Foreign Bank Subsidiaries' Default Risk during the Global Crisis
What Factors Help Insulate Affiliates from their Parents?*
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
  • | 2 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

This paper examines the association between the default risk of foreign bank subsidiaries in developing countries and their parents during the global financial crisis, with the purpose of determining the size and sign of this correlation and, more importantly, understanding what factors can help insulate affiliates from their parents. We find evidence of a significant and robust positive correlation between parent banks’ and foreign subsidiaries’ default risk. This correlation is lower for subsidiaries that have a higher share of retail deposit funding and that are more independently managed from their parents. Host country bank regulations also influence the extent to which shocks to the parents affect the subsidiaries’ default risk. In particular, the correlation between the default risk of subsidiaries and their parents is lower for subsidiaries operating in countries that impose higher capital, reserve, provisioning, and disclosure requirements, and tougher restrictions on bank activities.

Abstract

This paper examines the association between the default risk of foreign bank subsidiaries in developing countries and their parents during the global financial crisis, with the purpose of determining the size and sign of this correlation and, more importantly, understanding what factors can help insulate affiliates from their parents. We find evidence of a significant and robust positive correlation between parent banks’ and foreign subsidiaries’ default risk. This correlation is lower for subsidiaries that have a higher share of retail deposit funding and that are more independently managed from their parents. Host country bank regulations also influence the extent to which shocks to the parents affect the subsidiaries’ default risk. In particular, the correlation between the default risk of subsidiaries and their parents is lower for subsidiaries operating in countries that impose higher capital, reserve, provisioning, and disclosure requirements, and tougher restrictions on bank activities.

I. Introduction

Since the late 1990s, the importance of multinational banks has grown dramatically. Between 1999 and 2009 the average share of bank assets held by foreign banks in developing countries rose from 26 percent to 46 percent.1 The bulk of the pre-global financial crisis evidence analyzing the consequences of this significant transformation in bank ownership suggests that foreign bank participation brought many benefits to developing countries including financial stability.2 In contrast, by highlighting the role of multinational banks in the transmission of shocks across countries, the recent Global Financial Crisis (GFC), reignited the debate on the benefits of having global banks operating in the domestic banking sector and on how jurisdictions where such banks operate should deal with them.

In this context, our paper focuses on two specific related policy relevant questions. First, we investigate whether there was a significant positive correlation during the global financial crisis between parent banks’ default risk (i.e., the likelihood that a bank would fail and not be able to meet its debt obligations) and the default risk of their foreign affiliates in developing countries. This association is at the center of much of the policy debate surrounding the destabilizing impact of parent banks on host countries’ banking systems. A priori, there are a number of reasons to expect this association to be positive. Given the evidence on the existence of an internal capital market, where a global bank can “move” capital and liquidity across the locations where it operates (de Haas and van Lelyveld 2010, Cetorelli and Goldberg 2012, Cerutti and Claessens 2016), we would expect that when the default risk of the parent rises, in order to cover the losses and meet regulatory capital requirements at home, the global bank might channel funds and capital from its subsidiaries and, in the process, weaken the financial health of subsidiaries, causing their default risk to rise. Furthermore, even if subsidiaries’ capital and liquidity remain intact, when the parent bank faces a shock, the perception that the subsidiaries no longer could obtain support from the parent if it were needed (i.e., the loss of the implicit guarantees) could lead to a drop in the market value of the subsidiaries and a corresponding rise in their default risk. Also, the default risk of a parent and that of its subsidiaries might be correlated if these institutions share a common business model or focus and are exposed to similar shocks. Although the aforementioned factors suggest a positive relationship between the default risk of parent banks and their subsidiaries, the magnitude and the significance of this relationship remain unclear. The GFC represents a unique opportunity for corroborating the existence of a potential positive correlation between the default risk of the parent and that of its subsidiaries, and for measuring its size. In this sense, our paper provides empirical evidence that helps shed light on this important policy question.

Second, we investigate the factors that amplify or dampen the correlation between the foreign bank parents’ and their subsidiaries’ changes in default risks. In particular, we analyze the role of subsidiary financial characteristics (such as capital and funding structure) and the impact of host country bank regulations (such as bank capital, reserve requirements, bank activities). The question of how host regulators can limit the transmission of shocks to the affiliates of foreign banks that operate in their countries is very important given the significant presence of foreign banks in many developing economies and the issue is related to the recent discussion on ring-fencing (see Song 2004, Cerutti et al. 2010, Cerutti and Schmieder 2014, and D’Hulster 2014). In cross-border banking, ring-fencing refers to restrictions (whether regulatory or supervisory) on internal transfers of banks’ capital, liquidity, and profitability across jurisdictions within the same international banking group. To our knowledge, this is the first paper to examine the association between the default risk of foreign parents and their subsidiaries and the impact of subsidiary characteristics and host country regulations in limiting this correlation.

We use data for 93 publicly listed foreign bank subsidiaries, operating in 36 host developing countries and owned by 41 parent bank groups, headquartered in 24 home countries, during the period from September 2008 to December 2009, to compute the measure of default risk for global banks and their affiliates. In particular, we estimate the weekly correlation between parents’ and subsidiaries’ distance to default and investigate the factors that affect this correlation. Distance to default, which is based on Merton’s (1974) structural credit risk model, is the difference between the market asset value of the bank and the face value of its debt, scaled by the standard deviation of the bank’s asset value. Hence, distance to default is inversely related to default risk. Our focus on developing countries as hosts of foreign bank subsidiaries is driven by the fact that these countries were not at the core of the global financial crisis, allowing us to better identify factors that might help insulate affiliates from their parents potentially in trouble.

