The impact of Agricultural Policies on Agricultural Output in Sub-Saharan Africa (A study of Eight Sub-Saharan African Countries)

Master Thesis

The impact of Agricultural Policies on Agricultural Output in Sub-Saharan Africa (A study of Eight Sub-Saharan African Countries)

Name : Joshua M. Kimulu Student Number:s1822721 Rijksuniversiteit Groningen Faculty of Economics and Business

Submitted to:

1

^st

Supervisor: Dr. D. J. Bezemer

2

^nd

Supervisor: Dr. D.H.M. Akkermans

2 Abstract

This thesis investigates what sort of agricultural policies are good for agricultural output in Sub- Saharan African countries (SSA). Literature review from the available agricultural policies with data suggests that agricultural research and development and infrastructure have a positive and significant effect on agricultural output. Using a panel data for a sample of eight countries for the years 1981 through 2000, I empirically test what sort of agricultural policies are good for agricultural productivity is SSA. The estimation results were in contrary to our hypothesis that these policies have significant impact on agricultural output, even after lagging agricultural research and development to take into account that its benefits accrue with time.

Key words: Agricultural policies, Agricultural output, Agricultural Research and Development,

Infrastructure

3 Table of Contents

ABSTRACT ... 2

1. INTRODUCTION ... 5

1.1 R

ESEARCH

Q

UESTION

... 6

1.2 J

USTIFICATION FOR THE STUDY

... 6

1.3 O

VERVIEW OF

A

GRICULTURAL

P

OLICIES

S

UB

-S

AHARAN

-A

FRICA

... 7

2. THEORETICAL BACKGROUND ... 9

2.1 N

EW GROWTH THEORY

... 9

3. LITERATURE REVIEW ... 11

3.1 A

GRICULTURAL

R

ESEARCH AND

D

EVELOPMENT

... 11

3.1.1 Agricultural Research and Development in Kenya. ... 12

3.1.2 Agricultural Research and Development in Uganda ... 13

3.1.3 Agricultural Research and Development in Tanzania ... 14

3.1.4 Agricultural Research and Development in Burundi ... 15

3.1.5 Agricultural Research and Development in Ghana ... 16

3.1.6 Agricultural Research and Development in Botswana ... 17

3.1.7 Agricultural Research and Development in Ethiopia ... 18

3.2 I

NFRASTRACTURE

P

OLICY

... 20

3.2.1 Infrastracture in Kenya ... 20

3.2.2 Infrastracture in Uganda ... 21

3.2.3 Infrastracture in Tanzania ... 21

3.2.4 Infrastracture in Burundi ... 22

3.2.5 Infrastracture in Ghana ... 22

3.2.6 Infrastracture in Botswana ... 23

3.2.7 Infrastracture in Ethiopia ... 23

4. PAST EMPIRICAL RESEARCH... 25

4.1 A

LENE AND

C

AULIBALY

(2009): T

HE

I

MPACT OF

A

GRICULTURAL

R

ESEARCH ON

P

RODUCTIVITY AND

P

OVERTY IN

S

UB

-S

AHARAN

A

FRICA

. ... 25

4.2 T

ERUEL AND

K

URODA

(2005): P

UBLIC

I

NFRASTRACTURE AND

P

RODUCTIVIT

G

ROWTH IN

P

HILIPPINE

A

GRICULTURE

, 1974-2000 ... 25

4.3 A

NTLE

(1983): I

NFRASTRACTURE AND

A

GGREGATE

A

GRICULTURAL

P

RODUCTIVITY

: I

NTERNATIONAL

E

VIDENCE

... 26

4.4 H

U AND

A

NTE

(1993).

A

GRICULTURAL

P

OLICY AND

P

RODUCTIVITY

: I

NTERNATIONAL

E

VIDENCE

... 26

4

5. METHODOLOGY ... 27

5.1 E

CONOMIC MODEL

... 27

5.2 D

ATA

... 28

5.2.1 Dependent Variable ... 28

5.2.2 Independent variables ... 28

5.2.3 Control Variables ... 29

5.3 D

IAGNOSTIC CHECKS

... 30

5.3.1 Outliers ... 30

5.3.2 Mean and Standard Deviation ... 30

5.3.3 Multicollinearity. ... 31

5.3.4 Hausman test ... 32

5.3.5 Autocorrelation ... 32

5.3.6 Heteroskedasticity test ... 33

6. ANALYSIS ... 34

6.1 R

EGRESSION RESULTS

... 34

7. DISCUSSION, CONCLUSION AND LIMITATIONS ... 36

7.1 D

ISCUSSION

... 36

7.2 C

ONCLUSION

... 37

APPENDIXES ... 38

REFERENCE: ... 44

5 1. Introduction

Agricultural productivity is widely recognized as a major determinant of economic growth in Sub-Saharan Africa (SSA). The sector accounts for about 35 percent of the continents GDP, 40 percent of its exports and about 70 percent of employment (Townsend, 1999). These countries are characterized by low levels of farm productivity, limited growth of non-farm employment and high rates of population growth. Improving agricultural productivity is crucial to economic development in many developing countries as lower food prices can directly raise living standards especially those of the poor people who spend a large share of their income on food, and indirectly keep real wage costs low in the industrial sector, thereby fostering investment and economic transformation.

Hayami (1970) has argued that productivity growth in agricultural sector is essential if agricultural output is to grow at a sufficiently rapid rate to meet the demands for food and raw materials that typically accompany urbanization and industrialization. Indeed, the economic growth and development of these countries is closely tied to agriculture, but the striking feature is that many countries in order to promote economic growth, neglected agriculture and decided to promote industrialization. However, Madhin and Bruce (1999) doubt on the path taken by these countries and argue that lessons emerging from the East Asia are that transformation to industrialization cannot be achieved without substantial increases in agricultural productivity.

It is of general consensus that no country has been able to sustain a rapid transition out of

poverty without raising productivity in its agricultural sector. During the last four decades the

agricultural performance of Sub-Saharan African countries has continuously declined, from

1961-64 to 1995-98 per capita agricultural production has fallen by 13 percent in Africa, where

as in Asia it has grown by 69% (Binswanger, 2006). The number of people living with income of

less than a dollar between 1981 and 2001 in the East and South Asia declined from 1234 million

to 702 million, and at the same time the number of people living with less than a dollar in Sub-

Saharan Africa rose from 164 million to 316 million. Binswanger (2006) argues that much of

these enormous differences in poverty outcomes were the consequences of sharp differences in

the performance of agriculture in the different regions. The question is why agricultural

development is still lagging behind in sub-Saharan Africa? Much of the blame for the poor

6

performance of the sector is attributed to the policies used, particularly government controls on agricultural production (Nyangito & Okello, 1998).

