Quality of Financial Sector Regulation and Supervision Around the World
Author: Mr. Martin Cihakhttps://orcid.org/0000-0002-3571-9707 and Alexander F. Tieman1
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Authors’ E-Mail Addresses: mcihak@imf.org; atieman@imf.org
Publication Date:
01 Aug 2008
eISBN:
9781451870480
Language:
English
Keywords:
WP; KKM governance variable; financial system; regulatory governance; EU country; IOSCO standard
Download PDF (2.3 MB)

The paper analyzes the quality of financial sector regulation and supervision around the globe. Unlike studies that collect and analyze data on regulation and supervision "on the books," this study also analyzes available information on supervisory implementation, making use of data from IMF-World Bank assessments of compliance with international standards and codes. Incorporating supervisory implementation into the study provides an improved means of assessing countries' regulatory systems. We find that countries' regulatory frameworks score on average one notch below full compliance with the standards (on a 4-notch scale). There are substantial differences in the quality of regulatory and supervisory frameworks across countries, with the income level being a major factor.

Abstract

The paper analyzes the quality of financial sector regulation and supervision around the globe. Unlike studies that collect and analyze data on regulation and supervision "on the books," this study also analyzes available information on supervisory implementation, making use of data from IMF-World Bank assessments of compliance with international standards and codes. Incorporating supervisory implementation into the study provides an improved means of assessing countries' regulatory systems. We find that countries' regulatory frameworks score on average one notch below full compliance with the standards (on a 4-notch scale). There are substantial differences in the quality of regulatory and supervisory frameworks across countries, with the income level being a major factor.

I. Introduction

How good is financial sector regulation and supervision around the world? That is a rather grand question, given that there are many different regulatory frameworks around the world, operating in different institutional environments. But it is a valid and important question, because the ongoing financial globalization makes individual country financial systems much more closely linked, and substantial differences in regulatory and supervisory quality can become exposed in a cross-border crisis.

This paper addresses the question about the quality of regulation and supervision around the globe by using data from IMF-World Bank assessments of countries’ compliance with international standards and codes.2 Unlike some of the existing databases and studies that collect only information on regulation and supervision “on the books,” this study, by making use of the underlying data, can also assess the practical implementation of regulation and supervision.3

Our main findings are that (i) on average, countries’ regulatory frameworks score one notch below full compliance with the standards (on a 4-notch scale); (ii) per capita income is significantly linked to cross-country differences in regulatory quality; (iii) higher regulatory quality in banking is correlated with better banking sector performance; and (iv) there are substantial differences in regulatory quality across regions, some but not all of which can be explained by differences in economic development. The finding that high-income countries are characterized by better supervisory structures needs to be put in a context. These countries usually have more developed and more complex financial systems. It is therefore possible that, despite the higher grades, the supervisory frameworks in high-income countries may still leave something to be desired. Indeed, the developments in the global financial system in 2007-08 suggest that the higher quality of supervisory systems in high-income countries may not have been sufficient given the complexity of their financial systems.

Measuring regulatory quality is a Herculean task. Regulation should aim at supporting the efficient allocation of resources across the economy in normal times. Arguably, the ultimate test of a well-functioning regulatory framework is whether it contributes to the financial system’s intermediation capacity, while decreasing the likelihood and costs of systemic financial crises. However, achievement of these goals is next to impossible to measure, because they are either very broad (“efficient allocation”) or involve analyzing causality in “tail events.” This paper uses an alternative approach to measuring regulatory quality: it analyzes the data from assessments of compliance with international standards and codes aimed at identifying good supervisory practices. Specifically, we examine a unique dataset derived from assessments of regulatory and supervisory frameworks around the globe carried out under the IMF-World Bank Financial Sector Assessment Program (FSAP). The FSAP has so far covered about two-thirds of the IMF’s 185 member countries, and is therefore an important source of broadly comparable information on the quality of supervisory frameworks. This paper analyzes the findings from the FSAP assessments in a comprehensive way, relying on a combination of quantitative and qualitative approaches.