Our empirical findings show that foreign bank subsidiaries’ distance to default are significantly correlated with their parent banks’ distance to default, even when we account for the distance to default of other banks and firms in the home and host countries, as well as for global factors that may influence subsidiaries’ distance to default. In particular, the correlation in the distance to default between foreign subsidiaries and their parents ranges between 0.3 and 0.2, after we include subsidiary fixed effects, macro factors, and controls for distance to default of other banks and firms in the home and host countries. This finding is robust to the sample of banks considered and to the way we calculate the distance to default measure. Also, we find that certain subsidiary characteristics influence the correlation in the distance to default between subsidiaries and parents. In particular, this correlation is lower for subsidiaries that have a higher share of retail deposit funding and for those that are more independently managed from their parents. These results hold even after controlling for host country – parent distance to default interactions, which capture the impact of the institutional/regulatory environment. Finally, the regulatory regime in place in the host countries also affects the extent to which shocks to the parents’ distance to default influence subsidiaries. In particular, the correlation between the distance to default of the subsidiaries and the parents is lower for subsidiaries operating in host countries that impose higher capital, reserve, provisioning and disclosure requirements, and tougher restrictions on bank activities.

Our paper is related to the literature that explores how foreign banks transmit shocks. Most of the existing studies have examined the transmission of shocks through the lending channel: i.e., how shocks to the parent banks lead foreign bank subsidiaries to reduce the loans they make in host jurisdictions during a crisis. In particular, a number of papers, including some before the recent global crisis, have documented that lending by foreign bank affiliates’ declines when parent banks’ financial conditions deteriorate. Peek and Rosengren (2000) offer evidence based on the behavior of Japanese banks operating in the US during the 1990s Japanese crisis. Schnabl (2012) studies the lending behavior of foreign bank affiliates in Peru in the aftermath of the 1998 Russian crisis. In the context of the recent global crisis, Claessens and van Horen (2013), Choi et al. (2013), and de Haas and van Horen (2013), among others, show that foreign bank lending across countries declined more than domestic bank lending during the period 2008-2009. The last two studies, in particular, find that foreign bank affiliates whose parents’ relied more on wholesale funding (in the case of de Haas and van Horen 2013) and had lower capital ratios (in the case of Choi et al. 2013) experienced a sharper decline in lending.

Our paper contributes to the literature that examines the extent to which foreign banks transmit external shocks by focusing on a previously unexamined mechanism: the correlation in default risks between a global bank and its subsidiaries. This is an important channel because default risk is arguably the ultimate risk that matters for banking stability and the risk that supervisors are likely to care about the most. Default risk has implications for both bank debtors and borrowers since a failing bank is one that will not be able to pay its depositors or to continue to make loans. Similarly, bank failures can have significant implications for governments that have to shoulder the cost of paying insured depositors if a bank fails or of recapitalizing the bank if it is deemed too large to fail. Regardless of how a bank failure is resolved, the economic losses associated with such an event are likely to be larger than those caused by a bank that temporarily diminishes its lending as a result of a shock. Hence, our analysis of the correlation between the default risk of global banks and their subsidiaries and the factors that can affect this correlation is very relevant and makes a useful contribution to the existing literature. Moreover, since we measure default risk based on daily stock market data, we are able to use high frequency and forward-looking information to analyze the transmission of shocks, which is also another advantage relative to the literature that uses lower frequency loan data.

The rest of the paper is structured as follows. Section 2 describes our data. Section 3 details the empirical methodology we use to (a) calculate the distance to default of parent banks and, separately, their subsidiaries, (b) estimate the correlation between the distance to default of the parents and their subsidiaries, and (c) investigate the factors that affect the correlation between the default risk of parent banks and subsidiaries. Section 4 presents results from our econometric analysis. Section 5 concludes.

II. Data

We assembled an original and extensive database of stock market prices and balance sheet characteristics for both publicly traded parent banks and their publicly traded subsidiaries in developing countries. Our sample consists of 93 publicly listed foreign subsidiaries, operating in 36 host developing countries and owned by 41 parent bank groups, headquartered in 24 home countries (see Table 1). Our period of analysis is the peak of the global crisis: from September 2008 to December 2009.

Table 1:

List of foreign bank parents and subsidiaries in the sample

This table lists the subsidiary banks, their host countries, their parents and the parents’ home countries used in the analysis.

article image
article image
article image
article image
article image

Even though the presence of foreign banks has increased in recent decades, the final sample of subsidiaries that we were able to include in the analysis is smaller as the result of two constraints. First, most foreign subsidiaries are not listed in the stock market, since they are privately held. To identify the sample of foreign bank subsidiaries listed in developing countries’ stock markets, we checked Bloomberg, Compustat, and also conducted many web-searches (including the webpages of the bank regulators and the stock markets in developing countries), and then identified their controlling groups (using Bankscope, Bankers’ Almanacs, central banks/regulatory agencies’ information, and web-searches). As a result, we were able to identify about 167 listed banks operating in about 44 host countries where foreign banks held important ownership stakes. Second, of the banks that are listed in host countries’ stock markets, there are several cases that are not traded often, since parent banks control most shares (e.g., 98 percent or more ownership), reducing the final sample to only 93 subsidiaries. The median ownership stake in the sample is about 61 percent.3 The limitations to fulfilling the necessary data requirements for our analysis does not seem to bias the representation of the final sample, which covers 36 host countries.

The dataset used in the analysis also includes stock market prices for all other banks and firms in the home and host countries that are used to construct default risk control variables in the regressions. We use daily stock market information from Compustat Global for international banks and firms and stock market information from CRSP for U.S. banks and firms. Bank level variables are constructed from Bankscope, a commercial database of banks’ financial statements produced by Bureau Van Dijk.4 Since we are interested in how bank characteristics affect the correlation between parents’ and their foreign subsidiaries’ default risks, and since there have been significant changes to bank balance sheets during the crisis, we construct and use bank-level variables measured prior to the crisis (as of December 2006). For each bank, we calculate relative bank size (bank assets to total system assets), capital ratio (regulatory capital to risk weighted assets), equity ratio (equity to total assets), provisions (loan loss provisions divided by total loans), deposit funding (deposits divided by total funding), profitability (net income divided by total assets), and liquidity (liquid assets divided by total assets).5 We winsorize all financial variables at the 1st and 99th percentile level of their distributions to reduce the influence of outliers and potential data errors.

