Does Financial Openness Lead to Deeper Domestic Financial Markets?

Size: px

Start display at page:

Download "Does Financial Openness Lead to Deeper Domestic Financial Markets?"

Chester McKinney
6 years ago
Views:

1 Does Financial Openness Lead to Deeper Domestic Financial Markets? FPD Academy Award Seminar The World Bank July 28, 2010 César Calderón (The World Bank) Megumi Kubota (University of York)

2 Motivation Salient features of the global economy: Rapid integration of advanced countries and emerging market economies (EMs) to world capital markets. Important changes in the patterns of saving and investment across the world. Deepening of financial markets across industrial countries and EMs.

3 Motivation Rapid increase in financial openness (FO) and domestic financial development (DFD) over the last 25 years

4 Motivation FO leads to faster DFD through different channels: Eliminating financial repression and shifting interest rates to clearing-market competitive levels (Baldwin and Forslid, 2000) Raising the degree of bank competition i.e. adoption of more sophisticated banking techniques (Levine, 1996; Caprio and Honohan, 1999). Displacing inefficient financial intermediaries and creating pressures for a financial reform agenda (Stulz, 1999; Stiglitz, 2000; Claessens et al. 2001; Chinn and Ito, 2006).

5 Motivation FO, on the other hand, may lead to: Risky behavior by banks and trigger boom-bust cycles (Allen and Gale, 1999; Schneider and Tornell, 2004; Tornell and Westermann, 2005) Agency problems (Allen and Gale 2000). Buy risky assets during lending booms Bubbles may burst into banking crisis and recessions Financial liberalization may generate short-run tensions (Kaminsky and Schmukler, 2008)

6 Motivation The effect of FO on DFD is theoretically ambiguous. It becomes an empirical issue! Prior: The effect of FO on DFD depends on some structural features of the economy. Institutional framework Legal infrastructure

7 Goal Goal of the paper: Test whether rising FO has led to improvement in domestic financial markets. Test whether the effects of FO on DFD may depend upon the country s characteristics the degree of institutional quality, investor protection, and trade openness

8 Contribution Contribution to the literature: First paper to evaluate the effects on DFD of FO using outcome measures of cross-border asset trade rather than indicators of the lack of restrictions to perform cross-border trade Complements existing evidence that more intense financial linkages may lead to deeper domestic financial markets after controlling for the legal and institutional framework of the country

9 Rising FO enhances DFD Main Findings It enlarges size/activity of financial intermediaries Expansion of credit and higher assets It improves the efficiency in the banking system Reduction of net interest margins It contributes to the deepening of local stock markets and private bond markets Raising stock market capitalization and total value traded Increasing private bond market capitalization Structure of foreign assets and liabilities appears to play a role! Accumulation of risky vs. riskless assets

10 Main Findings Response of DFD to higher FO depends upon: The level of institutional quality (IQ) The degree of investor protection (IP) International trade linkages of the domestic country (TO) DFD enhanced by FO in countries with moderate to higher levels of IQ, IP, TO Bank size/activity (i.e. credit, deposits, assets) Bank efficiency (i.e. net interest margin) Bond markets (i.e. private bond market capitalization)

11 Main Findings Rising FO deepens domestic financial markets. (+) impact on bank credit, bank assets, net interest margins, capitalization of stock markets and private bond markets. Structure of foreign assets and liabilities appears to play a role! Accumulation of risky vs. riskless assets Response of DFD to higher FO depends upon: The level of institutional quality The degree of investor protection International trade linkages of the domestic country.

