CASADE

The Decline and Rise of Agricultural Productivity in Sub-Saharan Africa Since 1961

 Steven Block,  Tufts University.

Agricultural productivity growth in sub-Saharan Africa has been a qualified success. Total factor productivity growth has increased rapidly since the early 1980s. By the early 2000s, average annual TFP growth was roughly four times faster than it had been 25 years earlier. This period of accelerated growth, however, followed nearly 20 years of declining rates of TFP growth subsequent to independence in the early 1960s. Average agricultural TFP growth for sub-Saharan Africa was 0.14% per year during 1960 – 84, and increased to 1.24% per year from 1985 – 2002. The average over this period was approximately 0.6% per year, which accounts for 36% of the increase in total crop output over this period. These highly aggregated results conceal substantial regional and country-level variation. Expenditures on agricultural R&D, along with the reform of macroeconomic and sectoral policies shaping agricultural incentives, have played a substantial role in explaining both the decline and the rise in agricultural productivity. The case study of Ghana clearly reflects these broader findings.

The author wishes to thank the National Bureau of Economic Research Africa Project for supporting this research. He is particularly grateful to Keith Fuglie and Will Masters for detailed comments and suggestions, along with the feedback of participants at the NBER Africa Successes Project conference in Accra, Ghana (July 2010). The author also thanks Marina Dimova for her able assistance in constructing the dataset.

JEL codes: Q16, O13, O4

"Measuring technical change is of interest because, in a sense, it defines our wealth and puts limits on what we can accomplish...Since our ability to accumulate additional conventional resources...maybe limited, the growth of the economy and of per capita income and wealth depends on the rate at which technological knowledge is expanding..." (Zvi Griliches, 1987, p. 1010).

Introduction

Agricultural productivity is central to the lives of most Africans. Two-thirds of the population of sub-Saharan Africa is rural, and the FAO counts nearly half of sub-Saharan Africa's rural population as "economically active" in agriculture. For some countries, such as Burundi, Rwanda, Uganda, and Burkina Faso, the rural population share approaches 85-90%, with 45-50% the total population counted as economically active in agriculture. Even among the most urbanized countries of sub-Saharan Africa, such as South Africa, one-third of the population remains rural. In addition, up to 80% of Africa's poor live in rural areas, nearly all of whom work primarily in agriculture (World Bank, 2000). For these producer groups, agricultural productivity is the key determinant of welfare, and agricultural productivity growth is the key hope for poverty reduction (at least in the short- to medium-term). Non-farm rural employment, too, is often closely linked to agriculture -- either directly (as in the marketing of agricultural inputs and outputs), or indirectly (as in the provision of other services in rural markets). The indirect benefits of agricultural productivity growth, in the form of lower food prices, are also critical to the welfare of Africa's rapidly expanding urban populations, the poorest of whom devote 60-70% of total expenditures to food (Sahn, et. al., 1997).

From a macroeconomic perspective, as well, agriculture continues to play a central role in sub-Saharan Africa, accounting for 15% of total value added (20%, excluding South Africa). Of course, every generalization about sub-Saharan Africa and masks the region’s vast heterogeneity. In Liberia, for example, agriculture accounts for 66% of total value added, while in other countries, such as oil-rich Angola, agriculture accounts for only 10% of the value added (World Bank, 2010). African organizations, themselves, highlight these issues. The Comprehensive Africa Agriculture Development Program of the New Partnership for Africa's Development has stated that, "High and sustained rates of agricultural growth, largely driven by productivity growth, will be necessary if African countries are to accelerate poverty reduction. This is because agricultural growth has powerful leverage effects on the rest of the economy...The poor performance of the agricultural sector explains much of the slow progress towards reducing poverty and hunger in Africa." (CAADP, 2006) Current efforts to promote a "new Green Revolution" in Africa face myriad environmental, institutional, and physical challenges in their quest to promote agricultural productivity growth in the region.

This paper provides new estimates of cross-country agricultural productivity growth in sub-Saharan Africa. The resulting picture is one of qualified success. Total factor productivity growth in African agriculture has accelerated dramatically since the early 1980s. By the early 2000s, average annual total factor productivity growth in African agriculture was over four times faster than it had been 25 years earlier. The success is qualified by the finding that much of this acceleration represents a recovery from the substantial decline in TFP growth rates during the 1960s and early 1970s. In addition, levels of output per hectare and per worker in African agriculture remain low by global standards. Among a range of potential explanations for agricultural productivity growth in agriculture, expenditures on agricultural R&D play a dominant role, followed by policy distortions at both the macroeconomic and sectoral levels. Improvements in the quality of the labor force, as indicated by average years of schooling, have also played a central role in driving productivity growth in African agriculture. Many of these findings gleaned from cross-country analysis, are also evident in this paper's more detailed examination of agricultural productivity in Ghana.

This paper is organized as follows. Section 2 reviews related studies. Section 3 describes data used in the cross-country analysis, as well as the approach used to aggregate agricultural output across multiple commodities. Section 4 provides a preliminary perspective on agricultural productivity trends in the form of partial productivity ratios (output per worker and per hectare). Sections 5 and 6 describe, respectively, my methodology for estimating total factor productivity growth and my results. Section 7 explores various explanations for the productivity results presented in the previous section. Section 8 presents a brief case study of agricultural productivity in Ghana, while Section 9 concludes.

2. Related Studies

Within the broader literature on cross-country agricultural productivity, relatively few papers have focused specifically on sub-Saharan Africa. Block (1994) was the first to report a recovery of aggregate agricultural TFP in sub-Saharan Africa during the 1980s, a result confirmed by a number of subsequent studies. Block attributed up to two-thirds of this recovery to investments in agricultural R&D and to macroeconomic policy reform. Frisvold and Ingram (1995) provide an early growth accounting exercise for land productivity, concluding that most of it (up to 1985) resulted from increased input use (labor, in particular). Thirtle, Hadley, and Townsend (1995) highlight the role of policy choices, finding that an index of real agricultural protection played a significant role in explaining TFP growth in African agriculture for the period 1971-86. Lusigi and Thirtle (1997) highlight the role of agricultural R&D in explaining TFP growth in Africa. They also highlight the role of increasing population pressure in driving increased agricultural productivity in Africa. Chan-Kang, et. al. (1999) focus on the determinants of labor productivity in a cross-country African setting. They, too, find land per unit of labor to be an important determinant of labor productivity.

Fulginiti, Perrin, and Yu (2004) estimate agricultural TFP growth for 41 sub-Saharan African countries from 1960 to 1999, finding an average TFP growth rate of 0.83% per year, and confirming the finding from Block (1994) of an acceleration of the agricultural TFP growth since the mid-1980s. Their analysis concentrates on the role of institutions in explaining this growth. They conclude that former British colonies experienced greater rates of TFP growth, while former Portuguese colonies experienced lower rates. They also found negative effects for political conflicts and wars, and positive effects resulting from political rights and civil liberties. Three more recent papers conclude this review.

Nin-Pratt and Yu (2008) reconfirm the acceleration of African agricultural TFP growth since the mid-1980s. They find, however, a negative average growth rate of agricultural TFP (- 0.15% per year) from 1964 to 2003, casting the recovery period as making up for negative productivity growth during the 1960s and 70s. Specifically, Nin-Pratt and Yu find that average TFP growth fell at the rate of -2% per year from the mid-1960s to the mid-1980s, then grew by 1.7% per year between 1985 and 2003. They, too, highlight the role policy change in explaining this reversal in performance. In particular, they find that an indicator of reforms associated with structural adjustment played a positive role. In addition, they find that agricultural productivity in East and Southern Africa benefited from the end of internal conflicts, and that agriculture in West Africa benefited from the devaluation of the CFA franc. They also provide suggestive evidence of the positive effect of investments in agricultural R&D. Alene (2010) also focuses on the contributions of R&D expenditures to productivity growth in African agriculture. In contrast to the average TFP growth rate reported by Nin-Pratt and Yu (2008), Alene finds an average TFP growth rate of 1.8% per year for the period 1970- 2004 (a difference that he attributes to an improved estimation technique). Alene finds strong positive effects of lagged R&D expenditure on agricultural productivity growth, arguing that rapid growth in R&D expenditures during the 1970s helped to explain strong productivity growth after the mid-1980s, while slower growth of R&D expenditures in the 1980s and early 1990s led to slower productivity growth since 2000. Alene (2010) also notes a 33% annual rate of return on investments in agricultural R&D in Africa.

Most recently, Fuglie (2010) examines agricultural productivity growth in sub-Saharan Africa from 1961 to 2006. His findings are mixed. While he reports an increased rate of growth in agricultural output during the 1990s and early 2000s, Fuglie finds that most of this growth in output is explained by expanding crop land rather than improved productivity. Fuglie (2010) stands out in this literature for his critical assessment of the standard data sources, for which he proposes various corrections. In contrast to previous studies, Fuglie does not find a general recovery of agricultural productivity in recent decades. For the period 1961-2006, he reports an average TFP growth rate of 0.58% per year, with the lowest rate occurring during the 1970s (- 0.18% per year), and the highest rate occurring during the 1990s (1.17% per year). Thus, recent estimates of the rate of agricultural TFP growth in Africa differ widely, though there is a general consensus surrounding a decline in productivity during the first two decades following independence and a recovery during the past two decades. These studies applied different methodologies to essentially the same data set, which may explain some of the conflicting findings cited above. As described below, the methodology applied in the present study differs from all of the studies cited above.

3. Data and Output Aggregation

This study combines data from a variety of sources. The core data on agricultural outputs and inputs are drawn from the FAO online database. While often regarded as being of limited quality, these data are ubiquitous in studies of international agricultural productivity, as they are the only comprehensive and detailed source of cross-country data over a long period of time. The central challenge in constructing a data set suitable for estimating a cross-country agricultural production function lies in aggregating the output of multiple agricultural commodities in a way that is comparable across both time and space. The fact that national-level data on key agricultural inputs -- land, labor, fertilizer, tractors, and livestock -- are provided as national totals, and not disaggregated by the crops to which they are applied, requires that agricultural output also be aggregated to the national level.

The most comprehensive discussion of agricultural output aggregation for international comparison is Craig, Pardey, and Roseboom (1991). Drawing on index number theory, they note that the ideal approach to aggregating multiple commodities for a given country and year would be to multiply a vector of base-year local commodity prices expressed in dollars by a vector of quantities of individual commodities. In particular, they specify that the best price weights would be those most specific to the economic activity and agents in question. Yet, even in the absence of data constraints, there is no perfect way to implement this ideal. The key dimensions of the problem, in practice, lie in choosing appropriate deflator's for comparisons over time, and in choosing appropriate exchange rates for comparisons across countries. Severe constraints on the availability of commodity-specific price data over time for each country in sub-Saharan Africa add to these challenges of constructing internationally and inter-temporally comparable agricultural output aggregates.

Given the availability of commodity-specific local currency-denominated prices over time, the standard approach for converting aggregate output in a given year into internationally comparable units of measure is to select a numeraire currency, and to use Purchasing Power Parity exchange rates for conversion.1            For its global agricultural data set, the FAO has calculated "agricultural exchange rates," or agricultural PPPs, that it applies in creating internationally comparable aggregates of agricultural output. In practice, virtually every study of international agricultural productivity (whether global or region-specific) simply uses these FAO data, based on PPP prices calculated from the global data set. In theory, however, as noted above, the best price weights to use in aggregating output are those that are most specific to the particular setting of concern.

The present study thus departs from standard practice by calculating a unique set of international commodity prices and PPP exchange rates specific to African agriculture.

In order to calculate the Africa-specific international prices and PPP exchange rates used to construct the data set for this study, I applied the Geary-Khamis method summarized by Rao (1993).

This method requires calculating both a reference set of international commodity prices based on relevant PPP exchange rates, and calculating the PPP exchange rates based on the Craig, Pardey, and Roseboom (1991) provide an extensive discussion of the trade-offs involved in first deflating and then converting each year aggregate output versus first converting in any deflating reference set of international commodity prices. This problem is described by a system of two simultaneous equations. In the first equation, the international reference price for commodity i is calculated as a function of its local currency price in each country j = 1,...,m converted by the PPP exchange rate for country j. In the second equation, the PPP exchange rate for country j is calculated as a function of the quantities and international reference prices for each commodity i = 1,...,n in country j. This is done for a given base year. These two equations can be solved iteratively, ultimately converging on a unique set of reference prices and PPP exchange rates for the specific countries and commodities to be studied. For purposes of this study, I calculated international prices and PPP exchange rates using prices and quantities for the n = 35 commodities in the m = 27 sub-Saharan African countries for which data were available from the FAO.2            I then applied these reference prices in aggregating output across these commodities for the full set of 44 sub-Saharan African countries for which commodity-specific output data were available. Output data for each commodity are net of quantities used for seed and feed.

The base year for these reference prices was 2006. I then created a Paasche-type output index, applying the 2006 prices to aggregate the commodity output data in each country for each year going back to 1961. The rationale for applying the Paasche approach was that the range and, in particular, the quality of the price data has tended to improve over time, and that the best data would thus be the most recent. Data for the other standard inputs to be used in estimating the agricultural production function are also drawn from the FAO database. The land measure is hectares of permanent and arable crop land; the labor measure is the number of economically active males and females in 2. Appendix 1 presents the list of commodities and countries used in calculating the Africa-specific international prices. Resulting output data for each country-year are available on request from the author.  I am grateful to Philip Pardey for suggesting this approach.

Each of these indicators of agricultural inputs falls short of the ideal data for measuring agricultural productivity. In discussing the measurement problems generically associated productivity analysis, Griliches (1960, 1987) has noted that proper estimation of production functions should be based the flow of services of capital (accounting for vintage) in constant prices, as well as on the flow of labor services (e.g., hours worked) weighting different types of labor by their marginal prices. Clearly, the input data available for African agriculture, consisting of counts of the number of tractors and the number of agricultural workers (issues of data quality aside), fall far short of this ideal. In particular, the assumption in the data that all of what is counted as agricultural labor is specifically on-farm labor contradicts micro-based evidence of significant non-farm rural activity (Liedholm, McPherson, and Chuta, 1994). Over- counting labor in this way may impose a downward bias on estimated TFP growth. There must also be substantial measurement error in fertilizer data that capture only inorganic fertilizer in a setting where manure is the primary source of added soil nutrients.

In short, the methodological tradeoffs and measurement errors inevitably associated with constructing both the output and the input data for African agriculture are substantial, and suggest the potential for significant noise and bias in estimates of total factor productivity. Yet, as demonstrated in the seminal work of Jorgenson and Griliches (1967), it is possible to mitigate these problems by introducing explicit controls for the quality of inputs. Craig, Pardey, and Roseboom (1997) note, for example, that up to 70% of total horsepower traction in African agriculture is provided by livestock.

As described below, the quality of inputs differs across countries and over time within countries. To the limited extent possible, it is important to control for these differences by including input quality adjustments in productivity estimates. Data used here to adjust for variations in land quality include the proportion of permanent and arable crop land that is irrigated, and annual rainfall. The former are drawn from data compiled by Sebastion (2007). The annual rainfall data used in this study are drawn from Mitchell, et. al. (2003) and Jefferson and O’Connell (2004), based on the crop-weighting scheme of Ramankutty and Foley (1998).5 Quality adjustments to the agricultural labor force generally rely on literacy rates. This study takes advantage of newly-released data on average years of schooling from Barro and Lee (2010). Additional data used in trying to decompose the productivity residual are described below.

4. Partial Productivity Ratios

Partial productivity ratios (output per worker, and output per hectare) provide a useful initial overview of both the level and growth rate of agricultural productivity. While these ratios share the analytical limitation of not controlling for changes in other inputs, they have the virtue of reflecting the general nature of technical change and agriculture as being predominantly either land- or labor-saving. The simplicity of partial productivity ratios may also be a benefit in a preliminary analysis of noisy and often low-quality data.

The welfare of Africa's agricultural labor force ultimately depends on increasing output per worker. Equation (2) illustrates the challenge to that process in an environment characterized by rapid population growth and limited land area. To the extent that population growth outpaces the rate of expansion of agricultural area, area per worker (A/L) declines, thus increasing the challenge of raising average labor productivity (Y/L) by means of increasing average yield (Y/A). This dynamic has been a major obstacle to agricultural development in sub-Saharan Africa.

