Rethinking Bank Regulation: Till Angels Govern
Published December 1, 2006 | December 2006 issue
By James R. Barth, Gerard Caprio Jr.
and Ross Levine
Cambridge University Press, 2006
Countries making a transition to democracy or a market economy receive a lot of free advice, and more than a few visits, from more experienced "friends." Some of this counsel covers the big ideas of our time: Learn how to write a constitution in six months or less! Now available, "Contract Enforcement and Property Rights for Dummies." Going on such a mission sounds important and exciting.
But given my experience and interests, my most likely ticket to a country in transition is banking supervision and regulation.1 With a turn of the political winds in, say, sub-Saharan Africa, I could be jetting off to help strengthen how a country monitors the financial condition of its banks.
Now, in truth, I've always had doubts about the importance, excitement and, particularly, the sequencing of such trips. Bestsellers tell me to trust my instincts, and my gut says that nations (and their advisers) should make private contract enforcement or anti-government corruption efforts a priority over strengthening bank supervisory powers. Colleagues who have gone on such trips have the same reaction.
But the economics of banking suggests that my gut could be wrong about this. Unfettered banking markets can, in theory, produce undesirable results, like bank panics. Bank regulation, done right, can foster the development of a healthy banking system. And a more robust banking system can contribute to greater economic growth and higher standards of living. We're back to big ideas.
In Rethinking Bank Regulation, James Barth, Gerard Caprio and Ross Levine take a serious, formal and important look at competing views of banking regulation. They place my anxiety about emphasizing bank regulation in the context of theories where rulers and their cronies have a “grabbing hand”—that is, where political leaders look more to their personal interests than to the public's. They then juxtapose this private interest view of regulation with one where regulation serves the public interest.
To judge between these conflicting views, Barth, Caprio and Levine collected survey data on supervisory practices, rules, organization and the like from roughly 150 countries. (Their most recent questionnaire had about 275 questions!) They then carried out a series of statistical tests to see how supervision affects outcomes that countries care about. They conclude that the standard features of banking supervision and regulation do not reduce—and may even increase—the chance that countries experience banking crises. Nor do these rules and regulations lead to more developed banking sectors or more efficient banks. These findings are, according to the authors, consistent with private interest views and the fact that "few countries have highly developed democratic institutions" (p. 13). In contrast, policies that enable private markets to better monitor banks and that encourage private actors to "discipline" banks are associated with desirable outcomes.
This stark finding has begun to work its way into the supervisory consciousness. In a recent speech, the head of supervision for the state of New York, Diana L. Taylor, referred to Rethinking Bank Regulation, saying that it concludes that "some [banking regulation] does not work very well."2 Taylor summarizes the book by quoting from it: The overriding message is that simply strengthening direct official oversight of banks may very well make things worse, not better, in the vast majority of countries"
I can envision observers of banking and banking supervisors citing such summary statements from the book to support it or dismiss it. Others may even try to boil it down further: "Banking regulation is not good!"
But, in fact, the book resists such definitive summaries, and by saying that I'm not implying, obliquely, that readers should avoid this book. I am glad I read it, and I believe it will influence banking research to come. Rethinking Bank Regulation should also influence policymaking; its authors argue compellingly that supervisory best practice in a well-developed country may prove harmful in a less-developed country. Moreover, its empirical approach to studying supervision sets a standard for those wishing to recommend one set of supervisory policies over another.
At the same time, the authors have taken on many challenges, some of which seem nearly insurmountable. Consider one. Their goal is to isolate the effect that supervision—even some fairly specific forms of supervision—has on outcomes like banking crises. But countries with supervisory regimes linked to crises could have other factors (for instance, corrupt governments) that are also associated with crises. And these nonsupervisory factors might even cause undesirable supervisory regimes. Barth, Caprio and Levine recognize these problems and use sophisticated statistical approaches to address them, but they must still provide significant caveats to their most important findings.
Ironically, this limitation is another reason that policymakers, their staffs and policy analysts ought to read the book. Otherwise, they will have an incomplete understanding of results that others are sure to cite. But enough of my exhortations. Barth, Caprio and Levine believe that regulation works best when it facilitates market forces, and providing reliable data is one way to do so. In that vein, I offer a few guiding questions and answers to give prospective readers a sense of whether they'd benefit from reading the book.