12 The Data Sample of 145 countries (24 industrial economies) Annual information, (unbalanced panel) Domestic financial development (Beck, Demirguc- Kunt, Levine, 2009) Size and activity of financial intermediaries Efficiency and structure of commercial banking Stock market development Bond market development

13 The Data Financial openness Outcome measures: Foreign assets and liabilities (Lane and Milesi-Ferretti, 2001, 2007) Policy measure: Index of financial openness (Chinn-Ito, 2007) Intensity of financial openness? Control variables PPP-adjusted real GDP Inflation Exchange rate regimes: dual and freely falling (Reinhart and Rogoff, 2004; Ilzetzki, Reinhart and Rogoff, 2008)

14 Methodology Challenges of panel data estimation Presence of unobserved period- and country-specific effects Most explanatory variables are likely endogenous. Need to control for reverse causality. Method: IV estimation Need: Control reverse causation in financial openness DFD may affect cross-border asset holdings. Instruments for FO (Faria et al. 2007): Institutional quality (summary index: rule of law, corruption, bureaucratic quality, and democratic accountability) Legal origin of the country (La Porta et al. 1998) Natural resource abundance, and area of the country Initial size of the country

15 Baseline Regressions Financial Integration and Domestic Financial Development Baseline IV Regression Analysis Sample of 145 countries, (annual observations) Dependent variable expressed as % of GDP, in logs Bank Bank Stock market Credit Assets Capitalization Explanatory variable [1] [2] [3] Financial Openness (FO) ** ** ** Foreign liabilities (% GDP, logs) (0.03) (0.02) (0.08) Control variables Income per capita, lagged ** ** ** (in logs) (0.05) (0.04) (0.13) Trade openness, lagged ** ** * (Exports and imports, % of GDP, logs) (0.04) (0.03) (0.11) Inflation, lagged ** ** ** (in logs) (0.03) (0.03) (0.07) Falling exchange rate ** ** ** (lagged) (0.04) (0.03) (0.08) Dual exchange rate regimes (lagged) (0.10) (0.08) (0.17) Observations R** Country- and time-effects Yes Yes Yes Numbers in parenthesis are robust standard errors. * (**) implies significance at the 10 (5) percent level.

16 Baseline Regressions Financial Integration and Domestic Financial Development Baseline IV Regression Analysis Sample of 145 countries, (annual observations) Dependent variable expressed as % of GDP, in logs Bank Stock Market Credit Deposits Assets Capitalization Traded value Foreign assets ** ** ** ** (% GDP, logs) (0.024) (0.019) (0.020) (0.080) (0.126) Nobs R** Foreign liabilities ** ** * ** ** (% GDP, logs) (0.025) (0.019) (0.021) (0.075) (0.121) Nobs R** Foreign assets and ** ** * ** ** liabilities (% GDP, logs) (0.027) (0.021) (0.022) (0.084) (0.136) Nobs R** We control for income per capita, trade openness, inflation, and exchange rate regimes. All these control variables are lagged one period. Numbers in parenthesis are robust standard errors. * (**) implies significance at the 10 (5) percent level.

17 Baseline Results: A Scorecard All Industrial Developing Dependent Variable Countries Countries Countries Activity of financial intermediaries Private credit by deposit money banks (%of GDP, logs ) [+] [+] [ 0 ] Private credit by financial system (% of GDP, logs ) [+] [+] [ 0 ] Bank deposits (% of GDP, logs ) [ - ] [ 0 ] [ 0 ] Financial deposits (% of GDP, logs ) [ - ] [ 0 ] [ 0 ] Size of financial intermediaries Total assets of deposit money banks (% of GDP, logs ) [+] [+] [ 0 ] Total assets of the financial system (% of GDP, logs ) [+] [ 0 ] [+] Liquid liabilities [ 0 ] [ 0 ] [ 0 ] Efficiency of financial intermediaries Overhead costs (logs ) [ 0 ] [ 0 ] [ 0 ] Net interest margin (logs ) [ - ] [ - ] [ - ] Bank concentration (logs ) [+] [ 0 ] [+] Stock markets Stock market capitalization (% of GDP, logs ) [+] [+] [+] Stock market total value traded (% of GDP, logs ) [+] [+] [ 0 ] Number of listed companies (logs) [ 0 ] [ - ] [+] Bond markets Private bond market capitalization (% of GDP, logs ) [ - ] [+] [ - ] Public bond market capitalization (% of GDP, logs ) [ - ] [+] [ 0 ] Note: IV coefficient estimate of foreign liabilities (as % of GDP). [+] denotes a positive and statistically significant relationship, [ - ] denotes a negative and statistically significant coefficient, and [ 0 ] indicates a lack of significant in their relationship.