Binswanger (2006) argues that the under-performance in the agricultural productivity is not caused by lack of knowledge, but usually by deliberate policies and institutions which tend to reduce agricultural profitability and dis-empower rural people especially the poor women. The turnaround from low to high growth in agriculture and economic development for most of the Sub-Saharan African countries was seen to lie in reforming the policies under the structural adjustment programmes (SAPs). The SAPs promoted by the World Bank and the International Monetary Fund (IMF) advocated for the reduction of governments intervention in the economy and the liberalization of economies where market forces and the private sector was to play the dominant role. Nyangito & Okello (1998) noted that the transition from government controlled policies to liberalized markets has been in operation since 1980s, but the impacts of the policies on agricultural production are not clearly understood. However, in contrast, Jaeger (1992) has argued that Sub-Saharan African countries that adopted and sustained these policies to raise farm incentives have had better agricultural performance in the 1980s, on average, than the countries whose policies continued to discriminate agriculture.

1.1 Research Question

The aim of this research is to investigate what sort of agricultural policies are good for growth in agricultural output in Sub-Saharan Africa.

Therefore, the research question to be studied in this paper is:

What sort of agricultural policies are good for growth in agricultural output in Sub-Saharan Africa?

1.2 Justification for the study

It is interesting to research on this question because of the poor performance of agricultural

performance in Sub-Saharan African countries, despite the fact that governments have put in

place various agricultural policies which should spur agricultural productivity. As mentioned

earlier, agriculture accounts for the bulk of GDP in these region, and it is the only region with

7

agriculture growing at a rate below overall population growth from 1965 to 1998, and at a lower rate than growth in the agricultural labor force from 1980 to 1998 (Kydd, 2004). This has led most of these countries to depend on food aid (Netherlands Development Cooperation, 1991). It has been argued that African governments choose agricultural policies on the basis of political rather than economic rationality, for example they choose to depress agricultural prices and invest in numerous inefficient rural development projects because, although such policies stifle agricultural growth and economic development, they yield political gains (Berry, 1993).

Anderson and Masters (2007) argue that 60 percent of SSA workforce is still employed in agriculture and more than 80 percent of the regions poorest households depends directly or indirectly on farming for their livelihoods, and agriculture and trade policies remain the key influences on the pace and direction of change in Africa. Given the importance and the critical role of the agricultural sector in the economies of Sub-Saharan African countries, with agriculture accounting for 12 percent of the world’s farmers, 16 percent of agricultural land and 28 percent of those living on less than US$1 a day in 2006 (Anderson and Masters, 2009), there is need to investigate what sort of agricultural policies are good for growth of agricultural output in order to facilitate the policy makers to adopt more growth-enhancing and poverty reducing agricultural policies. This can be the ultimate solution if agricultural output is to improve at a rate equal to or greater than the population growth rate to meet the demand for food and raw materials like the East Asian countries.

1.3 Overview of Agricultural Policies Sub-Saharan-Africa

In many Sub-Saharan African countries, inappropriate agricultural policies and taxation of agricultural exports has been widely considered as contributed to the economic problems that has led to the stagnation of Sub-Saharan African countries. Due to this reason, agricultural policy reforms have been a major part of the structural adjustment efforts (Sahn et al, 1996). Ellis (1992) has analyzed the agricultural policies in developing countries of which the Sub-Saharan African countries belongs to include the price policy, marketing policy, input policy, credit policy, mechanization policy, land reform policy, research policy and infrastructure policy.

These policies vary across different countries, at different levels of development and with

different kinds of economic problems. These policies has been attributed to the worsening

8

performance of the agricultural sector in Sub-Saharan Africa, and the turnaround from the poor

to good performance of the agricultural and economic development of these countries was seen

to lie in reforming these policies under the structural adjustment programmes (SAPs) initiated by

the World Bank (WB) and the International Monetary Fund (IMF). Bates (1981) has argued that

there is a political rationale for seemingly inconsistent agricultural policies in many developing

countries and politicians often pursue policies detrimental to overall agricultural growth not out

of misunderstanding of economics but because such policies benefit key groups of rural and

urban constituents.

9 2. Theoretical Background

It is necessary to examine the existing literature with respect to the link between agricultural output and agricultural policies in the process of economic growth. Theories of policy formation imply that agricultural policy is an endogenous variable in the process of economic growth, and therefore agricultural policy variables are assumed to be endogenous to agricultural production (Hu and Ante, 1993).

2.1 New growth theory

In growth theory, the most basic proposition is that in order to sustain a positive growth rate of output per capita in the long run, there must be continual advances in technological knowledge in the form of new goods, new markets and new processes. This proposition is demonstrated using neoclassical growth model which shows that if there were no technological progress, then the effects of diminishing returns would eventually cause economic to cease. The basic building block of the neoclassical model is an aggregate production function exhibiting constant returns in labor and reproducible capita. The function expresses how much output can be produced, given the aggregate capital of stock, under a given state of knowledge, with a given range of available techniques, and a given array of different capital and intermediate inputs (Aghion and Howitt, 1998). The natural starting point for growth analysis is to specify an aggregate production function which describes how domestic output flow is generated from a stock of production factors. Thus, the production function is written as:

Y=AF (K, L)

Where K is the capital stock, L is the stock of labor, and A is a productivity factor which reflects

the existing stock of knowledge and the resulting efficiency of capital and labor in producing

final output. Aghion and Darlauf (2007) argue that growth in output results from the

accumulation of production factors K and L, and from the increases in productivity factor A. The

contribution of new growth theory is to explain productivity growth, that is, the growth in A as

resulting from innovations. Accordingly, the Cobb-Douglas aggregate production function is

written as:

10

Y = A K

^β1

L

^β2

. The new growth theory introduces research and development sector and then models production of new technologies (Romer, 2006). It argues that devoting more resources to research yields more discoveries and this is what the production function captures. The theory argues that both R&D and goods production is assumed to be generalized Cobb-Douglas function; that is, they are power functions. In Cobb-Douglas production function the relationship between the output and inputs is nonlinear, however when log-transformed we obtain the following econometric model:

lnY = lnA + β

1

lnK + β

2

lnL + ε

The aim of this research is to use the concept of the aggregate production function under the new

growth theory to research the effect of agricultural policies on aggregate agricultural output. The

Hicks-neutral productivity level captures the agricultural research and development of each

country’s agricultural sector in accordance with the new growth theory.

