WPS7053 Policy Research Working Paper 7053 Foreign Bank Subsidiaries’ Default Risk During the Global Crisis What Factors Help Insulate Affiliates from Their Parents? Deniz Anginer Eugenio Cerutti María Soledad Martínez Pería Development Research Group Finance and Private Sector Development Team October 2014 Policy Research Working Paper 7053 Abstract This paper examines the association between the default and profitability ratios and that are more independently risk of foreign bank subsidiaries and their parents during managed from their parents. Host country regulations the global financial crisis, with the purpose of under- also influence the extent to which shocks to the parents standing what factors can help insulate affiliates from affect the subsidiaries’ default risk. In particular, the cor- their parents. The paper finds evidence of a significant relation between the default risk of the subsidiary and the positive correlation between parent banks’ and foreign parent is lower for subsidiaries operating in countries that subsidiaries’ default risk. This correlation is lower for sub- impose higher capital, reserve, provisioning, and disclosure sidiaries that have higher capital, retail deposit funding, requirements and tougher restrictions on bank activities. This paper is a product of the Finance and Private Sector Development Team, Development Research Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at mmartinezperia@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Foreign Bank Subsidiaries' Default Risk During the Global Crisis: What Factors Help Insulate Affiliates from Their Parents? * Deniz Anginer, Eugenio Cerutti, and María Soledad Martínez Pería † JEL Classifications: F36, G11, G12, G15 Keywords: Banking crises, default risk, ring-fencing, bank subsidiaries, distance to default, Merton model. * Violeta Gutkowski and Pedro Juarros provided excellent research assistance. We are grateful to Charlie Calomiris, Stijn Claessens, Giovanni Dell’ariccia and other participants in the IMF Research Department Macro-Financial Division Brown Bag Seminar for comments and suggestions. † Anginer, Virginia Tech and World Bank (on-leave), danginer@worldbank.org; Cerutti, International Monetary Fund, ecerutti@imf.org; Martinez Peria, World Bank, mmartinezperia@worldbank.org. 1. 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 crisis evidence analyzing the consequences of this significant transformation in bank ownership suggests that foreign bank participation brought many benefits to developing countries, especially in terms of bank competition and efficiency. 2 The recent global financial crisis, however, highlighted the role of multinational banks in the transmission of shocks across countries. Most of the research has focused on transmission through the lending channel – how foreign bank lending behaved during the crisis. 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. 1 Data from the World Bank Regulation and Supervision surveys at http://econ.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/0,,contentMDK:20345037~pagePK: 64214825~piPK:64214943~theSitePK:469382,00.html 2 For a review of the literature see Cull and Martinez Peria (2008). 2 Rather than focus on how parent banks transmit shocks to their affiliates through the lending channel, this paper explores the association between parent banks’ and subsidiaries’ default risks during the recent crisis. More specifically, we examine whether an increase in the default risk of a parent bank is positively correlated with a similar rise in its foreign subsidiaries’ default risk. By focusing on the correlation between the default risk of parents and subsidiaries, we believe that we are answering a more fundamental question regarding the role of foreign banks in transmitting shocks, since default risk is a broader, more forward looking concept that is more intrinsically related to financial stability. We also examine the factors that dampen or amplify the correlation between parent banks’ and foreign subsidiaries’ default risks. In particular, we examine the role of subsidiary financial characteristics (such as capital and funding structure) and the impact of host country bank regulations (e.g., pertaining to bank capital, reserve requirements, bank activities, etc.). The question of how host regulators can limit the transmission of shocks to the affiliates of foreign banks that operate in their country is related to the recent discussion on ring-fencing (see Song, 2004; Cerutti et al., 2010; Cerutti and Schmieder, 2014; 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 impact of subsidiary characteristics and host country regulations in limiting the association between the default risk of a parent and its subsidiaries. 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. We estimate the weekly correlation 3 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 a subsidiary’s distance to default is significantly correlated with the parent bank’s 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. 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 higher capital, retail deposit funding, and profitability ratios and for subsidiaries that are more independently managed from their parents. 