18 Robustness analysis Our baseline estimates are robust to dropping from our sample: Sharp variations in financial variables High inflation episodes Does the structure of external assets and liabilities matter? Rising equity-related liability holdings leads to expansion of credit and deposits Rising debt liability holdings is associated with contractions in credit and deposits.

19 Role of Structural Features Test whether the impact of FO on DFD depends upon: Degree of institutional quality (IQ) Investor protection (IP) Extent of international trade integration (TO) We model the coefficient γ it as: FD γ it it = γ 0 + γ γ γ 1 it + 1 it + 1 FOit Prior: γ 1, γ 2, γ 3 >0 IQ IP TO it

20 Role of Structural Features Financial Integration and Domestic Financial Development: Role of Institutions, Investor Protection and Trade Openness Sample of 145 countries, (annual observations) Dependent variable expressed as % of GDP, in logs Bank credit Bank deposit Stock market capitalization Foreign Foreign Foreign Foreign Foreign Foreign Foreign Foreign Foreign Assets Liabilities A & L Assets Liabilities A & L Assets Liabilities A & L [1] [2] [3] [4] [5] [6] [7] [8] [9] Financial openness (FO) ** ** ** ** ** ** as % of GDP (in logs) (0.28) (0.25) (0.25) (0.20) (0.19) (0.19) (1.04) (0.93) (1.04) FO x Institutional quality ** ** ** ** ** ** (0.04) (0.02) (0.02) (0.03) (0.02) (0.02) (0.11) (0.06) (0.11) FO x Investor protection ** ** ** ** ** ** ** ** (0.12) (0.17) (0.17) (0.08) (0.13) (0.13) (0.63) (0.80) (0.63) FO x Trade openness ** ** ** ** ** ** (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.08) (0.08) (0.08) Nobs R** We control for income per capita, trade openness, inflation, and exchange rate regimes. All these control variables are lagged one period. Numbers in parenthesis are robust standard errors. * (**) implies significance at the 10 (5) percent level.

21 Role of Structural Features Foreign Liabilities (FL) Dependent variable FL FL x IQ FL x IP FL x TO Nobs. R**2 Activity of financial intermediaries Private credit by deposit money banks ** ** ** ** Private credit by financial system ** ** ** ** Bank deposits ** ** ** ** Financial deposits ** ** ** ** Size of financial intermediaries Total assets of deposit money banks ** ** ** ** Total assets of the financial system ** ** ** ** Liquid liabilities ** ** ** ** Efficiency of financial intermediaries Overhead costs Net interest margin ** ** * Bank concentration ** ** * Stock markets Stock market capitalization Stock market total value traded Number of listed companies ** ** ** Bond markets Private bond market capitalization ** ** ** ** Public bond market capitalization ** ** ** ** Numbers in parenthesis are robust standard errors. * (**) implies significance at the 10 (5) percent level.