11 3. Literature Review

3.1 Agricultural Research and Development

The majority of African countries depend on agriculture for their economic livelihood, and agricultural research and development (R&D) plays a vital role in building food security and economic stability in these regions yet government spending patterns provides evidence that many of the countries throughout Africa have shifted public investment priorities away from agricultural research. Agricultural research in Africa dates back to 1900 with much of the work conducted at the botanical gardens throughout the region, and with political independence in the late 1950s and early 1960s most African countries inherited agricultural research structures that operated as a regionalized system. Many inherited agricultural research infrastructures under- went repeated reshuffling, mergers, relocations and sub-divisions, but this institutional reorganization were driven by political motivation rather than the effort to improve agricultural research performance (Beintima and Strads, 2006).

Investments in public agricultural R&D throughout Africa increased in the 1960s and 1970s, and the trend reversed in the 1980s with the growing levels of international indebtedness and programmes of structural adjustment that curtailed public spending in general (Pardey et al.

1995). The annual growth rate in public agricultural R&D spending in Africa declined from 2 percent in the 1970s to 1.3percent in the 1980s and to only 0.8 percent in the 1990s (Pardey et al, 1997). He further states that in 2000 Africa invested in agricultural research $0.70 for every $100 of agricultural output, lower than the $0.84 in 1981. Furthermore, according to WDR (2008) public agricultural R&D spending in Sub-Saharan fell in nearly half the 27 countries with data, and the share of agricultural GDP invested in R&D fell on average for the whole region.

Improvement in agricultural R&D has been found to be a prerequisite for increasing agricultural

productivity and generating income for farmers and the rural workforce in Africa, where 75

percent of the population live in the rural areas. Increasing agricultural productivity and food

security requires improved technologies and their broad dissemination, and these can occur

through agricultural R&D (Beintema and Stads, 2004). In support of these, WDR (2008) finds

12

that agricultural productivity improvements have been closely linked to investments in agricultural R&D and the returns have been high including Sub-Saharan Africa.

Lio and Liu (2009) have argued that farmers in developing countries cannot achieve high levels of agricultural productivity because of lack of modern agricultural technologies, and therefore in order to improve the agricultural performance more effort should be made to enhance the capacity of agricultural research institutions. In line with these, Eicher and Staatz (1998) concurs that existing research facilities in Africa are not employed at full capacity because they are staffed with research workers with limited scientific and technical training because of inadequate financial, logistical and administrative support. When research is poorly funded, agricultural technologies cannot be improved, and there will be no downstream farm income increase, rural employment generation, reduction in food prices, establishment of agro-based industries, and economic growth. Alene and Coulibaly (2009) using a three-stage least estimates to estimate the impact of agricultural research on productivity and poverty in SSA over the period 1980-2003 found that agricultural research contributes significantly to productivity in SSA. They found that doubling agricultural research expenditure per hectare of agricultural land would lead to a 38 percent increase in agricultural productivity, with the implied absolute effects being even larger.

The research also demonstrated that agricultural research generates benefits in excess of expenditure with impressive aggregate rate of return of 55 percent.

3.1.1 Agricultural Research and Development in Kenya.

In Kenya, agriculture accounts for about 30 percent of GDP, 75 percent of the total employment and 70 percent of the exports that earn foreign exchange. Organized agricultural research and development in Kenya dates back 1903 when the department of agriculture of the colonial government established the first of its experimental station (Mussukuya, 1988). In 1963 when Kenya gained independence from the British there were 23 research centers employing 150 researchers all spread over the main ecological zones of the country. The government reorganized all agricultural R&D into a number of semi-autonomous parastatal institutes and by 1986 there were 43 national agricultural research stations and sub-stations (Mussukuya, 1988;

Lipton, 1988).

13

During the period 1971-2000, public agricultural R&D spending grew annually by 4 percent, though most of this growth occurred during the 1980s (Beintema et al, 2003). As presented in Figure 1 below, government spending on agricultural R&D increased during the 1980s and there was a decrease in the late1990s. This is attributed to the introduction of the Structural Adjustment Progrmmes (SAPs) introduced by the World Bank, when funding became increasingly scarce, irregular and donor dependent (Beintema and Stads, 2004). Nyoro and Jayne (1999) argue that agricultural R&D increase agricultural productivity, especially in Kenya where land density has forced smallholder farmers to transfer inappropriate technologies into new environments

Figure 1:

Source: ASTI database.

3.1.2 Agricultural Research and Development in Uganda

The responsibility for agricultural R&D in Uganda until World War II was mainly with the local

colonial government which created several regional agricultural research organizations in East

Africa which complemented or partially replaced the existing research institutions. Following the

years after independence in 1962, all the national research agencies were transferred to the

national government and there were no major organizational changes until the 1980s (Beintema

and Tizikara, 2002). Agricultural R&D was mostly focused on export commodities, although

gradually it broadened to include food crop research. Agricultural research was severely affected

14

during the late 1970s due to civil war, but the government strived to revamp agricultural R&D during the 1980s as part of the efforts to rebuild the economy. Government spending on agricultural R&D for the period 1995-2000 as depicted in Figure 2 below shows that government spending on agricultural R&D increased during the mid 1990s and declined in the late 1990s.

The trend however reversed towards 2000 and this was as a result of World Bank funding to the National Agricultural Research Organization (NARO) through the first Agricultural Research and Training Project (Beintema and Tizikara, 2002).

Fan et al. (2004) using district data for 1992, 1995 and 1999 estimated the effects of different types of government expenditure on agricultural growth in Uganda. They found that government spending on agricultural R&D improved agricultural production substantially and it had the largest measured returns to growth in agricultural production. However, as a percentage of GDP, agricultural growth in Uganda remained relatively stable between 0.35-0.5 percent in the past two decades, which is low compared to the average for other African countries which was about 0.85 percent in the late 1990s and extremely low when compared to most Asian and Latin American countries which averaged about 1 percent (Fan et al., 2004).

Figure 2:

Source: ASTI database.

