رتبه بندی اعتباری مستقل: گناه ورای تردید معقول؟

This paper questions the view that credit rating agencies aggravated the East Asian crisis by excessively downgrading those countries. I find that ratings are, if anything, sticky rather than procyclical. Assigned ratings exceeded predicted ratings before the crisis, mostly matched predicted ratings during the crisis period, and did not increase as much as predictions in the period after the crisis. Ratings are also found to react to non-macroeconomic factors such as lagged spreads and a country’s default history. Therefore it is questionable that ratings exacerbate the boom-bust cycle if they are simply reacting to news, whether macroeconomic or market.

In the aftermath of the Asian crisis of 1997 followed by Russia in 1998 and Brazil in 1999, much attention by the media and policymakers has turned on sovereign credit ratings. Rating agencies have been criticized for failing to predict the Asian crisis, and for exacerbating the crisis when they downgraded the countries in the midst of the financial turmoil. The International Monetary Fund (1998)1 highlights how the rating agencies reacted late when they downgraded the Asian countries. No sovereign credit rating was downgraded throughout 1996 and the first half of 1997 for the East Asian countries, with the exception of Thailand by Moody’s in April 1997. During the crisis, Indonesia, Korea and Thailand were downgraded to below investment-grade. A recent paper by Ferri et al. (1999), hereafter FLS, argues that in addition to failing to predict the Asian crisis, the credit rating agencies unduly amplified the crisis when they excessively downgraded the countries later and more than the worsening in their economic fundamentals would justify. This would occur if as a result the cost of borrowing increased and if the potential pool of investors declined due to statuary requirements.2 The question motivating this paper therefore is whether ratings have “tremendous power to influence market expectations on a country” (as pointed out by FLS) or whether they are simply reacting, without contributing, to news. Despite their alleged non-effectiveness and possible guilt, the Basel II proposals give increased prominence to ratings in bank capital requirements. The most recent proposals will allow banks to use ratings by either credit rating agencies or their own internal assessment in determining the amount of capital they need to put aside for different types of loans. The rationale seems to be that although credit ratings have performed worse than their aim, they are still a second-best solution. Because they are readily available, their use will probably improve the current Basel standards that do not accurately account for risk.3 More generally, sovereign credit ratings have important implications for international capital flows and for the linkages between company ratings and sovereign ratings. The objective of this paper is to investigate the behavior of sovereign credit ratings. I focus on the East Asian crisis and whether the rating agencies aggravated the crisis by excessively downgrading those countries. The key advantage of this study is the extension of the period of analysis to the post-crisis period, 1999–2001, allowing for a comparison of pre- and post-crisis rating behavior. In order to understand how ratings behave, it is first important to ask what do the sovereign ratings assigned by the agencies really capture? The two main rating agencies are US-based, Moody’s and Standard & Poor’s that have been publishing ratings since 1909 and 1923, respectively. Only recently have they begun rating sovereigns, and only in this past decade for many developing countries. Clearly, the ratings they offer must be of some value since investors pay to subscribe to their credit reports. Rating agencies are not specific neither about what determines their ratings nor their rating procedure. Ratings are meant to provide an estimate of the probability that borrowers will not fulfill the obligations in their debt issues. While this all seems plausible at the level of say US firms, it is problematic at the sovereign level. If we take the statement that ratings are supposed to capture the default probability of sovereign debt at face value, the question is whether this is true ex post. Do countries with lower ratings default more? The problem is that countries, unlike firms, rarely default because of the availability of international emergency credit (most of their problems being liquidity and not solvency related) and the high cost of future credit should they default. Therefore it is difficult to empirically assess this proposition. Table 1 illustrates the defaults on foreign currency bond debt as compiled by Standard & Poor’s. It shows the minimum rating assigned in the year prior to default and the ensuing rating during default. There are a number of striking features. First, only 15 countries have defaulted on foreign currency bond debt since 1975 (although many more have defaulted on bank debt). Second, the paucity of rating figures (only 7 were rated by Moody’s and 5 by S&P at the time of default) shows that defaulting countries were not even likely to be rated which would seem to defy the point of being rated. Third, while all the ratings were below investment-grade prior and during the default period, only Argentina, Ecuador, Pakistan and Russia were downgraded by S&P to “default status” during the supposed default period. In fact, Pakistan and Venezuela were not even downgraded by Moody’s (although this might reflect dating problems, because Moody’s did downgrade Pakistan from a B2 to a B3 on 5/28/1998 and to a Caa1 on 10/23/1998 while the default year was recorded as 1999. Likewise, they downgraded Venezuela from a Ba1 to a Ba2 on 4/8/1994). Nonetheless, it is troubling that even in the periods of default, many of the ratings were not systematically rated as such. The other important feature of Table 1 is what it does not contain, all those countries rated at similar and even lower ratings that did not subsequently default. It all goes back to the question of what do sovereign ratings try to capture? This paper explores this question by revisiting the FLS results. I find that ratings are sticky rather than procyclical.4 Assigned ratings exceeded predicted ratings in the run up to the crisis, mostly matched predicted ratings during the crisis period, and did not increase by the amount suggested by predictions in the recent period after the crisis.5 This study includes data from the post-crisis period, which helps reveal the inertia in ratings. Therefore they capture the crisis but are over-conservative in the period following the crisis. It takes a sufficient amount of either bad (or good) news to change in the direction of the news. Ratings are also found to react to non-macroeconomic factors such as lagged spreads and a country’s default history. Although ratings appear to lag financial markets, this does not imply that they have considerable market impact. While the first part of the statement is true when ratings are sticky (and I find evidence of this), the next part does not follow. Ratings are probably not contributing much to new information in the market. Unfortunately without a decent instrument for ratings, the discussion will not be closed. This is not to say that statuary requirements that extend the use of sovereign ratings will be harmless. Biases can occur if ratings are used in ways that distort capital allocations. The remainder of this paper is organized as follows. Section 2 reviews the literature. Section 3 describes the data and empirical techniques and presents empirical results. Section 4 concludes.