Are international standards and codes good measures of supervisory quality? A full theoretical discussion of what constitutes an optimal supervisory framework would go well beyond the scope of this paper. In fact, rigorous theoretical work on what constitutes good prudential regulation and supervision is limited, especially for non-bank financial institutions, and remains a topic for future research.4 For the purpose of this paper, let us just say that we acknowledge at the outset that compliance scores do not necessarily give the full picture of supervisory quality, but nevertheless: (i) the standards assessments are results of detailed consultations among top international experts; (ii) the gradings have proven useful in previous research that tried to explain cross-country differences in financial sector performance (e.g., Podpiera, 2004).

The paper contributes to the literature in two important ways: (i) it provides an analysis of prudential frameworks around the world that covers the practical implementation of regulation; and (ii) it covers all the key segments, i.e., banking, insurance, and securities regulation. Substantial work has been done on analyzing banking sector laws and regulations (in particular, through the work of Barth, Caprio, and Levine, 2006). However, that work has focused only on regulations “on the books” and on those pertaining to banks. Regulatory quality is assessed in the World Bank governance database (see Kaufman, Kraay, and Mastruzzi, 2007), but this is a survey-based broad measure of regulation in general, and does not specifically focus on financial sector regulation. A global analysis of the quality of regulation in the securities area was carried out by Carvajal and Elliot (2007); our paper uses a similar dataset, but in addition to securities regulation also covers banking and insurance supervision.

The structure of the paper is as follows. Section II explains the data and methodology being used. Section III provides a basic overview of the data on compliance with the various core principles. Section IV presents the results of the regression analysis that tries to explain the factors behind cross-country differences in regulatory quality. Section V analyzes some additional relevant findings from the FSAP program, which were not captured in the assessments of standards and codes. Section VI concludes.

II. Data and Methodology

To analyze the quality of regulation and supervision, this paper uses data on countries’ observance of internationally accepted standards in banking, insurance, and securities regulation (“standards”). The data are unique, because they reflect not only the laws and regulations “on the books” (data on that are widely available, for example, through the work of Barth, Caprio, and Levine, 2006); their key feature is that they reflect also detailed expert assessments of the practical implementation “in the field.”

The standards covered in this analysis are the Basel Core Principles for Effective Banking Supervision (BCP); the Insurance Core Principles (ICP), issued by the International Association of Insurance Supervisors (IAIS); and the International Organization of Securities Commissions’s (IOSCO’s) Objectives and Principles of Securities Regulation.5 The BCP contains 25 Core Principles and the ICP and IOSCO standards also comprise a number of principles (Tables 1, 2, and3). Although the terminology differs, the extent to which each principle in the three standards is observed is rated on a four-point scale, ranging from fully observed to nonobserved.6 Most of the assessments used in the analysis were prepared as part of an FSAP, under which the gradings are normally confidential, and therefore the analysis in this paper is presented so as to respect this confidentiality.7 The sample comprises all countries for which formal assessments are available (see Figure 4 for an overview).

Figure 1.
Figure 1.

Overall Compliance with Standards and Codes

Citation: IMF Working Papers 2008, 190; 10.5089/9781451870480.001.A001

Source: Authors’ calculations, based on IMF’s standards and codes database.
Figure 2.
Figure 2.

Overall Compliance by Income Level

Citation: IMF Working Papers 2008, 190; 10.5089/9781451870480.001.A001

Source: Authors’ calculations, based on IMF’s standards and codes database.
Figure 3.
Figure 3.

Overall Compliance by Geographic Region

Citation: IMF Working Papers 2008, 190; 10.5089/9781451870480.001.A001

Source: Authors, based on IMF’s standards and codes database.
Figure 4.
Figure 4.
Figure 4.
Figure 4.

Coverage of BCP, IAIS, and IOSCO Assessments, as of December 31, 2007

Citation: IMF Working Papers 2008, 190; 10.5089/9781451870480.001.A001

Source: IMF’s standards and codes database; help from the Map Design Unit of the World Bank gratefully acknowledged.
Table 1.

Banking Supervision ‘Dictionary’ 1/

article image

See http://www.imf.org/external/standards/index.htm for details on assessment methodology.

See IMF (2004) for more details. REG is regulatory governance, PRF is prudential framework, REP are regulatory practices, and FIN is financial integrity and safety nets.

Table 2.

Insurance Supervision ‘Dictionary’ 1/

article image

See http://www.imf.org/external/standards/index.htm for details on assessment methodology.