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 subsidiary and the parent is lower for subsidiaries operating in host countries that impose higher capital, reserve, provisioning and disclosure requirements and tougher restrictions on bank activities. 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 4 and 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. 2. 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. Our period of analysis is the peak of the global crisis: from September 2008 to December 2009. Table 1 presents a list of all the parent and subsidiary banks we consider in our analysis. 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. 3 Second, of the ones 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). In this context, we identified about 167 banks in about 44 host countries where foreign banks held important ownership stakes, but we found enough information for only 93 cases. The median ownership stake in the sample is about 61 percent. 4 The limitations to 3 As a robustness check, we extend the analysis to non-traded banks following the approach outlined in Falkenheim and Pennachi (2003). See section 4.1 and Table 7. 4 In 28 of the 93 foreign subsidiaries included in the sample, the identified parent banks seem to directly control less than 50 percent of the shares. We include them in the analysis because the identified parent banks are portrayed as 5 fulfilling the necessary data requirements for our analysis do not seem to bias the representation of the final sample, which covers 36 host countries. The data set 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 comprehensive commercial database of banks' financial statements produced by Bureau Van Dijk. 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, strategic partners, and they often have indirect control in the subsidiaries. In unreported results, we verify that the degree of ownership is not a significant factor explaining the relative strength of transmission of default risk from parents to affiliates. 5 Note that in the ratios mentioned here assets and equity are measured as book values. We also tried running regressions using the market value of assets and equity computed from the Merton model discussed above and found similar results. 6 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 refers 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 6 The 2007 survey covers the 2005-2006 period. 7 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 forces 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. We also control for macro factors that may affect the changes in credit risk of the parents and their subsidiaries. ∆VIX 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). ∆DEF is the change in the default spread measured as the difference in Baa-Aaa yields of US firms. Data come 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 3 presents correlations across the variables. 3. Empirical Methodology 3.1. Computing distance to default measures between parents and subsidiaries Our 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 8 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: = − (1 ) − − (2 ) + �1 − − � 2 (1) log � � + � − + � 2 1 = ; 2 = 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: − (1 ) = (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 7 The Merton (1974) distance to default measure has been shown to be a good predictor of defaults, outperforming accounting-based models (Campbell, Hilscher and Szilagyi, 2008; Hillegeist, Keating, Cram, and Lundstedt, 2004; and Bharath and Shumway, 2008). 8 In the event of default, equity holders receive nothing. If the company does not default, equity holders receive the market value of the assets conditional on no default and pay the face value of liabilities. In equation (1), N(d2) is the risk-neutral probability of default and − (2 ) is the discounted face value of the liabilities. (1 ) is the discounted value of assets conditional on the firm not defaulting. 9 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 equals 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 2 � � + � − − � 2 (3) = √ As a robustness check, we compute two alternative measures of Merton’s distance to default: (a) a simplified version of the Merton formula applied to the sample of listed subsidiaries and (b) a synthetic ‘market comparable’ measure to extend the analysis to the sample of non-listed subsidiaries. The simplified approach follows Byström (2006) and 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: 9 Since during recessions and market downturns the risk premium increases, as a robustness check we also computed distance to default values using a 12% equity risk premium. The correlation in levels of distance to default values using the two different equity premium values is 96%. The correlation in changes of distance to default is 99%. 10 The default probability is the normal transform of the distance to default measure and is defined as PD = F (–dd), where F is the cumulative distribution function of a standard normal. 11 For large values of leverage, the formula further simplifies to 1/sE. Atkeson, Eisfeldt and Weill (2013) show theoretically that one can approximate a firm’s distance to insolvency using data on the inverse of the volatility of that firm’s equity returns. 