Question 1: Do you find data on banking supervision relevant to your job or interests?
A book titled Rethinking Bank Regulation isn't aimed at the average reader, but the question still seems a fair one for determining its audience. And interest in banking-related data will go some way in defining that audience. Almost at page one the authors note that "[o]ur main contribution is to assemble a new database on bank regulation and supervision around the world, so that researchers and policy makers can for the first time compare what countries actually do and assess which policies work best to improve human welfare" (p. 4). I agree, and I believe this justifies my reading the book. I particularly appreciate that the authors have included a CD of their database along with the book. Commenting on supervisory rules across countries absent these data would be hard to imagine—not that absence of data has ever prevented such comments, including my own.
Here are a few statistics from the book that provide a sense for the data and for how differently countries approach banking and banking supervision.
- Governments in low-income countries (e.g., Chad and Cambodia) reject nearly half of the applications submitted by banks seeking to enter their banking markets. The rejection percentage for high-income countries is around 5 percent.
- About the same number of countries (69) set up their central banks as the sole bank supervisory agency as do not provide the central bank with any supervisory power (61).
- The percentage of banking assets in a country controlled by government-owned banks remains high in a few countries (at least at the time of the survey). Government-owned banks in China, Turkmenistan and Algeria, for example, controlled nearly 100 percent of banking assets. In 50 of 136 countries, however, the percentage of banking assets controlled via government banks is zero, and the median is about 5 percent. (Rounding must explain the 0 percent for the United States; the Bank of North Dakota is the self-reported only state-owned bank in the United States.3)
More interesting than these random facts are the groupings that countries seem to fall into. Those countries with mainly government-owned banks also tend to restrict the activities of banks in their country more than countries where banking assets are controlled by private domestic and/or foreign banks. And the countries with more restrictions on banks are generally the same as those rejecting new entrants into the banking market. Barth, Caprio and Levine combine the survey data with other characteristics of countries and find further linkages. For example, the restrictive, government banking countries also tend to have higher levels of corruption as assessed by a World Bank working paper.4
Question 2: Do you read compilations and/or reference books?
Rethinking Bank Regulation is a reference book in at least three ways.
- The book contains the data just mentioned.
- The authors hold the book out as a collection of essays. As noted in the introduction, "This book does not have to be read from beginning to end. ... [A]lthough the chapters fit together, we believe they stand sufficiently alone so that readers with different backgrounds and interests can use the book to achieve different goals" (p. 17). But to the degree that statement encourages readers to skip the thorough discussion of methodology, I don't agree. Since Barth, Caprio and Levine make a point of highlighting potential concerns about their work, anyone tempted to cite just the findings should read the whole thing.
- The book brings together related work issued previously.5 Not surprisingly, it captures work by Barth, Caprio and Levine. In addition, all of the authors worked for or are affiliated with the World Bank. Economists at the World Bank have churned out a significant number of papers on similar topics, and the book brings together the findings of much of that work as well.
Whether or not these traits are selling points for the book depends on your working style and your experience with and opinion of that literature.
Question 3. Which Nobel laureate most informs your view of banking regulation: Stigler (1982), Stiglitz (2001) or both?
Joseph Stiglitz gained fame as an economist for, among other accomplishments, modeling how markets, particularly credit markets, can fail to operate optimally when all parties to a transaction do not have the same level of information. This problem of "asymmetric information" can lead to undesired outcomes that could be improved, in theory, by government intervention. As the Nobel committee put it, Stiglitz and his co-authors "have time and again substantiated that economic models may be quite misleading if they disregard informational asymmetries. Their common message has been that in the perspective of asymmetric information, many markets take on a completely different guise, as do the conclusions regarding appropriate forms of public-sector regulation."6
In Rethinking Bank Regulation, Stiglitz seems to be the poster child for the public interest view of banking regulation. Adherents to this view, according to Barth, Caprio and Levine, believe that as long as a market failure exists, government intervention can lead to improvements to society. They write, "[M]any economists (e.g., Atkinson and Stiglitz, 1980) stress that, in the presence of information and transaction costs, governments can improve social welfare by ameliorating these market imperfections through coordinating the activities of many of society's actors. Thus, some stress that a powerful government is the only mechanism for easing severe market imperfections" (p. 179, emphasis added).