For sub-Saharan Africa as a whole over this entire period the average annual growth rate of output per worker has been only 0.41%, despite an average annual growth rate of 1.24% in output per hectare. As suggested by equation (2), the limited ability of yield growth in African agriculture to drive growth in average labor productivity has been driven by the increasing population density of rural Africa, where the annual growth of the agricultural labor force has outpaced area expansion by 0.83% per year from 1961 to 2007. Yet, recent years demonstrate a more optimistic trend. For the period 2001 to 2007, the growth rate of average labor productivity in African agriculture has increased dramatically (to over 2% per year) relative to previous periods -- an advance aided by a reversal of the historical trend towards declining area per worker.

In their seminal study of agricultural development, Hayami and Ruttan (1985) also developed a useful and intuitive graphical presentation of partial productivity ratios. Their graphical representation of equation (2) simultaneously relates changes over time in average land and labor productivity by measuring average land productivity along the vertical axis and average labor productivity along the horizontal axis. Changes in output per hectare and output per worker over a given period can be illustrated by drawing an arrow between the relevant beginning and ending coordinates in that space. Scaling the axes in logarithms conveniently implies that movements along any 45° line represent equal rates of change in both land and labor productivity. From equation (2), it follows that such equal rates of change imply a constant level of area per worker. Thus, each 45° line in this space represents a unique and constant level of A/L. Partial productivity paths steeper than 45° reflect increased rural population density over time.

Timmer (1988) provides various interpretations of movements over time in this space. He notes, for example that a movement due north (indicating growth in yield with no growth in average output per worker) may indicate population growth matched by increased yields through higher labor inputs and technical change, but no improvement in rural living standards. Movements to the northwest might suggest population growth faster than technical change in raising yields, with a consequent deterioration in rural living standards. In contrast, movements due east in this space might reflect a declining agricultural workforce with no changes in yields, but with new mechanical technologies needed to maintain output with fewer workers, hence increasing average labor productivity and rural welfare.

During the second half of this period sub-Saharan Africa reflects a slightly increased rate of growth in average labor productivity, that progress remains quite small by comparison with the other countries illustrated in Figure 1.            Note as well, that those countries with the most rapid increases in agricultural labor productivity have followed paths shallower than the 45° lines, indicating increases in area per worker over time. West Africa, too, made substantial progress in increasing agricultural labor productivity beginning in the early 1980s. In contrast, Sahelian countries began with the lowest level of average labor productivity in 1961/65, and saw that level decline consistently (along with yields) until at least the early 1980s. Similarly, countries in Middle Africa experienced slow declines in agricultural labor productivity until the early 1990s, while countries in Eastern Africa experienced consistent but relatively slow increases in both land and labor productivity over most of the period. These contrasting experiences, even at the regional level, illustrate the great heterogeneity of African agriculture. This heterogeneity pertains both to conditions and to rates of progress over time. (Note, for example, the substantially greater level of average area per worker in southern Africa as compared with Eastern Africa.)

Some countries, such as Nigeria, Côte d'Ivoire, and Benin experienced significant growth in average labor productivity accompanied by moderate growth in crop yield, while other countries, such as Togo, Niger, and Liberia experienced gains in crop yield accompanied by small reductions in average labor productivity. At the same time, Figure 3a depicts rapid declines in agricultural labor productivity in Senegal, Gambia, and Guinea-Bissau. Among countries in Eastern Africa, there was the predominant tendency towards moderate gains in crop yield accompanied by slow growth in output per worker. Figures 3c and 3d depict a similar Note here that the East African countries begin with relatively high levels of rural population density (reflected in their position along a higher 45° line) follow a relatively steep path over time, indicating a tendency towards land- saving technical change. This is consistent with the Induced Innovation Hypothesis, associated with Hayami and Ruttan (1985).

In general, these patterns (particularly at the level of regional disaggregation) conform to what is known of events on the ground. Gabre-Mahdin and Haggblade (2004) provide an interesting perspective on successes in African agriculture. They conducted a survey of over 100 experts working in various areas related to African agriculture (two-thirds of whom were Africans), asking them to identify the most important factors in advancing African agriculture. The majority (62%) pointed to successes tied to specific commodities; 21% identified activities such as policy reform and enhancement of soil fertility; and, 16% cited successful institution- building efforts as the primary drivers of African agriculture. Maize breeding (followed by cassava breeding) was the most widely-cited contributor. Byerlee and Jewell (1997) report that most of the successes in breeding, releasing, in adopting improved maize varieties was in East and Southern Africa. Between 1966 and 1990, Byerlee and Jewell note the release of over 300 improved varieties and hybrids by national maize research programs.

The release of hybrid maize in Africa dates back to the early 1930s in Zimbabwe (then Southren Rhodesia), though there were no major successes until the release in Zimbabwe of the variety SR52 in 1960. Successful hybrid maize releases followed shortly thereafter in Kenya. Byerlee and Jewell (1997) report widely varying results for the adoption of maize hybrids and improved open-pollinated varieties. By 1990, nearly all of Zimbabwe's maize area was planted to hybrids, as was 70% of Kenya's maize area, and 77% of Zambia's maize area. At the same time however, Malawian farmers had planted only 14% of maize area to improve varieties, similar to the 18% of Mozambique's maize area, and 13-29% of Ethiopia's maize area under improve varieties. Byerlee and Jewell also note that even in countries with substantial areas devoted to improved maize varieties, yield gains were often moderated by declining soil fertility combined with extremely limited application of chemical fertilizer. Kumwenda, et. al. (1997) cite declining soil fertility as the most widespread limitation on both yield improvement in the sustainability of the maize-based production systems in Southern and Eastern Africa.

Gabre-Madhin and Haggblade’s (2004) survey reinforces the specific success of maize breeding programs in East and Southern Africa, where by the turn of the century, they reported that 58% of maize area planted to improved hybrids with yields gains of about 40% over local varieties. In contrast, only about 20% of total maize area in West and Central Africa were planted to improve varieties. Those regions were more dominated by improved open-pollinating varieties, with output gains of 15-45% over local varieties.

Evenson and Gollin (2003) track the annual rate of varietal releases for all improved crop varieties. While not disaggregating by regions within Africa, they do report a near doubling of the number of average annual releases between 1976-80 and 1981-85, from 23 to 43.2 (and to 50 per year by the early 1990s). This accelerated release of improved crop varieties coincides with the acceleration in the growth of both partial productivity ratios. Other important sources of success in African agriculture cited in the survey included breeding to combat mosaic virus in cassava, as well as improvements in the yield and drought- resistance of that crop (which is particularly important in West and Central Africa); expansion of horticultural and flower exports from East and Southern Africa; rapid growth of cotton production and exports from West Africa (the Sahelian countries in particular); and, improved breeding of bananas in Central Africa. Among activity-led successes, Gabre-Madhin and Haggblade’s survey noted soil fertility enhancement, such as alley cropping in West Africa and improved water management techniques in Southern Africa. Respondents also noted the positive effects of market reforms, currency devaluation, and improved institutions as contributors to Africa's improved agricultural performance. Partial productivity ratios, while indicative of broad trends in the rate and nature of productivity growth, are limited by their lack of control for potentially confounding changes in other inputs. The remaining sections of this paper thus turn to the estimation of total factor productivity growth in African agriculture.

5. Measuring Total Factor Productivity Growth in Agriculture: Methodology

The rate of growth of total factor productivity (TFP) is conventionally defined as the difference between the rate of growth of real product and the rate of growth of real factor input. Assuming, as in Solow (1957), competitive factor markets and constant returns to scale in the aggregate production function, a change in total factor productivity can be measured as a vertical shift in the production function. A variety of methodological approaches have evolved for estimating total factor productivity growth, including the construction of TFP indices (such as the Tornquist-Theil), data envelopment analysis (based on the non-parametric Malmquist index), and stochastic frontier analysis, in addition to the econometric estimation of the aggregate production function. TFP estimation in the present study is based on the latter approach of estimating the aggregate agricultural production function for a panel of African countries. One Tornquist-Theil indices require detailed factor price data that are unavailable for African agriculture. Stochastic frontier approaches derive their results entirely by imposing very strong conditions on the error structure of the estimated production function -- an approach that seems particularly ill-suited to the present setting, which is key benefit of a parametric approach is that it helps to impose order in an otherwise noisy data set.

Specifying the aggregate agricultural production function requires numerous choices, beginning with functional form. I adopt the Cobb-Douglas functional form, which has been repeatedly validated in agricultural studies (Griliches, 1964; Hayami and Ruttan, 1985), as has been the assumption of constant returns to scale (Hayami and Ruttan, 1985). The "traditional" inputs included in virtually every cross-country study of agricultural productivity include: land, labor, fertilizer, tractors, and livestock. As noted above, available data for each of these inputs almost certainly include significant measurement error. In addition, as emphasized in the early studies of US agriculture by Griliches (1963, 1964), and for the US economy as a whole by Jorgenson and Griliches (1967), much of what might mistakenly be attributed to TFP growth may in reality be changes over time in the quality of inputs.

Whether one puts such adjustments for input quality in the production function or in the residual is an interesting question. Griliches (1960) takes an agnostic approach, suggesting "Whether or not we want the input measures to cover all possible quality changes is a semantic rather than a substantive issue. Hybrid seed corn can be viewed either as improvement in the quality of seed or as ‘technical change.’ Since we are interested in explaining the growth of agricultural output, it does not matter much whether we put it into the ‘input change’ category or the ‘productivity change’ category as long as we put it somewhere and know where it is." The data envelopment analysis approach, while often used in recent studies of agricultural productivity (Lusigi and Thirtle, 1997; Fulginiti, Perrin, and Yu, 2004; and Nin-Pratt and Yu, 2008; and Alene, 2010, among others), is also problematic. Heady, Alauddin, and Prasada Rao (2010), along with Nin-Pratt, et. al. (2003), note that DEA studies of agricultural TFP often produce anomalous and implausible results. The DEA approach measures countries' progress relative to a productivity frontier, which depends arbitrarily on the number and selection of countries included in the sample, and which is poorly suited to distinguish between TFP growth, noisy data, and measurement error. Coelli, Prasada Rao, O’Donnell, and Battese (2005) discuss the relative merits of these approaches.

Specification

The dependent variable in my aggregate production function is crop output aggregated (as described above) based on the Africa-specific international commodity prices and PPP exchange rates calculated for this study. The resulting TFP estimates are thus limited to crop agriculture. This, too, reflects a departure from most of the literature, which typically includes both crop and livestock output (summed) for aggregate output. The median share by value of livestock output in total agricultural output over the entire sample is 0.21, though this share varies by region and country.            The mean livestock share in total agricultural output is highest in the five countries included from southern Africa (0.48), and lowest among the ten included (non- Sahelian) countries of western Africa (0.17). For certain countries, including Botswana, Sudan, Mali, Mauritania, and Namibia, livestock output accounts for greater than half of the value of total agricultural output. For such countries, excluding livestock is a potentially significant omission. Yet, that omission brings with it the broader benefit of more accurate aggregation of output (based on Africa-specific data, which are not available for livestock output). On average this omission is relatively small. (Appendix 2 demonstrates the robustness of my main results compared against those derived from using a broader output aggregate that includes livestock.)

There is also a more theoretical reason for excluding livestock from the output aggregate, arising largely from the construction and interpretation of the production function itself. As typically specified, with inputs including tractors, fertilizer, livestock (used both for traction and as a source of manure), the production function conceptually describes specifically crop output. The estimated coefficients on these inputs are interpreted as production elasticities and serve as input weights for productivity measurement. This interpretation of estimated coefficients for tractors and fertilizer in particular is clouded by the inclusion of livestock in the dependent variable. Indeed, by comparison with crop agriculture, livestock production is less labor intensive and more land intensive, thus blurring the interpretation of those coefficients, as well. Yet, excluding livestock from the dependent variable does come at the cost of under- emphasizing integrated crop-livestock production systems that have become increasingly common in Africa. Available cross-country data on inputs and output in agriculture provide no perfect match between what is included on the left- and right-hand sides of the production function. For instance, while I can (and do) eliminate permanent pasture from my measure of land, the labor variable still includes labor applied to livestock production.

Prior to specifying and estimating the cross-country production function, it is useful to present the growth rates of output and inputs. Table 3 presents these growth rates, distinguishing the periods before and after 1985. Crop output for the entire period 1961 to 2007 grew at an average rate of just over 2% per year, accelerating post-1985. Growth of the agricultural labor force was also stable, at about 1.65% per year. Agricultural area also expanded at a relatively stable 0.85% per year. What is striking, however, is the dramatic reversal in the growth rates of the number of tractors and tons of chemical fertilizers pre- and post-1985, a break-point that may reflect the widespread onset of structural adjustment and related reforms. From 1961 to 1984, the average growth rate for tractors and fertilizer were just over 7% and 6%, respectively; yet, post-1985, consumption of both fell at an average rate of 0.5% per year.

 

7. Explanations for Total Factor Productivity Growth in Agriculture

This section considers several potential explanations for productivity growth in African crop agriculture, including: expenditures on agricultural research and development, infrastructure (roads), the effects of civil war, and incentives (agricultural and macroeconomic policy distortions). Severe data constraints, however, preclude a complete decomposition in which all of these potential explanations are considered together. The best one can do, then, is to compare the baseline TFP residual (net of adjustments for input quality) individually against each of these potential explanations. In each case, it is necessary to re-estimate the "baseline" TFP growth rate based on the sample of observations available for each potential explanation of productivity growth. This approach provides estimates of the share of TFP growth explained by each of these factors; yet, these results will not be strictly additive across the potential explanations (as the explanatory variables are not orthogonal to one another), and the generalizability of these results must be qualified (as each decomposition must be estimated over a slightly different sub-sample of the full data set). It may be reasonable, then, to think of the following results as reflecting upper-bounds on the role of any individual explanation for productivity growth. As in my previous accounting for input quality adjustments, my approach to measuring the contribution of a given explanatory variable to TFP growth is first to estimate the quality- adjusted production function with and without the additional variable, and then to calculate the percentage difference in the means of the resulting non-parametric TFP growth paths as the contribution of that variable to TFP growth.

Agricultural R&D

Ultimately, measured productivity growth is intended to reflect a deeper process of technological change. Expenditures on agricultural R&D are thus a potentially important driver of productivity growth, as numerous studies have shown for Africa and for other developing and developed regions (most recently for Africa, Alene, 2010). Data on agricultural research expenditures for 27 sub-Saharan African countries since 1971 have been collected by the Agricultural Science and Technology Indicators (ASTI) Initiative, housed at the International Food Policy Research Institute Beintema and Stads (2006) describe the rapid post- independence growth in funding for agricultural R&D in Africa, followed by slower growth in research expenditures during the 1980s, and near stagnation during 1990s.

By region, the average growth rate of R&D expenditures from 1971 to 2000 has been greatest in East Africa -- exceeding the growth rate of expenditures in West Africa by a factor of nearly eight. These are annual expenditures by governments in each country. They thus reflect a flow of inputs into R&D. While much of the national funding for agricultural R&D in Africa is donor-funded, these data do not include the benefits for any given country of expenditures by the international agricultural research centers. Thus, to the extent that national funding and the benefits of international research are correlated, the present estimates may be biased upwards. Substantial lags exist between the time expenditures on R&D occur and the time they affect productivity. Alene (2010) examines alternative lag structures on R&D expenditures, with lags ranging from 2 to 16 years. His finding that the maximum effect of agricultural R&D occurs around lag 10 leads him to conclude that the slowdown in agricultural TFP growth during the 1990s is partially explained by the reduced growth rate of agricultural R&D expenditures in the 1980s. This is consistent with the prediction by Block (1995), which also found that agricultural research expenditures, lagged by ten years, were significant in explaining the recovery of African agricultural productivity during the 1980s (but which expressed concern for the future impact of reduced R&D expenditures by the late 1980s).