10 log (X/(VE+X)) / (X/(VE+X)-1)×sE. (4) The synthetic ‘market comparable’ measure follows Falkenheim and Pennachi (2003). We use publicly traded banks to identify a statistical relationship between accounting variables (which are available for all banks) and the two key variables that feed into the structural credit risk calculation of the Merton model, namely, market leverage and volatility of asset returns. We then use these statistical relationships and the accounting information to compute market leverage and asset return volatility for all non-publicly traded banks. The key assumption of this methodology is that non-public banks would have on average the same credit risk profiles conditional on observable accounting variables as publicly listed banks. Appendix 1 provides a detailed explanation of the methodology. This approach has two limitations. First, there could be a selection bias in the estimation of credit risk for non-publicly traded banks. Since publicly traded banks can be significantly different than non-publicly traded banks along a number of dimensions, the relationship between credit risk and observable accounting variables may not be the same for non-traded banks. Second, we are interested in correlations and use weekly estimates of credit risk for subsidiary banks and their parents. There may not be enough time- series variation in the weekly coefficient estimates from cross-sectional regressions of publicly traded banks and the observable accounting variables. 3.2. 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: 11 0 1 2 3 4 ∆_, = + ∆_, + ∆_ + ∆_ + ∆ (5) 5 + ∆ + + , where ∆_, is the weekly change in distance to default of the subsidiary i in week t; ∆_, is the weekly change in distance to default of the parent of subsidiary i; ∆_ and _ 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. ∆ is the change in interest rate spread between Bbb and Aaa rated companies and is included to capture innovations in the default risk premium. ∆ 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. We also include subsidiary fixed effects, , 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 ∆_, . In particular, we estimate equation (6) below: 0 1 2 3 4 ∆_, = + ∆_, + ∆_ + ∆_ + ∆ + 5 ∆ + 6 × ∆_, + 7 × ∆_, + (6) + , where are foreign subsidiary characteristics (size, liquidity, funding structure, capital, etc.) computed as of December 2006, as described above. are host country regulations measured as of 2006. The errors are clustered at the host country level. 12 On average, we expect to find a positive correlation between foreign banks’ subsidiaries and parent banks’ distance to default, after controlling for country level averages in the distance to default of all companies in the home and host countries and for global factors such as the VIX index. In other words, we expect that the financial health of the parent will be associated with that of the subsidiary. How large this correlation will be is an empirical question that we hope to address. A number of theoretical papers emphasize the role of capital, profitability, asset liquidity and funding structure as potential buffers in absorbing liquidity and economic shocks (Allen and Gale, 2000; Repullo, 2004; Von Thadden, 2004; Diamond and Rajan, 2005; Cifuentes et al., 2004). We expect that the correlation between the distance to default of the subsidiary and the parent will be higher for more fragile foreign subsidiaries (i.e., those lacking the means to absorb shocks at the beginning of the crisis). At the same time, we expect the 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. 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) 12. 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 12 Both variables, geographical distance and cultural proximity, were taken from Mayer and Zignago (2011). See http://www.cepii.fr/anglaisgraph/bdd/distances.htm. 13 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. 13 4. Results 4.1. The correlation between the parent and subsidiary distance to default Figure 1 shows the average changes in Merton 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. Table 4 shows that foreign bank subsidiaries’ distance to default, calculated following Merton’s model, is significantly correlated with parent banks’ distance to default, even when we control for the average distance to default among all companies in the home and host countries, 13 Related to the idea that cultural proximity might result in better treatment for some subsidiaries, Giannetti and Yafeh (2012) show that cultural proximity affects financial contracts in a large dataset of international syndicated bank loans. For example, they find that lead banks offer larger loans at a lower interest rate to more culturally close borrowers. 14 respectively, and when we account for global factors like the VIX and the corporate credit spread. The correlation in the distance to default between foreign subsidiaries and their parents varies between a maximum of 0.45, when no other variables are added, and a minimum of 0.27 when we include all controls. 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. We conduct three variants of Table 4 as robustness checks. First, we show the same estimations as in Table 4, but excluding some parents banks which own many of the subsidiaries in our sample (see Table 5). 