3.1.3 Agricultural Research and Development in Tanzania

Agricultural research in Tanzania was initiated in the late nineteenth century by the German

colonial powers, but in the 1920s under the British rule, agricultural R&D was virtually

abandoned but re-established a decade later as part of the department of agriculture and

15

veterinary services. Until World War II, agricultural research was the domain of the colonial government, which created several research organizations in the East Africa that complemented or replaced the existing facilities. Tanzania inherited a relatively well research infrastructure at independence in 1961 which depended heavily on British researchers and favored research on export commodities (Beintema et al. 2003).

Following independence, investment in agricultural R&D rapidly developed resulting to the establishment of several research stations which existed until the collapse of the East African Community in 1977. Government spending on agricultural R&D for the period 1996-2000 as shown in figure 3 below show that spending on R&D has been erratic. The expenditure increased from 1997 to 1998, and declined in 1999, with positive improvements in 2000 which was attributed to loans for agricultural R&D from World Bank (Beintema et al. 2003). Agricultural R&D expenditure in Tanzania doubled between 1996 and 2000 in constant dollars, and as a percentage of AgGDP it increased from a low of 0.2 percent in 1996 to 0.4 percent in 2000. Fan et al. (2005) found that investment in agricultural R&D in Tanzania have a large impact on rural poverty, raising about forty persons out of poverty per million shillings spent, and it had the largest impact on incomes with an average benefit/cost ratio of about 12.

Figure 3:

Source: ASTI database.

3.1.4 Agricultural Research and Development in Burundi

Agricultural research in Burundi dates to the early days of Belgian colonization which began

soon after World War II, with research stations being build throughout the Belgian colonies.

16

Research was undertaken on export crops but contrary to the pattern that prevailed in other parts of Africa attention was also given to research on food crops (Magalhaes et al., 2003). An extensive network of 36 research stations was build across the Belgian colonies. Following independence in 1962 from Belgium, the Institute of Agronomic Sciences of Burundi (ISABU) was created to provide agricultural R&D in Burundi. As figure 4 shows government spending on agricultural R&D declined in 1993 to very low levels with the onset of the civil war, but since 1997 government funding has recovered slightly with donor funding reappearing but total levels remain far below those before the outbreak of the civil war (Magalhaes et al., 2003). Research by Alene and Coulibaly (2009) in their estimation of the impact of agricultural research to agricultural productivity found that it would generate a 28 percent rate of return in Burundi.

Figure 4:

Source: ASTI database.

3.1.5 Agricultural Research and Development in Ghana

Agricultural Research and Development in Ghana dates back in the 1890 when Government

Botanical Gardens were established to carry out research primarily on oil palm, cocoa and rubber

(Stads and Gogo, 2004). Various agricultural research stations were established between 1900

and 1910, which formed the basis for the department of agriculture. In addition several regional

research stations were established throughout the British West Africa in the late 1940s and 1950s

(Roseboom and Pardey, 1994). With independence in 1957 from Britain the West African Cocoa

17

Research Institute was nationalized and Ghana Council for Scientific and Industrial Research established to coordinate scientific research in Ghana.

The CSIR oversees 13 research agencies out of which 9 focuses on agricultural research and development. Initial agricultural research in Ghana was focused on Cocoa but over the years it has broadened to include other crops ((Stads and Gogo, 2004). Agricultural research has remained unchanged since 1970s, although legislation passed in 1996 heralded a shift towards commercialization of agricultural research (Roseboom and Pardey, 1994). Government agricultural expenditure on R&D as shown in figure 6 below increased in the 1980s with stagnation in the 1990s

Fig. 6

Source: ASTI database.

3.1.6 Agricultural Research and Development in Botswana

Agricultural R&D in Botswana dates back 1930s with the establishment of experimental stations

that focused on variety of crops and livestock. Until the 1960s agricultural research was

undertaken by a division within the Department of Agricultural Extension Services (Beintema et

al. 2004). Agricultural research grew considerably following independence and several research

stations were created under the Ministry of Agriculture. At independence research focused on

18

crop production, but in the 1970s research broadened to include livestock research which encompassed 17 research stations (Roseboom and Pardey, 1993). Agricultural research expanded during the 1970s and 1980s due to large donor-funded research projects which included FAO fertilizer project and Dry Land Farm Research Scheme. Currently, the Department of Agricultural Research (DAR) is the principal agricultural research organization whose research mandate includes crops, livestock, rangeland and natural resources research (Roseboom and Pardey, 1993). Botswana’s agricultural research expenditure increased in the 1990s, as depicted in figure 7 below largely as a result of increased government spending (Beintema et al. 2004).

Fig. 7

Source: ASTI database.

3.1.7 Agricultural Research and Development in Ethiopia

Agricultural research in Ethiopia compared to other African countries started relatively late with

limited research established in 1947. In 1953 the Zeit Agricultural Research Centre (DZARC)

was established and it remained the Ethiopian Agricultural research facility until 1966 when the

Institute of Agricultural Research (IAR) was established as a semi-autonomous Institute

(Beintema and Solomon, 2003). The institute subsumed the limited and scattered research

19

activities of the Ministry of Agriculture and conducted research on crops, livestock and natural resources. In 1977 the IAR research programmes were restructured to emphasize regional research with a centre identified for each regional zone (Roseboom and Pardey, 1994). A number of federal research centers were established in the 1970s, which underwent significant reforms in the 1990s. There were 41 agencies engaged in agricultural research in the late 1990s with the Ethiopian Agricultural Research Organization (EARO) accounting for two-thirds of the agricultural spending. Since the early 1990s, total investments in agricultural research have almost doubled as a result of increased government contributions as shown in figure 7 below.

Figure 7

Source: ASTI database.

The economic theory and the literature outlined encourage the assumption that the impact to be studied is of a positive nature i.e. the higher the agricultural R&D the higher the agricultural output. Consequently, we therefore test the following hypothesis:

Hypothesis 1: Agricultural Research and Development have a positive and significant impact on

agricultural output in Sub-Saharan African countries.

20 3.2 Infrastructure Policy

Productivity increase in agriculture is an effective driver of economic growth and poverty reduction, and such productivity increase depends on good infrastructure, well functioning domestic markets, appropriate institutions, and access to appropriate technology (Andersen and Shimokawa, 2006). Fan and Zhang (2004) in support of these argues that investment in infrastructure is essential to increase farmers’ access to input and output markets, to stimulate the rural non-farm economy and vitalize rural towns, to increase consumer demand in rural areas.

Akyuz (2001) argue that in Africa the major policy failure was not that of taxation of agriculture heavily, but rather that they did not invest in the rural sector to increase productivity.