Host country-level variables are collected from a number of sources. We use data from the World Bank 2007 Bank Regulation and Supervision survey to construct different indexes of bank regulation and supervision, following the methodology proposed by Barth et al. (2001, 2013).6 Capital regulation captures the amount of capital banks must hold and the stringency of regulations on the nature and source of capital. It is an index ranging in value from 0 to 10, with higher values indicating greater stringency. Activity restrictions is an index that measures the degree to which the national regulatory authorities restrict banks from engaging in securities, insurance, and real estate activities. Securities activities refer to underwriting, brokering, dealing and all aspects of the mutual fund industry. Insurance activities include underwriting and selling, and real estate activities refer to investment, development, and management. The activities restrictions index takes values from 3 (where each of the three activities is permitted) to 12 (where each activity is prohibited). Disclosure requirements is an index that captures the type of information banks must disclose about their financial condition. It indicates whether the income statement includes accrued or unpaid interest or principal on nonperforming loans, whether banks are required to produce consolidated financial statements, and whether bank directors are legally liable if information disclosed is erroneous or misleading. The variable ranges from 0 to 3, with higher values indicating greater bank disclosure. Diversification requirements is an index which measures whether regulations support geographical asset diversification. It is based on two variables: whether there are explicit, verifiable, and quantifiable guidelines for asset diversification and whether banks are prohibited from making loans abroad. The index takes values from 0 to 2 with higher values indicating more diversification. Loan classification stringency measures the actual minimum number of days beyond which a loan in arrears must be classified as sub-standard, doubtful, or loss. Provisioning stringency measures the minimum provisions (as a percentage of loans) required as a loan is successively classified as sub-standard, doubtful, and lastly as loss. Supervisory powers is an index measuring supervisory authorities’ power and authority to take specific preventive and corrective actions. The measure ranges from 0 to 14, where larger numbers indicate greater supervisory powers. Prompt corrective powers measures the extent to which the law establishes predetermined levels of bank solvency deterioration that force automatic enforcement actions such as intervention, and the extent to which supervisors have the requisite suitable powers to do so. The index ranges from 0 to 6 with higher values indicating more promptness in responding to problems. Reserve requirements is the average level of reserves that banks are required to hold relative to their deposits and other short-term liabilities. Financial outflow restrictions is the average of three binary variables measuring bank restrictions on lending to non-residents, maintaining accounts abroad, and on banks’ investment abroad. This variable comes from the IMF Annual Report on Exchange Arrangements and Restrictions.

In some estimations, we also control for macro factors that may affect the changes in default risk of the parents and their subsidiaries. ΔVIX is the change in the Chicago Board Options Exchange Volatility Index, which measures the 30-day expected volatility calculated from implied volatilities from S&P 500 index options. VIX data is obtained from the Chicago Board Options Exchange (CBOE). ΔDEF is the change in the default spread measured as the difference in Baa-Aaa yields of US firms. Data comes from the interest rate data releases from the Federal Reserve Board. Table 2 lists all the variables used in our analysis, provides their definition, data sources and descriptive statistics.

Table 2:

Variable definition and descriptive statistics

This table lists the definitions, sources and the summary statistics for the variables used in this study

article image
article image

III. Empirical Methodology

3.A. Computing distance to default measures between parents and subsidiaries

Our main measure of default risk is the distance to default that comes from the structural credit risk model of Merton (1974). Distance to default is computed as the difference between the market asset value of the bank and the face value of its debt, scaled by the standard deviation of the bank’s asset value.7 In the Merton (1974) model, the market equity value of a bank is modeled as a call option on the company’s assets:

VE=VAeDivTN(d1)XerTN(d2)+(1eDivT)VAd1=log(VAX)+(rDiv+sA22)TsAT;d2=d1sAT(1)

In equation (1), VE is the market value of a bank’s equity. VA is the market value of the bank’s assets. X is the face value of debt maturing at time T.8 r is the risk-free rate and Div is the dividend rate expressed in terms of VA. sA is the volatility of the value of assets, which is related to equity volatility (SE) through the following equation:

sE=VAeDivTN(d1)sAVE(2)

We simultaneously solve the above two equations to find the values of VA and sA. We use the market value of equity for VE and total liabilities to proxy for the face value of debt X. Since the accounting information is on an annual basis, we linearly interpolate the values for all dates over the period, using beginning and end of year values for accounting items. The interpolation method has the advantage of producing a smooth implied asset value process and avoids jumps in the implied default probabilities at year end (Bartram et al. 2007). sE is the standard deviation of daily equity returns over the past 3 months. In calculating the standard deviation, we require each bank to have at least 45 non-missing daily returns over the previous three months. T is the horizon over which default risk is computed and is set to one year. r is the one year US treasury yield, which we take to be the risk free rate. We use the Newton method to simultaneously solve the two equations above. For starting values for the unknown variables, we use VA = VE + X and sA = sEVE/(VE+X). We winsorize sE and VE/(VE+X) at the 1st and 99th percentile levels to reduce the influence of outliers. After we determine asset values VA, we follow Campbell, Hilscher and Szilagyi (2008) and assign asset return m to be equal to the equity premium (6%).9 Merton’s distance to default (dd) is finally computed as:10

dd=log(VAX)+(mDivsA22)TsAT(3)

As a robustness check, we compute a simplified version of the Merton formula following Byström (2006). This measure does not rely on distributional assumptions and makes the default risk less sensitive to the leverage ratio at very high levels equity volatility. Byström (2006) shows that, when applied to a sample of US firms, the simplified model provides the same relative default risk rankings as the Merton model.11 The simplified formula we use is given by:

log(X/(VE+X))/(X/(VE+X)-1)×sE.(4)

