sequence.

Extension programmes are designed to reduce both the practice gap, G(P), and the institutions gap, G(I). Extension programmes are not the only activities that reduce these gaps. Providing market information to farmers and developing organized farm groups reduce G(I). Information and teaching reduce G(P). Research programmes are generally required to reduce G(R), although extension programmes can facilitate the reduction of G(R) via facilitating the importing and local modification of improved technology developed elsewhere. Research programmes in most developing countries also modify and adopt imported technologies and germplasm.

Two of the gaps are closely linked. When G(R) is closed (that is, when the BPBI yields go up), G(P) is opened.6 (This may happen with G(I) also, but to a lesser extent.) Further, it should be noted that the size of the gap is an index of the potential impact of research or extension. As extension succeeds in closing G(P), diminishing returns set in. Successful research opens up new potential by increasing G(P). The relative mix of teaching versus informing is also related to these gaps. When the BPBI yield level has been constant for some time, the G(P) gap is closed mostly by teaching. When BPBI is increased, as by "green revolution" rice and wheat varieties, information and testing advice play a larger role.

The pattern of gaps and yield levels across country groups is intended as a stylized pattern. It is roughly based on experience. For Type la countries, both G(I) and G(P) are depicted as large.7 These are traditional economies with not much new technology being produced that is relevant to them.

As economies move to Type 1b, improvements in institutions allow BP (and A) to rise even without new technology; BPBI remains unchanged. Extension can contribute to reducing both G(I) and G(P), and these contributions are qualitatively different from those required in more advanced country groups. There is little new technology (few new practices) in these countries, farmers have little schooling, and infrastructure is poor. The teaching and organizing activities of extension dominate here.

As economies move to the Type 1c category, some new technology has been introduced (BPBI has risen), and the institutions gap has been further reduced. The practices gap, G(P), has been both opened (because BPBI increased) and closed because of continued teaching and because new practices now can be extended.

Figure 1. Sequence of crop yields for five technology infrastructure types. Yield levels for each type areas follows: A = actual; BP = best practice; BPBI = best practice, best infrastructure; BPBIRP = best practice, best infrastructure, research potential. These yield levels define the following gaps: G(P) = practice gap between the best practice (BP) yield and actual (A) farmers' yields; G(I) infrastructure-institutions gap between the best institutions, best practice (BPBI) yield and best practice (BP) yield; G(R) = research gap between the research potential yield (BPBIRP) and the best practice, best institutions (BPBI) yield.

As economies move further up the technology infrastructure scale, the institutions gap is further reduced and extension's role in reducing G(I) becomes small. However, G(R) is closed by national and international research programmes - and the BPBI yields rise, providing extension with more new practices to extend. The private sector grows in importance and markets are improved.

A note on statistical methods and issues for economic evaluation

The studies under review in this chapter sought to measure the impact of public agricultural extension programmes' activities in the following four areas: (1) farmer knowledge of technology and farm practices; (2) adoption or use of technology and practices; (3) farmer productivity and efficiency; and (4) farm output supply and factor demand.

Estimation of extension impact is subject to a number of problems which are also faced in the evaluation of other public sector investments. The approach commonly used is a statistical analysis relying on data measuring extension activities at the farm level. Alternatively, statistical analysis can be undertaken where observations refer to aggregate extension services supplied to a given region in a specific time period.

Studies assessing extension impact at the individual farm level that use a farm-level measure of extension may be affected by two basic estimation problems. The first is the problem of statistical "endogeneity" in extension-farmer interactions. 8 Early studies seeking to measure the impact of agricultural extension by identifying the extension variable as some form of extension contact often treated the extension contact as being unrelated to the farmers' actions and characteristics. However, it is likely that one of the characteristics of more productive farmers is the desire to acquire information about changing farm conditions or new technologies. Such farmers may be inclined to attend more demonstration days, read more literature, and seek extension contact. Analogously, extension agents themselves may also seek contacts with better farmers who would be good performers even in the absence of extension contacts.

In such cases, the extension contact variable is endogenous, and the estimates of extension impact on farmers' performance are likely to be biased upward, because some of the better performance credited to extension would in fact be the result of the superior attributes of the group which interacts with extension. The problem of endogeneity can, in principle, be handled econometrically by using two-stage procedures or simultaneous equations approaches, but this has been done in only a few of the studies undertaken so far.

