Free Email Notification
Moral Hazard in IMF Loans
How Big a Concern?
Kenneth S. Rogoff
It is an article of faith among many IMF critics that industrial country taxpayers bear the costs of IMF loans—the critics call them "bailouts"—to countries in crisis. The standard scenario is supposedly this: A developing country borrows money; it runs into repayment difficulties; capital markets dry up; the IMF ponies up a giant loan; the country is saved from the consequences of its profligacy; the private creditors get paid off in full; the IMF gets left holding the bag; and, thus, taxpayers in industrial countries, which are the IMF's main shareholders, end up footing the bill. Insidiously, the prospect of future "bailouts" fuels not only excessive lending to emerging market countries at interest rates that do not adequately reflect underlying risks but also irresponsible policies. Thus, the critique goes, the IMF helps cause and exacerbate crises of the kind that the organization was designed to prevent and ameliorate.
I will be the first to concede that this "moral hazard" theory of IMF lending is clever, having spent many years in the 1980s studying it and writing papers about it (including a 1988 piece with Jeremy Bulow of Stanford University that appeared, of all places, in IMF Staff Papers. Who says the IMF never listens to its critics?). Concerns over bailouts and the resultant moral hazard have certainly become influential in policy circles, particularly since the emerging market crises of the 1990s. But where is the empirical evidence that moral hazard in IMF lending is important, at least in this crude form?
It is certainly not easy to show. The first inconvenient fact is that IMF loans have had a stubborn habit of being repaid in full. Although some countries have gone into arrears, almost all have eventually repaid the IMF: the actual realized historical default rate is virtually nil. If we live in a world where virtually all countries will always repay their IMF loans in full, the IMF moral hazard theory is a bust, at least in its most worrisome form. But haven't IMF rescue packages sometimes had the effect of helping private creditors? Perhaps to some extent, yes, but the losses suffered by international investors and creditors, including foreign banks, during crises that the IMF has helped resolve—as a result of currency crashes, asset price declines, and debtor defaults—suggest that IMF lending has hardly provided a blanket to private lenders.
Wait, though: might the good repayment record of IMF loans be due to "evergreening?" That is, is it possible that the IMF has avoided default to date only by continually relending principal plus interest, pushing defaults off to a future day of reckoning? If so, then the IMF's excellent repayment record might be an accounting sham that, some day, will hit the books (or hit the rocks?) with the same force as WorldCom or Enron. But the data are not entirely friendly to this possibility either. The IMF's Olivier Jeanne and Jeromin Zettelmeyer have studied the frequency and duration of all IMF lending cycles since 1948, using a model to estimate the probability of evergreening. They find that, for emerging market countries, the proportion of cycles that register as "suspicious" (of evergreening) is only 5 percent and statistically insignificant.
Of course, the jury is still out on the loans from the 1990s "bailouts" that prompted cries of moral hazard in the first place. But with the exception of Indonesia, all recipient countries have started to pay down their debts to the IMF significantly.
Hold it. What if the evergreening takes a more subtle and hidden form, as suggested in a 1992 study I did with Jeremy Bulow and Afonso Bevilaqua? What if the debtor country looks at IMF loans, World Bank loans, and other official loans as all coming from one big pot (creditor-country taxpayers' pockets). In this case, there is a Ponzi scheme, but it involves complex interlinked bookkeeping across various international lenders and aid agencies that one must look at very closely to detect its true nature. Perhaps there is something to this view, but we could not prove it; we did find individual cases to support the interlinked Ponzi scheme conjecture, but, overall, the evidence was not convincing.
We also tried looking at secondary market spreads on bank debt, the predominant form of financing prior to the 1990s. By looking at spreads, one can potentially ask whether markets envision the possibility of future defaults on IMF and World Bank loans, even if (virtually) none have occurred to date. There are lots of complications in trying to conduct such a test, and our regressions were not entirely conclusive. More recent studies using bond spreads do seem to find some evidence of moral hazard before the Russian debt crisis of 1998, but no evidence thereafter. This break in pattern, if it exists, is consistent with the hypothesis that, by allowing Russia to default on its debt in 1998, the IMF established a tougher reputation.
It is so hard to find firm historical evidence supporting the bailout view that one might worry the IMF's repayment record is perhaps too sparkling—that the institution has not taken enough risks in the past. I would say not. As an outsider during the Mexican and Asian crises, I saw the IMF stick its neck out. These packages worked, however, and early postmortems that concluded that the IMF's handling of the Asian crisis was broadly and seriously at fault turned out to be way overblown. Today, Asia is the fastest-growing region in the world. And there is no doubt that the IMF has continued to take significant risks to help its member countries.
So what could be the scale of the costs of moral hazard associated with the IMF's operations? First, consider an analogy. In the mid-1930s, the United States became the first country to establish a broad-based system of official insurance for bank deposits. This invention appears to have been a significant factor in the subsequent stability of the U.S. banking system and an important factor in banks' role as an engine of growth. Deposit insurance, like IMF lending, induces some moral hazard. But it was not until the savings and loan crisis of the 1980s that moral hazard engendered any significant fiscal cost. The ultimate cost of the savings and loan bailout amounted to some 3 percent of GDP. Was this such a stiff price for 50 years of stability? Of course, some other countries have fared worse, with bailouts amounting to 10 percent of GDP and more.
What is the cost to shareholder taxpayers of an IMF "bailout"? Let's take Jeanne and Zettelmeyer's estimate of a 5 percent default rate. Suppose the typical emerging market country receives an IMF loan of 10 percent of GDP (Korea and Mexico, both considered massive bailouts, each received less than 5 percent of GDP.) Assuming the recipient accounts for 1 percent of global GDP (as the largest emerging market countries do), then the cost of the IMF loan subsidy amounts to only 1/2 of one 10,000th of global GDP. It's not exactly a crushing burden in a world where industrial countries are working toward an aid goal of 0.7 percent of GNP.
Overall, therefore, the evidence on the existence of moral hazard in IMF loans is somewhat mixed, despite the excellent repayment record to date. But the evidence does not suggest that the moral hazard element has been nearly as large as conjectured by some critics (including me, as an academic in the 1980s). At this writing, the IMF has many major packages out, and these may be different. The charge of evergreening is being made by some critics of recent IMF programs, though of course it is much too early to tell. This round, the moral hazard theorists may at last be proved right. Then again, they may just have to wait a few more decades.