Who then represents the competing private interest view? George Stigler would be a good place to start, and the authors do so. Stigler hypothesized that "in practice, some regulations protect firms, organizations and professional and occupational groups—i.e., producer interests—instead of the general public that, according to stated motives, they were intended to protect."7 This view has long made economists a bit skeptical about the ability of regulators to correct market failure.
The work of another Nobel Prize winner, James Buchanan (1986), also comes to mind. According to the Nobel committee, Buchanan argued that "[i]ndividuals who behave selfishly on markets can hardly behave wholly altruistically in political life. This results in analyses which indicate that political parties or authorities that to at least some extent act out of self-interest, will try to obtain as many votes as possible in order to reach positions of power or receive large budget allocations."8
Stigler, Buchanan and others have thus shown that we cannot take for granted that governments will act in the interests of society as a whole. Governments may act at the behest of interest groups or firms, taking actions that enrich them. Or they may act for the benefit of government employees or policymakers. Barth, Caprio and Levine's data analysis leads them to conclude that Stigler and Buchanan have it right. The structure of banking regulation and policy in most of the world ultimately benefits the private interests of government officials, bankers and perhaps others. If you don't find that view credible no matter what data or analysis you read or if you were already convinced of that view before picking the book up, you may find Rethinking Bank Regulation less than useful.
Given the dates of the Nobel prizes just mentioned, the private interest view is clearly not novel. Indeed, textbooks, which typically offer a lagging indicator on ideas, reflect the private interest view; the first textbook on policy analysis I open (also the only one I own) includes a chapter on "Limits to Public Intervention: Government Failures."9 As Barth, Caprio and Levine make clear, these views are also not novel when it comes to the specifics of banking. "Many others have cautioned against relying on official supervision and urged policy makers to focus on strengthening private monitoring" (p. 308).10 In part because these arguments seem well-established, I found the dueling theories part of the book excessive, although certainly not so distracting as to prevent me from reading it. The strength of the book is its data, descriptions and empirical bent. University of Chicago Professor John Cochrane, speaking generally (not with regard to Rethinking Bank Regulation), captures my complaint better than I can:
In most papers, the "main result" is empirical. There may be some theory or a model but if you (or the editor!) ask "does this paper expand our knowledge of economic theory?" the answer is "no." The theory is there to help understand the empirical work. Following the rule, then, the theory must be the minimum required for the reader to understand the empirical results.11
Moreover, as expressed by Barth, Caprio and Levine, the public interest view seems more like their straw man than a true competing argument. Elected officials and supervisors espouse the public interest view on occasion, but do many economists hold it? Does Stiglitz, who sees many more market failures than I do, really believe that perceived market failures always justify some government action?
I have not carefully reviewed Stiglitz's work on this topic, but consider the following statement from an article he wrote laying out his view of banking regulation:
The problem is that all too often, regulation goes beyond efforts to enhance the safety and soundness of banking systems. There are, for instance, frequent attempts to repress competition that are not justified by the objective of increasing franchise value to enhance prudential behavior. Incumbents in any industry try to protect themselves against the pressures of new entrants. Although a wide array of arguments are brought to bear in support of such contentions, it should be clear that these are typically no more than self-serving arguments of special interest groups attempting to maintain their monopoly rents.12
Question Four: Do you have strong feelings about the statistical tools that many economists use to explain cross-country differences?
Barth, Caprio and Levine use "regressions"—a statistical tool that links something you care about (e.g., test scores, inflation) with potential explainers (e.g., how much you study, how much money the government prints)—to analyze their cross-country data. This method falls in a long tradition used to explain why some countries grow faster than others. Levine, in particular, has been an important contributor to this literature in examining links between the financial sector of an economy and economic growth. In Rethinking Bank Regulation, the authors use regressions to explain how differences in bank regulation influence banking crises, the robustness of the banking system and the like.