Large amounts of investment were needed in infrastructures to ensure sustainable productivity growth and raise agricultural surplus, but there was very little public investment in these areas.

This is in contrary to the experiences of East Asian countries where green revolution which was achieved in the 1960s and 1970s was largely made possible through significant public investments in rural infrastructure. Quibria (2002) finds that even though like many other developing economies the Asian economies imposed both explicit and implicit taxes on agriculture, they also provided various public goods, such as rural roads and infrastructure.

Ashok and Balasubramanian (2006) emphasize that in Africa, rural road construction has been found to be associated with increases in agricultural production, especially in non-food export crops, expanded use of agricultural credit, increases in land value and expansion of rural markets.

3.2.1 Infrastructure in Kenya

Thurlow (2007) found that in Kenya apart from direct agricultural investments like irrigation and

research and extension (R&E), poor market access and inadequate infrastructure were binding

constraints to agricultural growth and rural development. Infrastructure reduces agricultural

transactions costs significantly, hence reducing poverty and encouraging growth beyond rural

areas. Thurlow (2007) states that expenditure on roads in Kenya has increased slightly, but while

it is impossible to isolate rural roads from the figure, it is reasonable to conclude that total

expenditure on agriculture and infrastructure has declined over the last decade. The authors used

CGE model to estimate the impact of investing in infrastructure and agricultural productivity in

21

Kenya. They found that each Ksh.1.0 spend on rural infrastructure during 2006-15 causes GDP to increase by Ksh.3.0. Nyoro and Jayne (1999) found that high transportation cost in Kenya contributed to high fertilizer prices, lower output prices for farmers and hence lower incentives to invest in productivity-enhancing technologies. Therefore, lowering Kenya’s high transport costs through improvements in rural infrastructure, especially roads, is not only important for improving access to input and output markets, but it is also found to indirectly enhance the productivity of nontraded crops. The road density in Kenya increased only slightly as figure 5 shows, which contributed to the poor performance of the agricultural sector among other factors.

3.2.2 Infrastructure in Uganda

Fan et al. (2004) finds that government spending on infrastructure in Uganda had substantial marginal impact on rural poverty reduction. Like many African countries, Uganda depends on agriculture as the main source of economic activity, with 95 percent of the poor concentrated in rural areas and with agriculture acting as their source of livelihood. Access to infrastructure has far reaching implications for development in general and agricultural production in particular, and Uganda is no exception to this phenomenon (Fan et al., 2004). However, the more recent Road Sector Development Programme has focused on rehabilitation, maintenance and selective upgrading of existing roads, emphasizing on the main paved and gravel roads. Fan et al., (2004) states that the programme has led average distance of households to tarred roads to fall from 32 km in 1997 to 22km in 1999/2000 and 60 percent of feeder roads were rehabilitated and improved. This facilitated the livelihood of rural communities through delivery of farm inputs, marketing of agricultural produce and other social and administrative services. Uganda has experienced increase in the density of roads as presented in figure 5 below, which is be attributed to governments’ commitments to improve rural infrastructure.

3.2.3 Infrastructure in Tanzania

Fan et al. (2005) investigated the public investment in roads among other items in Tanzania,

where agriculture contributes about 45% of GDP and employs 80% of the population. They

further state that the real agricultural growth achieved during the past decade has not been

sufficient to bring about sufficient reduction in the number of rural poor. Fan et al. (2005) finds

that public expenditure on roads and transport system has increased over the years, but the total

22

length of available roads remains low. Rural roads account for more than 60% of the total road length and less than one percent of rural roads are paved. The authors further found that by 2000, 38 percent of the truck roads remained unpaved and there remained large regional variation in access to road infrastructure. As presented in figure 5 below, the road density has showed a dismal trend during the period 1981-2000, and these may explain the poor performance of the agricultural sector.

3.2.4 Infrastructure in Burundi

One of the factors explaining the lack of dynamism associated with agricultural activities is the acute shortage of support in infrastructure (PRSP, 2006. P.41). Burundi is one of the world’s poorest countries, with per capita income at US$83 at the end of 2004, with agriculture accounting 94 percent of the working population and providing 95 percent of the food supply.

The country emerged from a decade of conflict, which decimated agriculture and infrastructure.

The infrastructural trend of the road density in Burundi as presented in figure 5 below shows a gloomy picture in a country which almost entirely depends on agriculture, with a very slight increase in the road density.

3.2.5 Infrastructure in Ghana

Road transport which forms the major means of distributing agricultural products within Ghana remains to be adequately developed. About half the road network is in good condition and about a third of the feeder road network is also in good condition (Japan Bank for International Cooperation, 2008). A significant proportion of roads in Ghana are not paved and most resources are channeled to building than into road maintenance. The quality of the road system has improved over the years, with 67 percent in good condition, 16 percent in fair condition and 17 percent in poor condition by the year 2008 as compared with 29 percent which was in good condition, 17 percent fair and 54 percent poor in the year 2003. However, transport cost remains a major challenge to doing business in Ghana, which led the Government of Ghana to seek donor support particularly in the 1980s and 1990s to support infrastructural development (JBIC, 2008).

The poor state of infrastructure in Ghana as in many African countries does not promote

effective and competitive agricultural production, trade and development. As presented in figure

23

5 below, there has been a slight increase in road density, which cannot adequately support the agricultural development.

3.2.6 Infrastructure in Botswana

One of the main impediments to higher growth in Botswana is lack of essential infrastructure including transport infrastructure and telecommunication. The total road network is estimated at around 24,455 km, of which primary and secondary roads account for 8,916 km. Of the latter, 6,116 km are paved, 1,501 km are graveled and the remaining 1,299 are dirt or sand roads (ADB 2005). Funding for road maintenance has been insufficient, and as a result the existing road network has been deteriorating. Similarly, as depicted in figure 5 below the road density has not shown any meaningful improvement to support agricultural production.

3.2.7 Infrastructure in Ethiopia

Ethiopia’s road network expanded from 24,970 km in 1997 to 36,496 km in 2004 for an average

annual growth of 6 percent. Road density has increased from 24.1 km per 1000 sq km in 1997 to

33.2 km per 1000 sq km in 2004. About 65 percent of the road network is in a good condition,

but the density and quality of the transport network remain inadequate to support economic

development, especially given the importance of agriculture (ADB, 2005). In rural area only 17

percent of the population lives within two km of an all weather road, and the road density as

showed in figure 5 below is very low which some authors argue is among the lowest in the

world.