The second source of potential bias is the problem of indirect or secondary information flows where knowledge which originates from extension contacts is passed on to other farmers who do not directly interact with extension personnel. The extension of interfarm communications is substantial, as shown in Birkhaeuser, Evenson, and Feder (1991), where data on farmers' sources of information were reviewed. This review showed that most farmers in areas receiving extension services report that other farmers are the main source of information. Except for the contact farmers in T&V extension areas who are singled out for extension contact by the nature of the programme, direct contact with extension personnel was typically not the major source of information to farmers. Information may, of course, be diffused (to other farmers) from farmers who were informed by extension agents. In such cases, there may be little difference in performance between farmers interacting directly with extension and other farmers, and an estimate of extension impact based on individual extension contacts would erroneously indicate zero extension effect. Generally the presence of interfarmer communications tends to cause an understatement of extension effects when the approach of defining extension impact by the number of direct contacts is used.

In the studies reviewed below, data from farm surveys and secondary sources on farmer awareness, adaptation, and productivity were related to the provision of extension services in different regions and time periods.9 Productivity is typically measured as production per unit of all inputs (including labour, land, and fertilizer), although in some studies an aggregate production function approach was used.10

Estimated coefficients measure the marginal product of extension - the added production due to a one-unit addition to extension services supplied. The extension variables also typically have a time dimension. The adoption of improved practices will typically occur at some rate in the absence of extension services, depending on schooling and infrastructure. Extension both accelerates practice adoption and affects the long-run level of practice adoption. For Type la and 1b economies, extension may have a strong level effect if it is effective. For more advanced economies, the extension impact is primarily a speed-up effect. Most studies find speed-up periods of three to five years. Recent studies for Africa (Kenya and Burkina Faso) find significant level effects, implying that extension impacts in these economies are long-term impacts.

Knowledge of the timing weights and the marginal product allows the calculation of a marginal product "stream" over future years associated with an investment in time (t). This stream can be discounted to find the marginal internal rate of return (r) to the investment." This rate of return estimate is the measure used to compare extension studies in the next section. It can be interacted as the interest rate that investors (typically taxpayers) receive from investing in this activity.

Estimates of economic impacts: A summary

Table 2 summarizes estimates of economic impact from 57 economic studies undertaken in seven African countries (Burkina Faso, Cote d'Ivoire, Botswana, Nigeria, Ethiopia, Kenya, Malawi), seven Asian countries (Bangladesh, Indonesia, Malaysia, Nepal, Philippines, South Korea, Thailand), three Latin American countries (Brazil, Paraguay, Peru), and the United States and Japan. The studies are grouped into several categories: the distribution of the estimates by level of statistical significance and, where reported, by level of rate of return to extension. (Note that some studies reported more than one estimated impact. Many studies did not calculate returns, and returns are reported only when the estimated coefficient had a high level of statistical significance.)

Of the 174 estimated impact coefficients, 59 (one-third) are reported to be not significant. Very few of these were actually negative. This set of studies, however, cannot be said to be fully representative of the regions or types of extension programmes. Quite possibly a number of studies that found little or no extension impact were not reported.

Awareness (Knowledge) Studies

Six studies of extension's impact on awareness and knowledge were undertaken. Three of these (India, Kenya, Burkina Faso) (see Table 3) examined the impact of T&V management on awareness of recommended practices. These studies find strong evidence that extension does create awareness and knowledge and that T&V management makes extension more effective in doing so.

Adoption Studies

Nine studies of adoption of farm practices were undertaken. All sought to determine the impact of extension in accelerating adoption. This evidence is somewhat less conclusive than the awareness evidence.

Most studies found that farm size and farmers' schooling also determined adoption rates. Most studies found evidence for some extension impact on adoption. The T&V studies found that T&V enhanced the extension effect. Two of the studies (Kenya and Burkina Faso) linked practice adoption to productivity. Both found that extension accelerated adoption and led to productivity change.

Productivity Studies

Forty-two studies reported estimates of extension impacts on farm productivity: 25 used farm survey data, and 17 used aggregated data, such as district-level data. Sixteen of the 25 farm survey studies used a farm-specific extension variable, usually a contact with extension. As noted earlier, these variables are highly vulnerable to the endogeneity problem. It is interesting to note that this category of studies actually had the highest proportion of insignificant estimates.