Cross-country regressions of this sort are subject to a number of criticisms, many of which the authors themselves discuss.
One concern flows from their absence of a mathematical model-based theory. Absent a theoretical model, some economists—particularly those from the "Minnesota" school—would surely argue that the results found by Barth, Caprio and Levine defy ready interpretation as they could be consistent with multiple models.13
A related and particularly important concern is the difficulty of disentangling the effects of bank regulation and other factors on key outcomes. As the authors recognize:
[S]ome may note that there is a positive correlation between policemen and crime. This does not imply policemen cause crime. Rather, it implies that societies engage policemen to curb crime. In the case of bank regulation, inefficient, unstable, and corrupt banking systems may attract regulators to address these ills. From this perspective, it is wrong to attribute poor banking to regulatory restrictions on bank activities and competition, official supervisory power, and government ownership of banks (p. 253).
Barth, Caprio and Levine use a variety of statistical techniques to address concerns about the "third factor" and cause-and-effect confusion. However, these corrections face important limitations and raise reasonable doubts about robustness of findings.14 The authors and others have called for different approaches, especially country-specific studies, to try to untangle the thicket of potential causes of bad outcomes.15
Finally, the authors list several important but fairly specific concerns with the context in which these tests occur.
- They reflect rules on the books rather than actual practices. Given their aim to "compare what countries do," that's a significant weakness.
- They feed data from different time periods into the models.
- They aggregate survey data into a variety of summary measures which may or may not make sense. Also, the variables measured may or may not be good proxies for what the authors really want to measure.16
While these concerns fall outside any doubts about cross-country regressions per se, they add to the overarching caution one should take when examining the results.
None of these concerns means that Barth, Caprio and Levine are wrong. The burden of proof falls on those who want to challenge their results. And by using standard, frequently-published-in-esteemed-journals approaches, the authors have set a high bar.
At the same time, these critiques mean that we cannot be entirely confident that the authors have it right, a fact which they do not hide.17 Readers simply repeating their findings without thinking carefully about the caveats raised do themselves a disservice. Those who want to apply these findings (based on analysis of many countries typically less developed than the United States) to the United States should be especially careful to read the authors' analysis thoroughly.
Question Five: Do you believe that economics simply proves the obvious?
Some disparage economic research for simply proving the obvious. Newspapers reported that economist James Tobin, a pioneer in explaining the importance of diversified asset allocation, won the Nobel Prize for determining that all eggs should not go in one basket.18 Such criticism misses the mark. Economic research has shown that popular predictions of cause and effect often get it wrong. Mandating lower rents can lead to less low-cost housing, limiting trade to help a country's citizens can actually reduce their welfare and so on.
Even confirming the "obvious" has benefits. Sometimes we simply need to remind ourselves and our policymakers about the simple truths. Human nature seems to encourage equating the complicated with the effective when, in fact, implementing the simple plan may prove more desirable.
So what important (if obvious) points did this book leave with me?
Barth, Caprio and Levine stress the importance of tailoring banking policy to the institutional and political setting of a country. Where governments loot, for example, do not establish banking policies that facilitate the crime. I doubt banking regulation provides the key tool for looting, but why make it easier for autocrats to steal? The authors also confront institutions that issue best-practice guidance for banking regulation. These best-practice lists shouldn't offer a fig leaf to countries. Rather, to the degree to which they do not already, these lists should clarify the political, economic and legal foundations and the infrastructure a country must have before adopting the best practices in bank regulation.
Barth, Caprio and Levine also force readers to take a "zero-based" review of banking policies. In the United States, new banking policies seem to receive the bulk of whatever benefit/cost review occurs; banking supervisors look over existing regulations to determine if and how they could become more efficient. But such reviews do not typically lead to the elimination or significant scaling back of existing regulatory policies, such as those governing entry into banking.
While I don't read Barth, Caprio and Levine as proving the failure of U.S. bank regulatory policies, they do challenge the standard review process. They ask the key question about existing regulations: Are they making us better off? Supervisors have an obligation to take the question seriously.
Done poorly, supervision can retard the standard of living. It seems particularly tragic to get supervision wrong in developing countries where living standards already languish. How then to figure out what constitutes "right" supervision?