3.B. Estimating the size and determinants of the correlation between parents and subsidiaries

To examine the correlation between the foreign bank parents’ and their subsidiaries’ changes in distance to default, we estimate equation (5) below:

Δdd_Subi,t=β0+β1Δdd_Parenti,t+β2Δdd_Homet+β3Δdd_Hostt+β4ΔDEFt+β5ΔVIXt+γi+ɛi,t

where Δdd_Subi,t is the weekly change in distance to default of the subsidiary i in week t; Δdd_Parenti,t is the weekly change in distance to default of the parent of subsidiary i; Δdd_Hostt and dd_Homet are changes in average distance to defaults of all the publicly traded banks and companies in the host country and home or parent bank country (excluding the foreign subsidiary and parent banks in question), respectively. These variables are included to control for the overall financial health of host and home companies, respectively. ΔDEFt is the change in interest rate spread between Bbb and Aaa rated companies and is included to capture innovations in the default risk premium. ΔVIXt is the weekly change in the VIX volatility index. This variable is included to capture innovations in macro volatility that affect all banks in our sample. In some estimations, we replace ΔDEFt and ΔVIXt with time dummies that are able to capture the impact of global factors that can influence parent banks and their subsidiaries throughout the world. Finally, we also include subsidiary fixed effects, γi, to control for time invariant heterogeneity across subsidiaries.

Since we are interested in uncovering factors that may amplify or dampen the correlation between the foreign bank parents’ and their subsidiaries’ changes in distance to default, we include firm level characteristics and country level regulations in the regression specified in (5) and interactions of these variables with Δdd+_Parenti,t. In particular, we estimate equation (6) below:

ΔddSubi,t=β0+β1ΔddParenti,t+β2ΔddHomet+β3ΔddHostt+β4ΔDEFt+β5ΔVIXt+β6Xi×ΔddParenti,t+β7Mc×Δdd_Parenti,t+γi+ɛi,t(6)

where Xi are foreign subsidiary characteristics (size, liquidity, funding structure, capital, etc.) computed as of December 2006, as described above. Mc are host country regulations measured as of 2006. The errors are clustered at the host country level.

As discussed in the introduction, either due to the existence of an internal capital market by which a failing parent might reallocate capital and funding from its subsidiaries, weakening their financial heath, or because of the perception that a weaker parent will be less able to support a subsidiary in trouble down the line, or because both subsidiaries and parents might share a similar business model and be similarly affected by shocks, we expect to find a positive correlation between foreign banks’ subsidiaries and parent banks’ distance to default. The size of this correlation is an empirical question that we hope to address. At the same time, we hope to ascertain the extent of cross-sectional heterogeneity in the default correlation between the parents and their subsidiaries. In particular, we consider the role of subsidiary financial characteristics as well as host country bank regulations in dampening the effect of shocks from the parent banks.

Following the theoretical literature on financial contagion, we focus on four sets of bank characteristics that can potentially influence the default correlation between the parents and their subsidiaries. In particular, a number of theoretical papers emphasize the role of bank size in the transmission of liquidity and economic shocks. Others emphasize the role of capital, profitability and liquidity as potential buffers in absorbing these shocks.12 We expect subsidiaries that are more profitable and that have high quality assets on their balance sheets prior to the financial crisis to be better positioned to absorb shocks from their parents. We use provisions to proxy for asset quality and return on assets to proxy for profitability. We also expect larger banks in host countries to exhibit a lower correlation with the default risk of their parents, as they are more likely to benefit from potential too-big-to-fail guarantees (Acharya, Anginer and Warburton 2014) and more likely to be subject to more stringent oversight by local supervisors.13 We use total assets as a measure of bank size.

Theoretical papers emphasize the importance of both asset and funding liquidity in the transmission of shocks in banking systems. Reliance on non-deposit funding has been shown to be a significant driver of systemic fragility (Adrian and Brunnermeir 2016, Anginer and Demirguc-Kunt 2014a). The volume and price of wholesale funding can adjust quickly resulting in liquidity problems especially for banks that hold significant amount of illiquid assets on their balance sheets (Brunnermeir and Pederson 2009).14 We use local deposit funding and liquid assets as a percentage of total assets to proxy for funding and asset liquidity, respectively. Theoretical papers also emphasize the role of capital as a buffer in absorbing earnings and liquidity shocks (Repullo 2004 and Von Thadden 2004).15 We use both tangible equity and regulatory capital over risk-weighted assets as measures of capital.16

In addition to the subsidiary financial characteristics, we also explore the importance of two measures of distance/proximity between the subsidiary and the parent: geographical distance (log of distance, measured in kilometers, between the parent/home country and the host country) and cultural proximity (as measured by whether the subsidiary and the parent have a common official language).17 A priori, we expect geographical distance to reduce the correlation between the parent and subsidiary distance to default, since geographically distant subsidiaries may be less integrated into the parent group and the parent bank might find it difficult to exercise control over the local management of the subsidiary (Rajan, Servaes, and Zingales 2000; de Haas and van Horen 2013). When it comes to cultural proximity, the impact might be more ambiguous. On the one hand, cultural distance might operate like geographical distance and reduce the correlation between subsidiaries’ and parents’ default risk because more distant subsidiaries might be harder to monitor. On the other hand, cultural proximity might reduce the correlation between subsidiaries and parents if these subsidiaries are granted more independence because the parent is more comfortable decentralizing some control when it is more familiar with the culture and business environment in the host country.18

In addition to the bank level characteristics described above, we also investigate the influence of measures of the host country regulatory environment on the default correlation between the parent banks and their subsidiaries. Prior literature suggests that countries with efficient supervision and monitoring of financial institutions respond better to shocks to their banking systems (Anginer and Demirguc-Kunt 2014b, and Demirguc-Kunt and Detragiache 1999). In this sense, we expect the default risk correlation between the subsidiary and the parent to be lower for subsidiaries operating in countries where the regulatory authorities impose tighter regulatory regimes, which de jure or de facto help to ring fence the foreign subsidiary from a parent in distress. In particular, we examine the influence of four sets of regulatory and supervisory rules that were in place prior to the financial crisis.