In contrast, the nine studies that relied on an extension supply variable such as the number of extension staff made available in a region or to a group of farms have a high proportion of highly significant estimates. The T&V studies were in this category, and they generally found a T&V management enhancement effect. Two of these studies used two-stage procedures to predict adoption or membership in T&V groups and found that the extension impact was in general realized via its effect on practice adoption and on T&V group participation.

The 17 studies based on aggregate data in most cases included variables measuring research, schooling, and infrastructure in addition to extension variables. Almost all found evidence for an extension impact. The studies that used interaction variables between extension and farmers' schooling generally found a net substitution relationship. Higher levels of farmer schooling reduced the impact of extension, and vice versa. The studies that examined the research-extension interaction generally did not find a significant interaction except in the U.S. studies.

Estimates by Period and by Country Group

There are no differences in the distribution of significant estimates or rates of return by period.

In an earlier section of this paper, I noted that the technological and institutional setting in which extension operates affects its design and impact. The estimates classified by country group show two things. First, they show considerable variation, with a substantial range of significance being reported. Second, they show a difference between Type 1 and Type 2 developing economies. For the Type 1 economies, 45 of 105 (43 percent) of reported estimates have a high level of statistical significance. For Type 2 countries, 38 of 56 (68 percent) of the estimates have a high level of statistical significance. Thus it appears, on the one hand, that it is possible to design effective extension programmes for the Type la and 1b economies where new technology is being developed at a slow rate. The T&V studies confirm this. But it also appears that the more dynamic technological environments of the Type 2 economies provide a setting for broader effectiveness of extension programmes.

Lessons

Perhaps the overriding lesson is associated with the range of results reported. Clearly, many extension programmes have been highly effective in aiding farmers to achieve higher productivity. It also appears that some programmes have not done so. Probably many programmes are underperforming and many lack the design and management discipline to be effective. Many extension programmes have large investments in monitoring and evaluation units. None of these, to my knowledge, has yet produced estimates of economic impact. Virtually all of the studies under review were conducted by economists in academic programmes. It clearly is important to extension policy makers that the competence of their monitoring and evaluation programmes be upgraded to enable more effective evaluation. At the same time, it continues to be important that independent evaluations of the type reviewed here be continued.

A second lesson appears to be available from the evidence by country type. Extension systems are most effective when researchers are effective. But research programmes are not effective everywhere. The evidence of effective extension programmes in the Type la and 1b countries, which are concentrated in sub-Saharan Africa, shows that properly designed and managed extension programmes can be effective in these environments. The two studies of T&V-managed programmes in the region stress that they have been effective.

A final lesson is that, in the long run, extension has its highest payoff in Type 2 economies where farmers have access to schooling, to new technology, and to extension. As conditions change, extension must change. The need for continuing evaluation is high at all levels.

Notes

Table 2. Summary: 57 Economic Studies of Extension for Selected Countries'

Type or Category

Number of Studies

Distribution by Level of Statistical Significance

Distribution by Returns Estimates

Not Significant

Medium Significant

High Significant

Low

Medium

High

Awareness

6

7

2

27

nc

nc

nc

Adoption

9

16

8

17

nc

nc

nc

Productivity


Farm Observation:










Farm Contact

16

21

4

10

2

1

7



Extension Supply

9

11

3

21

1

1

12


Aggregate Observation

17

4

5

17

2

0

6


All Productivity

42

36

13

48

4

2

25

By Period


Before 1980

17

12

3

13

2

2

7


After 1980

40

47

20

79

2

2

18

By Country Group


1A-1B

9

16

6

24

1


9


1C

14

26

11

21

1

1

4


2A

12

5

3

28

2

2

3


2B-2C

13

8

2

10

1


4

Industrialized

9

4

1

9



5

Note: For statistical significance, the estimated "t" ratio is less then 1.5 for not significant, 1.5-2.0 for medium significance, greater then 2 for high significance. For rates of returns, low is 5-25 per cent, medium 26-50 per cent, high 50 per cent or greater.

* African countries: Burkina Faso, Cote D'Ivoire, Botswana, Nigeria, Ethiopia, Kenya, alawi; Asian countries: Bangladesh, Indonesia, Malaysia, Nepal, Philippines, South Korea, Thailand; Latin American countries: Brazil, Paraguay, Peru; the United States; Japan.