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 only between 0.26 and 0.29. Second, we repeat the types of estimations shown in Table 4, but using the simplified measure of distance to default proposed by Byström (2006). As shown in Table 6, the correlation between foreign subsidiaries and their parents is highly significant and ranges between 0.2 and 0.3. Finally, Table 7 shows the correlation between subsidiaries’ and parents’ synthetic ‘market comparable’ measure of distance to default following Falkenheim and Pennachi (2003). These regressions are done for the 310 foreign subsidiaries (both listed and unlisted) that operate in the host countries in our sample. In this case, the correlations between subsidiaries and parents continue to be significant, but are smaller in size, ranging from almost 0.09 to 0.19. The lower correlations are to be expected since, as mentioned earlier, we use annual accounting data and 15 weekly cross-sectional estimates to create the synthetic ‘market comparable’ measure which does not have as much time-series variation as the actual distance to default measure. 14 4.2 The factors that affect the correlation between the parent and subsidiary distance to default Table 8 explores how subsidiary characteristics affect the association between the Merton distance to default of foreign bank subsidiaries and that of their parents. We find that the parent bank’s distance to default has a smaller impact on the subsidiary’s distance to default when subsidiaries have higher capital, deposit funding, and profitability ratios. Also, the association between the parent and the subsidiary distance to default is lower for countries that are physically distant or culturally closer. Most of these results survive even when we combine all variables together as in column 8.10. 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 subsidiary from its 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 14 We have also replicated the analysis of publicly traded firms using synthetic measures of distance to default. We also find lower correlations between the parents and the subsidiaries for this smaller sample of publicly traded banks. 16 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 9 shows estimations allowing for threshold effects in the impact of subsidiary characteristics on the association between distance to default of foreign subsidiaries and their parents. Rather than interact the aforementioned correlation with continuous measures of affiliate characteristics, we use a dummy which equals one for those affiliates whose characteristics rank above the median. We find that for subsidiaries with capital, deposit funding, provisions, and geographical distance above the median, the correlation between the subsidiaries’ and the parents’ distance to default is lower. In particular, holding everything else constant, for subsidiaries with low capital ratios (i.e., those below the median) the correlation between the subsidiary and the parent distance to default equals 0.37, while for those above the median the correlation falls to 0.25. We find similar effects when comparing subsidiaries above and below the median funding and profitability ratios, as well as for those above and below the median geographical distance. Table 10 allows us to corroborate the significance of subsidiary characteristics (except for the capital ratio) even when we control for host country dummies and parent distance to default interactions, which are included but not reported. These interactions are intended to account for any host country factors that can mitigate the impact of the parent distance to default. As before, we find that for subsidiaries that have higher retail deposit funding ratios and that are culturally 17 closer to the parent, the correlation between the subsidiaries’ and the parents’ distance to default is lower. Table 11 investigates the impact of host country regulations on the correlation between foreign subsidiaries’ and parents’ distance to default. 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 15, parent banks’ distance to default has a smaller impact on the subsidiaries’ distance to default. 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 12 repeats the estimations of Table 11 but controlling for the specific subsidiary characteristics and variables that we found to be significant in Table 8. Even though our sample is reduced significantly when we do this, none of the results regarding the impact of host country regulations change. We continue to 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, parent banks’ distance to default has a smaller impact on the subsidiaries’ distance to default. 5. 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 15 We have also conducted estimations looking separately at restrictions on specific bank activities such as securities, investments, and real estate and have found that regulations restricting banks’ ability to engage in securities underwriting are the most significant in terms of lowering the correlations between the subsidiary and parent distance to default. 18 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 sample of banks considered and to the way we calculate the distance to default. 