The first set of host country regulations we examine relate to the stringency of capital, provisioning, and bad loan classification rules, as well as those pertaining to reserve requirements. Empirical evidence suggest that more stringent capital rules increase bank holdings of equity (Cihak et al. 2013), reduce bank risk-taking (Barth et al. 2004, and Laeven and Levine 2008) and attenuate systemic risk in financial systems (Anginer and Demirguc-Kunt 2014b). We expect more stringent capital and related rules to reduce the default risk correlation between the parent banks and their subsidiaries.

The second set of regulatory variables we consider relate to activity restrictions, diversification requirements, and restrictions on financial outflows from the host country. Financial contagion can be influenced by the extent to which host regulators restrict banks from engaging in certain business activities. Non-traditional banking activities such as trading, underwriting and investment banking can cause conflicts of interest (John et al. 1994) and increase risk-taking (Boyd et al. 1998, and Brunnermeier et al. 2012), making banks more fragile to outside shocks. Lack of guidelines to diversify risk and assets (Anginer, Demiguc-Kunt and Zhu 2014b) and lack of restrictions on intra-group cross-border asset outflows (or inflows of doubtful assets from the parent) can also increase the fragility of subsidiaries.

The quality and effectiveness of supervisors in host countries can also have an influence on the severity of outside shocks on local banks.19 More powerful supervisors can insert modularity into the financial system by allowing subsidiaries to operate on a standalone basis or by protecting subsidiaries against cross-border risks (Schwarcz 2013). Effective monitoring and supervision, however, requires authorities to have the ability to take timely corrective action (Barth et al 2004). Hence, our third set of regulatory variables relate to supervisory power as well as the legal authority of supervisors to take prompt corrective action.

Finally, information asymmetry provides a potential channel in which shocks can be propagated through the banking system.20 We expect shocks from the parent to have a less severe effect on host banks that have more transparent and informative balance sheets. Greater information availability and transparency would also allow for better private monitoring of financial institutions (Djankov, McLiesh and Shleifer 2007).21 We use disclosure requirements as a measure of transparency and of the information content of bank financial statements.

IV. Results

4.1. The correlation between the parent and subsidiary distance to default

Figure 1 shows the median changes in Merton’s distance to default for all parent banks and, separately, for all subsidiaries over the period September 2008 to December 2009. It is clear from the figure that there is a very high correlation between changes in parent and subsidiaries distance to default. Because this figure does not control for other factors that can jointly influence these variables, we turn next to our empirical estimations that control for global factors and for changes in the distance to default of all firms operating in the corresponding parent and host countries.

Figure 1:
Figure 1:

Changes in the distance to default of all multinational parent banks and their foreign bank subsidiaries, 2008-2009

Citation: IMF Working Papers 2016, 109; 10.5089/9781484382189.001.A001

Columns (2.1)-(2.4) in Table 3 show that foreign bank subsidiaries’ distance to default, measured following Merton (1974), is significantly correlated with parent banks’ distance to default, even when we control for the average distance to default of all companies in the home and host countries, respectively, and when we account for global factors like the VIX and the corporate credit spread. The correlation between the distance to default of foreign subsidiaries and their parents varies between a maximum of 0.45, when only subsidiary fixed effects are added, and a minimum of 0.27, when we include time and subsidiary dummies. This correlation is not only statistically and economically significant, but also it is almost twice as large as the correlation of the distance to default of the foreign subsidiaries vis-à-vis all companies in the home and host countries. When we use the simplified measure of distance to default proposed by Byström (2006) the correlation between foreign subsidiaries and their parents is still highly significant and ranges between 0.2 and 0.3.

Table 3:

The association between parent banks’ and subsidiaries’ distance to default

Regression results for the model Δdd_Subi,t = β0 + β1Δdd_Parenti,t + β2 Δdd_Homet + β3Δdd_Hostt + β4ΔDEFt + β5ΔVIXt + γi + εi,t are reported in this table. Δdd_Subi,t is the weekly change in distance to default of the subsidiary i in week t; Δdd_Parenti,t is the weekly change in distance to default of the parent of subsidiary i; Δdd_Hostt and Δdd_Homet are changes in average distance to defaults of all the publicly traded banks and companies in the host country and parent bank country (excluding the foreign subsidiary and parent banks in question), respectively. ΔDEFt is the change in interest rate spread between Bbb and Aaa rated companies. ΔVIXt is the weekly change in the VIX volatility index. Regressions also include subsidiary fixed effects, γi and estimations (2.4) and (2.8) replace ΔDEFt and ΔVIXt with time dummies to account for time fixed effects. Estimates (2.1) through (2.4) show estimations where the measure of distance to default follow Merton (1974), whereas estimations in column (2.5) through (2.8) use Byström (2006) distance to default measure. Standard errors are reported below coefficient estimates in parentheses and are clustered at the host country level. ***, ** and * indicate the 1%, 5%, and 10% level of significance, respectively.

article image

In unreported regressions, we assess whether excluding some parents banks, which own multiple subsidiaries in our sample and account for many of the observations (e.g., Barclays, BNP Paribas, Citibank, HSBC, ING, Society General, and Standard Charter among others), affects our results. We find that the estimates of the association between the distance to default of foreign bank subsidiaries and their parents do not change much when we exclude parent banks with multiple subsidiaries. The correlation varies between 0.26 and 0.29.

