Accountability and Economic Performance

Until recently it has not been regarded as feasible to investigate the relationship between public expenditure and its consequences through macro-statistical techniques. However, both improvements in data and in econometric technology are changing this assessment. The aim of this project is to tease out how those institutions that are measurable with macro data affect economic performance (in various dimensions), and how this relationship is in turn affected by resource rents. Some aspects are institutions are readily measurable with long time series, such as the extent of press freedom and aspects of constitutional design such as decentralization and checks and balances. Others, which have more direct pertinence to service delivery, are only measured in recent years but may permit some backward estimations. Resource rents can be estimated over a long time period, through new data that combines costs and revenues.  The dependent variable, the performance of public service delivery, can be proxied in various ways. One is to take specific components of spending, such as unit costs of schooling. Another is to investigate the overall impact of public spending on some aggregate measure of performance such as growth or life expectancy. We are currently investigating whether resource rents have adverse effects, whether these can be compensated by particular institutions, and whether the institutions themselves are endogenous to the resource rents.

To date, and during the period of the inception phase, work has started and advanced well on five technical papers and two policy versions are in course of publication. In a first piece, on commodity prices and growth, we investigate the growth consequences of commodity exporting using co-integration techniques which allow both long and short-run effects to be distinguished. This paper is the first to reconcile the short run results using VARs, as pioneered by Deaton and Miller, with the long-run cross-section literature pioneered by Sachs and Warner. The former found that there was no ‘resource curse’ whereas the latter found it to be severe. We find that indeed the short run effects of high commodity prices are benign for exporters, but that in the long run the effects are highly adverse. The long run adverse effects are, in turn, traced to initial conditions of governance. If initial governance is ‘good’ – at or above that of Botswana – there is no resource curse. Further, the resource curse is confined to non-agricultural commodities – revenues from the former accrue to private agents and so are not so adversely affected by poor governance. Finally, we are able to trace the key channels by which poor governance dissipates the resource revenues – excess public consumption, inadequate investment, and excess protectionism.

In second paper, on ‘Testing the NeoCon Agenda’, we take this analysis further and investigates whether democracy is an effective institutional transformation for adequate governance in resource-rich societies. It finds that, on the contrary, it tends to make performance significantly worse. However, this in turn depends upon the design of democracy. We distinguish between electoral competition and checks and balances, finding that the former is significantly detrimental in resource-rich societies, whereas the latter is distinctively helpful.

A third technical paper takes this further, investigating specifically the role of elections in policy reform, starting from initial conditions of very poor policies and governance. Taking a global approach, it finds that elections have ambiguous effects on the pace of reform. The ambiguity depends partly upon the frequency of elections and partly upon the depth of democracy. Elections can be a spur to reform, but if they are too frequent they significantly retard it. The ‘depth’ of democracy also has ambiguous effects. Partial democracy – scores in the range 0-4 on the widely-used Polity index – actually retard reform. This work was presented as the opening keynote at a WIDER conference in Helsinki in June.

The final two technical papers of this project investigated the politics of reform, but with a wider lens than the focus on elections. The fourth paper looked at the preconditions for reform, while the fifth investigated reforms that were already underway. One line of investigation in both papers is how the political economy of reform is affected by the injection of aid. The hypothesis, confirmed by the initial results, is that technical assistance promotes reform, both as a precondition and especially once it is underway, whereas money itself, if provided too early in the reform process, significantly retards it.