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 higher capital, deposit funding and profitability 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. However, it is important to note that this 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 19 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. 20 References Allen, F., Gale, D., 2000. “Bubbles and crises,” The Economic Journal, Vol. 110, pp. 236-255. Atkeson, A., Eisfeldt, A., Weill, P., 2014. “The market for OTC Derivatives,” NBER Working Paper No. 18912. 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. 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. Bartram, S., Gregory, B., Hund, J., 2007. “Estimating systemic risk in the international financial system,” Journal of Financial Economics, Vol. 86(3), pp. 835-869. Basel Committee on Banking Supervision, 2010. Report and recommendations of the Cross- Border Resolution Group, March, Basel. Bharath, S., Shumway, T., 2008. “Forecasting default with the Merton distance to default model,” Review of Financial Studies, Vol. 21 (3), pp. 1339-1369. Byström, H, 2006. “Merton unraveled: A flexible way of modelling default risk,” Journal of Alternative Investments, Vol. 8 (4), pp. 39-47. Campbell, J.Y., Hilscher, J., Szilagyi, J., 2008. “In search of distress risk,” Journal of Finance, Vol 63(6), pp. 2899-2939. Cerutti, E., Schmieder, C., 2014, “Ring fencing and consolidated banks’ stress tests,” Journal of Financial Stability, Vol. 11., pp. 1-12.. 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. 21 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. Cifuentes, R., Ferrucci, G., Shin, H., 2004. “Liquidity risk and contagion,” Journal of the European Economic Association, Vol. 3(2-3), pp. 556-566. Claessens, S., Van Horen, N., 2013. “Impact of foreign banks,” Journal of Financial Perspectives, Vol. 1(1), pp. 29-42. 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 P 5398. Washington, DC: World Bank. De Haas, R., Van Horen, N., 2013. “Running for the exit? International bank lending during a financial crisis,” Review of Financial Studies, Vol. 26(1), pp. 244-285 D’Hulster, K., 2014. “Ring-fencing cross border banks: How is it done and it is important?,” World Bank, mimeo. Diamond, D., Rajan, R., 2005. “Liquidity shortages and banking crises,” Journal of Finance, Vol. 60 (2), pp. 615–647. Giannetti, M., Yafeh, Y., 2012. “Do Cultural Differences Between Contracting Parties Matter? Evidence from Syndicated Bank Loans.” Management Science, 58, pp. 365–83. .Hauswald, R., Marquez, R., 2006. “Competition and strategic information acquisition in credit markets,” Review of Financial Studies, Vol. 19(3), pp. 967-1000. Hillegeist, S., Keating, E., Cram, E., Lundstedt, K., 2004. “Assessing the probability of bankruptcy,” Review of Accounting Studies, Vol. 9 (1). 22 Falkenheim, M., Pennacchi, G., 2003. “The cost of deposit insurance for privately held banks: A market comparable approach,” Journal of Financial Services Research, Vol. 24(2), pp. 121- 148. Fama, E. F., MacBeth, J. D., 1973. “Risk, return, and equilibrium: Empirical tests,” Journal of Political Economy, pp. 607-636. Hillegeist, S. A., Keating, E. K., Cram, D. P., Lundstedt, K. G., 2004. “Assessing the probability of bankruptcy,” Review of Accounting Studies, Vol. 9(1), pp. 5-34. Jaffee, D., Modigliani, F., 1969. “A theory and test of credit rationing,” American Economic Review, Vol. 59, pp. 850-872. Merton, R. C., 1974. “On the pricing of corporate debt: the risk structure of interest rates,” Journal of Finance, Vol. 29, pp. 449–470. Mayer, T., Zignago, S., 2011. “Notes on CEPII’s distances measures (GeoDist),” CEPII Working Paper 2011-25. Newey, W. K., West, K. D., 1987. “Hypothesis testing with efficient method of moments estimation,” International Economic Review, pp. 777-787. Peek, J., Rosengren, E., 2000, “Collateral damage: effects of the Japanese bank crisis on real activity in the United States,” American Economic Review, Vol. 90 (1), pp. 30-45. Rajan, R., Servaes, H., Zingales, L., 2000. “The cost of diversity: The diversification discount and inefficient investment,” The Journal of Finance, Vol. 55(1), pp. 35-80. Repullo, R., 2004. “Capital requirements, market power, and risk-taking in banking,” Journal of Financial Intermediation, Vol. 13, pp. 156–182. Schnabl, P., 2012. “The international transmission of bank liquidity shocks: Evidence from an emerging market,” Journal of Finance, Vol. 67(3), pp. 897–932. 23 Song, I., 2004. “Foreign bank supervision and challenges to emerging market supervisors,” IMF Working Paper 04/82. von Thadden, E.-L., 2004. “Bank capital adequacy regulation under the new Basel Accord,” Journal of Financial Intermediation, Vol. 13, pp. 90-95. 