Adding the 10-year lag of log agricultural R&D expenditures to the production function estimated above (net of input quality adjustments) results in a production elasticity of approximately 0.2 (P = 0.000), suggesting that doubling the level of agricultural R&D expenditures at time t would boost agricultural output per worker by 20% at time t+10 -- a substantial effect, and one that is consistent with studies that find high rates of return to

agricultural research expenditures in Africa (Alene, 2010).19            Including the 10-year lag of R&D expenditures limits the estimation period to1981-2000. For that period, the 10-year lag of R&D expenditures explains 75% of estimated TFP growth. Extending the estimation period back to 1976-2000 by including only the 5-year lag of R&D expenditures results in only a small reduction in the estimated production elasticity (to 0.18). In this case, agricultural R&D expenditures still explain 45% of estimated TFP growth.

Roads

The potential benefits of increased road density for agricultural productivity have been explored in a variety of developing-country settings.            These benefits, according to Zhang and Fan (2004) include: increased profitability of farming resulting from reduced transportation costs; greater purchases of inputs and marketing of output resulting from reduced transportation costs; and, the potential to shift land from low-value cereals to higher-value horticulture with reduced risks of perishability. Zhang and Fan (2004) demonstrate significant contributions of roads to crop TFP in rural India, as do Mendes, Teixeira, and Salvato (2009) for Brazil, and Suphannachart and Warr (2009) for Thailand, among many others. In a simulation model of Uganda, Gollin and Rogerson (2010) also find significant complementarities between road density and agricultural TFP. Most recently, Dorosh, et. al. (2010) provide evidence from sub- Saharan Africa that agricultural production is higher in areas with lower travel times to urban markets, and that adoption of modern technologies is negatively correlated with travel time to urban centers. Including R&D expenditures in the production function required excluding the country dummies, as virtually all of the variation in R&D expenditure is in the cross-section dimension of the data (rendering the “within” estimator impractical).

Such findings are consistent with both intuition and with the broadly held presumption that roads are a critical ingredient for growth in agricultural productivity in Africa. For instance, in its Framework for African Agricultural Productivity, the Comprehensive African Agriculture Development Program (2006, p. 16) presents it as a given that, "... investment in infrastructure, particularly rural feeder roads, can also lead to large productivity growth and poverty reduction efforts." It is difficult, however, to demonstrate this contribution with available cross-country country data. To account for the potential contributions of roads to agricultural TFP in Africa, I re- estimate my baseline semi-parametric production function to include countries' share of paved roads as a proportion of total roads. These roads data, drawn from the World Bank's World Development Indicators, are quite limited in their country coverage and only begin in 1990. The median paved road share for 1990-2007 was 16%. Perhaps owing to either the small sample size or to the general lack of paved roads, the estimated production elasticity for paved road share is effectively zero, and its inclusion makes virtually no difference to the estimated rate of TFP growth. Replacing the paved road share of total roads with the ratio of road kilometers to arable land does not change this result. One cannot conclude from this that the broad intuition regarding roads' potential contribution to agricultural TFP is wrong. Rather, available cross- country data and historical experience in Africa do not yet provide the expected statistical support for that intuition.

Civil War

Civil conflict has been endemic in much of sub-Saharan Africa in the post-independence period. Sambanis and Elbadawi (2000) report that between 1960 and 2000, 40% of sub-Saharan African countries had experienced at least one period of civil war, and that in the year 2000 alone 20% of sub-Saharan Africa's population lived in countries that were formally at war (with endemic low-intensity conflict in any other countries). They attributed this problem to high levels of poverty, failed political institutions, and economic dependence on natural resources. It is reasonable to suppose that endemic civil war (and perhaps even the expectation of civil war) could negatively affect agricultural productivity. Physical destruction of crops, damaged infrastructure inhibiting both the purchase of inputs and the marketing of outputs, the diversion and destruction of human capital, and the potential reticence of households to invest in agricultural improvement given the threat of these disruptions, could all lead to reductions in agricultural productivity. I test this hypothesis by including in the production function data on the incidence of civil wars, carefully constructed by Sambanis (2006).

A dummy variable equal to one during years of civil war enters the production function negatively, with a coefficient equal to -.04 (P = 0.11), suggesting that average crop output across the sample falls by 4% during years of civil war. Its effect on productivity is greater. Comparing the averages of the non-parametric TFP growth paths with and without the incidence of civil wars suggest that average TFP growth in African crop agriculture for the period 1960 to 2000 would have been over 11% greater in the absence of civil wars. This is the average effect based on the occurrence of civil war in 13% of the country-year observations included in the regression. A cautious interpretation of this result might consider the possibility that the incidence of civil war acts as a proxy for broader (and excluded) institutional failures. Given this qualification, one can gain additional insight into the effect of civil war on agricultural productivity in Africa by dividing the sample into observations with and without civil war, observing their distinct experiences over time as opposed to the average effect of civil war across the entire sample. This approach reveals that the average rate of agricultural TFP growth was 0.74 percentage points lower (and negative on average) in the presence of civil war. Figure 8 illustrates these differences, which (given the inclusion of country fixed effects) are identified by countries moving in or out of the state of civil war.

Macroeconomic Policy Distortions (Black Market Premium)

It is well-documented that African economies have historically experienced high degrees of distortion in macroeconomic policy. It has also been documented, first by Krueger, Schiff, and Valdes (1988), that macroeconomic distortions in developing countries have often imposed indirect taxes on agricultural producers in excess of their rates of direct taxation. That story highlighted the role of real exchange rates, which were often overvalued to the detriment of African farmers (who tended to produce import-competing tradables or exportables). By undermining agricultural incentives, macroeconomic policy distortions might also have affected agricultural productivity. To test that hypothesis, I use data on the black market premium for foreign currency, often employed as a proxy for such distortions. Over the period 1961-2004, the mean black market premium for sub-Saharan Africa was approximately 66% (though this mean falls to 30% if one excludes as outliers observations with black market premia greater than 500%).

The estimated coefficient on the log black market premium in the production function is not statistically different from zero, indicating that this proxy for macroeconomic distortions did It is possible that this over-estimates the difference between settings with and without civil war if TFP is under- estimated during civil wars. This could be the case if the data simply count the number of workers in the sector, some of whom are prevented from working by war. Yet, including the log black market premium in the specification accounts for 29% of measured TFP growth. Figure 9 illustrates this result. It is interesting to note that the productivity cost of this macroeconomic distortion diminishes over time relative to the baseline TFP growth path, given that black-market currency premia in Africa over this period fell on average by 12% per year (and was half the level post-1990 that had pertained pre-1990).

Agricultural Policy Distortions (Relative Rate of Assistance)

Producer incentives might also exert a substantial effect on agricultural productivity, particularly as regards farmers' choices on production intensity, crop mix, and input use. In a recent and major update to the earlier work by Krueger, Schiff, and Valdes (1988), the World Bank has released an extensive data set on trade-based agricultural price distortions (Anderson and Valenzuela, 2008). This data set provides commodity-specific indicators of the policy- induced divergence between domestic and international prices, covering 30 different commodities in 68 countries (including 13 countries from sub-Saharan Africa) since 1955. The key analytical building block of this data set is the nominal rate of assistance (NRA) for each commodity-year observation, essentially measuring the rate of tax or subsidy at the border. Anderson and Valenzuela (2008) also aggregate these nominal rates of assistance into agricultural and non-agricultural categories. By calculating the ratio of the rate of assistance to agricultural versus non-agricultural commodities, they create a relative rate of assistance (RRA) indicator, which measures the extent to which agriculture is either favored or disfavored by trade policy.

 

Ghana, in many ways, reflects the experience of sub-Saharan Africa over this period. The following section draws on the broader cross-country analysis to highlight key aspects and determinants of Ghana’s agricultural productivity. The black market premium and the first difference of the RRA are only loosely correlated (ρ = -0.11). While this negative correlation suggests that countries with distorted currency regimes also tended to discriminate against agriculture, the small magnitude of this correlation suggests that these two indicators do indeed reflect different impacts on agricultural productivity.

8. The Case of Ghana

This brief review is not intended to be a comprehensive analysis of Ghana’s agricultural productivity experience. Rather, the primary objective is to explore in greater detail key findings from the cross-country analysis regarding the drivers of productivity growth. A secondary objective of this brief review of Ghana is to highlight some the issues that arise in country-level analysis – issues that are generally invisible at the cross-country level, but which may suggest caution in interpreting of cross-country findings.

Partial & Total Factor Productivity in Ghana

Ghana typifies the decline and rise pattern of agricultural productivity seen in the broader African sample. Figure 11 summarizes Ghana’s experience as reflected in the time path of its partial productivity ratios. The first decade of independence saw small gains in crop yield combined with declining output per worker. The country’s decline into economic chaos during the 1970s is reflected in the rapid deterioration of both land and labor productivity depicted in Figure 11. For agriculture, the country’s economic nadir in 1983 was exacerbated by severe drought (starting in 1981), widespread bushfires, and the forced repatriation of one million Ghanaians from Nigeria. These negative trends were strikingly reversed in the early 1980s, leading to a sustained (and continuing) period of growth in the productivity of both land and labor. Clearly, looking only at a path connecting the first and last periods (from which we would conclude that the annual growth rates of average land and labor productivity were 1.35% and 0.6%, respectively) would obscure the dramatic decline and resurgence seen by tracing out successive 5-year period averages. The narrative of Ghana’s agricultural productivity is thus much more complex than would be implied by the moderate rates of growth in land and labor productivity observed on average over the period 1961 – 2007. The challenge is to explain the decline and rise.

The semi-parametric estimation approach developed above is not well-applied to a single country time series of only 40 observations. The estimated (input quality-adjusted) production elasticities are not statistically significant. Yet, controlling linearly for the conventional inputs results in a TFP growth path, depicted in Figure 12, which is statistically different from zero and suggests an average rate of crop TFP growth of 1.03% per year from 1961 – 2000. This pattern of TFP growth rates is also consistent with the pattern of partial productivity ratios for Ghana. For the period 1961 – 2000, aggregate crop output in Ghana grew at the average annual rate of 2.37%. Growth accounting thus suggests that a TFP growth rate of 1.03% accounts for approximately 43% of the growth in crop output.

One way to summarize the current levels of crop productivity is to compare current yields against potential yields. Such analysis by Ghana’s Ministry of Agriculture (2007) suggests that the yields gaps remain substantial. For example, average maize yield of 1.5 MT/Ha is reported to be 40% short of the achievable yield. Yield gaps calculated for other staple grains are reported on the same order of magnitude, while the yield gap for cassava in Ghana is reported to be 57.5% (Breisinger, et. al., 2008). The challenge is to identify the constraints to reducing these yield gaps.

One critical constraint to reducing the yield gap is the great heterogeneity of conditions that characterize agriculture in Ghana (and virtually every other country in sub-Saharan Africa). Figure 13 shows that Ghanaian agriculture is spread across six distinct agro-ecological zones, each listed here with its mean annual rainfall in millimeters: Rain Forest (2,200), Deciduous Forest (1,500), Transitional (1,300), Coastal (800), Guinea Savanna (1,100), and Sudan Savanna (1,000). These zones differ in their average annual rainfall by a factor of nearly four (Figure 14); unlike the first four zones, which have two growing seasons, the two Savanna zones have only one. Ghana’s agro-ecological zones also differ in their soil types and in the length of their growing seasons, as a result of which they also differ widely in the mix of crops produced. In addition, the productivity levels and growth rates for individual crops also vary widely across agro-ecological zones.

The greatest concentration of relatively high-yield maize production is in the southern Guinea savanna in transitional zones, while the greatest concentration of relatively low-yield maize production lies just south of there in the forest zone. Average yields in the former are approximately twice those of the latter. Cassava production is similarly widespread (with the exception of the northernmost savanna areas), with a spatial distribution of yields similar to that of maize. In contrast, sorghum is grown exclusively in the Guinea and Sudan savanna zones, and districts with vastly different yields border one another; while plantain is grown exclusively in the forest and coastal zones, with somewhat less spatial variation in yields.

R&D

The cross-country analysis identified expenditure on agricultural R&D as a key determinant of productivity growth. The diversity of agricultural conditions within Ghana multiplies the technical challenges to increasing agricultural productivity. Broadly, however, the relationship between R&D expenditures and TFP growth in Ghana is consistent with the cross- country evidence. While the poor estimation of the underlying production function renders the estimated TFP growth rates for Ghana as merely suggestive, their conformity with a 10-year lag of expenditures on agricultural R&D is striking.26            Figure 12 juxtaposes the growth path of crop TFP with R&D expenditures. The transition to positive rates of TFP growth in the early 1980s follows by roughly 10 years the increased expenditures on agricultural R&D of the early 1970s; the peak in TFP growth rates seen in the mid-1990s similarly follows the peak of R&D expenditures of the mid-1980s; and, the decline in TFP growth rates in the late 1990s also lags by approximately 10 years the reduced R&D expenditures of the late 1980s. The main anomaly to this pattern is that the reduced expenditures of the late 1970s and early 1980s are not reflected in the estimated TFP growth path.

R&D expenditure is a blunt proxy for specific research outputs. The main research output of interest here is improved varieties of staple grains. As Figure 15a demonstrates, maize is grown is all of Ghana’s agro-ecological zones. The diversity of growing conditions, however, implies that improved maize varieties must be adapted to specific settings. Ghana’s Crop Research Institute takes the lead in developing and releasing improved varieties. During the critical period of reversal in crop productivity trends, the Crop Research Institute, in collaboration with the International Maize and Wheat Improvement Center (CIMMYT), the International Institute of Tropical Agriculture (IITA), and the Canadian International Development Agency (CIDA) implemented the Ghana Grains Development Project. The TFP growth path for Ghana is not statistically different from zero when lagged R&D expenditures are included in the production function (though the sample falls to 19 years). 1984 and 1996, this project developed and released twelve improved varieties of maize (Morris, Tripp, and Dankyi, 1999). The project also promoted use of chemical fertilizers to complement these improved varieties, and recommended new planting strategies.

While these research advances created the potential for improved maize productivity, the real benefits came only with their widespread adoption. By 1997, a nation-wide survey found that 54% of farmers planted modern varieties of maize, though adoption rates varied widely across agro-ecological zones (the highest adoption rate, 69%, was in the coastal savanna, while the lowest rate, 38%, was in the Forest zone). Adoption of recommended planting strategies followed a similar pattern. Yet, only 21% of farmers adopted the recommended fertilizers (ranging from 36% in the Guinea Savanna to 9% in the Forest zone), and only 26% of the national maize crop (by area) received fertilizer. (Morris, Tripp, and Dankyi, 1999.) In 1997, approximately half of Ghana’s maize area was planted to modern varieties (ranging from 75% in the Coastal Savanna, to 33% in the Forest).

Adoption of improved maize was thus reasonably widespread, if unevenly so, across the country. On the supply side, one constraint to more widespread adoption of improved maize varieties was an inability of the Ghana Seed Company (a government entity) to multiply the improved seeds in sufficient quantity (Morris, et. al., 1999). On the demand side, Doss and Morris (2001) found that the key constraints to adoption were lack of access to land, labor, and credit. Jatoe, Al-Hassan, and Abatania (2005) found similar constraints to the adoption of improved sorghum varieties in northern Ghana, where 40% of farmers had adopted improved sorghum, but only 0.1% of total sorghum area was planted to modern varieties. The survey also found that 9% of farmers who adopted modern varieties subsequently “disadopted” them, along with nearly one-third of those who had tried fertilizer, and 13% of those who had adopted recommended management techniques.

More recently, Kwadzo, Ansah, Kuwornu, and Amegashie (2010) surveyed farmers in Ghana’s Eastern Region. They found that 83% of farmers had adopted improved maize, which covered 78% of maize area planted in the region. Yet, they also found that the yield potential of this adoption was not maximized because only 34% of farmers had also adopted nitrogen fertilizer, and that only 30% of maize area received fertilizer. They also found that the likelihood of adoption of improved maize was a positive function of both road access by farmers and the number of visits by extension agents.

Policy Interventions

Policy interventions – both macroeconomic and sectoral – were also found to play important roles in shaping agricultural productivity patterns in the African cross-section. In this regard, too, Ghana is representative. Ghana’s post-independence economic and policy experience is divided into two distinct periods. Following its auspicious emergence into independence in 1957 as an essentially middle- income country, Ghana’s economy spiraled gradually downward into chaos, reaching its nadir in the crisis of 1983. With the adoption of its well-known Economic Recovery Program in that year, the country entered an extended (and continuing) period of stable growth. The macroeconomic environment that ended in crisis was characterized by high inflation, large fiscal deficits, declining exports, and a black market premium on its currency that grew from 35% in the early 1970s to 367% in the late 1970s, to nearly 1300% in the early 1980s (World Bank data cited in Brooks, Croppenstedt, and Aggrey-Fynn, 2009). This history coincides cleanly with the sharp reversal of the partial productivity path  as well as with the transition to positive rates of TFP growth.