22 Role of Structural Features The Role of Structural Characteristics: Conditional response of domestic financial development to an increase in financial openness (Response to doubling our measure of financial openness ) Response of bank credit Response of bank deposits Response of bank assets Response of liquid liabilities Structural Increase in: Increase in: Increase in: Increase in: Characteristic lfay lfly lfaly lfay lfly lfaly lfay lfly lfaly lfay lfly lfaly Institutions Low (L) [-] [0] [0] [-] [-] [-] [-] [0] [0] [0] [0] [0] Medium [0] [+] [+] [-] [-] [-] [+] [+] [+] [+] [+] [+] High (L) [+] [+] [+] [+] [0] [+] [+] [+] [+] [+] [+] [+] H vs. L (p-value ) (0.000) (0.000) (0.002) (0.000) (0.000) (0.000) (0.000) (0.001) (0.008) (0.000) (0.000) (0.003) Investor protection Low (L) [-] [-] [-] [-] [-] [-] [-] [-] [-] [+] [-] [-] Medium [0] [+] [+] [-] [-] [-] [+] [+] [+] [+] [+] [+] High (L) [+] [+] [+] [0] [+] [0] [+] [+] [+] [+] [+] [+] H vs. L (p-value ) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.168) (0.000) (0.000) Trade openness Low (L) [0] [0] [0] [-] [-] [-] [0] [0] [0] [+] [-] [-] Medium [0] [+] [+] [-] [-] [-] [+] [+] [+] [+] [+] [+] High (L) [+] [+] [+] [0] [0] [-] [+] [+] [+] [+] [+] [+] H vs. L (p-value ) (0.015) (0.000) (0.004) (0.005) (0.000) (0.001) (0.107) (0.000) (0.010) (0.047) (0.000) (0.000) Notes: Low (L) and high (H) levels of structural variables are computed as the 25th and 75th percentile of the distribution of the corresponding dimensions (with the other dimensions being evaluated at their medians). The p-value tests the null of equality in the effects whereas the alternative tests whether H is greater than L effect.

23 Conditional Responses Bank credit response to FO conditional on the level of institutional quality Bank deposit response to FO conditional on the level of institutional quality Low Median High Low Median High Low Median High Low Median High Doubling FA/Y Doubling FL/Y Doubling FA/Y Doubling FL/Y Bank credit response to FO conditional on the level of investor protection Bank deposit response to FO conditional on the level of investor protection Low Median High Low Median High Low Median High Low Median High Doubling FA/Y Doubling FL/Y Doubling FA/Y Doubling FL/Y Notes: Low (L) and high (H) levels of structural variables are computed as the 25th and 75th percentile of the distribution of the corresponding dimensions (with the other dimensions being evaluated at their medians). The extreme values of the lines crossing the bars represent the 95% confidence interval.

24 Conditional Responses Bank asset response to FO conditional on the level of institutional quality 0.2% 0.0% Net interest margin response to FO conditional on the level of institutional quality % -0.4% -0.6% -0.8% % Low Median High Low Median High -1.2% Low Median High Low Median High Doubling FA/Y Doubling FL/Y Doubling FA/Y Doubling FL/Y Bank asset response to FO conditional on the level of investor protection 0.2% 0.0% -0.2% -0.4% -0.6% -0.8% -1.0% -1.2% -1.4% -1.6% Net interest margin response to FO conditional on the level of investor protection Low Median High Low Median High -1.8% Low Median High Low Median High Doubling FA/Y Doubling FL/Y Doubling FA/Y Doubling FL/Y Notes: Low (L) and high (H) levels of structural variables are computed as the 25th and 75th percentile of the distribution of the corresponding dimensions (with the other dimensions being evaluated at their medians). The extreme values of the lines crossing the bars represent the 95% confidence interval.

25 Robustness analysis II The estimates of our augmented model are robust to: Excluding sharp variations in financial variables (associated to lending booms, equity bubbles, etc.) Changes in the measure of institutional quality Political institutions, quality of institutions, socio-economic environment, conflict Changes in the measure of investor protection Disclosure index, director liability, and shareholder suits

26 Conclusions World economy characterized by: Rising financial globalization Rapid increase in cross-border asset trade Deepening of local financial markets How does FO affect DFD? Theoretically ambiguous Empirical issue Possible to determine causal impact?

27 Conclusions Rising FO appears to enhance DFD by: Raising bank credit and total assets of the banking system Reducing net interest margins Increasing stock market capitalization and total value traded Improving the capitalization of private bond markets Response of DFD to greater FO depends on institutions, investor protection and trade openness. E.g. Banking size and activity expands in response to greater FO in countries with moderate to high levels of institutions, investor protection and trade openness

Understanding the Growth of African Financial Markets

Introduction Facts Review Empirical model Conclusions Understanding the Growth of African Financial Markets University of Rennes 1 - International Monetary Fund 2009 AFRICAN ECONOMIC CONFERENCE November