Today’s guest post is from Roberto de Michele and Francesco De Simone, of the Inter-American Development Bank and Ugo Panizza of the Graduate Institute of International and Development Studies in Geneva.

A year ago, at a seminar at the Inter-American Development Bank (IDB), a representative from one of the major private credit rating agencies got everyone’s attention with a single slide. That slide showed a strong positive correlation between corruption perception indicators and sovereign risk ratings. The simple yet compelling message: corruption, or at least its perception, negatively affects a country’s perceived credit risk, in turn may raise the country’s borrowing cost.

What are we to make of this correlation? Does it indeed indicate a causal connection between corruption and high borrowing costs? If so, what are the implications for policymakers? Although there was some discussion of this issue in the academic literature a decade ago, the subject had not received much attention. Intrigued by this simple correlation, the IDB Transparency Fund sponsored a study of this topic, for which one of us (Ugo Panizza) served as principal investigator. That study, published last October, is available in English and Spanish on the IDB website. The main findings were as follows:

• First, there is indeed a strong statistical correlation between the main indicators of corruption and the sovereign risk ratings issued by the three main risk rating agencies, and this correlation holds even if one controls for a variety of macroeconomic variables. The correlation is substantively as well as statistically significant: An improvement of one standard deviation above the mean on the Worldwide Governance Indicators (WGI) control-of-corruption index is associated with an increase of approximately 2.5 points on the sovereign credit rating—an improvement greater than that associated with an equivalent improvement of other explanatory variables. (For example, a one-standard-deviation reduction in the debt/GDP ratio is associated with only a 0.5 point increase in the credit rating, while a one-standard-deviation reduction in the inflation rate is associated with an improvement of 0.75 points.)
• Second, however, the strength of the correlation is not as robust if one controls for other governance variables. This suggests that that corruption is a key component—but only one component—of a country’s overall governance environment, which in turn is one of the key determinants of sovereign risk.
• Third, despite this strong correlation between corruption indicators and sovereign risk rating, that correlation alone does not establish a causal link. The correlations in quantitative data alone would also be consistent with sovereign risk ratings influencing perceived corruption, or both being caused by some third factor for which the study was not able to control. That said, a closer examination of how the rating agencies determine sovereign risk scores reveals that the agencies do indeed take corruption perception indicators into account. Thus, the correlation between corruption perception indicators and sovereign risk ratings is in fact the result of a deliberate choice by the rating agencies.

From the IDB’s perspective, the results of this study have a number of potentially important implications for our work:

• First, the strong correlation between perceived corruption and credit ratings is likely to get senior policymakers—especially in finance ministries—to pay more attention to corruption issues, which in turn may generate more demand for attention to, and support for, anticorruption efforts.
• Second, while the increased attention to the importance of anticorruption is likely a good thing, the study results—and in particular the extent to which sovereign credit rating agencies explicitly incorporate corruption perception indexes into their rankings—also raise a serious concern about how these corruption indicators can generate perverse and short-term incentives. For example, governments may have incentives to take measures and policy actions that may impact perception indicators in the short term, but without really addressing the causes that foster corruption in the long term. Even worse, the connection between corruption perceptions and sovereign risk ratings can end up penalizing governments that take action to confront corruption, at least in the sort run: There is some evidence that when countries start to actively investigate corruption cases and as a result expose more widespread corruption—as in Brazil—the corruption perception indicator scores may end up getting worse, at least in the short term. Because of this, organizations like the IDB must do what they can to reaffirm the importance of transparency and anticorruption reforms that are sustainable in the long term. In recent years, the Bank, through policy-based loans in several countries in the region, has been supporting processes of transparency reforms that go precisely in this direction.
• Third, the study implies an important future research agenda for the future, one that will go beyond the purely statistical correlation and explore in detail the causal links between corruption and sovereign risk. There is a growing literature investigating the same issue but from a different perspective, showing that policy reforms that increase data transparency reduce the sovereign risk rating spread in emerging markets. We need more “event studies” to investigate how specific corruption scandals may (or may not) result in a downgrading by rating agencies. In other words, while we have made progress in understanding better the relationship between sovereign risk and corruption, but we still have a way to go.