The potential connections between macroeconomic distortions and agricultural productivity are direct. The dramatically overvalued exchange rates that characterized the late 1970s and early 1980s in Ghana directly undermined incentives for domestic producers of import-competing crops (such as maize and rice), as well as for export-crop producers (cocoa). The 90% real depreciation of the cedi between 1983 and 1987 helped to relieve prior macroeconomic discrimination against agriculture, improving incentive on the output side, yet also increasing the cost of imported inputs. In addition, economic reform included the elimination of numerous input subsidies that had contributed to the unsustainable fiscal deficits. Thus, for example, the removal of fertilizer subsidies in 1990 led to a 36% increase in the real price of fertilizer, while the prices of insecticides and fungicides tripled in real terms with the removal of their subsidies (Seini, 2002).

Policy reforms at the sectoral level were less ambiguous in their benefits for Ghana’s farmers. The period from independence to 1983 was characterized by high rates of agricultural taxation – both indirect (arising largely from the overvalued exchange rate), and direct. Subsequent to the liberalization of Ghana’s foreign exchange market and the devaluation of the cedi in 1984, agricultural taxation was primarily direct taxation. The example of cocoa taxation is notorious. The combination of an overvalued exchange rate and direct taxation in the form of low producer prices paid by the monopsonistic Ghana Cocoa Board was such that by 1983, farmers received about one-fifth of the FOB price of cocoa (Seini, 2002). With the subsequent devaluation and the reform of agricultural policies that accompanied the Economic Reform Program, cocoa farmers’ share of the FOB price had increased to 40% by 1995, and to 50% by 2001 (Brooks, Croppenstedt, and Aggrey-Fynn, 2009).

The nominal and relative rates of assistance (described above) provide a more general indicator of agricultural policy in Ghana. Average rates of taxation (measured relative to international prices) for agricultural tradables increased from approximately 17% in the early 1960s to 50% by the late 1970s. With the period of reform, these rates of taxation fell back to 17% by the late 1980s, and averaged just over 3% for 2000-04 (Brooks, Croppenstedt, and Aggrey-Fynn, 2009). Comparing this indicator to similar measures for non-agriculture provides an indicator of price discrimination of agriculture relative to non-agriculture (the “relative rate of assistance”). From this broader perspective, as well, one finds substantial and increasing discrimination against agriculture in the pre-reform period, with declining but persistent discrimination against agriculture in the post-reform period. Relative discrimination against agriculture averaged just over 6% in the early 1960s, increasing to approximately 25% in the late 1970s. While falling substantially during the period of economic reform, this indicator of relative discrimination was still 8% for 2000-04.

This brief review demonstrates that agricultural productivity growth in Ghana broadly reflects the cross-country experience of sub-Saharan Africa. The general pattern of post- independence decline followed by renewed productivity growth since the 1980s is clear in Ghana. The important roles of agricultural R&D expenditure and policy interventions seen in the broader cross-section are also clear in Ghana.

Cautionary Note

Even a brief country case study can serve the purpose of providing a cautionary note for the interpretation of cross-country findings. In particular, Ghana’s agro-ecological diversity is common in sub-Saharan Africa. From a technological perspective, this diversity greatly complicates current efforts to promote a new green revolution for Africa. Different crops are specific to different agro-ecological zones; and for ubiquitous crops such as maize, an improved variety that thrives in humid Evergreen zones of south-western Ghana may be inappropriate for planting in the arid zones of the northern savanna. An analysis that explains aggregate agricultural productivity at the country level based on total expenditures on agricultural R&D inevitably obscures the fact that both expenditures and productivity growth are likely to be quite unevenly distributed across the country. This diversity is even more obscured when that aggregate country-level analysis is merely part of a broader cross-country panel data set.

Looking within a particular country also enables a closer examination of the sources and quality of agricultural data. In the case of Ghana, Obirih-Opareh (2004) provides a critical examination of the methods applied by the Ministry of Food and Agriculture in compiling its national area and production data. He notes, for example, that most Ghanaian farmers do not keep their own records of area and production. In addition, Obirih-Opareh notes that most farmers mix numerous crops in a single field, further complicating the calculation of area and yield of individual crops, and that many farms are not accessible by road. As a result, production and area surveys must rely on limited and potentially poorly-measured samples. For export crops, such as cocoa, the situation is better. Similarly, consumption data for imported inputs such as chemical fertilizer, are also more reliable. Yet, Obirih-Opareh in general finds that the limited ability of the Government to undertake annual nation-wide surveys of complex and remote production systems often leads to statistical anomalies in the published data. He also notes that different international and national sources of published data on agricultural area and production in Ghana provide conflicting information. In this respect, too, Ghana is undoubtedly not unique in sub-Saharan Africa.

9. Conclusions

Agricultural productivity growth in sub-Saharan Africa has been a qualified success. Total factor productivity growth has increased rapidly since the early 1980s. By the early 2000s, average annual TFP growth was roughly four times faster than it had been 25 years earlier. This period of accelerated growth, however, followed nearly 20 years of declining rates of TFP growth subsequent to independence in the early 1960s. Average agricultural TFP growth for sub-Saharan Africa was 0.14% per year during 1960 – 84, and increased to 1.24% per year from 1985 – 2002. The average over this period was approximately 0.6% per year, which accounts for 36% of the increase in total crop output over this period. With the exception of East Africa, every region’s TFP growth rate was higher between the years1985 – 2002 than it had been during 1960 – 1984.

From among the long list of potential explanations for these trends, this paper considers several leading contenders. Data constraints on individual explanations preclude a unified and comprehensive decomposition of the productivity residual. It is clear, however, that expenditures on agricultural R&D, along with the reform of macroeconomic and sectoral policies shaping agricultural incentives have played a substantial role in explaining both the decline and the rise in agricultural productivity found in this paper.

The case study of Ghana clearly reflects these broader findings, and permits a more nuanced view of their effects. The case study also provides a brief window into the vast complexity of agricultural development in any single country, and in doing so, provides a cautionary note for the interpretation of aggregate cross-country results.

 

References

Alene, A. 2010. "Productivity Growth in the Effects of R&D in African Agriculture." Agricultural Economics, forthcoming.

Anderson, K., and E. Valenzuela. 2008. Estimates of Distortions to Agricultural Incentives, 1955 to 2007, core database at www.worldbank.org/agdistortions.

Barro, R. and J-W Lee. 2010. “A New Data Set of Educational Attainment in the World, 1950– 2010.” NBER Working Paper No. 15902. April. Cambridge, MA.

Bates, R. 1981. Markets and States in Tropical Africa. Berkeley, University of California Press.

_____ and S. Block. 2010. “The Political Economy of Agricultural Trade Interventions in Africa,” in K. Anderson, ed., The Political Economy of Agricultural Price Distortions. Cambridge, Cambridge University Press.

Beintema, N., and G-J. Stads. 2006. “Agricultural Research and Development in sub-Saharan Africa: An Era of Stagnation,” Background Paper, ASTI Initiative, International Food Policy Research Institute, Washington, D.C.

Binswanger, H., and R. Townsend. 2000. "Technological Change and Commercialization in Agriculture: The Effect on the Poor." World Bank Research Observer 6 (1): 57-80.

Block, S. 1994. "A New View of Agricultural Productivity in Sub-Saharan Africa," American Journal of Agricultural Economics, Vol. 76, No. 3, August.54

_____. 1995. "The Recovery of Agricultural Productivity in Sub-Saharan Africa," Food Policy, Vol. 20, No. 5, October.

Breisinger, C., X. Diao, J. Thurlow, and R. Al-Hassan. 2008. Agriculture for Development in Ghana: New Opportunities and Challenges. International Food Policy Research Institute Discussion Paper 784. Washington, D.C.

Brooks, J., A. Croppenstedt, and E. Aggrey-Flynn. 2009. "Ghana." In K. Anderson and W. Masters, eds., Distortions to Agricultural Incentives in Africa. Washington DC: World Bank.

Byerlee, D., and D. Jewell. 1997. "The Technological Foundation of the Revolution." In D. Byerlee and C. Eicher, eds., Africa's Emerging Maize Revolution. Boulder: Lynne Rienner Publishers.

CAADP, 2006. Framework for African Agricultural Productivity, the Comprehensive African Agriculture Development Program.

Chan-Kang, C., P.Pardey, S. Wood, J. Roseboom, and M. Cremers. 1999. "Reassessing Productivity Growth in African Agriculture." Selected paper for the annual meeting of the American Agricultural Economics Association, Nashville, Tennessee, August 8-11, 1999.

Coelli, T, D.S. Prasada Rao, C. O’Donnell, and G. Battese. 2005. An Introduction to Efficiency and Productivity Analysis. New York: Springer Press.

Craig, B., P. Pardey, and J. Roseboom. 1991. "Patterns of agricultural growth and economic development." In P. Pardey, J. Roseboom, and J. Anderson, eds., Agricultural Research 55

Policy: International Quantitative Perspectives. Cambridge: Cambridge University Press.

_____. 1997. "International Productivity Patterns: Accounting for Input Quality, Infrastructure, and Research." American Journal of Agricultural Economics 79:1064-76.

Dorosh, P., H-G Wang, L. You, and E. Schmidt. 2010. “Crop Production and Road Connectivity in Sub-Saharan Africa, A Spatial Analysis,” Policy Research Working Paper 5385, World Bank, Washington, D.C.

Doss, C., and M. Morris. 2001. “How does gender affect the adoption of agricultural innovations? The case of improved maize technology in Ghana." Agricultural Economics. Vol. 25: 27-39.

Evenson, R. and D. Gollin. 2003. "Assessing the Impact of the Green Revolution, 1960 to 2000." Science, Vol. 300 (2 May):758-62.

Frisvold, G. and K. Ingram. 1995. "Sources of Agricultural Productivity Growth and Stagnation in Sub-Saharan Africa." Agricultural Economics 4:169-80.

Fuglie, K. 2010. "Agricultural Productivity in Sub-Saharan Africa." Mimeo, Economic Research Service, US Department of Agriculture, Washington D. C.

Fulginiti, L., R. Perrin, and B. Yu. 2004. "Institutions and Agricultural Productivity in Sub- Saharan Africa." Agricultural Economics 4:169-80.

Gabre-Madhin, E., and S. Haggblade. 2004. “Successes in African Agriculture: Results of an Expert Survey." World Development, Vol. 32, No. 5:745-66.

Gollin, D. and R. Rogerson. 2010. “Agriculture, Roads, and Economic Development in Uganda.” NBER Working Paper 15863. March. Cambridge, MA.

Griliches, Zvi. 1960. "Measuring Inputs in Agriculture: a Critical Survey." Journal of Farm Economics XL22(5), December.

_____. 1963. "The Sources of Measured Productivity Growth: United States Agriculture, 1940- 1960." Journal of Political Economy, Vol. LXXI, No. 4, August.

_____. 1964. "Research Expenditures, Education, and the Aggregate Agricultural Production Function." American Economic Review, Vol. LIV, No. 6, December.

_____. 1987. “Productivity: Measurement Problems.” The New Palgrave: A dictionary of Economics, Vol. 4, pp. 1010-1013.

Hayami, Y. and V. Ruttan. 1985. International Development: an International Perspective. Baltimore MD.: Johns Hopkins University Press.

Headey, D., M. Alauddin,, and D.S. Prasada Rao. 2010. "Explaining Agricultural Productivity Growth: an International Perspective." Agricultural Economics 41:1-14.

Jatoe, J.B., R. Al-Hassan, and L. Abatania. 2005. "Factors Affecting the Adoption of Improved Sorghum Varieties among Farm Households in Northwest Ghana: A Probit Analysis." Ghana Journal of Development Studies. Vol. 2, No. 1: 37-50.

Jefferson, Philip N. and Stephen A. O'Connell. 2004. “Rainfall shocks and economic performance in four African countries.” In prep.

Jorgenson, D., and Z. Griliches, 1967. "The Explanation of Productivity Change." Review of Economic Studies.

Krueger, A., M. Schiff, and A. Valdes. 1988. “Agricultural Incentives in Developing Countries: Measuring the Effect of Sectoral and Economywide Policies." The World Bank Economic Review. Vol. 2, No. 3: 255-271.

Kumwenda, J., S. Waddington, S. Snapp, R. Jones, and M. Blackie. 1997. "Soil Fertility Management in Southern Africa." ." In D. Byerlee and C. Eicher, eds., Africa's Emerging Maize Revolution. Boulder: Lynne Rienner Publishers.

Kwadzo, G. T-M., W. Ansah, J.M.K. Kuwornu, and D.P.K. Amegashie. 2010. "Technology Package Adoption by Smallholder Maize Farmers: Acceptability Index and Logit Model Analysis." Mimeo. Department of Agricultural Economics, University of Ghana.

Leidholm, C., M.McPherson, and E. Chuta. 1994. “Small Enterprise Employment Growth in Rural Africa,” American Journal of Agricultural Economics, Vol. 76: 1177-1182.

Lokshin, M. 2006. “Difference-based semiparametric estimation of partial linear regression models,” The Stata Journal, Vol. 6, Num. 3, pp. 377-383.

Lusigi, A., and C. Thirtle. 1997. "Total Factor Productivity and the Effects of R&D in African Agriculture.” Journal of International Development 2:529-38.

Masters, W. and A. Garcia. 2010. “Agricultural Price Distortions and Stabilization,” in K. Anderson, ed., The Political Economy of Agricultural Price Distortions. Cambridge, Cambridge University Press.

Mendes, S., E. Teixeira, and M. Salvato. 2009. “Effect of Infrastructure Investments on Total Factor Productivity (TFP) in Brazilian Agriculture.” Contributed Paper prepared for presentation at the International Association of Agricultural Economists Conference, Beijing, China, August 16-22, 2009.

Ministry of Food and Agriculture, Government of Ghana. 2007. District Level Agricultural Production and Price Data. Accra.

Mitchell, T. D., et. al. 2003. “A comprehensive set of climate scenarios for Europe and the globe.” In prep.

Morris, M., R. Tripp, and A.A. Dankyi. 1999. Adaptation and Impacts of Improved Maize Production Technology: a Case Study of the Ghana Grains Development Project. Economics Program Paper 99-01. Mexico, D. F.: CIMMYT.

Nin-Pratt, A. and B. Yu. 2008. "An Updated Look at the Recovery of Agricultural Productivity in Sub-Saharan Africa." International Food Policy Research Institute Discussion Paper 00787, Washington, D.C.

Obirih-Opareh, N. 2004. “Quality Agricultural Statistical Analysis for Better Policy Planning and Analysis in Ghana.” The Empirical Economic Letters, Vol. 3, No. 6:299-310.

Ramankutty, N. and J. A. Foley. 1998. “Characterizing patterns of global land use: An analysis of global croplands data.” Global Biogeochemical Cycles, 12, 667-685.

Rao, D.S.P. 1993. Inter-country Comparisons of Agricultural Output and Productivity, Social and Economic Development Paper, No. 112, Food and Agriculture Organisation, Rome.

Sahn, D., P. Dorosh, and S.Younger. 1997. Structural Adjustment Reconsidered – Economic Policy and Poverty in Africa. Cambridge: Cambridge University Press.

Sambanis, N., and M. Doyle. 2006. Making War and Building Peace: United Nations Peace Operations. Princeton, NJ: PrincetonUniversity Press.

Sambanis, N. and I. Elbadawi. 2000. “Why Are There So Many Civil Wars in Africa? Understanding and Preventing Violent Conflict.” Journal of African Economies, Vol. 9 (3):244-69.

Sebastian, K. 2007. "GIS/Spatial Analysis Contribution to 2008 WDR: Technical Notes on Data & Methodologies." Background Paper for the World Development Report 2008, World Bank, Washington, D.C.

Seini, A. W. 2002. Agricultural Growth and Competitiveness Under Policy Reforms in Ghana. Technical Publication No. 61, Institute of Statistical, Social and Economic Research. University of Ghana.