See IMF (2004) for more details. REG is regulatory governance, PRF is prudential framework, REP are regulatory practices, and FIN is financial integrity and safety nets.

Table 3.

Securities Regulation ‘Dictionary’ 1/

article image

See http://www.imf.org/external/standards/index.htm for further details on the assessment methodology.

2/ See IMF (2004) for details. REG=regulatory governance, PRF=prudential framework, REP=reg. practices, and FIN=financial integrity, safety

Following the previous literature on this subject (e.g., IMF, 2004; Čihák and Podpiera, 2006, 2008; and Čihák and Tieman, 2007), we have used these calculations to process the input data:

  • Principle-by-principle gradings. For each standard and each principle, there is a grading on a four-point scale. This grading was transformed into a numeric value from 0 percent (nonobserved) to 100 percent (fully observed). The value of 67 percent corresponds to the “largely compliant” rating and the value of 33 percent to the “materially noncompliant” rating in the BCP assessment.

  • Summary grading. An unweighted average of the principle-by-principle gradings was calculated to arrive at a summary grading for each standard. This summary grading is also a number between 0 percent (nonobserved) and 100 percent (fully observed).

  • Component gradings. Given that the individual principles cover different subjects and that the composition of the principles differs for the three standards, it is easier to carry out cross-sectoral comparisons if the principles are aggregated into comparable groups that cover similar topics. As in IMF (2004), the principles are grouped into the following four components of a good supervisory framework: (i) regulatory governance, which includes the aims, independence, and accountability of regulators; (ii) prudential framework, which consists of regulations covering risk management, capital adequacy, internal controls, and corporate governance; (iii) regulatory practices, which include monitoring and supervision, enforcement, conglomerates, and licensing; and (iv) financial integrity and safety nets, including consumer protection and addressing financial crimes. Table 4 maps the individual principles into the four components. For each of the four components, an observance index was calculated as an unweighted average of the individual principles included in that component.

Table 4.

Financial Sector Standards and Their Four Main Components

article image
Source: Adapted from IMF (2004).

For each component, the upper row corresponds to the original (2000) IAIS standard, and the lower row corresponds to the revised (2003) IAIS standard.

It is important to emphasize that we did not originate the underlying gradings. The information used in this paper is a result of coordinated work by the IMF, the World Bank, and other cooperating institutions and their experts over a number of years. The only thing that we are doing as far as the input data are concerned is to convert them from the gradings on a 4-notch scale into gradings on a 0-100 scale and aggregating. All the underlying methodology for the assessments and some of the principle-by-principle gradings are available at the IMF’s website at http://www.imf.org/external/np/fsap/fsap.asp. The full principle-by-principle grading information is available for countries that agreed to publication of the detailed assessments;8 for other countries that have undergone the assessments, the data are not publicly available on an individual country basis, but are included in our calculations and presented on aggregate basis.

How precise are these assessment gradings? Grading is not an exact science, and there are some obvious limitations. In particular, individual assessments may be influenced by factors such as the assessors’ experience and the regulatory culture they are most familiar with. For each standard, there are methodology documents and assessors’ templates, which add a degree of direction to the assessments, but the assessments still have an element of subjectivity and require exercise of judgment on the part of the individual assessor. Also, standards and codes assessments are not designed to address political economy issues, or issues associated with financial regulation such as crisis management, the bankruptcy code, or deposit insurance. Finally, the assessments are a one-time measurement of the regulatory system and are therefore limited by time.

While we need to be mindful of these limitations, there are also reasons to be confident that these assessments capture relevant information. Each assessment is based on a standardized methodology and carried out by a team of senior international assessors, assembled by the IMF and the World Bank. The assessors are from a diversified pool of experts with different backgrounds, which limits the effect of individual assessors’ experience and background. The team spends several weeks in the assessed country, and several more months preparing a detailed report. To help ensure internal consistency and cross-country comparability, all assessments go through review at the IMF and World Bank headquarters. The review may not eliminate all inconsistencies, but it limits them substantially. And, as mentioned above, the gradings have been successfully used in previous literature (e.g., Podpiera, 2004; Carvajal and Elliott, 2007).