4.2 The factors that affect the correlation between the parent and subsidiary distance to default

Table 4 explores whether the association between the Merton distance to default of foreign bank subsidiaries and that of their parents changes depending on the value of different subsidiary characteristics, namely: size, capitalization, funding structure, liquidity, profitability, provisioning, and distance from the parent (both geographical and cultural). We find that parent banks’ distance to default is less correlated with the subsidiaries’ distance to default when subsidiaries have higher deposit funding ratios.22 Also, the association between the parents and the subsidiaries distance to default is lower for countries that are physically distant or culturally closer. In economic terms, we find that a one standard deviation increase in the deposit funding ratio (geographical distance) lowers the association between the distance to default of the parents and the subsidiaries from 0.28 (0.32) to 0.21 (0.27). Similarly, for subsidiaries that are culturally close to the parents the correlation between the parents and the subsidiaries is 0.33, while it is 0.13 for those that are not culturally similar.

Table 4:

The impact of subsidiaries’ characteristics on the association between foreign subsidiaries’ and parents’ distance to default

Regression results for the model Δdd_Subi,t = β0 + β1Δdd_Parenti,t + β2 Δdd_Homet + β3Δdd_Hostt + β4Xi × Δdd_Parenti,t + γi + δt + εi,t are reported in this table. Δdd_Subi,t is the weekly change in distance to default of the subsidiary i in week t; Δdd_Parent is the weekly change in distance to default of the parent of subsidiary i; Δdd_Hostt and Δdd_Homet are changes in average distance to defaults of all the publicly traded banks and companies in the host country and parent bank country, respectively. Xi are foreign subsidiary characteristics computed as of December 2006. These variables are described in detail in Table 2. Regressions also include subsidiary fixed effects, γi and time (weekly) fixed effects δt. Standard errors are reported below coefficient estimates in parentheses and are clustered at the host country level. ***, ** and * indicate the 1%, 5%, and 10% level of significance, respectively.

article image

How should we interpret the significance of geographical distance and cultural proximity? We view geographical distance and cultural proximity as proxies for more independent management of the subsidiaries from the parents. Though, we are unable to conclusively confirm this hypothesis, we are able to offer some suggestive evidence for a subset of banks. In particular, for 47 subsidiaries, using information obtained from banks’ annual reports, we were able to construct a proxy for subsidiary management independence from the parent: the share of declared independent board members (i.e., ratio of members identified as being independent because they own no or a small number of shares in the bank, are not clients or suppliers of the bank and do not have family members working in the bank). The share of independent board members interacted with the parent distance to default is positively and significantly correlated with the interaction of parent distance to default with the measures of geographical distance and cultural proximity: the correlation is 0.92 with geographical distance and 0.65 with cultural proximity. Hence, we interpret the negative interaction of the geographical distance and cultural proximity measures with the distance of default of the parent as suggestive of the fact that more independently managed subsidiaries exhibit a lower correlation between the measures of distance to default of the subsidiaries and parents.

Table 5 allows us to corroborate the significance of the subsidiary characteristics, even when we control for host country dummies-parent distance to default interactions, which are included but not reported. These interactions are intended to account for any host country factors that are non-time varying over our period of analysis (e.g., features of the institutional/regulatory environment) that can reduce the effect of the parent distance to default. As before, we find that for subsidiaries that have higher retail deposit funding ratios and that are culturally closer to the parent, the correlation between the subsidiaries’ and the parents’ distance to default is lower.

Table 5:

The impact of subsidiaries’ characteristics controlling for host country dummies-parent distance to default interactions

Regression results for the model Δdd_Subi,t = β0 + β1Δdd_Parenti,t + β2 Δdd_Homet + β3Δdd_Hostt + β4Xi × Δdd_Parenti,t + β5γi × Δdd_Parenti,t + δt + γi + εi,t are reported in this table. Δdd_Subi,t is the weekly change in distance to default of the subsidiary i in week t; Δdd_Parenti,t is the weekly change in distance to default of the parent of subsidiary i; Δdd_Hostt and Δdd_Homet are changes in average distance to defaults of all the publicly traded banks and companies in the host country and parent bank, respectively. Xi are foreign subsidiary characteristics computed as of December 2006. These variables are described in detail in Table 2. Regressions also include host country dummies-parent distance to default interactions, γi × Δdd_Parenti,t, which are included but not reported. Regressions also include subsidiary fixed effects, γi and time (weekly) fixed effects, γt. Standard errors are reported below coefficient estimates in parentheses and are clustered at the host country level. ***, ** and * indicate the 1%, 5%, and 10% level of significance, respectively.

article image

Table 6 investigates how the correlation between foreign subsidiaries’ and parents’ distance to default changes depending on the banking regulations adopted by host jurisdictions. We find that in host countries where regulators impose greater disclosure, capital, reserve, and provisioning requirements and where the range of activities banks can undertake is more limited,23 the correlation between the parent banks’ and the subsidiaries’ distance to default is lower. For example, a one standard deviation increase in the index of capital regulation lowers the correlation of the distance to defaults from 0.29 to 0.18. The economic impact for all other statistically significant variables is roughly of the same magnitude.