24 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. Parent Bank Name Home Country Subsidiary Bank Name Host Country Abu Dhabi Islamic Bank - Public Joint UNITED ARAB EMIRATES National Development Bank/Egyp EGYPT, ARAB REP. Stock Co. Albaraka Banking Group B.S.C. BAHRAIN Al Baraka Bank Egypt ESC EGYPT, ARAB REP. Albaraka Banking Group B.S.C. BAHRAIN Albaraka Turk Katilim Bankasi TURKEY Allied Irish Banks plc IRELAND Bulgarian American Credit Bank BULGARIA Arab Bank Plc JORDAN Arab Tunisian Bank TUNISIA Arab Banking Corporation BSC BAHRAIN Banco ABC Brasil SA BRAZIL Attijariwafa Bank MOROCCO Attijari Bank TUNISIA Australia & New Zealand Bankin AUSTRALIA Bank Pan Indonesia Tbk PT INDONESIA Australia and New Zealand Banking AUSTRALIA AMMB Holdings Bhd MALAYSIA Group Banco Bilbao Vizcaya Argentaria SA SPAIN Banco Bilbao Vizcaya Argentaria CHILE Banco Bilbao Vizcaya Argentaria SA SPAIN BBVA Banco Frances SA ARGENTINA Banco Bilbao Vizcaya Argentaria SA SPAIN BBVA Colombia SA COLOMBIA Banco Continental-BBVA Banco Banco Bilbao Vizcaya Argentaria SA SPAIN PERU Continental Banco Bilbao Vizcaya Argentaria SA SPAIN Banco Provincial VENEZUELA, RB Banco Comercial Português S.A. PORTUGAL Bank Millennium POLAND Banco Santander SA SPAIN Banco Santander Rio S.A. ARGENTINA Banco Santander SA SPAIN Banco Santander (Brasil) S.A. BRAZIL Banco Santander SA SPAIN Banco Santander Brasil SA/Braz BRAZIL Banco Santander SA SPAIN Banco Santander Chile CHILE Banco Santander SA SPAIN Banco Santander Colombia SA COLOMBIA Banco Santander SA SPAIN Attijariwafa Bank MOROCCO Bank of East Asia Ltd HONG KONG SAR, CHINA Affin Holdings Bhd MALAYSIA Bank of New York Mellon UNITED STATES Wing Hang Bank Ltd HONG KONG SAR, CHINA Bank of Nova Scotia - Scotiabank CANADA Scotiabank Sud Americano CHILE Bank of Nova Scotia - Scotiabank CANADA Scotia Group Jamaica Ltd JAMAICA Bank of Nova Scotia - Scotiabank CANADA Scotiabank Peru SAA PERU Scotiabank Trinidad & Tobago Bank of Nova Scotia - Scotiabank CANADA TRINIDAD AND TOBAGO Limited BARCLAYS PLC UNITED KINGDOM Barclays Bank of Botswana BOTSWANA BARCLAYS PLC UNITED KINGDOM Barclays Bank of Kenya Ltd KENYA BARCLAYS PLC UNITED KINGDOM ABSA Group Limited SOUTH AFRICA Banque Internationale pour le BNP Paribas FRANCE Commerce et l'Industrie de la Côte COTE D’IVOIRE d'Ivoire SA - BICICI BNP Paribas FRANCE Bank of Nanjing CHINA Banque Marocaine pour le BNP Paribas FRANCE MOROCCO Commerce et l'Industrie BMCI BNP Paribas FRANCE Union Bancaire pour le Commerc TUNISIA 25 Table 1: List of foreign bank parents and subsidiaries in the sample (continued) Parent Bank Name Home Country Subsidiary Bank Name Host Country BNP Paribas FRANCE Turk Ekonomi Bankasi A.S. TURKEY BNP Paribas FRANCE BNP Paribas Bank Polska SA POLAND BTA Bank JSC KAZAKHSTAN Sekerbank TAS TURKEY Caixabank SPAIN Bank of East Asia Ltd HONG KONG SAR, CHINA FirstCaribbean International Bank Canadian Imperial Bank of Commerce CANADA BARBADOS Limited CIMB Group Holdings Bhd MALAYSIA Bank CIMB Niaga Tbk PT INDONESIA Citigroup Inc. UNITED STATES Banco de Chile CHILE Citigroup Inc. UNITED STATES Bank Handlowy w Warszawie S.A. POLAND Citigroup Inc. UNITED STATES Akbank TAS TURKEY Commerzbank AG. GERMANY BRE Bank SA POLAND Commerzbank AG. GERMANY Bank Forum UKRAINE Crédit Agricole S.A. FRANCE Credit Agricole Egypt EGYPT, ARAB REP. Crédit Agricole S.A. FRANCE Crédit du Maroc MOROCCO Deutsche Bank AG GERMANY Hua Xia Bank co., Limited CHINA Dexia BELGIUM Denizbank A.S. TURKEY Dubai Bank PJSC UNITED ARAB EMIRATES Bankislami Pakistan Ltd PAKISTAN Hang Seng Bank Ltd HONG KONG SAR, CHINA Industrial Bank Co Ltd CHINA Shenzhen Development Bank Co., HSBC Holdings Plc. UNITED KINGDOM CHINA Ltd HSBC Holdings Plc. UNITED KINGDOM Hang Seng Bank Ltd. HONG KONG SAR, CHINA HSBC Holdings Plc. UNITED KINGDOM Bank Ekonomi Raharja Tbk PT INDONESIA HSBC Holdings Plc. UNITED KINGDOM HSBC Bank Malta Plc MALTA ING Groep NV NETHERLANDS Bank of Beijing Co Ltd CHINA ING Groep NV NETHERLANDS ING Vysya Bank Ltd INDIA ING Bank Slaski S.A. - Capital ING Groep NV NETHERLANDS POLAND Group ING Groep NV NETHERLANDS TMB Bank PCL THAILAND Privredna Banka Zagreb d.d- Intesa Sanpaolo ITALY CROATIA Privredna Banka Zagreb Group Intesa Sanpaolo ITALY Vseobecna Uverova Banka a.s. SLOVAKIA Intesa Sanpaolo ITALY Banco Patagonia SA ARGENTINA Ithmaar Bank B.S.C. BAHRAIN Faysal Bank Ltd PAKISTAN KBC GROEP NV/ KBC GROUPE SA BELGIUM Kredyt Bank SA POLAND Malayan Banking Bhd MALAYSIA MCB Bank Ltd PAKISTAN Mitsubishi UFJ Financial Group Inc. JAPAN Chong Hing Bank Limited HONG KONG SAR, CHINA Mitsubishi UFJ Financial Group Inc. JAPAN Dah Sing Banking Group Limited HONG KONG SAR, CHINA National Bank of Greece SA GREECE Stopanska Banka a.d. Skopje MACEDONIA FYROM National Bank of Greece SA GREECE Finansbank A.S. TURKEY Nomura Holdings Inc JAPAN Silkbank Ltd PAKISTAN Nordea Bank AB (Publ) SWEDEN Nordea Bank Polska SA POLAND 26 Table 1: List of foreign bank parents and subsidiaries in the sample (continued) Parent Bank Name Home Country Subsidiary Bank Name Host Country Oversea-Chinese Banking Corporation SINGAPORE Bank of Ningbo CHINA Limited OCBC Oversea-Chinese Banking Corporation SINGAPORE Bank OCBC Nisp Tbk PT INDONESIA Limited OCBC Raiffeisen Bank International AG AUSTRIA Raiffeisen Bank Aval UKRAINE Société Générale de Banques en Société Générale FRANCE COTE D’IVOIRE Côte d'Ivoire - SGBCI Société Générale FRANCE Komercni Banka CZECH REPUBLIC National Societe Generale Bank Société Générale FRANCE EGYPT, ARAB REP. SAE Société Générale FRANCE SG-SSB Limited GHANA Societe d'Equipement Domestique Société Générale FRANCE MOROCCO et Menager Société Générale FRANCE Ohridska Banka ad Ohrid MACEDONIA FYROM Société Générale FRANCE BRD-Groupe Societe Generale SA ROMANIA Société Générale FRANCE JSC Rosbank RUSSIAN FEDERATION Société Générale FRANCE Union Internationale de Banques TUNISIA Standard Chartered Bank Standard Chartered Plc. UNITED KINGDOM BOTSWANA Botswana Ltd Standard Chartered Plc. UNITED