A quarterly magazine of the IMF

December 2006, Volume 43, Number 4

Beyond the Ballot Box
Robert H. Bates

Political reform and policy reform in contemporary Africa

Development economists and political scientists point to a variety of factors to explain variations in growth in developing countries—among them geographic location, availability of natural resources, and the quality of political and economic institutions. They try to discover why one group of countries may have done well in the postcolonial period while others stagnated. Why, for instance, have the economic performances of Africa and Asia diverged so greatly when countries such as Ghana and Korea were at roughly similar levels of development in the 1960s?

In the late 1970s and 1980s, a consensus emerged that the answer boiled down to differences in economic policies. To encourage development, international financial institutions began to advocate a combination of economic policy reform, coupled with institutional change to make governments answerable to the people and therefore more likely to follow policies that benefited the majority. The result was a shift from military and authoritarian regimes to multiparty systems, particularly in Africa.

But improvements in economic policies did not necessarily follow. Indeed, the evidence suggests that while political competition reins in predatory governments, it may not encourage them to manage the economy better. In fact, the opposite may be true: governments subject to electoral risk may be less willing to adhere to macroeconomic (particularly fiscal) discipline. By introducing electoral competition, reformers may have stayed the hand of potential predators but also introduced politically driven budget cycles.

Using Africa as a case study, this article looks at the extent and pattern of African political reform, examines how politicians may turn political accountability to their advantage, and suggests why institutional change alone may not be enough to improve economic policies.

Impact of political change

For a long time, economists believed that good policies would follow good governance. Citizens would choose governments that delivered economic performance and depose those that attempted to line their own pockets or those of special interest groups. Economic models provided support for this argument. In 1973, for example, Robert Barro showed that, in principle at least, institutions that render officeholders accountable can generate incentives that influence economic policy in the ways claimed by political reformers.

More recently, a study we undertook (Humphreys and Bates, 2005) highlighted a second implication: the larger the decisive group of citizens—which we call the selectorate—the more inclined the government will be to furnish public rather than divisible goods in its effort to retain office. Intuitively, as the selectorate grows in size, the costs of "paying off" each member individually rise; it therefore becomes cheaper for the government to pay them off by financing a good that all can enjoy, such as a school or a hospital.

In an attempt to put this theory about good policies into practice, local champions of political reform joined with international financial institutions to seek political change in low-income countries. In no place was this truer than in Africa, where political change came quickly, particularly between 1985 and 1995 (see Charts 1 and 2). The percentage of countries with multiparty systems rose dramatically, and the percentage of military governments declined (although less sharply).

How successful was this experiment? Did political reform lead to effective policy reform, as expected? To address this question, we used two measures to explore the relationship of electoral competition to policy outcomes. Both reflect the assessments of informed observers and both offer insight into the policy choices of governments.

The first comes from the International Country Risk Guide (ICRG) and is produced by Political Risk Services (PRS). Each year, the PRS convenes a panel of international investors to rate governments on a series of dimensions relating to the political, economic, and financial risks faced by investors. The resulting score provides a measure of the tendency to make opportunistic use of public power. The measure, labeled simply "opportunism," combines ratings of the government's propensity to repudiate its financial obligations and its likelihood of expropriating private investments.

The second is the World Bank's Country Policy and Institutional Assessment (CPIA), which evaluates annually the conduct of governments that have loans outstanding to it. The Bank's rating covers governments' policy performance in 20 specific areas, grouped into four major categories. Scoring a country's performance in each area from 1 (low) to 5 (high), the Bank calculates an aggregate score, or CPIA, which is the unweighted average of the rating in each of the 20 areas. In essence, the CPIA provides a measure of the government's adherence to the Washington Consensus (Williamson, 1990 and 1994)—a set of reforms promoted by international institutions in the 1990s that emphasized macroeconomic (particularly fiscal) discipline, the importance of the market economy, and openness to trade and foreign investment. Although the measure contains several flaws, it provides an informed assessment of the government's efforts to generate a sustainable macroeconomic environment, free of major policy distortions. In addition to a measure of the size of the selectorate, we included among the independent variables a measure for electoral competition, the degree of unrest in the country, and the dependence on agriculture and oil, along with measures for wealth and GDP growth (see box).

