Labour Markets, Natural Resources and Poverty in East Africa
On completion of his thesis in 1976, Paul was appointed to a University Lectureship and a Fellowship at Keble College, where he tutored a number of future academic economists, most notably Tim Besley.
His research at Keble might have seemed at first to be completely unrelated to his doctoral work. He was commissioned to write a World Bank report on labour markets and poverty in Tanzania. This provided his first experience of fieldwork in Africa, and his first opportunity to conduct applied economics research, using household survey data at a time when such data were still unfashionable. The Tanzanian study was followed by a similar commission in Kenya, this time with Deepak Lal, and a study of Dutch Disease in Nigeria funded by the International Labour Organisation. In fact, the academic publications that resulted from this work (Labour and Poverty in Rural Tanzania (Collier et al. 1986), which won the Edgar Graham Prize, and Labour and Poverty in Kenya, 1900-1980 (Collier and Lal 1986)) were very much connected to international trade theory, stressing the importance of location and capability-based comparative advantage in explaining economic outcomes. The insight that theory explaining variation in the wealth of nations can be applied at a different scale, in order to explain variation in the wealth of peasant farmers, is important to the broader discipline, not just to development economics.[215]Despite the many differences between the studies of Kenya and Tanzania and the earlier work on customs unions, there are some similarities in the style of the economic analysis. In both cases, existing theory was extended by the addition of more moving parts, and this sometimes led to radically different policy conclusions. One example is the study of migration and unemployment in Tanzania (see Collier 1979b). The starting point for the analysis is the Harris-Todaro model, in which a persistent gap between urban formal-sector wages and rural wages creates an incentive for rural workers to migrate to cities in order to look for work (see Harris and Todaro 1970).
In equilibrium, there is a pool of urban workers who are either unemployed or underemployed in the informal sector, but the prospect of high-wage formal-sector employment keeps them from returning home to work. In such a world, a government policy that expands the urban formal sector will also increase the number of people who chose to migrate and are unemployed or work in the informal sector. Paul extended the model by making explicit the distinction between unemployment and informal sector employment, and by allowing for a heterogeneous labour force, distinguished by characteristics, such as education, that affect their ability to migrate. In the extended model, an expansion of the formal sector can induce a more-than-proportional increase in the reservation wage of migrants, and therefore a net fall in the total number of people who are unemployed or in the informal sector. Analysis of evidence from Tanzania between 1969 and 1975 indicates that the expansion of the formal sector did indeed tend to depress the total number of unemployed and informally employed migrants.The work on Kenya and Tanzania also included one of the first serious studies of the dynamics of African income distribution (see Collier and Lal 1984). This study showed that remittances from urban migrants to their rural families tended to mitigate rural income inequality. Remittances and other sources of non-farm income were particularly important as a source of funds for investment, given the absence of credit markets for small farmers. Here, in Kenya and Tanzania, the work showed an acute awareness of the ways in which individual markets and government policies had failed (see also Collier 1983), but the analysis of radical State failure that was to be so important in later work was still far over the horizon.
The research in Kenya and Tanzania was conducted during the middle of the coffee boom in the late 1970s. Coffee is the countries’ main cash crop, and a steep rise in world prices led to a large increase in coffee revenue, at least some of which made its way into the hands of peasant farmers.
The visible consequences of the boom, such as more bicycles on the street and better quality rooves, stimulated Paul's interest in the way in which economic policy might enhance or detract from the potential benefits of such a windfall. The offer of World Bank funding for further fieldwork in Kenya and Tanzania prompted him to make a career-changing decision. He resigned his Fellowship at Keble, which came with substantial undergraduate teaching responsibilities, in order to take up a less well-paid Fellowship at St Antony's College, with no such responsibilities and therefore the freedom to travel to East Africa for long periods. Since then, his academic life has been devoted to research, engagement with policymakers, and postgraduate education; those undergraduates who knew him as a tutor and lecturer are all now in their 50s.A large part of the fieldwork in East Africa involved tracking down official government publications that were not available abroad. Detective work and a certain amount of arm-twisting were required to locate data that had been stored in a single hard copy or tape. The efforts were not always successful— for example, the 1968 Tanzanian household budget survey had disappeared without trace—but the work provided the empirical foundation for two monographs co-authored with David Bevan and Jan Gunning, one with a microeconomic focus (Peasants and Governments (Bevan et al. 1989)) and one with a macroeconomic focus (Controlled Open Economies (Bevan et al. 1994)). Using East Africa as a case study, this work explores the consequences of government controls in small open economies that are subject to large external shocks.
Peasants and Governments is notable for its use of household panel data: to my knowledge, the only other panel dataset in development economics that dates back to the 1970s is from India (see Baulch 2011).[216] The construction of the panel involved locating and interviewing households who had taken part in previous government surveys.
The method was not perfect—for example, there was no way to control for the sample selection bias due to households who had migrated and were therefore untraceable—but it is among the earliest examples of a research method that has since become ubiquitous in development economics. By using panel data, Paul and his colleagues could track the response of small farmers to changing economic conditions during the coffee boom. Government policy in Kenya differed markedly from that in Tanzania, so the study also sheds light on the consequences of different policy regimes. In Kenya, the prices of most locally consumed goods were determined by the market, and although there was a wedge between the farmgate coffee price and the international price, the Coffee Board of Kenya, which acted as a middleman, passed on most of the international price increase to farmers. The panel data evidence indicated that farmers saved most of their temporary windfall, demonstrating some degree of forward-looking behaviour. Most of the saving was in the form of fixed capital investment (bicycles and rooves being two of the most visible examples), but large variations in the rates of return to different types of investment suggested the existence of information asymmetries or some other type of capital market failure. In Tanzania, most prices were controlled and the supply of goods was rationed, inhibiting the production of cash crops and delivering outcomes much further from the first best than in Kenya.Controlled Open Economies is among the first substantial empirical studies in development macroeconomics that take dynamics seriously, working through some of the macroeconomic implications of the results in Peasants and Governments. The analysis of Kenya focuses on the macroeconomic consequences of the coffee farmers' windfall income. As elaborated in Peasants and Governments, farmers saved in the form of physical assets, lacking access to reliable financial institutions. Much of this physical capital was not internationally tradeable, so there was a steep rise in the relative price of non- tradeable capital goods, in other words, a form of Dutch Disease.
The analysis of Tanzania focuses on the macroeconomic consequences of the extensive price controls there.The publication of Peasants and Governments and Controlled Open Economies coincided with the genesis of the Centre for the Study of African Economies at Oxford, which is now arguably the most influential research institute of its kind in the world. The Centre began with a small amount of industry funding in 1989, obtained through the success of the two monographs, and became sustainable with the award of a £2 million Economic and Social Research Council Grant over 1991-2001. The Centre owes its existence to Paul's vision and leadership, although the success of its early years was also due to the work of David Bevan. Subsequent directors of the Centre have included Jan Gunning, Marcel Fafchamps and, most recently, Stefan Dercon, who was one of Paul's first graduate students.
4