KINGDOM Bank Permata Tbk PT INDONESIA Standard Chartered Plc. UNITED KINGDOM Standard Chartered Bank Kenya KENYA Standard Chartered Bank Standard Chartered Plc. UNITED KINGDOM PAKISTAN (Pakistan) Standard Chartered Bank Zambia Standard Chartered Plc. UNITED KINGDOM ZAMBIA Plc-SCBZ Plc Unicredit Spa ITALY Zagrebacka Banka dd CROATIA Unicredit Spa ITALY Bank of Valletta PLC MALTA Bank Polska Kasa Opieki SA-Bank Unicredit Spa ITALY POLAND Pekao SA Unicredit Spa ITALY Yapi ve Kredi Bankasi AS TURKEY Joint-Stock Commercial Bank for Unicredit Spa ITALY UKRAINE Social Development - Ukrsotsbank 27 Table 2: Variable definition and descriptive statistics This table lists the definitions, sources and the summary statistics for the variables used in this study Variable Definition Source Mean Standard deviation ∆dd_Sub Change in subsidiaries’ Merton distance to default. Authors' calculation using bank data from Bankscope and -0.0019 0.0266 stock return information from Compustat/CSRP/Datastream ∆dd_Parent Change in parent banks’ Merton distance to default. Authors' calculation using bank data from Bankscope and -0.0054 0.0276 stock return information from Compustat/CSRP/Datastream ∆dd_Home Change in home countries’ average Merton distance to default. Authors' calculation using bank data from Bankscope and -0.0025 0.0222 This country average includes all listed firms and banks except stock return information from Compustat/CSRP/Datastream for the bank we are using. ∆dd_Host Change in host countries’ average Merton distance to default. Authors' calculation using bank data from Bankscope and -0.0016 0.0346 This country average includes all listed firms and banks except stock return information from Compustat/CSRP/Datastream for the bank we are using. ∆DEF Change in Bbb - Aaa spread Interest rate data releases from the Federal Reserve Board -0.0067 0.1413 ∆VIX Change of the CBOE VIX index, implied volatility index on the Chicago Board Options Exchange -0.0006 0.0597 S&P 500. Size Subsidiaries’ assets to banking systems’ assets ratio Bankscope 0.0579 0.0584 Capital ratio Subsidiaries’ capital ratio Bankscope 0.1484 0.0459 Equity assets Subsidiaries’ equity to total assets ratio Bankscope 0.0929 0.0388 Dep. Funding Subsidiaries’ deposits to total funding ratio Bankscope 0.8245 0.1229 Profitability Subsidiaries’ return on average assets (ROAA) Bankscope 0.0157 0.0146 Liquidity Subsidiaries’ liquid assets to total assets ratio Bankscope 0.2336 0.0998 Provisions Subsidiaries’ loan loss provision to total loans ratio Bankscope 0.0082 0.0141 Geographical distance Log of the distance between the capitals of the Parent and CEPII 8.0759 0.9032 Subsidiary countries Cultural distance Dummy=1 if home and host share a common language CEPII 0.3653 0.4816 Reserve Requirements Average reserve requirements Data come from World Bank Regulation and Supervision 14.481 15.713 Survey. Disclosure requirements Index variable that indicates whether the income statement Data come from World Bank Regulation and Supervision 2.751 0.443 includes accrued or unpaid interest or principal on nonperforming Survey. Index is constructed following Barth, Caprio, and loans, whether banks are required to produce consolidated Levine (2001, 2013) 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 more informative bank accounts. Activity restrictions Index variable that ranges from 3 to 12, with 12 indicating the Data come from World Bank Regulation and Supervision 8.417 2.344 highest restrictions on bank activities such as securities, Survey. Index is constructed following Barth, Caprio, and investment, and real estate. (For each type of activity: Levine (2001, 2013) Unrestricted=1, Permitted=2, Restricted=3, and Prohibited=4). 28 Table 2: Variable definition and descriptive statistics (continued) Variable Definition Source Mean Standard deviation Capital regulation Index captures the amount of capital banks must hold and the Data come from World Bank Regulation and Supervision 5.359 2.099 stringency of regulations on the nature and source of capital. Survey. Index is constructed following Barth, Caprio, and Ranges in value from 0 to 10, with higher values indicating Levine (2001, 2013) greater stringency Loan classification Measures the actual minimum number of days beyond which a Data come from World Bank Regulation and Supervision 475.722 164.39 loan in arrears must be classified as sub-standard, doubtful, or Survey. Index is constructed following Barth, Caprio, and loss. Levine (2001, 2013) Provisioning Measures the minimum provisions (as a percentage of loans) Data come from World Bank Regulation and Supervision 164.684 33.175 required as a loan is successively classified as sub-standard, Survey. Index is constructed following Barth, Caprio, and doubtful, and lastly as loss. Levine (2001, 2013) Diversification An index variable that ranges from zero to two, with higher Data come from World Bank Regulation and Supervision 1.326 0.558 values indicating more asset diversification. Survey. Index is constructed following Barth, Caprio, and Levine (2001, 2013) Supervisory powers An index variable that ranges from zero to fourteen, with fourteen Data come from World Bank Regulation and Supervision 11.939 2.566 indicating the highest power of the supervisory authorities Survey. Index is constructed following Barth, Caprio, and Levine (2001, 2013) Prompt corrective action Measures the extent to which the law establishes predetermined Data come from World Bank Regulation and Supervision 2.436 2.636 levels of bank solvency deterioration that forces automatic Survey. Index is constructed following Barth, Caprio, and enforcement actions such as intervention, and the extent to which Levine (2001, 2013) 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. Financial outflows Average of the financial sectors that involve mostly controlling Authors' calculation using data from IMF AREAR 0.630 0.331 restrictions outflows: lending to non-residents, maintenance of account abroad, and investment regulations, abroad by banks for the year 2007. 