The gradings are relatively robust with respect to time. When interpreting the results, one must bear in mind that the gradings are a series of country-by-country snapshots. The individual assessments were undertaken at different points during 2000-07, and some of the earlier gradings may have become outdated subsequent to the FSAP. The existence of lags between regulatory developments and a reassessment of the gradings means that older assessments are likely to underestimate the true quality of current regulation and supervision in a country. Interestingly, statistical tests do not suggest a strong link between the “vintage” of an assessment and the overall grading (specifically, we did not find a significant relationship between the year of the assessment and the overall grading, controlling for the country’s income per capita). Nonetheless, to address this concern, updated gradings were taken into account for those countries for which reassessments took place in the context of FSAP Updates; for others, the FSAP gradings represent the most recent assessment information available.

Supervisory frameworks include elements that are not easy to quantify, and information may be lost if the focus is solely on quantitative analysis. Each assessment therefore contains a rich set of underlying, qualitative information from the FSAPs. Moreover, not all principles are equally relevant in all countries, and there are issues that may not be captured by the standards assessments. To address this, the FSAP reports use the standards assessments in combination with other analytical tools to form an integrated analysis of the financial sector. The key messages from these overall analyses are surveyed in the next section of this paper.

III. An Introductory Analysis of the Quality of Supervision

A. Overall Findings

Based on the overall results for all economies in the sample (Figure 1 and Table 5), the regulatory quality seems to be remarkably similar in the three sectors. Both in banking and in insurance supervision, the average overall level of compliance is 67 percent, in both cases with a standard deviation of 19 percent. The average for securities is marginally (insignificantly) higher, with the same standard deviation. Given that a value of 67 percent corresponds to the third point on the four-point scale used in the three standards, it is possible to say that on average, the regulatory systems are “largely compliant.”9

Table 5.

Supervisory Quality by Level of Development

article image
Note: Standard deviation of the relevant index across the country sample.Source: Authors’ calculations based on IMF’s standards and codes database.

This overall assessment overlooks important differences among banking, insurance, and securities regulation in terms of the main regulatory components. Specifically, insurance regulation lags behind both banking and securities regulation in terms of regulatory governance. This reflects the fact that many insurance supervisory agencies lack clear autonomy from the government. In terms of prudential framework, banking scores somewhat higher than insurance and securities regulation, but the difference is small compared to the cross-country variation. Finally, in terms of both regulatory practices as well as financial integrity and safety nets, securities regulation is more aligned with international standards than banking and insurance.

These general comparisons, however, should not overlook important cross-country differences, which are indicated by the high standard deviation. In the next sub-section, we therefore focus on groups of countries by income level and region.

B. Differences by Income Level and Geographic Region

Differences by income level

There are important differences in the level of compliance based on a country’s income level (Figure 2).10 In all the three sectors, wealthy economies have clearly higher levels of compliance than poorer ones; the difference is bigger between high-income and medium-income economies than between medium-income and low-income economies.11 The text chart illustrates the relationship between compliance and income level for the overall gradings; Table 5 shows the numbers and confirms that this general conclusion also applies when one looks at the four main regulatory components (regulatory governance, prudential framework, regulatory practices, and financial integrity and safety nets).

Differences by geographical region

There are also notable regional differences in supervisory quality (Figure 3 and Table 6). In particular, European economies show, on average, higher levels of observance than economies in all the other four regions.12 A large part of this difference can be attributed to differences in income levels: the average (unweighted) GDP per capita in the European sample is $19,566, compared to $5,800 in the non-European sample.

Table 6.

Supervisory Quality by Geographic Region

article image
Source: Authors’ calculations based on IMF’s standards and codes database.

The difference in regulatory quality is even more notable if one compares European Union (EU) countries with non-EU countries, both European and non-European. The degree of observance of the three standards and their subcomponents was on average about 8 percentage points higher in the EU countries than in the non-EU countries (for details, see Čihák and Tieman, 2007).

EU member countries also show a more even level of observance than non-EU countries. Usual measures of cross-country variability (such as the standard deviation or the difference between minimum and maximum) suggest that EU member countries have a lower variability in quality of regulation and supervision. Cross-country variability in the EU tended to be higher in regulatory governance than in other aspects of the supervisory framework. Both in banking and in insurance, regulatory governance showed higher cross-country variability than the prudential framework, regulatory practices, or financial integrity and safety nets. In securities regulation, the prudential framework was the component with the highest cross-country variability.