24 Figure 8

Source: David Canning: A Database of World Infrastructure Stocks 1950-95

Therefore, in line with the literature, investments in infrastructure play a key role in the growth of agricultural productivity in SSA countries. Consequently, the following hypothesis is tested:

Hypothesis 2: Infrastructure policy has a positive and significant impact on agricultural output

in Sub-Saharan countries.

25 4. Past Empirical Research

4.1 Alene and Caulibaly (2009): The Impact of Agricultural Research on Productivity and Poverty in Sub-Saharan Africa.

Alene and Coulibaly (2009) investigated the impact of agricultural research on productivity in Sub-Sahara Africa for the period 1980-2003. Agricultural productivity was modelled as a function of lagged agricultural research expenditures and production factors including fertilizer, labour, machinery and irrigation. The authors lagged agricultural research expenditures because agricultural productivity equation takes into account the fact that agricultural research investments generate a flow of benefits over time. The authors found national and international research has a positive and significant effect on agricultural productivity, and the estimated total factor productivity elasticity with respect to agricultural research was 0.38. Doubling agricultural expenditures per hectare of agricultural land would lead to 38% increase in agricultural productivity. Alene and Coulibaly (2009) also found fertilizer, labour and agricultural machinery to be positively and significantly related to agricultural productivity whereas irrigation had insignificant effect on agricultural productivity.

4.2 Teruel and Kuroda (2005): Public Infrastructure and Productivity Growth in Philippine Agriculture, 1974-2000

Teruel and Kuroda (2005) investigated the impact of public infrastructure on the productivity

performance of Philippine agriculture for the period 1974-200 by applying translog cost-based

model. The authors used a simultaneous estimation procedure full information maximum

likelihood (FIML) and argued it is possible to estimate the translog cost function using the single

equation OLS, but use of FIML employs more information from the system of equations. The

variables used in the estimation of the translog cost model included quantities of output and land,

capital (agricultural machinery and animal labour) and intermediate input (fertilizer and seeds)

among others. Teruel and Kuroda (2005) found public infrastructure to be a substitute for labour

and intermediate inputs and it reduces production cost thereby enhancing agricultural

productivity.

26

4.3 Antle (1983): Infrastracture and Aggregate Agricultural Productivity: International Evidence

Antle (1983) investigated the effects of infrastructure on aggregate agricultural productivity on 47 LDCs and 19DCs. They used Cobb-Douglas production functional model. The variables used included land, labour, livestock, fertilizer, education, research and infrastructure. Antle (1983) found the coefficient on the infrastructure significant supporting the hypothesis that transport and communication infrastructure has a positive impact on aggregate agricultural productivity.

Education variable was found to be insignificant, although without the infrastructure variable it was significant. Further, they found that the infrastructure coefficient was larger than the research variable suggesting that the agricultural research explained less of the variation in agricultural production than the infrastructure variable.

4.4 Hu and Ante (1993). Agricultural Policy and Productivity: International Evidence

Hu and Ante (1993) investigated the impact of agricultural policies in 24 countries for the years

1960, 1970 and 1980 through Cobb-Douglas production function. They used Nominal Protection

Coefficient as a proxy for agricultural policy. They estimated the production model specifying

per farm output (Q) to be a function of labor, land, livestock, machinery and fertilizer and the

policy variable NPC. Their econometric results strongly supported the hypothesis that

agricultural policy has a positive and significant impact on agricultural productivity in both the

developed and less developed countries.

27 5. Methodology

5.1 Economic model

To test the hypothesis that agricultural policies of research and development and infrastructure have a positive and significant impact on agricultural output, a panel dataset consisting of eight Sub-Sahara African countries for the period 1981-2000 will be used by specifying the following model for countries i’s agricultural production function at time t as follows:

ln_agoutput

it

= β

0 +

β

1

ln_RandD

it

+ β

2

ln_infra

it

+ β

3

ln_labour

it

+ Β

4

ln_land

it

+β

5

ln_livestck

it

+ Β

6

ln_fert

it

+ β

7

ln_tractor

it

+ β

8

precp

it

+β

9

ln_temp

it

+ β

10

educ

it

+

+β

11

landlock2

_it

+ε (1) ln_agoutput

it

= β

0 +

β

1

ln_RandD

it

+ β

2

ln_infra

it

+ β

3

ln_labour

it

+ β

4

ln_fert

it

+β

5

ln_tractor

it

+ ε (2) ln_agoutput

_it=

=β

_{0 +}

β

1

ln_RandD

_it

+ β

₂

ln_infra

_it

+ β

3

lnlab_land

_it

+ β

4

lnfert_land

_it

+β

₅

ln_tractor

_it

+ ε (3) The rationale for including the variables in model (2) and (3) is based in terms of theoretical model according to the aggregate production function. The variable tractor represents capital (K) while the variable labour represents labour (L) in accordance with the Cobb-Douglas production function. Fertilizer variable represents intermediate inputs. The research variable captures the effect of investments in agricultural research on the Hicks-neutral productivity level of each country’s agricultural sector, and in accordance with the new growth theory. The infrastructure variable accounts for the difference across countries in transporting investments that increase agricultural productivity potential.

The aggregate agricultural production function takes the Cobb-Douglas form which has been

used by various authors in their specification of agricultural production (Hayami and Ruttan,

28

1970; Fulginiti and Perrin, 1992; Fulginiti and Perrin, 1993; Fulginiti and Perrin, 1999; Hu and Ante, 1993; Lio and Liu, 2008). The Cobb-Douglas production function is non-linear in its form and the variables have been put in their natural logarithm form to transform them from non- linear to linear form.

5.2 Data

¹

5.2.1 Dependent Variable

The dependent variable is the country’s total agricultural output (agoutput), measured by the agricultural value added in constant million 2000 US dollars. Using value added in the estimation of agricultural output is a common practice in the existing literature (Fulginiti and Perrin, 1993; Fulginiti and Perrin, 1999; Lio and Liu, 2008; Lio and Hu, 2009). The data on agricultural output was obtained from FAOSTAT 2009.