Solow, R. 1957. "Technological Change in the Aggregate Production Function." Review of Economics and Statistics, Vol. 39, No. 3, pp. 312-20.

Suphannachart, W. and P. Warr. 2009. “Research and Productivity in Thai Agriculture,” mimeo, Kasetsart University, Bangkok, Thailand.

Thirtle, C., Hadley, D. and Townsend, R. 1995. "Policy Induced Innovation in Sub-Saharan African Agriculture: a Multilateral Malmquist Productivity Index Approach." Development Policy Review 13 (4): 323-42.

Timmer, C.P. 1988. "The Agricultural Transformation." In H. Chenery and T.N. Srinivasan, eds., Handbook of Development Economics, Vo. 1, Amsterdam: North Holland Press.

World Bank. 2010. African Development Indicators. Washington, DC: World Bank.

_____. 2000. Can Africa Claim the 21st Century? Washington, DC: World Bank.

Yatchew, A. 2003. Semiparametric Regression for the Applied Econometrician. Cambridge: Cambridge University Press.

Zhang, X. and S. Fan. 2004. “How Productive is Infrastructure? A New Approach and Evidence from Rural India.” American Journal of Agricultural Economics, Vol. 86 (2):492-501.

---

by CASADE

An Aid-Institutions Paradox? A Review Essay on Aid Dependency and State Building in Sub-Saharan Africa

“The importance of public revenue to the underdeveloped countries can hardly be exaggerated if they are to achieve their hopes of accelerated progress.”

-Nicolas Kaldor, Foreign Affairs, January 1963.

“I have made revenue collection a frontline institution because it is the one which can emancipate us from begging, from disturbing friends… if we can get about 22 percent of GDP we should not need to disturb anybody by asking for aid….instead of coming here to bother you, give me this, give me this, I shall come here to greet you, to trade with you.”

-Yoweri Museveni, President of Uganda (which collects 11% of GDP in taxes and receives a further 11% of GDP in aid), Washington DC, September 21, 2005.

1. Introduction
 After a crisis of legitimacy throughout the 1990s, aid is popular again in the policy community. Several new studies have suggested that at least a doubling of overseas development assistance (ODA) from 2000 levels is necessary as a precondition for meeting international development targets (Zedillo Panel 2001; Devarajan et al. 2002). The Commission for Africa (2005) chaired by British Prime Minister Tony Blair called for an immediate $25 billion increase in aid to sub-Saharan Africa, with an additional $25 billion to come by 2015. This would constitute roughly a tripling of aid to the continent. Further, the UN’s Millennium Project (2005) has estimated that global ODA will need to rise even further than the previous estimates, reaching at least $195 billion by 2015 from current levels of some $79 billion in 2004. These calls for more ODA are echoed in various parts of the United Nations system, the World Bank, many NGOs, recipient countries, and even some European governments.

Many of the low-income countries targeted for substantial increases in aid already receive historically unprecedented flows. For instance, ODA to sub-Saharan Africa was the equivalent of 11.7 percent of the continent’s GNI in 2003 (excluding Nigeria and South Africa).2 Exactly half of the region’s 46 countries with data for 2003 received in excess of 10 percent of GNI in ODA, and 11 received more than 20 percent. Globally, there is a core set of roughly three dozen countries that have received a tenth of GNI or more in aid for at least the last two decades. This is a lengthy time period for receiving sizeable aid with few historical precedents. The large flows to Europe during the Marshall Plan lasted only a few years and never exceeded 3 percent of GDP of any receiving country (De Long and Eichengreen, 1991; O’Connell and Soludo, 2001). While substantial US support during the early Cold War to allies such as Korea and Taiwan tapered off within a decade, contemporary aid ratios in these three dozen countries have tended not to recede, but to grow larger over three decades.

Skepticism about the desirability of such aid increases has tended to emphasize economic and management issues. Some observers have expressed concerns about the capacity of low-income states to absorb large new flows in addition to the flows they already receive, and have pointed to the weak management capacities of governments, the dearth of good new projects and programs to fund, or the ambiguous association between aid and measurable development outcomes (White, 1998; Burnside and Dollar, 2000). Other observers have worried about the macro-economic impact of large aid increases; they have pointed to “Dutch disease” effects on small economies (see for example Heller, 2005, Rajan and Subramanian, 2005). Relatively less critical attention has been paid to the potential effects of large increases in aid on public institutions in low-income countries.

Yet institutional issues have recently returned to the foreground in debates on economic development. The critical importance of sound public institutions to the development process has become an article of faith, not only among political scientists (for example, Herbst, 1990, Haggard, 1990; Evans 1995), who could be supposed to have professional reasons to argue for the importance of institutions, but also has emerged more recently as a consensus among economists (for example, Rodrik, 2003; Ndulu and O’Connell, 1999; Acemoglu et al., 2004). Sachs’ (2005) view that good institutions are entirely a result of development, rather than their cause, is now a minority view.

Aid is thought to work best in environments with high quality public institutions, presumably as part of a capable ‘developmental’ state (among a large literature see Burnside and Dollar, 2000; World Bank, 1998). Increasingly, measures of institutions are an explicit factor for aid disbursement and allocation. Thus, ‘institutional development’ is frequently an independent variable thought to affect the efficiency of aid, and thus a legitimate factor in selecting aid recipients and determining allocation strategies. This suggests that aid should be “selectively” focused on countries that are thought to most effectively use resources to engage in poverty reduction. Such logic underlies IDA’s performance-based allocation process and the Millennium Challenge Account, a new US aid program that explicitly targets assistance to countries that are thought best able to use additional resources (Radelet, 2003).

In many respects, this new approach is at odds with the more traditional argument that one of the primary purposes of aid (if not its most important) should be to build effective indigenous public institutions. By this formulation, institutional development is thought to be a dependent variable, affected by targeted aid. In contrast to the ‘selectivity’ philosophy, this older doctrine has been to channel aid instead to places with the greatest need for improved public institutions with the idea that aid itself will help to improve the institutional environment. This approach underlies growing donor efforts at so-called ‘capacity building’ and the ‘big push’ on aid first popular in the 1950s and 1960s and now advocated by the UN and others (Easterly, 2005). The Commission for Africa report wavers back and forth between these two views of aid and institutions. It recognizes the importance of good institutions to making aid effective, in part because it argues that improved institutions will allow absorption of the much larger aid flows it advocates. But it also believes that these large increases can serve to leverage a much greater commitment on the part of African governments to improving the domestic institutions important to growth and poverty reduction.

Does aid necessarily help to develop public institutions and state capacity, or can there be an aid-institutions paradox? In this essay, we review an emerging literature that explores the potential effects of large amounts of aid on institutional development, including some of the most basic functions of the state such as the ability to collect revenues. Given the current debates regarding large new infusions of additional aid, an analysis of the institutional effects of aid is particularly timely. Because Africa presents the greatest challenges to development, and is the region most aid dependent, we especially look at the aid-institutions relationship in that region. Many political scientists now argue that public institutions in the region are poorly suited to promote economic development because of neo-patrimonial tendencies (Callaghy, 1988; Sandbrook, 1992; Chabal and Daloz, 1999; van de Walle, 2001). In the poorly integrated and fragmented states of the region, political leaders have relied on systematic clientelism and the private appropriation of state resources for political ends. As a result, government resources have not been utilized primarily to promote economic development, as political elites have acted in a predatory fashion to maintain themselves in power. There are other reasons for which the economies of sub-Saharan Africa have failed to gain economic development, but it is now widely conceded that these political dynamics have constituted a significant brake on growth. As a result, it is thus possible that inflows of external resources like aid could be a disincentive to state transformation. Does aid, and the manner in which it is given, encourage the transition from patrimonialism and predation to rational developmental states?

It is far from impossible that certain types of aid could undermine long-term institutional development, despite donors’ sincere intentions. Such a paradox is, of course, not new to the development literature. The so called ‘resource curse’ has long posited that unearned income undermines incentives to build local institutions and perhaps a social contract with the population (see Ross, 1999 for an excellent review; Karl, 1997; Birdsall and Subramanian, 2004). Natural resources represent an unearned rent accruing to governments; it is argued that this rent can have a negative and anti-developmental effect on the economy, public institutions, and even on the government’s relationship with the citizenry. We will argue that aid can have many of the same dysfunctional effects as natural resources; that is, there can be an ‘aid curse’ as well that might create perverse incentives and lead to anti- developmental outcomes.

To analyze these issues, this review essay seeks to integrate two disciplinary literatures that have too long ignored each other. On the whole, political scientists have been remarkably oblivious to the political dynamics created by foreign aid, particularly in low-income countries where it is today the leading sector of economic activity and might thus be thought to have a significant impact on the local political economy. For their part, economists have mostly ignored a long tradition in the political science literature which establishes a historical link between the state’s revenues and its political and institutional attributes. The following section lays out the context for these questions, and explains why they are particularly relevant for today’s debates about aid and development. Section 3 reviews the well-known macro-economic effects of large volumes of aid, and focuses on the institutional implications of these effects. Section 4 explores the potential negative effects of large aid flows on institution building through its effect on local bureaucratic and policy-making dynamics. Section 5 then examines the literature on state revenues and its relationship to foreign aid. The historical linkages between state revenue collection and state-building are considered in Section 6. A theme that emerges in the second half of the essay is the low quality of the available data on state revenues, particularly for Africa. At present, data deficiencies unfortunately prevent the formal empirical testing of many of the hypotheses developed in the essay. Nonetheless, the possibility of the existence of an “aid-institutions paradox” is significant, and Section 7 discusses the policy implications of our findings before concluding.

2. Aid and the development debates
Aid clearly can be useful and has certainly contributed to economic development and improvements in quality of life variables in many countries. Evidence for successful aid is particularly strong in targeted programs with defined objectives (see Levine 2004 for examples in global public health). But, at the same time, and especially at very high levels over a sustained period, aid could also have distorting effects on some of the very outcomes donors hope to encourage through aid, such as policy ownership, fiscal sustainability, institutional development, and, ultimately, autonomous long-term economic growth.

One way to consider this problem is to think of aid as a subsidy. As such, aid is supposed to provide temporary financial assistance in order to encourage certain long-term behaviors: revenue collection, investment in physical and human capital, and the establishment of the institutions of a developmental state. There are clearly some cases where aid-as-subsidy has played this role, for example in South Korea or Botswana, where foreign assistance supported local efforts to do these things and the country gradually was weaned off aid. At the same time, there are many, indeed dozens, of other cases where aid is neither temporary, nor seeming to assist countries in fulfilling these roles. Instead, it could be argued that the subsidy has in fact discouraged revenue collection, distorted expenditure decision-making, and undermined the incentives to build state capacity. In these cases, aid could be viewed as not only a crutch delaying institutional development, but as potentially undercutting those efforts.

This possibility of harmful aid dynamics seems particularly acute in sub-Saharan Africa, where some countries have now entered into their third and fourth decades of receiving substantial volumes of aid. Much of this aid has also included explicit capacity building technical assistance from donors. The World Bank alone provided Africa with 70 civil service reform projects between 1987 and 1997, for instance (Levy and Kpundeh, 2005, p. v), while a recent internal Bank evaluation estimates that over a quarter of all Bank credits to the region is explicitly devoted to capacity building (OED 2005, p. 9). Technical assistance to central banks seems to have been successful in building institutional capacity, but such examples appear more the exception than the rule. Many experts argue that state capacity has improved little during this period, and point to specific cases of clear decline (see van de Walle 2001; 2005).

In some cases, the lack of progress on capacity building can be attributed to political instability. After all, at any given time in the last three decades, over a dozen economies in the region have been subject to violent civil conflict (see for example Collier, 2005) and the emergence of warlord rule in the context of the collapse of the central state (Reno, 1998). Long periods of political stress, conflict and state collapse, continue to have a significant impact on state capacity, even after the return to political stability because of their long-term institutional effects, notably on the supply of trained manpower. Perhaps more striking is the slow pace of institution-building in relatively stable political systems. Indeed, a substantial literature has documented the pervasive weakness of the central state in sub- Saharan Africa, which often exercises weak if any effective sovereignty over much of its territory, and has less legitimacy than a variety of sub-national and private governance structures that compete with it for popular support (Herbst, 2000; Englebert, 2000; Jackson and Rotberg, 1982). It has become fashionable in the donor community to blame this surprisingly slow pace of state capacity building on the nature of African bureaucracies, which are argued to be patrimonial and corrupt, and thus not particularly interested in the provision of public goods essential to development (Levy and Kpundeh, 2005; OED, 2005). But even if one accepts this diagnostic, the question remains, why has the large volume of aid devoted to capacity building not had a bigger impact on improving these public institutions, and transforming them into, using the Weberian terminology, more ‘rational- legal’ bureaucracies?

3. Aid, fiscal policy, and macroeconomic outcomes
A number of potential negative effects of large aid volumes on institutional development can be identified. Much of the focus from economists has been on macroeconomic imbalances caused by large volumes of aid. One central issue has been the possibility of large ODA inflows affecting the real exchange rate and undermining the competitiveness of the export sector— the so called ‘Dutch disease’ (most recently, see Rajan and Subramanian, 2005). Management of the real exchange rate is arguably rendered even more difficult by ODA volatility, which also is thought to have negative effects (see below). Dutch disease-type effects have been noted in a number of African aid recipients (see Younger, 1992 on Ghana; Adam and Bevan 2003). Experiences from Uganda (Atingi-Ego, 2005; Nkusu, 2004) and other countries suggest that an active central bank can manage these exchange rate appreciations and, for the most part, mitigate pernicious effects on competitiveness, but nonetheless, a number of country episodes suggests that in fact a large volume of aid can and does undermine competitiveness.

Another set of economic concerns emphasize the role of aid within the budget process itself, with most studies suggesting that foreign aid can undermine the ability of recipient governments to budget appropriately. Several have implicated the volatility of aid flows as the source of distortions. In a 37-country survey, Bulir and Lane (2002) found that aid is more volatile than domestic fiscal revenues and that this volatility lessens any potential positive benefits of aid on recipients. McGillivray and Morrisey (2000b) found the volatility of aid often leads to poor budgeting and underestimation of revenues, particularly since aid commitments tend to overestimate actual disbursements. Similarly, Heller and Gupta (2002) argue that the fiscal uncertainty of dependence on external assistance makes long-term planning extremely difficult.

Beyond volatility, there have also been some questions about perverse incentives of aid on the process of economic policymaking. Brautigam and Knack (2004), for example, found that high levels of aid serve as a “soft budget constraint”: the access to foreign resources convinces decision makers that budgets are flexible and encourages fiscal indiscipline. Two case studies looking at Ghana found that as donor financing increased, so did disparities between budgeted expenditures and actual spending, suggesting that the budget process was increasingly directed toward satisfying external donors rather than reflecting actual public spending preferences. Killick (2004) thus described Ghana’s “budgetary façade” and Pradhan (1996) similarly called the budget a “deceptive mirage”, in which aid was distorting both the budget process itself and the government’s ‘ownership’ of the country’s purported development agenda.

A number of observers have examined the impact of large volumes of aid on the mix of public expenditure and the overall spending levels. A number of papers suggest that aid results in excessive and unsustainable levels of government consumption, also leading potentially to macro-imbalances. Khan and Hoshino (1992) found aid to be generally treated as an increase in income leading to higher government consumption, but that the some public investment is also financed by aid. In a broad literature review, McGillvray and Morrissey (2000a) found that aid tends to be associated with government spending increases in excess of the value of the aid, although there is no clear answer on the impact of aid on consumption versus investment. This was reinforced in McGillvray and Morrissey (2000b) where they concluded that aid leads to increases in expenditure not financed by the corresponding increase in revenue. More recently, Remmer (2004) also found that aid leads to overall increases in government spending.

How might these possible macro effects of aid negatively impact public institutions? The potential loss of competitiveness means lower exports and economic growth, fewer jobs, and increased dependence on external assistance. Resource volatility contributes to macro- economic instability, which complicates public policy making in vital areas such as budgeting and planning, and tilt public spending toward consumption rather than investment. These can exact a negative effect on the quality of the civil service, public services, and infrastructure, all indirectly undermining the ability of the state to transition from patrimonialism to a more ‘developmental’ path.