Despite the relatively favorable performance vis-à-vis non-EU countries, compliance in the EU countries was far from perfect. The overall level of compliance ranged from 79 percent in insurance to 85 percent in securities regulation.13 For example, in banking, two EU countries observed all core principles, but no principle was observed fully across the EU. On average, there were about nine less-than-fully-compliant EU countries for each principle. For some principles, more than half of EU countries were less than fully compliant.

Looking at the other regions, Europe is followed (in terms of supervisory quality) by Asia and the Pacific, the Middle East and Central Asia, the Western Hemisphere, and Africa. Again, one should note that these aggregate differences are driven to a large extent by differences in income per capita, and also that there is substantial variability within each of the regions.

The lowest values are generally reached in the area of financial integrity and safety nets. This is particularly true in African banking supervision (42 percent), in Middle East and Central Asian insurance supervision (40 percent), and in African securities regulation (45 percent). The highest levels of observance are generally reached in Europe, in particular in regulatory governance in banking (82 percent) and in regulatory practices in securities regulation (83 percent).

C. Main Strengths and Areas for Improvement

This section highlights the main strengths and the main areas for improvement that are relatively consistently highlighted in the BCP, ICP, and IOSCO assessments. The discussion is based on a detailed principle-by-principle analysis of the results, a summary of which is provided in Figure 5 (banking), Figure 6 (insurance), and Figure 7 (securities). A more detailed discussion is provided in the Appendix.

Figure 5.
Figure 5.

Compliance in Detail: Basel Core Principles

Citation: IMF Working Papers 2008, 190; 10.5089/9781451870480.001.A001

Source: authors’ calculations, based on IMF’s standards and codes database.
Figure 6.
Figure 6.

Compliance in Detail: IAIS Principles

Citation: IMF Working Papers 2008, 190; 10.5089/9781451870480.001.A001

Source: authors’ calculations, based on IMF’s standards and codes database.
Figure 7.
Figure 7.

Compliance in Detail: IOSCO’s Objectives and Principles

Citation: IMF Working Papers 2008, 190; 10.5089/9781451870480.001.A001

Source: authors’ calculations, based on IMF’s standards and codes database.

Each of the figures consists of the following two closely related charts: the first one shows the average degree of compliance for each principle across all the countries in the sample; the second one shows the percentage of countries that were less-than-fully compliant with the particular principle.14 So, for example, for the first Basel core principle (objectives, autonomy, powers, and resources of banking supervisors), the first chart in Figure 5 illustrates that the average degree of compliance over the whole sample was 73 percent, while the second chart illustrates that 61 percent of countries were less than fully compliant with this principle (the exact numbers are available upon request).15 As one could expect, there is a negative correlation between the average degree of compliance and the percentage of less-than-fully compliant countries, but the correlation is lower (in absolute value) than 1, so each of the charts has a separate informational value.

The specific strengths and areas for attention include the following:

  • In banking regulation, the strongest observance has been recorded in the area of licensing, specifically in the principle that deals with definition of permissible activities for the licensed institutions. In contrast, the areas most in need of improvement included supervision of other risks; connected lending; issues related to money laundering; supervisory objectives, autonomy, powers, and resources; remedial measures; and consolidated supervision.

  • In insurance regulation, the highest degree compliance has been recorded in the areas of confidentiality and winding-up. In contrast, areas with low observance include market conduct issues to internal controls, derivatives and off-balance-sheet items, organization of the supervisor, corporate governance, assets, onsite inspection, and licensing.

  • In securities regulation, the areas of strength include clarity and consistency of the regulatory process, professional standards, the appropriate use of self-regulatory organizations, and the rules on transparency of trading. The areas for improvement relate to enforcement powers and compliance program; capital and other prudential requirements; powers, resources, and capacity; and operational independence and accountability.

IV. Regression Analysis

Quality of regulation and supervision is positively correlated with the level of economic development. This finding, indicated in previous literature (e.g., Čihák and Podpiera, 2006, 2008), has been confirmed by our study (Figure 8).

Figure 8.
Figure 8.

Regressions of Compliance on Income Per Capita

Citation: IMF Working Papers 2008, 190; 10.5089/9781451870480.001.A001

Source: authors’ calculations, based on IMF’s standards and codes database.