Table 3. Source of Awareness of Extension Recommended Practices

Study

Study of (Per cent)

Extension

Research

Private

Stuff

Centres

Sector

Taiwan (Lionberger & Chang 1970)

Shangfung

36

na

na

Liupao

24

na

na

Paraguay (Evenson 1988)

Western Regions

21

1

10

India (Federal et al. 1986)

T&V farmers

71

0

0

T&V group farmers

27

6

4

Non-T&V farmers

24

3

5

Burkina Paso (Bindlish et al. 1993)

T&V group members

74

na

5

Non-T&V farmers

36

na

12

1. A number of countries have moved from one country group to another in recent years. All of today's Type 2b countries were in the 2a or 1c categories twenty years ago

2. Agricultural extension programme development was also undertaken earlier than research investment in most countries (Judd, Boyce, & Evenson, 1986).

3. The perception that extension staff services were bargain-priced probably led to overstaffing of many extension programmes.

4. The recognition of the teaching component of extension has been growing in recent years, especially in the T&V-managed systems.

5. See Evenson (1986) for an earlier version of this gap analysis and a discussion of the economics of extension.

6. This gap-opening phenomenon is the source of new potential gains from extension and explains why extension programmes are demanded over long periods of time. If the BPBI yields remain stagnant in advanced countries, extension has a very limited role to play.

7. Schultz (1964), in his influential book Transforming Traditional Agriculture, stated that, in traditional agricultural economies where BPBI yields had been constant, farmers were "poor but efficient." In terms of Figure 1, he was saying that G(P) was actually not very large and that the potential for yield improvement from extension was also low. He effectively said that a rise in BPBI yields was required to create potential for extension to be effective.

8. The term endogeneity is a statistical term. Endogenous variables are chosen or controlled by he units being studied (for example, farmers). Exogenous variables are not chosen by the units. Exogenous variables can "cause" endogenous variables. Endogenous variables cannot be said to cause other endogenous variables.

9. The basic statistical model used in these studies

In this expression, Z may alternatively be a measure of awareness, knowledge, testing activity, adoption, or farm productivity. Extension (EXT), schooling (SCH), research (RES), and other variables are the independent (exogenous) variables that determine the endogenous variable, Z. Interaction variables are often included to test for substitution-complementary relationships.

10. In the production function studies, inputs are included as independent variables:

, etc.

When a production function formulation is used, the interpretation of the coefficient b' measures the change in output, holding inputs (land, fertilizer, etc.) constant.

11. This internal rate of return is the rate r for which the following equation holds:

References

Benor, D., Harrison, J. Q., & Baxter, M. (1984). Agricultural extension: The training and visit system. Washington, DC: World Bank.

Bindlish, V., & Evenson, R. E. (1993). Evaluation of the performance of T&V extension in Kenya. World Bank Agricultural and Rural Development Series No. 7. Washington, DC: World Bank.

Bindlish, V., Gbetibouo, M., & Evenson, R. (1993). Evaluation of T&V extension in Burkina Faso. World Bank Technical Paper No. 226. Washington DC: African Technical Department.

Birkhaeuser, D., Evenson, R. E., & Feder, G. (1991). The economic impact of agricultural extension: A review. Economic Development and Cultural Change, 39, 607-650.

Evenson, R. E. (1986). The economics of extension. In G. Jones (Ed.), Investing in rural extension: Strategies and goals (p. 65-91). Amsterdam: Elsevier Applied Sciences Publishers.

Evenson, R. E. (1988). Estimated economic consequences of PIDAP I and PIDAP II programs for crop production. Unpublished manuscript. New Haven, CT, Yale University, Economic Growth Center.

Evenson, R. E., & Westphal, L. E. (1994). Technological change and technology strategy. UNU/INTECH Working Paper No. 12.

Feder, G., & Slade, R. (1986). The impact of agricultural extension: The training and visit system in India. The World Bank Research Observer, 1, 139-161.

Feder, G., Slade, R., & Sundaram, A. (1986). The training and visit extension system: An analysis of operations and effects. Agricultural Administration, 21, 48.

Judd, A., Boyce, J., & Evenson, R. E. (1986). Investing in agricultural supply: The determinants of agricultural research and extension investment. Economic Development and Cultural Change, 35, 77-113.

Lionberger, H., & Chang, H. C. (1970). Farm information/or modernizing agriculture: The Taiwan system (p. 282-283). New York: Praeger.

Schultz, T. W. (1964). Transforming traditional agriculture. New Haven, CT: Yale University Press.

Swanson, B. E., & Claar, J. B. (1984). The history and development of agricultural extension. In Agricultural extension: A reference manual. Washington, DC: FAO.


Previous Page Top of Page Next Page