Barth, Caprio and Levine take on that important task empirically, and they substantially improve on efforts that have relied on little more than intuition, assertions or anecdote. After reading this book, no one could justify the application of common supervisory standards to developing countries simply because "rich countries do it."
That does not mean that the authors have had the last word. Opinions on the effect of one policy on a country's outcomes when the outcomes and the other policies of the country come wrapped in a hard-to-unwind ball of rubber bands come with great uncertainty. Rethinking Bank Regulation is no exception. Changing supervisory processes without changing underlying legal, political and institutional structures might not offer much hope to a country. Even before Rethinking Bank Regulation was published, the authors' cross-country data and analysis had a high profile in banking economics. I expect that this book will raise that profile still further and will interest even more analysts in trying to improve banking supervision. Greater riches should result.
1 For simplicity's sake, I will use the terms "regulation" and "supervision" interchangeably.
4 See Daniel Kaufmann, Aart Kraay and Pablo Zoido-Lobaton, "Governance Matters," Policy Research Working Paper 2196, World Bank, 1999.
5 For examples, see James R. Barth, Gerard Caprio Jr. and Ross Levine, "Bank Regulation: What Really Works," Milken Institute Review 7, Fourth Quarter 2005, 56-76; James R. Barth, Gerard Caprio Jr. and Ross Levine, "Bank Regulation and Supervision: What Works Best?" Journal of Financial Intermediation 12, April 2004, 205-48; Thorsten Beck, Asli Demirguc-Kunt and Ross Levine, "Bank Supervision and Corruption in Lending," Journal of Monetary Economics 53, November 2006, 2131-2163; James R. Barth, Gerard Caprio Jr. and Ross Levine, "Banking Systems Around the Globe: Do Regulation and Ownership Affect Performance and Stability?" in Financial Supervision and Regulation: What Works and What Doesn't? ed. Frederick Mishkin, National Bureau of Economic Research, 2001.
6 See Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2001, Information for the Public.
9 See David L. Weimer and Aidan R. Vining, Policy Analysis: Concepts and Practices, Prentice Hall, 1998.
12 See Joseph E. Stiglitz, "Principles of Financial Regulation: A Dynamic Portfolio Approach," World Bank Research Observer 16, Spring 2001, p. 16.
13 For an example of such a critique, see Peter J. Klenow and Andrés Rodríguez-Clare, "Economic Growth: A Review Essay," Journal of Monetary Economics 40, December 1997, 597-617.
14 For example, see Steven Durlauf, Paul A. Johnson and Jonathan R.W. Temple, "Growth Econometrics," in Handbook of Economic Growth, Vol. 1A, ed. Philippe Aghion and Steven N. Durlauf, 2005, North-Holland, pp. 638-40. For a discussion of similar issues in a different book review, see Daron Acemoglu, "Constitutions, Politics and Economics: A Review Essay on Persson and Tabellini's The Economics Effects of Constitutions," in Journal of Economic Literature 43, December 2005, 1025-48.
15 See, for example, the case study of regulation in Luigi Guiso, Paola Sapienza and Luigi Zingales, "The Cost of Banking Regulation," NBER Working Paper 12501, August 2006. They take a case study instead of a cross-country view, arguing, "These two views of regulation are hard to disentangle empirically. According to the benign view of regulation, governments intervene more where markets fail more. Hence, any attempt to estimate the effect of bank regulation would spuriously attribute a negative effect to bank regulation unless the pre-existing degree of market failure is controlled for (an almost impossible task)."
16 For example, Barth, Caprio and Levine include the extent of credit ratings for large banks in their index of "private monitoring." Surely credit ratings reflect private monitoring, but might a county's economic size and integration with other large economies play a determinative role in this variable? Similarly, this index assumes higher private monitoring if a country has no explicit deposit insurance scheme, but might that reflect an implied government backing? See page 139 of the book for a description of this index.
17 In reference to their measuring policies on paper and not necessarily actual or optimal practice on the ground, the authors argue, "This shortcoming may reduce confidence in our econometric work, and if the primary result of this book is 'merely' to encourage greater effort to measure supervisory effectiveness, we will be pleased" (p. 11).