The rest of this paper addresses more direct and, we argue, more significant but less well- documented negative institutional effects of large volumes of aid. Much of the literature cited in this section describes dysfunctional economic outcomes but does not really explain them. To do so, we need to turn to institutional factors, which we begin to do in the next section.

4. Donor practices and institutional change
In addition to macroeconomic and fiscal effects, there are costs of aid related to the structures, practices, and procedures of the current international aid system. These include a longstanding and well-known list of common complaints about aid: volatility and uncertainty of ODA flows; fragmentation of donor efforts; project proliferation and duplication; conflicting or dominant donor agendas; competition for staff; and high administrative and oversight costs (Among many, see Cohen, 1992; Berg, 1993; Brautigam and Knack, 2004; Knack and Rahman, 2004; van de Walle, 2005). Birdsall (2004) lists many of these as the “seven deadly sins” of the aid business. Such practices are argued to have substantial costs for public administration. For instance, the proliferation of donors and projects constitute a substantial burden for the small number of qualified public officials, who spend much of their time attending to donor concerns and managing aid activities rather than promoting the development of the country—that is, when they do not exit altogether from the civil service to go work for better wages in donor and NGO organizations. Management of donor visits (“missions”) became such a problem in Tanzania, that the country was forced to declare a ‘mission holiday’, a four-month period when they take a break from visiting delegations to focus on budget preparation. Similarly, aid volatility and project proliferation complicate effective government control over budgets and development planning. Much, if not most, aid is not integrated into national budgets, thus posing real sustainability problems, and they are often implemented through parallel structures that cream the best staff from the civil service, and make government coordination of policy much more difficult.

 

Far from helping to develop effective state bureaucracies, certain aid practices can in fact serve to reinforce the patrimonial element within recipient governments at the expense of the legal-rational. Projects provide for the allocation of all sorts of discretionary goods to be politicized and patrimonialized, including expensive four-wheels drive cars, scholarships, decisions over where to place schools and roads, and so on. The common practice of paying cash ‘sitting fees’ for civil servants attending donor-funded workshops, where the daily rates can exceed regular monthly salaries, even turns training into a rent to be distributed. More broadly, when donor projects are poorly integrated into national budgetary processes, and not subject to much transparency or effective control, it is argued, they help sustain anti- developmental practices within the state apparatus. Because local officials are not included in policy planning, they often come to view aid projects as little more than a set of scarce private goods to be allocated. Aid dependence thus leads to a situation in which bureaucrats are often not rewarded for focusing on their core developmental functions but rather on getting money from donors. Technocrats, who are specialized in budget management and/or planning, say, are less rewarded than bureaucrats who are adept at interacting with donor organizations and accessing their resources. If these two are not part of the same skill set, the wrong kind of individual expertise may be rewarded and over time, real developmental capacities may atrophy within the administration.

Particularly pernicious for state institutions in Africa has been the combination of high aid flows and economic crisis, both sustained over a long period of time. As development policy has come to be dominated by repeated fiscal crises and driven by short-term adjustment and debt management, the patrimonial attraction of aid resources has been accentuated. In countries where power means access to state privileges and rents, and political systems are sustained by complex clientelist relationships, aid and the scarce goods it provides become all the more desirable for the political management of economic crises (van de Walle, 2001). In short, states who find it difficult to meet civil service payrolls are more likely to politicize aid funded sitting fees, per diems and scholarships to study abroad. Indeed, they will endeavor to turn aid into a mechanism to increase government consumption rather than public investment.

It also seems reasonable to surmise that the larger the relative aid flows, the more these problems are likely to be exacerbated. In many countries of sub-Saharan Africa, aid flows are such that aid dynamics simply dominate local development efforts. Moss and Subramanian (2005) identify 22 low-income countries, 16 of which are in sub-Saharan Africa where ODA inflows are equivalent to at least half of total government expenditure. In twelve poor countries, of which ten are African, the ratio of ODA to government expenditure was 75 percent or more. Looking at a slightly earlier period, Brautigam and Knack (2004) find roughly similar numbers of aid intensity. Because most of the concerns listed in this section are directly related to the way in which aid is delivered and administered, in theory at least, many of these are fixable through changes on the donor side. These shortfalls have long been identified and some efforts are underway to address them, such as donor pooling or using budget support instead of project aid (Eifert and Gelb, 2005). There are also several large institutional attempts to improve the efficiency of aid delivery, such as various programs by the OECD’s Development Assistance Committee or the Paris High Level Forum on Aid Harmonization and Alignment.

In practice, however, these inefficiencies exist because of very real political or bureaucratic constraints, and progress in reducing them has proven to be slow and uneven. In many ways, these problems seem to be actually getting worse; for instance, the number of distinct projects funded by donors has nearly tripled since 1995 (Roodman, forthcoming). Perhaps most worrying, there appear to be few incentives for either donors or recipients to change their practices. As Brautigam and Knack (2004) argue, “political elites have little incentive to change a situation in which large amounts of aid provide exceptional resources for patronage and many fringe benefits” (2004, p. 263). Moreover, until very recently, the government’s performance did not appear to affect whether or not it received aid, so there appeared to be little or no cost to misusing aid. Alesina and Weder (2002) actually find that high levels of corruption within recipient countries were positively correlated with aid flows through out the 1990s. On the other hand, incentives to improve aid effectiveness appear less important within donor organizations than other concerns, related to bureaucratic incentives within aid agencies, and to the importance of commercial, foreign policy and ideological objectives on the part of donor governments (Easterly, 2003).

5. The aid-revenue relationship
A third institutional effect of aid that has been posited in the literature concerns its impact on state revenues. This is an important issue since the ability of the state to collect revenues is critically linked to state capacity, while the central role of revenue collection in political development and state-building has long been accepted. Schumpeter was perhaps the first to argue that a country’s tax system fundamentally reflects its political institutions (Schumpeter, 1918/1991). Reliance on citizens for raising public revenues, as opposed to unearned income via offshore extraction or external assistance, is considered an essential ingredient to establishing accountability between the state and society.

The contemporary literature suggests that taxation is a useful indicator of state capacity. Because revenues are necessary to fund the state’s activities in a sustainable manner, the size and consistency of government revenues can tell us a lot about the level of capacity that exists within the state apparatus. This has long been argued in the aid literature. According to Kaldor (1963), the key determination of whether a state moves from aid dependency to economic self-sufficiency is the degree to which the state learns how to tax, thereby leading to a lessening of the need for aid. According to Bauer (1976), foreign aid displaces the processes of institutional maturation essential to development, including the capacity of the state to collect revenue. Azam et al. (1999) arrive at a similar conclusion, claiming that the ability of a state to remove itself from reliance on aid will depend on the degree to which the state engages in learning-by-doing in the public sector, a process that is greatly affected by the level of aid in relation to overall revenue and the initial institutional conditions.

Early empirical tax effort studies focused on a small set of variables considered the main determinants of tax effort, most commonly measured as tax revenue as a share of GDP (Lotz and Morrs, 1967; Celliah, 1971, Celliah et al. 1975; Tanzi, 1981). The tax effort literature typically considers the level of development and the economic structure as primary determinants of tax shares: GDP per capita, the degree of openness of the economy, the agricultural and/or industrial share of GDP, and in some cases population growth (Tanzi, 1992; Leuthold 1991; Stotsky and WoldeMariam, 1997; Ghura 1998). In the tax effort literature, foreign aid is generally expected to reduce tax shares since aid provides an alternative, non-earned source of revenue for governments in addition to tax revenue (Ghura, 1998; Remmer, 2004; Brautigam and Knack, 2004). Consequently, a government that receives significant amounts of aid is thought to have less incentive to tax and improve its tax administration. That is, foreign aid may be used as a substitute for domestic revenue mobilization whilst allowing the same level of expenditure (Heller, 1975; Kimbrough, 1986). Not only may aid inflows lead to lower tax effort, but they may also slow down the development of domestic institutions such as the tax administration in recipient countries (Brautigam and Knack, 2004).

A negative relationship between tax revenue and aid is certainly suggested by Figure 1 that shows the four-year averages of tax revenue (excluding trade taxes) as a share of GDP against the four-year averages of aid as a share of GNI for 55 low and lower middle income countries for 1972-1999, using the standard IMF data on government finances. The point of this figure is merely to show the simple correlation between tax collection and aid receipts. There are many cases of low aid and low tax (those toward the bottom left). There are also a moderate number of high aid and low tax (bottom right) and low aid-high tax countries (upper left). However, there are no incidences at all of high aid (>10% of GNI) and high tax (>18% of GDP). These thresholds are of course arbitrary and the figure does not imply any causation, but it does indicate that based on historical experience, high levels of both do not occur at the same time.

What is the econometric evidence on behalf of this relationship? It is actually more ambiguous than one might think. Leuthold (1991) examines the effect of the standard variables, controlling for the economic structure and the level of development, and aid on tax revenue using panel data. All the standard variables are statistically significant and have the anticipated signs, whereas foreign aid has the expected negative sign but is not significant. Using data from Stotsky and WoldeMariam (1997), Brautigam (2000, p. 48) found that 71 percent of the African countries with aid/GDP above 10 percent in 1995 had lower than expected tax effort. Ghura (1998) examined the determinants of tax revenue for 39 sub-Saharan African countries between 1985 and 1996 and found foreign aid to have a significant, negative impact on tax shares. Teera and Hudson (2004) use data for 116 developed and developing countries for 1975-1998, but find aid to be insignificant. A panel study of 120 middle- and low-income countries over the period 1970-1999 by Remmer (2004) finds that aid dependency reduces tax revenue mobilization. The dependent variable is the change in tax revenue as a share of GDP, and the explanatory variables include the standard variables and three different measures of aid dependency (aid/GNI, aid/government expenditures, and aid/imports). All three aid dependency measures are negatively related to tax shares although only aid as a share of imports is significant. When reducing the sample to the period 1980-1999, both the aid share in imports and the aid share in government expenditures have a significant negative impact on tax effort.

 

One of the most recent studies focusing on the revenue response to foreign aid inflows separates total net aid into grants and loans to test if the impact of grants on domestic revenue is different from that of (concessional) loans (Gupta et al. 2004). This study suggests that some governments may consider grants to be a free substitute for tax revenue. By contrast, loans must be repaid, which provides incentives for governments to at least maintain tax revenues at current levels if not to increase them (Brautigam, 2000). Gupta et al. use the standard variables controlling for the economic structure and level of development in a panel of 107 developing countries over the period 1970-2002, but augment the model by adding grants and loans separately and a corruption variable (the International Country Risk Guide (ICRG) corruption index), to test their separate impact on the aid- revenue relationship. In their baseline regression the overall effect of total aid (grants and loans) on domestic revenue is negative and significant. When total aid is split into grants and loans, grants have a significant, negative effect on revenue while loans have a significant, positive impact. In their extended model, corruption is found to reduce revenue. To further test the negative effect of corruption on revenues, Gupta et al. rank countries according to their score on the ICRG corruption index and test the impact of grants versus loans on revenues in a sample consisting of the relatively corrupt countries. Their results suggest that countries with weaker institutions are likely to suffer a larger negative impact of grants on revenue than countries with better institutions.

In addition to the cross-sectional time-series studies there are several country case studies for which the results are more mixed. A negative relationship between aid and domestic revenue mobilization was found in Pakistan (Franco-Rodriguez et al., 1998), Zambia (Fagernas and Roberts, 2004a) and Cote d’Ivoire (McGillivray and Outtara, 2003). By contrast, a positive relationship between aid and revenue collection was found in Indonesia (Pack and Pack, 1990), Ghana (Osei et al, 2003), and Uganda and Malawi (Fagernas and Roberts, 2004b, 2004c). Generally, thus, the literature finds a negative relationship between aid and revenue collection, but this is not a conclusive result. For all of the studies, there are considerable concerns about the quality of the data and also the sensitivity of the results to specification changes, which make firm causal conclusions about the aid-revenue relationship impossible. Nevertheless, the typical explanatory channels are the replacement of tax with aid in the short term and the disincentives and moral hazard faced by aid-dependent governments to build tax administration and institutional capacity over the long-term.

To further test the hypothesis that there is a negative causal relationship between high levels of aid and domestic tax effort, we would need reliable revenue data for a broad range of countries over an extended time period. Ideally, we would also have disaggregated revenue data to be able to strip out the possible differential effects of aid on trade taxes versus direct taxation and other forms of revenue collection. This would allow us to isolate the revenues which require more state capacity to collect. Unfortunately, much of this data does not exist, especially for the set of countries that have been highly aid-dependent (indeed, the lack of data itself suggests capacity gaps). In addition to data shortages, much of the existing fiscal data is of extremely poor quality, particularly among low-income countries. This limits our ability to empirically analyze the hypothesis with any degree of confidence and of course also raises caution about drawing firm conclusions from any of the data-driven assessments.

To summarize this section: a clear bivariate relationship appears to exist between high levels of aid and low levels of taxation. Poor data quality do not, however, allow us complete confidence that this relationship is not confounded by other factors that high aid-dependent countries have in common, such as low levels of economic activity and industrialization, that are also associated with poor revenue extraction. In addition, it is very hard to establish the exact nature of the link between state capacity and levels of state revenue. Intuitively, revenue generation is so central to state survival that one would think that states would not voluntarily abstain from collecting revenues it was able to collect. However, it is not completely possible to reject the alternative hypothesis that states choose not to seek revenues they have the capacity to collect because they are able to receive the equivalent revenues from foreign aid. To gain insights into this issue, it is perhaps useful to turn to an older political science literature on state building, which explores the impact of revenue collection on regime type.

6. Aid, Accountability, and the Political Regime
A political regime can be defined as the set of institutions which determine the nature of political power, and which structures the relationship between the government and the citizenry. A third and perhaps most critical set of negative institutional effects of aid can be identified as those that influence the political regime in a way that discourages the establishment of rational developmental states. The hypothesis here is that large sustained aid flows fundamentally alter the relationship between government elites and local citizens. Any kind of external financial flow changes the incentives faced by recipient government officials and their citizens, regardless of the precise nature of donor practices. That is, aid flows themselves, separate from particular inefficiencies in the aid system, can affect the evolution of state-society relations. If donors are providing the majority of public finance and governments are primarily accountable to those external agencies, then it may simply not be possible to also expect a credible social contract to develop between the state and its citizens. Using the current terminology, aid may undercut the very principles the aid industry intends to promote: ownership, accountability, and participation.

 

Large aid flows can result in a reduction in governmental accountability because governing elites no longer need to ensure the support of their publics and the assent of their legislatures when they do not need to raise revenues from the local economy, as long as they keep the donors happy and willing to provide alternative sources of funding. Although governments typically complain about conditions, it is still easier to manage donor demands than the slow and politically difficult task of building or improving domestic revenue collection. A reliance on aid as a substitute for local resources means the flow of revenues to the state is not affected by government efficiency, so there will be a tendency for governments to underinvest in developmental capacity. This moral hazard effect of aid dependence is borne out empirically, as high aid is associated with decreased quality of governance (for example, Brautigam and Knack, 2004; Knack, 2000). Heller and Gupta (2002) also argue that aid creates moral hazard because it reduces the incentive to adopt good policies and reform inefficient institutions, and thus weakens the government’s developmental performance and encourages rent-seeking.

The link between the loss of accountability and aid is particularly striking in Africa since the region combines sharp economic decline with a relatively high level of political stability, at least if the latter is defined as the ability of office holders to remain in power. Thus, since 1980, the average African leader has remained in power just under 12 years, more than three times longer than democratically elected leaders in the prosperous democracies of the West (van de Walle, 2001). The absence of accountability is then not a manner of speech, but a practical reality: it is literally true that African governments avoid accountability for their performance.

A long-term decline in governmental accountability also appears to have a direct impact on the degree of democracy prevailing in the system. Qualitatively, Moore (1998) has argued that countries which rely on a greater proportion of ‘unearned’ income will tend to be less democratic and have less effective institutional mechanisms and accountability. Simply put, the actions of such governments typically indicate they do not have to worry as much about maintaining legitimacy because they do not collect revenues from their own population. Guyer (1992) and others have made exactly such an argument about Nigeria, while more recently the negative links between aid dependency and low levels of democratic rule have been argued in a number of country case studies (see Hoffman and Gibson, 2005 on Tanzania; Hanlon, 1991 on Mozambique). In these dynamics, aid can be compared with a natural resource, such as oil, that provides unearned rents for the government. An earlier literature on ‘rentier’ states in the Middle East had indeed argued that oil resources have allowed governments to resist pressures to democratize (Anderson, 1987; Chaudhry, 1997).