Table 6:

The impact of host countries’ regulations on the association between foreign subsidiaries’ and parents’ distance to default

Regression results for the model Δdd_Subi,t = β0 + β1Δdd_Parenti,t + β2 Δdd_Homet + β3Δdd_Hostt + β6Mc × Δdd_Parenti,t + δt + γi + εi,t are reported in this table. Δdd_Subi,t is the weekly change in distance to default of the subsidiary i in week t; Δdd_Parent is the weekly change in distance to default of the parent of subsidiary i; Δdd_Hostt and Δdd_Homet are changes in average distance to defaults of all the publicly traded banks and companies in the host country and parent bank country (excluding the foreign subsidiary and parent banks in question), respectively. Mc are subsidiary host country bank regulations measured as of December, 2006. These variables are described in detail in Table 2. Regressions also include subsidiary fixed effects, γi and time (weekly) fixed effects, δt. Standard errors are reported below coefficient estimates in parentheses and are clustered at the host country level. ***, ** and * indicate the 1%, 5%, and 10% level of significance, respectively.

article image

V. Conclusions

While many papers have examined how foreign bank parent conditions affect lending by their overseas subsidiaries, this paper is the first to analyze the correlation between parents’ and subsidiaries’ default risk. More importantly, we also analyze the subsidiary characteristics and policies that can dampen or amplify this correlation. These issues are important because they allow host countries to assess how exposed they are to shocks affecting multinational banks and what factors can help reduce this exposure.

Our analysis shows that there is a statistically and economically significant positive correlation between parents’ and subsidiaries’ distance to default. This finding is robust to the way we calculate the distance to default and to the sample of banks considered. Also, we find that the correlation in the distance to default between subsidiaries and parents varies with certain subsidiary characteristics. In particular, this correlation is lower for subsidiaries that have higher deposit funding ratios and that are more independently managed from the parent. Finally, the regulatory system in place in the host country also influences the extent to which shocks to the parents distance to default influence subsidiaries. In particular, the correlation between the distance to default of the subsidiary and the parent is lower for subsidiaries operating in countries that impose higher capital, reserve, provisioning, and disclosure requirements and tougher restrictions on bank activities.

From an individual host country’s policy perspective, our findings indicate that tighter host banking regulations seem to help insulate foreign subsidiaries from changes in the default risk of parent banks during crises. This could lead domestic regulators towards trying to minimize default correlations with the parent banks, while, at the same time, trying to minimize the costs from more regulation in the host country. This tradeoff would be present especially if the tighter regulation needed to insulate foreign subsidiaries from the parent also needed to cover domestic banks (e.g. due to uniformity of treatment principles). However, it is important to note that this solution may not necessarily be optimal from a global perspective. First, ring fencing measures taken by authorities in one country could increase stress on the banking group’s legal entities in other jurisdictions or for the banking group as a whole. Second, ring fencing may create inefficiencies in the allocation of capital and liquidity within multinational bank groups. These potential downsides from ring fencing practices by host regulators have been highlighted in the Basel Committee’s Report and Recommendations of the Cross-Border Bank Resolution Group (CBBRG). Furthermore, in light of the concerns about ring fencing practices, the CBRG has called for the establishment of a credible framework for cooperation across national supervisors and for uniform mechanisms for the resolution of cross-border banking groups to help avoid unilateral and likely more costly solutions.

VI. References

  • Acharya, V., Anginer, D., Warburton, A.J., 2014. “The End of Market Discipline? Investor Expectations of Implicit Government Guarantees.” http://risk.econ.queensu.ca/wpcontent/uploads/2013/10/Acharya-et-al-End-of-Market-Discipline-20141.pdf

    • Search Google Scholar
    • Export Citation
  • Adrian, T., Brunnermeier, M., 2016. “CoVaR”. American Economic Review, Forthcoming.

  • Allen, F., Gale, D., 2000. “Bubbles and crises,” The Economic Journal 110, 236255.

  • Anginer, D., Cerutti, E., Martinez Peria, M.S., 2014. “Foreign bank subsidiaries’ default risk during the global crisis: What factors help insulate affiliates from their parents?World Bank Policy Research Working Paper Series 7053.

    • Search Google Scholar
    • Export Citation
  • Anginer, D., Demirguc-Kunt, A., 2014a. “Bank capital and systemic stability?World Bank Policy Research Working Paper Series 6948.

  • Anginer, D., Demirguc-Kunt, A., 2014b. “Has the global banking system become more fragile over time?Journal of Financial Stability 13, 202213.

    • Search Google Scholar
    • Export Citation
  • Atkeson, A., Eisfeldt, A., Weill, P., 2014. “The market for OTC Derivatives,” NBER Working Paper No. 18912.

  • Barlevy, G., Veronesi, P., 2003. “Rational Panics and Stock Market Crashes,” Journal of Economic Theory, 110, 234263.

  • Barth, J., Caprio, G., Levine, R., 2001. “The regulation and supervision of banks around the world - a new database,” World Bank Policy Research Working Paper 2588.

    • Search Google Scholar
    • Export Citation
  • Barth, J., Caprio, G., Levine, R., 2013. “Bank regulation and supervision in 180 countries from 1999 to 2011.” National Bureau of Economic Research Working Paper 18733.

    • Search Google Scholar
    • Export Citation
  • Barth, J., Caprio, G., Levine, R., 2004. “Bank regulation and supervision: What works best,” Journal of Financial Intermediation 12, 205248.

    • Search Google Scholar
    • Export Citation
  • Bartram, S., Gregory, B., Hund, J., 2007. “Estimating systemic risk in the international financial system,” Journal of Financial Economics 86(3), 835869.

    • Search Google Scholar
    • Export Citation
  • Basel Committee on Banking Supervision, 2010. Report and recommendations of the Cross-Border Resolution Group, March, Basel.

  • Berger, A., Bouwman, C., 2013. “How does capital affect bank performance during financial crises?Journal of Financial Economics 109, 146176.

    • Search Google Scholar
    • Export Citation
  • Bharath, S., Shumway, T., 2008. “Forecasting default with the Merton distance to default model,” Review of Financial Studies, 21 (3), 13391369.

    • Search Google Scholar
    • Export Citation
  • Boyd, J., Chang, C., Smith, B., 1998. “Moral hazard under commercial and universal banking,” Journal of Money, Credit, and Banking 30, 426468.

    • Search Google Scholar
    • Export Citation
  • Brunnermeier, M., Pedersen, L., 2009. “Market liquidity and funding liquidity.” Review of Financial Studies 22, 22012238.

