This paper addresses the determinants of public expenditure policies, by reviewing theories and empirical investigations of what features explain the budget process and how the various attributes of actors—including politicians, bureaucrats, interest groups, and donors—and of institutions and political and economic governance environments affect the prioritization of public investments. It draws conclusions with regard to the determinants of agricultural public investments.
Studies that explicitly examine the budget process as it pertains to agricultural ministries and agencies in developing countries question the relevance of the formal budget process to understanding how decisions are actually made. There exists a body of work, adhering to the so-called garbage can budgeting model, which, although clearly rejecting the notion of a textbook budget process in empirical reality, seems to also reject the notion that there are any systematic politico-economic or other influences on how public expenditures are apportioned across competing needs. The budgetary model of incrementalism, at the other extreme, models budget makers as backward looking and changes in budget allocation as incremental. Another body of literature focuses on nature in which budgetary trade-offs are or are not made. The passage of the final budget—however it is arrived at—is not the end of the budget process, as there is still the implementation and execution of that budget. Discrepancies between the approved agricultural budget and the executed budget in the sector can come in the form of leakages, or they can occur because of a lack of capacity to execute or because of changing priorities mid–fiscal year.
One conception of the budget allocation process prevalent in the economics literature offers an economistic view of public resource allocation undertaken by a benevolent and autocratic (in the sense of unencumbered) social planner seeking to maximize aggregate welfare. Other distinct branches have developed within the public choice literature, including those that depart from the notion of an unencumbered policymaker. One such branch of this literature analyzes budget outcomes emerging from the interface between budget-maximizing bureaucrats and vote-seeking politicians. In the collective action literature, characteristics of interest groups (in the broadest sense of the term) affect these groups’ ability to press for public policies, including agricultural investments, subsidies, and other public interventions, that are favorable to them. An interesting phenomenon in policy processes is the seeming existence of a status quo bias among policymakers, such that policies that have outlived their usefulness, such as agricultural input subsidies, appear to often fail to be discontinued. A diverse literature examines donors as actors, with their sets of incentives, constraints, and preferences, and the reach of their influence over public spending in developing countries.
It is often difficult to attribute to policymakers’ actions the creation or improvement of certain services. Incorrect or imperfect attribution, in turn, dampens policymakers’ (political) incentives to undertake effort in improving services and infrastructure and drives the prioritization of investments. The extent to which attribution is achieved depends on the visibility of the investments, and on the length of lag between the time when resources are allocated to provide a good or service and the time when the good or service is created. This helps explain the underinvestment in agricultural research.
Areas of public spending in which large infrastructural or other capital investments are undertaken lend themselves more to rent-seeking activities by public officials. The prevalence of corruption in a society thus affects the composition of public spending, by increasing aggregate public investments, although the quality of these investments will be lower. The effect of wider political governance features of countries on the composition of public investments is more complex, nonlinear and not fully conclusive in the literature.