Of course, it might be argued that aid comes with more strings attached than oil, and that donors can affect governmental behavior by setting conditions on their aid. However imperfectly, donors do not condone government corruption, incompetence or authoritarianism. Though inconsistently, donors have promoted democracy and better governance in Africa, at least since the end of the Cold War (van de Walle, 2001). Donors have also sought to explicitly promote accountability and participation, using intensified oversight of accounts and conditionality, such as the insistence on a poverty reduction strategy paper (PRSP) process, which subjects the national budget to multiple rounds of consultation with civil society groups. But the PRSPs and conditionality, and other donor processes (even if better enforced), cannot fundamentally replace government accountability towards its citizens in an equally legitimate way, no matter how well-intentioned or vigilant the donors.

A large flow of aid over a sustained period also can undermine popular participation. On the one hand, the assent of the population is less important to governments that receive large amounts of external support. They will devote less time and resources to explaining and defending policy decisions to their citizens, and will underfund the kinds of public institutions that encourage popular participation. On the other hand, the decline in ownership brought about by the externalization of decision-making necessarily results in departicipation. If citizens believe that their leaders respond to pressures from London, Paris or Washington, they will not devote as much time pressing demands on the local legislature and executive. More to the point, they may view the local legislature as the place to press for favors and patronage, rather than for policy outcomes, and this will once again tend to reinforce the patrimonial elements in the local political economy.

The African contemporary record is certainly compatible with such an interpretation. The region is characterized by strong presidential rule, as well as weak and pliant legislatures (van de Walle, 2001; Joseph, 2003; Barkan and Gibson, 2005), and frail civil society organizations. The absence of participatory checks on the executive branch of government in the region can tentatively at least in part be ascribed to the high volume of aid governments receive. Indeed, the relationship with the donors may well have served to reinforce tendencies which other structural factors were already creating. Many early observers had noted the low levels of participation in African political systems following independence (Kasfir, 1971; Collier, 1982), while the tendency of these countries to produce highly presidential political systems, with powerful executives and impotent legislatures has also long been related in the literature (Schatzberg, 2001; Bratton and van de Walle, 1997). The weakness of civil society in the region has also been described by numerous observers (Ndegwa, 1995; Harbeson et al., 1994).

To be sure, political systems with stronger traditions of both vertical and horizontal domestic accountability might have been affected differently by a large volume of aid. But in Africa’s post-colonial regimes, these aid flows inevitably enhanced an evolution already under way. Poorly integrated political communities with substantial ethnic fragmentation, and a small (or non-existent) middle class to buttress democratic rule were more likely to fall prey to authoritarian rulers relying on clientelism to remain in power. A small number of countries, such as Botswana, avoided the worst of these pitfalls (Acemoglu et al, 2003; Lewis, 1993), in part thanks to unusual leadership. But for most the post-independence period was characterized by the emergence of regimes that enjoyed little popular legitimacy and needed a combination of systematic clientelism and various repressive political instruments to remain in power. These types of regimes were comforted in these tendencies by steady increases in aid that seemed automatic throughout the 1970s and 1980s. Governments could undertake profoundly anti-developmental actions and not threaten their relationship with the western donors. Thus, the move to authoritarian and military government in the 1960s did not reduce aid. The replacement of independent civil service commissions and merit-based promotions by politicized presidential control of the civil service was similarly condoned, not least because donors found it convenient to rely on a large number of foreign experts, to palliate weaknesses in the civil service (Berg, 1993). Disastrous nationalization of private firms owned by foreigners for the benefit of political cronies close to the president in countries as diverse as Nigeria and Zaire did not prevent aid to continue its upward trajectory (See Rood, 1976; Callaghy, 1984 on Zaire; Biersteker, 1987 on Nigeria).

A comparison with the historical experience in the West is instructive here. Much scholarly work has closely linked democratic development to the evolution of taxation (see Ross, 2004 for an excellent overview). Historians of the emergence of strong democratic states in the West emphasize the link between the progressive growth of democratic and accountable government, on the one hand, and the emergence of a state apparatus that had both the capacity and the legitimacy to extract an increasing amount of revenue from society, on the other. In an influential essay, North and Weingast (1989) showed that the emergence of Parliamentary sovereignty in Britain with the Glorious Revolution of 1688 dramatically increased the ability of the British government to raise taxes, and ensured the country’s military and economic success in the 18th and 19th century.

Ardant (1975) showed that the European states that were able to finance and then win wars were the states that were able to build their extractive capacity, but also to gain the assent of their populations, often by extending political rights. In Tilly’s famous aphorism, “the state made war and war made the state”, the need to finance wars motivated states both to build their extractive capacity, but also to maintain their own legitimacy (Tilly, 1975, 1985). In short, taxation is important because it is essential to democratic governance, but also because it holds the key to state building and state survival. The comparison with the low income countries in Africa is instructive, since they did not have to fight international wars to ensure their survival (Herbst, 2000; Jackson and Rotberg, 1982).

In the 20th and 21st centuries, the basic functions of a developmental low-income state are to raise revenues and make effective expenditures in order to promote development. Successful developing countries—Korea, Taiwan, Botswana—have typically been solidly extractive states, with above average tax effort ratios. Not all of these states have enjoyed democratic governance during the early phases of their development, but almost invariably, their governments enjoyed substantial political legitimacy, and none were highly repressive. Of course, most of the low–income states of Africa typically have low tax effort ratios. In recent years, donors have financed a sharp increase in social services provision in sub- Saharan Africa, notably in health and education. Historically, such increases in provision have been the hallmark of democratic governments, or at least of governments facing substantial participatory pressures. Thus, in the West, the rise of public education coincided in general terms with the introduction of the electoral franchise. Pressures from below encouraged governments who wanted to remain in power to provide more services to its citizens. In most African countries, however, the pressures have been external. In fact, a substantial proportion of the national development effort is not integrated into the national budget and does not concern the government. It is not uncommon for donors to fund over half of the country’s public investment budget, while foreign NGOs with their local partners can be providing from a third to half of the social services available to African citizens in some countries (Semboja and Therkildsen, 1995). Indeed, somewhat ironically, African governments have found it politically convenient to blame the donors and NGOs for unpopular sectoral policies, poor social services and negative economic outcomes, as if these were not among their core responsibilities.

 

There are similar differences in the rise of civil society. In the West, an emerging middle class sought to build a counter-weight to the state and its organizations, and the result was a wide variety of membership organizations, unions and clubs with an independent basis of power, that with time were able to increase the accountability of the central state (Hall,1995). In Africa, the absence of economic growth long undermined the development of an indigenous independent civil society. In recent years, there has been a flowering of small non-state actors and some of them were instrumental in the emergence of democratic movements that did topple some authoritarian governments in the early 1990s (Bratton and van de Walle, 1997; Harbeson et al., 1994). Yet, there remain very few membership organizations in the region, and many of the bigger NGOs that have emerged are mostly funded by the donors, typically to help undertake donor initiatives in the social sectors. Because these organizations receive funding from the donors, they are less likely to seek to build up their own memberships, or autonomy. Because they help donors implement projects that governments fail to undertake, they actually help governments escape accountability for their developmental failures.

As a result of these different dynamics, governments have escaped accountability and have been allowed to focus their resources on non-developmental expenditures that help them remain in office. Van de Walle (2001) shows that there have been substantial increases in the size of defense expenditures in the region for instance, and the number and size of public offices such as parliamentary bodies, ministerial cabinets, national commissions, and provincial governments have steadily risen through two decades of economic crisis, even as the share of aid in the funding of development has sharply increased. Paradoxically, as a result, in Africa, the extension of social services has often been accompanied by a decline of participation, low governance quality and an increase in clientelistic behavior. If we agree with Fox (1994) and others that the process of democratic consolidation in low-income states requires a transition from clientelism to citizenship, in which governments engage in participatory contractual exchange relationships with the population, then donors efforts may have paradoxically negative effects on citizen-government relations in the region.

7. Conclusions
Our review of the literature suggests that there are reasons to believe that a large and sustained volume of aid can have negative effects on the development of good public institutions in low income countries. We have reviewed different bodies of literature that suggest that the current aid system may have undercut incentives for revenue collection and negatively affected public governance in Africa. In addition, we have examined a political science literature that finds both anti-development governance patterns across most of sub- Saharan Africa and strong historical evidence that revenue generation is central to the idea of accountability and the establishment of state institutions. Combined, they suggest that aid may undermine the development of effective state structures.

There are many gaps in the data needed to prove these tentative claims. Also, state revenues are an imperfect indicator for state capacity, since states are able to get revenues in many different ways, only some of which involve much extractive capacity. Nonetheless, the analysis does suggest that an aid-institutions paradox, whereby high levels of aid can have a negative effect on local institutions, is a potentially serious concern. Given the possibility for substantially more aid flowing to Africa in the near future, scope for such a harmful dynamic is likely to be exacerbated. A quarter of a century ago, the World Bank issued its so-called Berg Report (World Bank,1981), which called for a doubling of aid to address its many economic and social problems. It must be particularly distressing to the development community how many of those problems persist, despite the fact that increases in aid were considerably higher than those hoped. This fact alone should encourage skepticism about the current proposals that a sharp increase in aid volume will have the intended effects in the region. It is not at all clear that the current aid practices – with the negative effects on institutions described above – will or can be reformed. But, as we have argued, there are good reasons to believe that high levels of aid over a prolonged period is likely to have negative institutional effects, at the very least, if the current aid delivery modalities are not substantially reformed.

How much is too much aid? We have studiously avoided this difficult question until the end of the paper. The same Berg (1997, 2000) suggested that aid starts to have negative effects on local institutions when aid flows reach 5 percent of GDP, which would mean that the overwhelming majority of states in the region are negatively affected. A more recent and thorough review of aid absorption (Clemens and Radelet, 2003) find the ‘saturation point’ (where additional aid would produce zero economic impact) highly dependent on local conditions, but ranging from 15-45 percent of GDP. Surely, the incentive dynamics raised by this easy come into play well before such an extreme level is reached.

Our analysis is in no way meant to disparage the desirability of general increases in aid flows, however, or suggest that additional aid could not necessarily be spent without producing the negative institutional effects. Our findings do not cover a range of activities that might be donor financed which could have positive institutional effects, such as debt relief, peacekeeping, and regional security arrangements. Similarly, we join other analysts who have advocated substantial increases in funding for regional and global public goods, such as agricultural research or anti-malaria research. All the available evidence on the likely impact of the eradication of endemic diseases in the region suggest current funding levels to be inadequate (Ferroni and Mody, 2002), and a substantially larger flow of resources would be unlikely to have the kinds of negative institutional effects described here.

In sum, it seems likely that the extra public dollars now being proposed for traditional development assistance might well be better spent for other types of assistance that would in the long run have a greater impact on the development of the region. However, an historical view of the complex evolution of state institutions suggests that not only are they critical to producing developmental outcomes, but that donors should be unambiguously aware that their assistance can have perverse effects on some of the very outcomes they hope to encourage.

 

References:
Acemoglu, Daron, Simon Johnson, and James Robinson. 2004. “Institutions As the

Fundamental Cause of Long-Run Growth.” NBER Working Paper

10481.Cambridge, MA: National Bureau of Economic Research.

Acemoglu, Daron, Simon Johnson, and James Robinson. 2003. “African Success Story: Botswana.” In Dani Rodrik (ed.), Search of Prosperity: Analytical Narratives on Economic Growth. Princeton, NJ: Princeton University Press.

Adam, Christopher S. and David L. Bevan. 2003. “Aid, Public Expenditure and Dutch Disease.” CSAE Working Paper 2003-02. Oxford, UK: Center for the Study of African Economies.

Aguierre, Carlos A., Peter S. Griffith, and M. Zuhtu Yucelik. 1981. “Tax Policy and Administration.” IMF Occasional Paper 8. Washington, DC: International Monetary Fund.

Alesina, Alberto and Beatrice Weder. 2002. “Do Corrupt Governments Receive less Foreign

Aid?” American Economic Review 92(4).

Anderson, Lisa. 1987. “The State in the Middle East and North Africa.” Comparative

Politics 20(1).

Ardant, Gabriel. 1975. “Financial Policy and Economic Infrastructure of Modern States and Nations.” In Charles Tilly (ed.), The Formation of National States in Western Europe. Princeton, NJ: Princeton University Press.

Atingi-Ego, Michael. 2005. “Budget Support, Aid Dependency, and Dutch Disease: The Case of Uganda.” Paper presented at Practitioners’ Forum on Budget Support, Cape Town, South Africa, May 5-6.

Azam, Jean-Paul, Shantayanan Devarajan, and Stephen A. O'Connell. 1999. “Aid Dependence Reconsidered.” World Bank Policy Research Working Paper 2144. Washington, DC: World Bank.

Bauer P.T. 1976. Dissent on Development: Studies and Debates in Development Economics.

Cambridge, MA: Harvard University Press.

Berg Elliott. 1997. “Dilemmas in Donor Aid Strategies.” In Gwin Catherine. and Joan M.

Nelson (eds.), Perspectives on Aid and Development. Washington DC: Overseas Development Council.

Berg, Elliot. 2000. “Aid and Failed Reform: the Case of Public Sector Management.” In Finn Tarp (ed.), Foreign Aid and Development: Lessons Learnt and Directions for the Future. London, UK: Routledge.

Berg, Elliott. 1993. Rethinking Technical Co-operation. New York, NY: United Nations Development Program.

Biersteker, Thomas. 1987. Multinationals, the State and Control of the Nigerian Economy.Princeton, NJ: Princeton University Press.

Bird, Richard M., Jorge Martinez-Vazquez and Benno Torgler. 2004. “Societal Institutions and Tax Effort in Developing Countries.” International Studies Program Working Paper 04-06. Atlanta, GA: Andrew Young School of Policy Studies, Georgia State University.

Birdsall, Nancy and Subramanian. 2004. “Saving Iraq From Its Oil.” Foreign Affairs, July/August.

Birdsall, Nancy. 2004. “Seven Deadly Sins: Reflections on Donor Failings.” CGD Working Paper 50. Washington DC: The Center for Global Development.

Birdsall, Nancy. 2005. “Underfunded Regionalism in the Developing World.” In Inge Kaul and Pedro Conceição (eds.), The New Public Finance: Responding to Global Challenges. New York, NY: Oxford University Press.

Bratton, Michael and Nicolas van de Walle. 1997. Democratic Experiments in Africa: Regime Transitions in Comparative Perspective. New York, NY: Cambridge University Press.

Brautigam, Deborah and Knack, Steven. 2004. “Foreign Aid, Institutions, and Governance in Sub-Saharan Africa.” Economic Development and Cultural Change 52, January.

Brautigam, Deborah. 2000. Aid Dependence and Governance. Stockholm, Sweden: Almqvist & Wiksell.

Bulir, Ales and Lane, Timothy. 2002. “Aid and Fiscal Management.” In Sanjeev Gupta, Benedict Clements, and Gabriela Inchauste (eds.), Helping Countries Develop. The Role of Fiscal Policy. Washington, DC: International Monetary Fund.

Burnside, Craig and David Dollar. 2000. “Aid, Policies and Growth.” American Economic Review 90(4).

Callaghy, Thomas M. 1988. “The State and the Development of Capitalism in Africa: Theoretical, Historical and Comparative Reflections.” In Donald Rothchild and Naomi Chazan (eds.), The Precarious Balance: State and Society in Africa. Boulder, CO: Westview Press.

Callaghy, Thomas M. 1984. The State Society Struggle: Zaire in Comparative Perspective. New York, NY: Columbia University Press.

Celliah, Raja J. 1971. “Trends in Taxation in Developing Countries.” IMF Staff Papers 18(2). Washington, DC: International Monetary Fund.

Celliah, Raja J., H. J. Baas, and Margaret R. Kelly. 1975. “Tax Ratios and Tax Effort in Developing Countries, 1969-1971.” IMF Staff Papers 22(1). Washington, DC: International Monetary Fund.