We have calculated correlations between the “supervisory quality” data studied here, and the regulatory database by Barth, Caprio, and Levine (2006, henceforth BCL). The main difference between the two is that the latter focuses on regulations “on the books,” while the former includes both regulations on the books and their actual implementation in practice. We have found correlations between the two datasets low, all below 50 percent, and many in the 20-30 percent range (Table 7). This is consistent with the notion that the two datasets describe different phenomena, given that the standards and codes database incorporates also practical implementation of regulations.

Table 7.

Correlation of Standards and Codes Assessments with the BCL Database 1/

article image
Sources: IMF standards codes database and World Bank BCL Database

Explanatory Note. The questions used from the BCL database are:

Q 3.4 “What is the actual risk-adjusted capital ratio in banks as of yearend 2005, using the 1988 Basel Accord definitions?”Q 3.8.1 “What fraction of the banking system’s assets is in banks that are 50 percent or more government owned as of yearend 2005?”Q 3.8.2 “What fraction of the banking system’s assets is in banks that are 50 percent or more foreign owned as of yearend 2005?”Q 6.1 “Can the supervisory authority force a bank to change its internal organizational structure?1=Yes, 0 = No”Q 6.2 “Has this power been utilized in the last 5 years? 1=Yes, 0 = No”Q 9.4 “What is the ratio of nonperforming loans to total assets as of year-end 2005?”Q 12.4 “How many professional bank supervisors are there in total?”Q 12.5 “How many onsite examinations per bank were performed in the last five years?”

We have calculated correlations between all BCP, IAIS, and IOSCO categories, and the World Bank governance database by Kaufman, Kraay, and Mastruzzi (2007, henceforth KKM) as of 2002 and 2006. This database reflects the views on governance by experts from the public sector, private sector, and non-government organizations, as well as thousands of citizen and firm survey respondents worldwide. With many correlations between 50 and 60 percent, a few readings higher than 60 percent, and about half below 50 percent, the correlation between these survey data and our regulatory quality indicators is higher than the correlation between the regulatory database of BCL and our supervisory quality data (Tables 8and 9).16 Apparently, as both our dataset and the KKM governance data aim to measure implementation and actual practice, the relationship between these data is closer than the relationship between implementation and regulation on the books, as measured by BCL.

Table 8.

Correlation of Standards and Codes Assessments with KKM 2002 Database 1/

article image
Sources: IMF’s standards and codes database, and World Bank’s KKM database.

The KKM data relate to Voice and Accountability, Political Stability, Government Effectiveness, Regulatory Quality, the Rule of Law, and the Control of Corruption.

Table 9.

Correlation of Standards and Codes Assessments with KKM 2006 Database 1/

article image
Sources: IMF’s standards and codes database; and World Bank’s KKM database

The KKM data relate to Voice and Accountability, Political Stability, Government Effectiveness, Regulatory Quality, the Rule of Law, and the Control of Corruption.

We tested which macroeconomic variable best explained standards and codes. As expected, we found this to be gross domestic product per capita. It performed better than alternative variables, such as domestic credit and claims to the private sector, or proxies for the size of the financial sector,17 which all exhibited lower explanatory power.

Subsequently, we gauged whether information embedded in the standards and codes add information to that contained in a country’s level of development (as measured by its GDP per capita), and its governance regime (as measured by the individual KKM indicators). To do so, we regressed all standards and codes on the individual KKM governance variables and per capita GDP, using OLS, i.e. we estimated:18

SCij=α+β⋅KKMi+γ⋅GDPj+ε,

with

SCij= standards and codes variable i for country j ,i = regulatory governance, prudential framework, regulatory practice, financial integrity, for all three standards (BCP, IAIS, and IOSCO),

KKMi= KKM variable i for country j, i = Voice and Accountability, Political Stability, Government Effectiveness, Regulatory Quality, Rule of Law, Control of Corruption,

GDPj= GDP per capita in country j

We ran these regressions for all standards and codes being considered, and for the KKM data from both 2002 and 2006. For GDP per capita we use the value from the same years as the KKM data (i.e. either 2002 or 2006).19 This results in matrices of coefficients and t-values (Tables 10 and 11), discussed below:20

Table 10.