  • Brunnermeier, M. K., Dong, G., Palia, D., 2012, “Banks’ non-interest income and systemic risk,” Working paper, Princeton University.

    • Search Google Scholar
    • Export Citation
  • Brunnermeier, M., Gorton, G., Krishnamurthy, A., 2012. “Risk topography”. NBER Macroeconomics Annual, 2011, 26, 149176.

  • Byström, H, 2006. “Merton unraveled: A flexible way of modelling default risk,” Journal of Alternative Investments 8(4), 3947.

  • Campbell, J.Y., Hilscher, J., Szilagyi, J., 2008. “In search of distress risk,” Journal of Finance, 63(6), 28992939.

  • Cerutti, E., and S. Claessens, 2016. “The Great Cross-Border Bank Deleveraging: Supply Side Characteristics and Intra-Group Frictions.” Review of Finance, forthcoming.

    • Search Google Scholar
    • Export Citation
  • Cerutti, E., Ilyina, C., Makarova, Y., Schmieder, C., 2010. “Bankers without borders? Implications of ring-fencing for European cross-border banks.” IMF WP No. 10/247.

    • Search Google Scholar
    • Export Citation
  • Cerutti, E., Schmieder, C., 2014, “Ring fencing and consolidated banks’ stress tests,” Journal of Financial Stability 11, 112.

  • Cetorelli, N., Golberg, L., 2012. “Liquidity management of U.S. global banks: Internal capital markets in the Great Recession,” Journal of International Economics 88(2), 299311.

    • Search Google Scholar
    • Export Citation
  • Choi, M., Martinez Peria, M.S., Gutierrez, E., 2013. “Dissecting foreign bank lending behavior during the 2008-2009 Crisis,” World Bank Policy Research Working Paper 6674.

    • Search Google Scholar
    • Export Citation
  • Cifuentes, R., Ferrucci, G., Shin, H., 2004. “Liquidity risk and contagion,” Journal of the European Economic Association 3(2-3), 556566.

    • Search Google Scholar
    • Export Citation
  • Cihak, M, Demirguc-Kunt, A, Martinez Peria, M., Mohseni-Cheraghlou, A., 2013. “Bank regulation and supervision in the context of the global crisis,” Journal of Financial Stability, 9(4), 733746.

    • Search Google Scholar
    • Export Citation
  • Claessens, S., Van Horen, N., 2013. “Impact of foreign banks,” Journal of Financial Perspectives 1(1), 2942.

  • Crystal, J. S., Dages, B.G., Goldberg, L.S., 2001, “Does foreign ownership contribute to sounder banks in emerging markets? The Latin American experience.” Open Doors: Foreign Participation in Financial Systems in Developing Countries, 1921.

    • Search Google Scholar
    • Export Citation
  • Crystal, J. S., Dages, B.G., Goldberg, L.S., 2002, “Has foreign bank entry led to sounder banks in Latin America?Federal Reserve Bank of New York, Current Issues in Economic and Finance, 8, No.1.

    • Search Google Scholar
    • Export Citation
  • Cull, R., Martinez Peria, M.S., 2010. “Foreign bank participation in developing countries: what do we know about the drivers and consequences of this phenomenon?Policy Research Working Paper 5398. Washington, DC: World Bank.

    • Search Google Scholar
    • Export Citation
  • Demirguc-Kunt, A., Detragiache, E., 1999. “Monitoring banking sector fragility: a multivariate logit approach with an application to the 1996-97 banking crises,” Policy Research Working Paper Series 2085, The World Bank.

    • Search Google Scholar
    • Export Citation
  • Demirguc-Kunt, A., Huizinga, H., 2013. “Are banks too big to fail or too big to save? International evidence from equity prices and CDS spreads.” Journal of Banking and Finance 37 (3), 875894.

    • Search Google Scholar
    • Export Citation
  • Demirgüç-Kunt, A., Levine, R., Min, H., 1998. “Opening to foreign banks: Issues of stability, efficiency and growth,” in Seongtae Lee, (ed.), The Implications of Globalization of World Financial Markets, Seoul, Korea: The Bank of Korea, 1998, 83115.

    • Search Google Scholar
    • Export Citation
  • Demirguc-Kunt, A., Merrouche, O., Detragiache, E. 2013, “Bank capital: Lessons from the financial crisis”, Journal of Money, Credit and Banking 45 (6), 11471164.

    • Search Google Scholar
    • Export Citation
  • De Haas, R., van Horen, N., 2013. “Running for the exit? International bank lending during a financial crisis,” Review of Financial Studies 26(1), 244285.

    • Search Google Scholar
    • Export Citation
  • De Haas, R., van Lelyveld, I., 2006. “Foreign banks and credit stability in Central and Eastern Europe. A panel data analysis.” Journal of Banking and Finance 30, 192752.

    • Search Google Scholar
    • Export Citation
  • De Haas, R., van Lelyveld, I., 2010. “Internal capital market and lending by multinational bank subsidiaries.” Journal of Financial Intermediation 19, 125.

    • Search Google Scholar
    • Export Citation
  • De Haas, R., van Lelyveld, I., 2014. “Multinational Banks and the Global Financial Crisis: Weathering the Perfect Storm?Journal of Money, Credit and Banking, 46, 333364.

    • Search Google Scholar
    • Export Citation
  • Detragiache, E., Gupta, P., 2006, “Foreign banks in emerging market crises: Evidence from Malaysia.” Journal of Financial Stability 2(3), 21742.

    • Search Google Scholar
    • Export Citation
  • D’Hulster, K., 2014. “Ring-fencing cross border banks: How is it done and it is important?,” World Bank, mimeo.

  • D’Hulster, K., Otker-Robe, I., 2015Ring-fencing cross-border banks: An effective supervisory response?,” Journal of Banking Regulation 16(3), 169187.