Chabal, Patrick and Jean Pierre Daloz. 1999. Africa Works. London, UK: James Currey. Chaudhry, Kiren Aziz. 1997. The Price of Wealth: Economics and Institutions in the Middle

East. New York, NY: Cornell University Press.

 

Clemens, Michael and Steven Radelet. 2003. “Absorptive Capacity: How Much Is Too Much?” In Steven Radelet, Challenging Foreign Aid: A Policymaker’s Guide to the Millennium Challenge Account. Washington, DC: Center for Global Development.

 

Cohen, John. 1992. “Foreign Advisors and Capacity Building: the Case of Kenya.” Public

Administration and Development 12.

 

Collier, Paul and Anke Hoeffler 2004. “Greed and Grievance in Civil War.” Oxford

Economic Papers 56.

 

Collier, Ruth Berins. 1982. Regimes in Tropical Africa: Changing Forms of Supremacy,

1945-1975. Berkeley, CA: University of California Press

 

Commission for Africa. 2005. Our Common Interest. Report of the Commission for Africa.

<http://www.commissionforafrica.org/english/report/thereport/english/11-03-

05_cr_report.pdf>

 

De Long, J. Bradford and Barry Eichengreen. 1991. “The Marshall Plan: History’s Most Successful Structural Adjustment Program.” NBER Working Paper 3899. Cambridge, MA: National Bureau of Economic Research.

 

Devarajan, Shantayanan, Margaret J. Miller, and Eric V. Swanson. 2002. “Goals for Development: History, Prospects and Costs.” World Bank Policy Research Working Paper 2819. Washington, DC: World Bank.

 

North, Douglass and Barry Weingast. 1989. “Constitutions and Commitment: The Evolution of Institutions Governing Public Choice in Seventeenth-Century England.” Journal

of Economic History 49(4).

 

Easterly, William. 2005. “Reliving the ‘50s: the Big Push, Poverty Traps, and Takeoffs in Economic Development.” CGD Working Paper 65. Washington, DC: Center for Global Development.

 

Easterly, William. 2003. “The Cartel of Good Intentions: The Problem of Bureaucracy in

Foreign Aid.” Policy Reform 5.

 

Eifert, Benn and Alan Gelb. 2005. “Improving the Dynamics of Aid: Towards More Predictable Budget Support.” Paper presented at Practitioners’ Forum on Budget Support, Cape Town, South Africa, May 5-6.

 

Englebert, Pierre. 2000. State Legitimacy and Development in Africa. Boulder, CO: Lynne

Rienner Publishers.

 

Evans, Peter. 1995. Embedded Autonomy: States and Industrial Transformation. Princeton, NJ: Princeton University Press.

 

Fagernas, Sonja and John Roberts. 2004a. “The Fiscal Effects of Aid in Zambia.” ESAU Working Paper 10. London, UK: Economics and Statistics Analysis Unit, Overseas Development Institute.

 

Fagernas, Sonja and John Roberts. 2004b. “The Fiscal Effects of Aid in Malawi.” ESAU Working Paper 7. London, UK: Economics and Statistics Analysis Unit, Overseas Development Institute.

 

Fagernas, Sonja and John Roberts. 2004c. “The Fiscal Effects of Aid in Uganda.” ESAU Working Paper 9. London, UK: Economics and Statistics Analysis Unit, Overseas Development Institute.

 

Ferroni, Marco and Ashoka Mody, (eds.). 2002. International Public Goods: Incentives, Measurement and Financing. Boston, MA: Kluwer Academic Publishers.

 

Fox, Jonathan. 1994. “The Difficult Transition from Clientelism to Citizenship.” World

Politics 46(2).

 

Franco-Rodriguez, Susana, Mark McGillivray, and Oliver Morrissey. 1998. “Aid and the

Public Sector in Pakistan: Evidence with Endogenous Aid.” World Development,

26(7).

 

Franco-Rodriguez, Susana. 2000. “Recent Advances in Fiscal Response Models with an

Application to Costa Rica.” Journal of International Development 12(3).

 

Ghura, Dhaneshwar. 1998. “Tax Revenue in Sub-Saharan Africa: Effects of Economic Policies and Corruption.” IMF Working Paper 98/135. Washington, DC: International Monetary Fund.

 

Gupta, Sanjeev, Benedict Clements, Alexander Pivovarsky, and Erwin Tiongson. 2004. “Foreign Aid and Revenue Response: Does the Composition of Foreign Aid Matter?” In Sanjeev Gupta, Benedict Clements and Gabriela Inchauste (eds.),. Helping Countries Develop. The Role of Fiscal Policy. Washington, DC: International Monetary Fund.

 

Guyer, Jane. 1992. “Representation Without Taxation: An Essay on Democracy in Rural

Nigeria, 1952-1990.” African Studies Review 35(1).

 

Haggard, Stephan. 1990. Pathways from the Periphery. Ithaca, NY: Cornell University

Press.

 

Hall, John, A. (ed.). 1995. Civil Society: Theory, History and Comparison. Cambridge, UK: Polity Press.

Hanlon, Joseph. 1991. Mozambique: Who Calls the Shots? Indiana University Press. Harbeson, John W., Donald Rothchild and Naomi Chazan (eds.). 1994. Civil Society and the

State in Africa. Boulder, CO: Lynne Rienner Press.

 

Heller, Peter S. and Sanjeev Gupta. 2002. “Challenges in Expanding Development Assistance.” IMF Policy Discussion Paper 02/5. Washington DC: International Monetary Fund.

 

Heller, Peter, S. 2005. “Pity the Finance Minister”: Issues in Managing a Substantial Scaling Up of Aid Flows.” IMF Working Paper 05/180. Washington, DC: International Monetary Fund.

 

Heller, Peter S. 1975. “A Model of Public Fiscal Behaviour in Developing Countries: Aid, Investment, and Taxation.” American Economic Review 65(3).

 

Herbst, Jeffrey. 1990. “The Structural Adjustment of Politics in Africa.” World Development

18(7).

 

Herbst, Jeffrey. 2000. States and Power in Africa: Comparative Lessons in Authority and

Control. Princeton, NJ: Princeton University Press.

 

Hoffman, Barak, D. and Clark C. Gibson. 2005. “Fiscal Governance and Public Services: Evidence from Tanzania and Zambia.”

<http://www.polisci.ucla.edu/wgape/papers/9_GibsonHoffman.pdf>

 

Jackson, Robert H. and Carl G. Rotberg. 1982. “Why Africa’s Weak States Persist: The

Empirical and the Juridical in Statehood.” World Politics 35(1).

 

Joseph, Richard A. 1997. “Democratization in Africa After 1989: Comparative and

Theoretical Perspectives.” Comparative Politics 29(3).

 

Joseph, Richard A. “Africa: States in Crisis.” Journal of Democracy 14(3).

 

Kaldor, Nicholas. 1963. “Will Underdeveloped Countries Learn to Tax?” Foreign Affairs, January.

 

Kapstein Ethan. 2004. “Behavioral Foundations of Democracy and Development.” CGD Working Paper 52. Washington, DC: Center for Global Development.

 

Karl, Terry Lynn. 1997. The Paradox of Plenty. Oil Booms and Petro States. Studies in

International Political Economy 26.

 

Kasfir, Nelson. 1971. The Shrinking Political Arena: Participation and Ethnicity in Africa

Politics. Berkeley, CA: University of California Press.

 

Khan, Haider, Ali and Eiichi Hoshino. 1992. “Impact of Foreign Aid on the Fiscal Behavior of LDC Governments,” World Development, 20(10).

 

Killick, Tony. 2004. “What Drives Change in Ghana? A Political-Economy View of

Economic Prospects.” Mimeo.

 

Kimbrough, Kent P. 1986. “Foreign Aid and Optimal Fiscal Policy.” Canadian Journal of

Economics 19(1).

 

Knack, Stephen, and Aminur Rahman. “Donor Fragmentation and Bureaucratic Quality in Aid Recipients.” World Bank Policy Research Working 3186. Washington, DC: World Bank.

 

Knack, Stephen. 2000. “Aid Dependence and the Quality of Governance: a Cross-Country Empirical Analysis.” World Bank Policy Research Working Paper 2396. Washington DC: World Bank.

 

Leuthold, Jane H. 1991. “Tax Shares in Developing Countries: A Panel Study.” Journal of

Development Economics 35(1).

 

Levine, Ruth and the What Works Working Group with Molly Kinder. 2004. Millions Saved: Proven Successes in Global Health. Washington, DC: Center for Global Development.

 

Levy, Brian and Sahr Kpundeh, (eds.). 2005. State Capacity in Africa: New Approaches, Emerging Lessons. Washington, DC: World Bank Institute.

 

Lewis, Stephen. 1993. “Policy Making and Economic Performance: Botswana in Comparative Perspective.” In Stephen John Stedman (ed.), Botswana: the Political Economy of Democratic Development. Boulder, CO: Lynne Rienner Publishers.

 

Lotz, Jorgen R. and Elliott R. Morss. 1967. “Measuring ‘Tax Effort’ in Developing

Countries.” IMF Staff Papers, 14(3). Washington, DC: International Monetary Fund.

 

McGillivray, Mark and Bazoumana Outtara. 2003. “Aid, Debt Burden and Government Fiscal Behaviour in Cote d’Ivoire.” CREDIT Research Paper 03/05. Nottingham, UK: Centre for Research in Economic Development and International Trade, University of Nottingham.

 

McGillvray, Mark and Oliver Morrissey. 2000a. “A Review of Evidence on the Fiscal Effects of Aid.” CREDIT Research Paper No. 01/13. Nottingham, UK: Centre for Research in Economic Development and International Trade, University of Nottingham.

 

McGillvray, Mark and Oliver Morrissey. 2000b. “Aid Illusion and Public Sector Fiscal Behaviour.” CREDIT Research Paper No. 00/9. Nottingham, UK: Centre for Research in Economic Development and International Trade, University of Nottingham.

 

Moore, Mick. 1998. “Death Without Taxes: Democracy, State Capacity, and Aid Dependence in the Fourth World.” In Gordon White and Mark Robinson (eds.), Towards a Democratic Developmental State. Oxford, UK: Oxford University Press.

 

Moss, Todd and Arvind Subramanian. 2005. “After the Big Push? Fiscal and Institutional Implications of Large Aid Increases.” CGD Working Paper 71. Washington, DC: Center for Global Development.

Musgrave, Richard A. 1969. Fiscal Systems. London, UK: Yale University Press. Ndegwa, Stephen. 1995. The Two Faces of Civil Society: NGOs and Politics in Africa.

West Hartford, CT: Kumarian Press.

 

Ndulu, Benno J. and Stephen A. O’Connell. 1999 “Governance and Growth in Sub-Saharan

Africa.” Journal of Economic Perspectives 13(3).

 

Nkusu, Mwanza. 2004. “Financing Uganda’s Poverty Reduction Strategy: Is Aid Causing More Pain than Gain?” IMF Working Paper 04/170. Washington DC: International Monetary Fund.

 

O’Connell, Stephen, A. and Charles C. Soludo. 2001. “Aid Intensity in Africa.” World

Development 29(9).

 

Odedokun, Matthew. 2003. “Economics and Politics of Official Loans versus Grants.

Panoramic Issues and Empirical Evidence.” WIDER Discussion Paper No. 2003/04. Helsinki, Finland: United Nations University World Institute for Development Economics Research.

 

OECD-DAC database. 2005. <http://www.oecd.org/dataoecd/50/17/5037721.htm>

 

OED. 2005. Capacity Building in Africa: An OED Evaluation of Bank Support.

Washington, DC: World Bank.

 

Osei, Robert, Oliver Morrissey, and Tim Lloyd. 2003. “Modelling the Fiscal Effects of Aid: An Impulse Response Analysis for Ghana.” CREDIT Research Paper 03/10. Nottingham, UK: Centre for Research in Economic Development and International Trade, University of Nottingham.

 

Pack, Howard and Janet Pack. 1990. “Is Foreign Aid Fungible? The Case of Indonesia.” The

Economic Journal 100(399).

 

Pradhan, Sanjay. 1996. “Evaluating Public Spending: A Framework for Public Expenditure

Reviews.” World Bank Discussion Paper 323. Washington DC: World Bank.

 

Radelet, Steven. 2003. Challenging Foreign Aid: A Policymakers Guide to the Millennium

Challenge Account. Washington, DC: Center for Global Development.

 

Rajan, Raghuram G., and Arvind Subramanian. 2005. “What Undermines Aid’s Impact on Growth?” IMF Working Paper 05/126. Washington, DC: International Monetary Fund.

 

Remmer, Karen. 2004. “Does Foreign Aid Promote the Expansion of Government?”

American Journal of Political Science, 48(1).

 

Reno, William. 1998. Warlord Politics and African States. Boulder, CO: Lynne Rienner

Publishers.

 

Rodrik, Dani (ed.). 2003. In Search of Prosperity: Analytical Narratives on Economic

Growth. Princeton, NJ: Princeton University Press.

 

Rood, Leslie L. 1976. “Nationalization and Indigenization in Africa.” Journal of Modern

African Studies 14.

Roodman, David. Forthcoming. “Aid Project Proliferation: Theory and Evidence.” Draft. Ross, Michael. 1999. “The Political Economy of the Resource Curse.” World Politics 51. Ross, Michael. 2004. “Does Taxation Lead to Representation?” British Journal of Political

Science 34.

 

Sachs, Jeffrey. 2005. The End of Poverty: Economic Possibilities for our Time. New York, NY: Penguin Press.

 

Sandbrook, Richard. 1992. The Politics of Africa's Economic Recovery. Cambridge, UK: Cambridge University Press

 

Schatzberg, Michael. 2001. Political Legitimacy in Middle Africa: Father, Family, Food.

Bloomington, IN: Indiana University Press.

 

Schumpeter, Joseph, A. 1918/1991. “The Crisis of the Tax State.” In Richard A. Swedberg (ed.), Joseph A. Schumpeter. The Economics and Sociology of Capitalism. Princeton, NJ: Princeton University Press.

 

Semboja, Joseph and Ole Therkildsen (eds.). 1995. Service Provision Under Stress in East

Africa. London, UK: James Currey.

 

Stotsky, Janet G. and Asegedech WoldeMariam. 1997.”Tax Effort in Sub-Saharan Africa.” IMF Working Paper 97/107. Washington, DC: International Monetary Fund.

 

Tanzi, Vito. 1981. “A Statistical Evaluation of Taxation in Sub-Saharan Africa.” IMF Occasional Paper 8. Washington, DC: International Monetary Fund.

 

Tanzi, Vito. 1992. “Structural Factors and Tax Revenue in Developing Countries: A Decade of Evidence.” In Ian Goldin and Alan L. Winters (eds.), Open Economies: Structural Adjustment and Agriculture. New York: Cambridge University Press.

 

Teera, Joweria M. and John Hudson. 2004. “Tax performance: A Comparative Study.”

Journal of International Development 16(6).

 

Tilly, Charles. 1975. The Formation of National States in Western Europe. Princeton, NJ: Princeton University Press.

 

Tilly, Charles.1985. “War Making and State-Making as Organized Crime.” In Peter Evans, Dietrich Rueschemeyer and Theda Skocpol (eds.), Bringing the State Back In. Cambridge, UK: Cambridge University Press.

 

van de Walle, Nicolas. 2001. African Economies and the Politics of Permanent Crisis, 1979-

1999. Cambridge, UK: Cambridge University Press.

 

van de Walle, Nicolas. 2005. Overcoming Stagnation in Aid-Dependent Countries.

Washington, DC: Center for Global Development.

White, Howard. 1998. Aid and Macro-Economic Performance. Basingstoke: Macmillan. World Bank. 1981. Accelerated Development in Sub-Saharan Africa. Washington DC:

World Bank

 

World Bank. 1998. Assessing Aid – What Work, What Doesn’t, and Why. Washington, DC: World Bank.

 

World Bank. 2005. World Development Indicators. Washington, DC.

 

Younger, Steven. 1992. “Aid and the Dutch Disease: Macroeconomic Management When

Everybody Loves You.” World Development 20(11).

 

Zedillo panel. 2001. Report of the High-Level Panel on Financing for Development. New

York, NY: United Nations.

---

by CASADE