Regressions of Standards and Codes on KKM 2002 Variables and GDP Per Capita

article image
Source: Authors’ estimates.
Table 11.

Regressions of Standards and Codes on KKM 2006 Variables and GDP Per Capita

article image
Source: Authors’ estimates.

  • For the BCP, the coefficient estimates for GDP per capita always exhibit the expected positive sign, and are often significant at the 5 percent uncertainty level (in 80 percent of the cases, both for the 2002 and the 2006 data). At the same time, the KKM governance variables are also significant most of the time. We interpret this as indication that the BCP captures information contained in the level of development, which is not contained in the World Bank governance data. We paid specific attention to KKM’s “regulatory quality” variable, which comes closest to our BCP data. For the BCP regressions including this explanatory variable, significance at the 5 percent uncertainty level for this variable is found in about 50 percent of the cases for 2006 data, and about 75 percent of the cases for 2002 data, while the GDP per capita variable is almost always significant. Taken together, these results suggest rather clearly that the BCP contains information beyond the regulatory governance data.

  • For IAIS and IOSCO, the results are less strong. Still, around a half of the regressions featured GDP per capita with a coefficient with the expected sign and significantly different from zero at the 5 percent uncertainty level. In particular, this is the case for IAIS Regulatory Practice, and IOSCO Prudential Framework and Financial Integrity. The KKM governance variables were not significant most of the time. We interpret this as the BCP “core” banking regulations being better reflected in the general regulatory quality variable of KKM, whereas the ‘more obscure’ insurance and securities regulations feature less prominently in the survey responses of the KKM study.

We have also performed the above regressions for regions (following the IMF’s World Economic Outlook classification, i.e., Europe, Middle East and Central Asia, Asia and Pacific, Africa, and Western Hemisphere)21 and by income level (following the World Bank country classification through the Atlas method,22 and adding the OECD as a separate group). The results generally show less significance, reflecting the much smaller samples. For the BCP regression, however, the GDP per capita variable is significant at the 5 percent uncertainty level in a number of cases, specifically for both the low income and the high income (and OECD) countries. It therefore seems that the results for the BCP are driven in particular by low- and high-income countries.

A similar approach for the BCL regulation and supervision database shows even more striking results (Table 12). For the questions that we consider, in almost all cases GDP per capita is another significant variable. This is the counterpart of the fact that many of the variables from the BCL database turn out not to be significant in the regression. This is at least partly due to the fact that, in this WB database, there are many missing values. Still, the regression analysis identifies several interesting patterns. First, while higher government ownership of banks is associated with a lower BCP ranking (significant in 2 out of 4 regressions), higher foreign ownership is weakly associated with a better BCP ranking (not significant). This indicates that market discipline, in the form of bank ownership by parties not directly under the influence of government has a positive relationship with banking supervision quality, as measured by BCP. Second, the supervisor’s authority to ask a bank to change its structure is associated with higher BCP scores (significant). A possible explanation is that this is probably the case in only the better regulated systems. Third, a higher ratio of nonperforming loans to total assets is associated with lower BCP scores (significant in only 1 out of 4 regressions). This could be because systems with high nonperforming loan ratios are generally the ones that are less well-run.

Table 12.

Regressions of Standards and Codes on BCL Variables and GDP Per Capita

article image
Sources: IMF standards codes database and World Bank BCL Database

Explanatory Note. The questions used from the BCL database are:

Q 3.4 “What is the actual risk-adjusted capital ratio in banks as of yearend 2005, using the 1988 Basel Accord definitions?”Q 3.8.1 “What fraction of the banking system’s assets is in banks that are 50 percent or more government owned as of yearend 2005?”Q 3.8.2 “What fraction of the banking system’s assets is in banks that are 50 percent or more foreign owned as of yearend 2005?”Q 6.1 “Can the supervisory authority force a bank to change its internal organizational structure?1=Yes, 0 = No”Q 6.2 “Has this power been utilized in the last 5 years? 1=Yes, 0 = No”Q 9.4 “What is the ratio of nonperforming loans to total assets as of year-end 2005?”Q 12.4 “How many professional bank supervisors are there in total?”Q 12.5 “How many onsite examinations per bank were performed in the last five years?”