5.2.2 Independent variables

The main explanatory variables are agricultural research and development and infrastructure policies. Agricultural research and development is measured by expenditure in million 2005 US dollars. Infrastructure is measured by road density (km/1000 km square). The independent variables include five essential agricultural inputs of labor, land, livestock, machinery, and fertilizer which have been used by much of the literature on agricultural production. Labour is measured by thousands of participants in economically active population in agriculture while land is measured in thousands of hectares of arable land and permanent cropland (Alene and Coulibaly, 2009). Livestock is measured by thousands of cow equivalent as calculated by Hayami and Ruttan (1970)

²

. Machinery is measured by the number of agricultural tractors while fertilizer is measured by the sum of nitrogen, potash and phosphate content of various fertilizers consumed in metric tons. The data for labour was averaged for five years (FAOSTAT, 2009).

1

For more synthesis of data description and sources refer to appendix 1

2

Conversion factors for livestock equivalent units: 1.1 for camels;1.0 for buffalo, horses and mules;0.8 for

asses;0.2 for pigs;0.1 for sheep and goats;0.01 for poultry.

29 5.2.3 Control Variables

In addition to the agricultural inputs, control variables are used in the agricultural production function. Control variables for climatic condition are represented by annual precipitation and temperature. Precipitation is measured by annual precipitation in millimeters while temperature is measured by daily mean temperature, in degrees Celsius. In addition, education is included in the model. Education is measured by the combined primary, secondary and tertiary gross enrolment ration. Previous study by Bravo-Ortega and Lederman (2004) found a positive and insignificant coefficient on the education variable. Data was obtained from UNDP (various years). A dummy variable landlock is included. It takes the value of one if the country is landlocked and zero otherwise. Data on landlockness was obtained from CIA fact book.

In line with the hypothesis that agricultural policies have a positive and significant impact on agricultural productivity, the expectation is that the coefficients of both agricultural policies are positive and significant in line with theory and literature. Agricultural research and development has positive impact on agricultural output (Beintema and Stads, 2004; Alene and Coulibaly, 2009). Similarly, infrastructure is expected to be positive and significant on agricultural production (Andersen and Shimokawa, 2006; Fan and Zhang, 2004). All the essential agricultural inputs of labor, land, livestock, machinery, and fertilizer are expected to be positive and significant (Kudaligama and Yanagida, 2000; Alene and Coulibaly, 2009). Precipitation and temperature are presumed to have a positive impact on agricultural productivity (Weibe, 2003), while education is expected to be positive but insignificant on aggregate agricultural productivity. (Appleton and Balihuta, 1996; Bravo-Ortega and Lederman, 2004). Landlock is expected to be negative. Faye et al (2004) found that landlockness has a substantial negative impacts on country’s economic development.

The panel data set ran for the period 1981-2000, which depended on the availability of data especially for agricultural research and development which is not yet available for the period beyond 2000. Similarly, the countries were chosen on the basis of their availability of data, and also on their representation of the three SSA regions

³

.

3

See the list of countries in Appendix 9

30 5.3 Diagnostic checks

Before presenting the results of the regressions, several tests will be performed.

5.3.1 Outliers

An Outlier is an observation that is much different (either very small or very large) in relation to the observations in the sample. The inclusion of such an observation can substantially alter the results of regression analysis (Gujarati, 2003). Outliers were studied using histogram to identify and remove the far outliers. This will prevent the results from being driven by these outliers.

5.3.2 Mean and Standard Deviation

Table 1: Mean and Standard Deviation in absolute terms

Variable Obs Mean Std. Dev. Min Max

agtp 150 1624. 601 1322. 328 94.9 6024.7

ard 101 13. 81584 13. 96467 .9 69.2

rinf 124 152. 8183 156. 7542 13.71 603.97

labour 148 6858. 074 5803. 448 283 26292

land 148 4487. 216 3468. 305 231 10676

livestk 148 6642. 155 8276. 687 238.9 37392.1

fert 144 30891.9 46823.01 131 177549

tractor 148 4607. 703 3756. 234 60 16300

precp 160 920. 945 289. 9876 201.7 1471.4

temp 160 22. 67687 2. 583601 18 28.1

educ 80 40. 7375 15. 20638 14 71

Table 2: Mean and Standard Deviation in relative terms

Variable Obs Mean Std. Dev. Min Max

rel_agtp 150 0.3447 0.1329 0.0225 0.5658

rel_ard 101 0.0025 0.0014 0.0005 0.0063

rinf 124 152.8183 156.7542 13.7100 603.9700

rel_labour 148 0.9157 0.1876 0.5389 1.3719

land 148 4487.216 3468.305 231 10676

rel_livestk 140 1.1815 1.0935 0.1228 5.5945

rel_fert 136 4.6645 5.5128 0.0482 25.2378

rel_tractor 140 0.8917 0.5677 0.0331 1.8080

precp 160 920.9450 289.9876 201.7000 1471.4000

temp 160 22.67687 2.583601 18 28.1

educ 80 40.7375 15.20638 14 71

31

The mean and standard deviation are as presented in table 1 and 2 above. The mean of agricultural output is 0.3447, indicating that agriculture contributes about 35 percent of the country’s GDP where as the mean of labour is 0.9157 indicating that about 92 percent of the total labour force are engaged in agriculture. This indicates that agriculture is an inefficient activity with 92 percent of the total labour force engaged in agriculture but only contributing 35 percent of the total GDP. This is confirmed by Townsend (1999) who argues that agriculture accounts for about 35 percent of Sub-Saharan African countries GDP and by Beintema and Strads (2006) who find that agriculture in Africa employs over 80 percent of the labour force.

Similarly, the mean of agricultural research and development is 0.0025, indicating that only 0.2 percent of the country’s total expenditure is channelled to agricultural research and development.

This is also confirmed by Pardey et al, (1997) who argue that the annual growth rate in public agricultural R&D spending in Africa declined from 2 percent in the 1970s to 1.3percent in the 1980s and to only 0.8 percent in the 1990s. Furthermore, this results are confirmed by WDR (2008) which reports that public agricultural R&D spending in Sub-Saharan fell in nearly half the 27 countries with data, and the share of agricultural GDP invested in R&D fell on average for the whole region The average density of infrastructure is 152 km which has been argued by Akyuz (2001) to be very low and is the major policy failure in Africa in contrast with East Asian countries.

5.3.3 Multicollinearity.

Multicollinearity exists when several variables correlate with each other (they move together in

systematic ways). When the correlation coefficients cut-off is (-) 0.8 or (-) 0.9, this indicates the

presence of multicollinearity ((Hill, et al.2001). To test for multicollinearity, a correlation matrix

was used. The correlation matrix for model 1 shows that most of the variables are correlated to

one another as shown in Appendix 2, most likely because the variables are correlated to the size

of the economy and land size. To deal with the problem of the multicollinearity, some variables

were dropped and those left were in accordance with the production function as shown in model

2. Even after dropping some variables, the multicollinearity still existed as shown in the

correlation matrix in appendix 3. To deal with the problem further, Labour and fertilizer were

32

expressed as per unit of land. The correlation matrix for model 3 which is the basis for our research as shown in appendix 4 shows no multicollinearity problem.

