Comment and Discussion

Comment by George G. Kaufman: This paper describes a massive, ambitious, formidable, and important project. The project is intended to help governments, regulators, and consultants worldwide design appropriate supervisory, regulatory, and safety net structures for banking and the financial system that will maximize efficiency and safety and stimulate economic growth and development. By providing relevant data for many countries in a massive database, the authors hope to encourage rigorous assessments of which specific regulatory and supervisory standards actually matter for bank performance and stability.

In their oral presentation and forthcoming working paper, the authors note that policymakers today are often generating best-practice recommendations based on little solid evidence and, where evidence is available, mostly from industrial countries, such as the United States.³¹ These may not be readily transferable, particularly to developing economies. Therefore, there is an urgent need for a large-scale, wide-reaching empirical study of different systems.

Their working paper attempts to fill that void. Data are obtained from several surveys that cover more than 100 countries, although not all countries are used in all summaries and analyses. Numerous organizational, regulatory, and public policy characteristics are grouped into nine categories that are perceived to be useful in explaining both measures of bank performance and the probability of a bank crisis.

At the end of their paper, the authors derive eight tentative conclusions and two broad themes. The conclusions are as follows:

--Restrictions on bank activities are detrimental to bank performance and stability.

--Tighter restrictions on entry, particularly of foreign banks, are detrimental to bank performance and stability.

--Capital adequacy regulation is not closely linked to bank performance and stability.

--More generous depositor insurance schemes are closely linked to banking crises.

--Measures of official supervisory indicators are not closely related to bank performance.

--Banking diversification is strongly and positively related to bank stability.

--Efforts to encourage private monitoring of banking enhance bank performance and reduce the probability of bank crises.

--Public ownership of banks is negatively related to performance and stability.

The two themes are as follows:

--Regulatory supervisory strategies that promote the private sector are successful.

--Diversification of bank income streams benefits stability.

In addition, while not a theme and "remaining subject to revision," the authors conclude, "These initial findings do not lend much support for an excessive reliance on official regulation and supervision of bank capital, loan classification, provisioning policies, among other official actions." I return to this later.

Let me start by posing three questions: Do I buy all or most of these conclusions and themes? With two major exceptions, I do. Did I believe these before listening to their presentation? On the whole, yes. Did their presentation cause me to change my mind on any of these issues based on the data or arguments presented? The answer is not really. And I will focus on my hesitation to do so. In particular, I will try to explain my skepticism on the two somewhat counterintuitive "not closely" related conclusions for the roles of capital regulation and official supervision. If I have questions, others are likely to have them as well, and I offer these comments in the spirit that it is better to clarify and possibly modify these conclusions now while they are still preliminary, than later when they are cast in stone.

Database and Methodology

The massiveness of the database and project makes it difficult to digest at one sitting, so I may not be perfectly accurate in all my remarks. Nevertheless, given the important uses that are likely to be made of the database, it is worth taking the time to examine it and to understand the following:

--The reliability of the information

--The comparability of the data across countries

--Whether the correct information is collected to answer the particular questions

--Whether the model in their presentation is specified correctly given the database, particularly with respect to possible omitted variables across countries.

DATA RELIABILITY...