In short, I argue that the case for the guilt of sovereign credit rating agencies is not tenable. This paper was written to investigate the view of Ferri et al. (1999) and others that credit rating agencies have a strong impact on market expectations, thereby damaging the macroeconomic fundamentals of a country and aggravating crises. This paper suggests a more cautious view. There is little support for assigned ratings being excessively conservative during the crisis period from 1997 to 1998. When not including country fixed effects, ratings are found to be sticky rather than procyclical.33 While there is support for the FLS finding that predicted ratings were lower than assigned ratings during the period prior to the crisis, this is consistent with both the FLS and inertia views. That ratings are not found to be predicted higher than assigned during the crisis period weakens the FLS view. The additional advantage of this study is that it extends the sample period to the post-crisis period from 1999 to 2001. This helps reveal inertia in ratings. Predicted ratings are found to be higher than assigned ratings during this period (when not including country fixed effects). Therefore they capture the crisis but remain over-conservative after the crisis. Ratings adjust only when there is a sufficiently large divergence of predicted ratings from assigned ratings. I also find that in a model where ratings react passively to market sentiment (as proxied by spreads), then the drying up of credit might be attributed to excessive downgrading, even though it would have occurred regardless. Ratings are also found to be influenced by a country’s default history, which may be capturing political factors that are value-relevant. Although ratings may to the most part be reacting to macroeconomic and market news, this does not imply that their use in statuary requirements is benign. This particularly deserves attention with the introduction of ratings into the latest Basel II proposal for determining risk weights for bank loans. This point is well made by Kashyap and Stein (2004) who argue that a single time-invariant risk curve that maps credit-risk measures (such as ratings) into capital charges is sub-optimal. They make a case for having a family of risk curves such that the capital charge is reduced when economy-wide bank capital is scarce, such as during recessions. Others, such as The Economist (2/6/2003), call for taking credit ratings completely out of financial regulation and using instead the spreads at which secondary market debt securities trade. More careful study is needed prior to 2007 when Basel II is expected to take effect so that the new capital standards do not exacerbate business cycle fluctuations. There are good reasons for capital requirements to be countercyclical, but this does not imply that ratings can be expected to be countercyclical. Instead, their use for regulatory reasons should be carefully evaluated.