5.3.4 Hausman test

To check for any correlation between the error component and the regressors we conduct the Hausman test. The Hausman tests the null hypothesis that the coefficients estimated by the efficient random effects estimator are the same as the ones estimated by the consistent fixed effects estimator. If they are insignificant (p-value, prob>chi2 larger than .05) we use random effects. If it is significant p-value, however, we should use fixed effects. The Chi2=17.78 and p- value=0.0032. According to the Hausman test, we reject the hypothesis that the difference in coefficient is not systematic and therefore we can conclude that random effects is inconsistent, therefore we use fixed effects. In order to test for the inclusion of time effects in our fixed effects regression, we run fixed effects including the time effects. We reject the null hypothesis that the time effects are not jointly significant if p-value is smaller than 10%, and as a consequence our fixed effects regression should include time effects. The p-value is 0.12%, which suggests that the null hypothesis should be rejected and leads us to conclude that our fixed effects regression should include time effects.

5.3.5 Autocorrelation

Since panel data have time-period dimension, the possible problem of autocorrelation associated with time series analyses should be taken into consideration. Autocorrelation refers to the situation in which the time series data observations follow a natural ordering through time, generating a possibility that the successive errors will be correlated with each other (Hill, et al.

2001). It is a violation of the assumption that the errors are uncorrelated and independent. If

autocorrelation is ignored, the research could overstate or understate the reliability of the least

square estimates. To examine the presence of autocorrelation, the Breusch-Godfrey test is

implemented. According to Breusch-Godfrey test, there is no autocorrelation as shown in

appendix 5. The Chi2= 29.662194 and the p-value of ehat_1 is 0.183 which suggests that we

accept the null hypothesis of no autocorrelation.

33 5.3.6 Heteroskedasticity test

As panel data has the cross-sectional nature, a relative large number of cross-sections might result a biased estimated variance. Heteroskedasticity refers to the situation in which the random variables have different variances. Under heteroskedasticity OLS estimator is inefficient, and therefore not the best linear unbiased estimator. Also, the standard errors of the OLS estimates and the t-statistic are not correct any more (Hill et al, 2001).

To explore the presence of this problem, I conduct the Goldfeld-Quandt test. According to the

Goldfeld-Quandt test, p-value =0.5 which suggests that the hypothesis of homoskedasticity is not

rejected. Therefore, heteroskedasticity is not a problem in the model.

34 6. Analysis

6.1 Regression results

To test the hypothesis that agricultural policies have a positive and significant impact on agricultural output in Sub-Saharan African countries, we estimated two regressions. Regression one was estimated with time effects but without lagged research and development whereas regression two we included time effects and lagged research and development as shown in appendix 6, which is in line with our theory. The lagged research and development was included in regression two because investments in agricultural R&D generate a flow of benefits over time.

Suprisingly, the regression coefficients of both agricultural research and development and infrastructure were negative and insignificant, which is contrary to our expectations.

Consequently, our hypothesis one and two that agricultural policies have a positive and significant impact on agricultural output were not supported by our regression results. The contradicting results to our expectation may be due to low data on the agricultural policies. The database contained many missing gaps, and this may have as a result led to the negative and insignificant coefficients. Similarly, the coefficients of the agricultural inputs were also negative and insignificant except labour which was negative and significant at 10 percent, which is also contrary to our expectation. The contradicting results similarly may be due to low data on the agricultural inputs. Also, in regression one the coefficient on agricultural research and development may be negative because investments in agricultural research and development do not have immediate impact on agricultural output. However, even after lagging agricultural research as shown in appendix 6 in regression 2, the coefficient was still negative and insignificant which is as a result of too low data.

The coefficients for the time effects from 1988 to 2000 are positive and significant indicating that these years had a positive impact on agricultural output in Sub-Saharan African countries.

This may be a result of the implementation of the Structural Adjustment Programmes (SAPs) which were introduced by the World Bank(WB) and the International Monetary Fund (IMF).

The SAPs were introduced in the 1980s but the implementation started in the late 1980s. This is

supported by Jaeger (1992) and Townsend (1999) who argued that Sub-Saharan African

countries that adopted and sustained these policies to raise farm incentives have had a better

35

agricultural performance in the 1980s and 1990s on average than countries which continued to

discriminate agriculture.

36 7. Discussion, Conclusion and Limitations

7.1 Discussion

Theory predicts that agricultural policies have a positive and significant impact on agricultural output (Ante, 1983; Hu and Ante, 1993; Teruel and Kuroda ,2005; Alene and Caulibaly, 2009).

We estimated the model using the Cobb-Douglas production function which is in line with the theory of agricultural production (Romer, 2006) to test the impact of agricultural policies on agricultural output. Similarly, we included important agricultural inputs which has been used by various outhers in their estimation of agricultural production. The econometric test indicated that time effects should be used with fixed effects in the estimation of the model, and therefore we included year dummies in the estimation. We specified three models as shown in section 5.1 above. The first model with five agricultural inputs in addition to climatic conditions and dummy variable showed multicollinearity problem as indicated in correlation matrix in appendix 2, which could have lead to insignificant coefficients. In model 2 some variables were dropped in line with the recommendation of Gujarati (2001), and those left were in accordance with the theory of agricultural production. Model 2 still indicated multicollinearity problem and therefore labour and fertilizer as indicated in model 3 were expressed as per unit of land, which forms the basis of our regression model.

According to the results of our estimation, both the agricultural policies of agricultural research

and development and infrastructure were found to have no significant impact on agricultural

output, which is contrary to theory and also the findings of Alene and Caulibaly (2009), Teruel

and Kuroda (2005) and Hu and Ante (1993). The reoson for this deviation may be as a result of

too low data and also small sample size. We lagged agricultural research and dvelopment in

accordance with the recommendation of Alene and Caulibaly (2009) who argue that agricultural

research and development investments benefits flow over time, but the estimation results did not

support our hypothesis that agricultural research and development have a positive and significant

impact on agricultural output. This as well can be explained by the low data, and again that

african countries are underdeveloped to absorb the benefits of agricultural research and

development investments.