Our regressions showed that, first, both the ICRG risk ratings and the World Bank CPIA scores bear a large and significant relationship with subsequent rates of economic growth, meaning that the two measures capture behaviors that matter. Thus, the poorer the risk and CPIA scores, the lower the subsequent growth.

Second, controlling for the impact of other independent variables, we found that an increase in the level of electoral competition is associated with significantly lower ratings for opportunism—decreases of sufficient magnitude to associate with a rise in growth rates of approximately one-fourth of a percentage point.

Third, we also found, however, that when governments face risks from electoral competition, they then choose policies that distort the macroeconomy and so receive lower ratings from the World Bank. The results were robust to a variety of specifications, including the introduction of fixed effects (to control for omitted variables) and instrumental variables (to control for endogeneity, or outside factors).

Political business cycles?

What do these results mean? Upon reflection, they suggest that although accountability models may capture key elements of the political process, so too may models of political business cycles, that is, business cycles tied to the political process, a subject examined extensively by Steven Block and his associates, Karen Ferree and Smita Singh (2003).

Using data from 47 African states over the period of political reform, 1980–94, Block (2001) sought to determine whether the introduction of elections affected the policy choices of governments. He found significant evidence of preelection increases in the growth of the money supply and decreases in the nominal rate of interest and the exchange rate in countries outside the CFA franc zone. The procyclical impact of these changes emerges after the election, when the rate of money growth turns negative and the interest and exchange rates rise. Ahead of the election, public expenditures and government consumption rise significantly, along with the level of government debt. Inflation in preelection periods remains higher than expected, and it rises significantly in the postelection period.

Block, Ferree, and Singh next narrowed the focus to presidential elections in which opposition parties could challenge incumbent regimes. For years in which there were competitive elections, they found the following:

• The share of government consumption in GDP increases on the order of 1.5 percent, which is significantly greater than that for noncompetitive elections.
• Competitive elections are associated with a nearly 13 percent increase in the rate of real money growth over that observed in noncompetitive election years.
• Competitive elections are typically followed by devaluations in the order of 20 percent greater than after noncompetitive elections—an adjustment made necessary by the imbalances introduced by the changes in fiscal and monetary policy and by the postponements of adjustments in the pre-election periods.

In no case do they find noncompetitive elections yielding evidence of politically induced business cycles; evidence of political shocks is confined to elections in which there is political competition.

The evidence thus suggests that while introducing political competition may stay the hand of predatory governments, it fails to inspire them to manage the macroeconomy better. In fact, the opposite may be true: governments subject to electoral risk may be less willing to adhere to macroeconomic (particularly fiscal) discipline. This helps explain the appearance of politically induced business cycles in Africa in recent years.

The study also shows that the relationship between electoral competition and macroeconomic distortions peaks with the so-called founding election—the election that occurred at the time of transition from authoritarian rule. Thereafter, the relationship remains but declines in magnitude. The macroeconomic disruption that initially characterized the introduction of institutional reform thus may have represented "transitional dynamics" rather than a steady state. Only the analysis of data drawn from subsequent periods can resolve this issue.

The role of information

Why is it that elections can result in worse economic policies? The answer may be that voters don't realize that they are being manipulated or realize only after some time. For the accountability theory to work, voters must be informed about such issues as the level of prices and employment.

Employment figures are difficult to come by in Africa, however; indeed, given the size of the subsistence sector and the informal economy, employment is difficult to define. Nor are average price levels common knowledge; even when calculated with precision, their release is too long delayed to assist in the making of voting decisions. It is therefore difficult for citizens to monitor the management of the national economy, to judge the economic performance of incumbents, and, thus, to be able to implement punishment strategies. Knowing that, politicians would have less reason to practice macroeconomic restraint.

That said, although the assumption of complete information works against the accountability model, it works in favor of models of the politically induced business cycle. Such models suggest that politicians' ability to exploit the voters' lack of information enables the former to manipulate the economy in ways that, albeit economically harmful, are politically advantageous. This characteristic of Africa's political economies—the lack of information—helps to explain why political reform failed to produce a more stable macroeconomic environment in the 1990s.

The introduction of political accountability in Africa thus achieved mixed results in its early stages. It reduced the level of opportunistic predation by government leaders, it appears, but failed to increase the quality of macroeconomic policies. Africa's experiment with political reform leaves us with the question: Did the macroeconomic disturbances that followed the introduction of party competition represent the costs of transition? Or did they signal the beginning of a new policy regime?