29 Table 3: Correlations across variables This table shows Pearson correlations of the variables used in this study. Panel A displays correlations of bank level subsidiary characteristics. Panel B displays correlations of host country regulations. Panel A: Subsidiary characteristics Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 ∆dd_Sub 1.000 2 ∆dd_Parent 0.500 1.000 3 ∆dd_Home 0.454 0.647 1.000 4 ∆dd_Host 0.384 0.384 0.500 1.000 5 ∆DEF -0.258 -0.340 -0.298 -0.204 1.000 6 ∆VIX -0.183 -0.224 -0.224 -0.142 0.207 1.000 7 Size -0.042 -0.076 -0.040 -0.023 0.007 0.016 1.000 8 Capital ratio -0.018 -0.003 -0.015 -0.030 0.010 0.017 0.035 1.000 9 Equity assets -0.057 0.024 -0.002 -0.031 -0.011 0.018 0.083 0.623 1.000 10 Dep. Funding -0.022 -0.022 -0.026 0.009 0.004 0.008 0.126 0.141 0.029 1.000 11 Profitability -0.028 -0.036 -0.049 -0.018 0.015 0.025 0.375 0.327 0.347 0.204 1.000 12 Liquidity 0.003 0.000 -0.004 -0.013 0.016 0.009 0.002 0.397 0.190 0.200 -0.042 1.000 13 Geo. Distance 0.041 0.051 0.012 0.045 -0.003 -0.006 -0.196 0.037 -0.070 0.294 0.117 -0.123 1.000 14 Cultural Prox. 0.019 -0.020 -0.020 0.012 0.009 0.013 0.284 -0.033 -0.022 0.246 0.427 -0.157 0.368 1.000 15 Provisions 0.032 0.062 0.016 0.039 -0.007 0.001 -0.072 -0.156 0.110 -0.201 -0.088 -0.309 0.209 0.043 1.000 Panel B: Host country regulations Variable 1 2 3 4 5 6 7 8 9 10 1 Reserve requirement 1.000 2 Disclosure 0.084 1.000 3 Financial outflows 0.303 0.095 1.000 4 Activity restrictions -0.137 0.071 0.419 1.000 5 Capital regulation -0.154 -0.036 0.129 0.672 1.000 6 Loan classification 0.308 0.228 0.597 0.278 0.194 1.000 7 Provisioning stringency -0.053 0.011 0.038 0.328 0.299 0.157 1.000 8 Diversification 0.216 0.082 -0.314 -0.374 0.066 0.322 -0.140 1.000 9 Supervisory power 0.191 -0.012 0.473 0.306 0.249 0.345 0.101 0.141 1.000 10 Prompt corrective action 0.242 -0.117 0.360 0.445 0.344 0.292 0.180 0.062 0.640 1.000 30 Table 4: The association between parent banks’ and subsidiaries’ Merton’s distance to default Regression results for the model ∆ _, = 0 + 1 ∆_, + 2 ∆ _ + 3 ∆ _ + 4 ∆ + 5 ∆ + + , are reported in this table. ∆_, is the weekly change in distance to default of the subsidiary i in week t; ∆ _, is the weekly change in distance to default of the parent of subsidiary i; ∆_ and ∆_ 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. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. Regressions also include subsidiary fixed effects, . 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. Variables (4.1) (4.2) (4.3) (4.4) (4.5) ∆dd_Parent 0.446*** 0.292*** 0.276*** 0.282*** 0.271*** (0.041) (0.034) (0.032) (0.033) (0.032) ∆dd_Home 0.188*** 0.182*** 0.182*** 0.175*** (0.051) (0.048) (0.049) (0.047) ∆dd_Host 0.142*** 0.136*** 0.138*** 0.135*** (0.028) (0.027) (0.027) (0.027) ∆DEF -0.013*** -0.011*** (0.004) (0.004) ∆VIX -0.0305* -0.0270* (0.0152) (0.0147) Observations 3,786 3,663 3,581 3,581 3,581 Subsidiary fixed effects Yes Yes Yes Yes Yes R-squared 0.216 0.276 0.285 0.285 0.288 Number of Subsidiaries 93 93 93 93 93 31 Table 5: The association between parent banks’ and subsidiaries’ Merton’s distance to default excluding parents with multiple subsidiaries Regression results for the model ∆ _, = 0 + 1 ∆ _, + 2 ∆_ + 3 ∆ _ + 4 ∆ + 5 ∆ + + , are reported in this table. ∆ _, is the weekly change in distance to default of the subsidiary i in week t; ∆ _, is the weekly change in distance to default of the parent of subsidiary i; ∆ _ and ∆ _ 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. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. Regressions also include subsidiary fixed effects, . Each column from 5.1 to 5.14 shows regression results excluding the bank specified in each corresponding column. 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. Variables (5.1) (5.2) (5.3) (5.4) (5.5) (5.6) (5.7) (5.8) (5.9) (5.10) (5.11) (5.12) (5.13) (5.14) Banco Bank of Mitsubish Bilbao Banco Standard Bank of National HSBC Bank Nova BNP Citigroup ING Societe UniCredit i UFJ Vizcaya Santander Barclays Plc Chartered East Asia Bank of Holdings Excluded Scotia - Paribas INC Groep NV Generale SpA Financial Argentaria SA Plc Ltd Greece SA PLC Scotiabank Group SA Inc-Kabu ∆dd_Parent 0.276*** 0.269*** 0.270*** 0.273*** 0.260*** 0.271*** 0.266*** 0.258*** 0.287*** 0.262*** 0.272*** 0.270*** 0.270*** 0.280*** (0.0317) (0.0339) (0.0327) (0.0326) (0.0351) (0.0321) (0.0313) (0.0336) (0.0340) (0.0327) (0.0327) (0.0323) (0.0331) (0.0329) ∆dd_Home 0.175*** 0.174*** 0.173*** 0.180*** 0.185*** 0.176*** 0.182*** 0.166*** 0.174*** 0.175*** 0.175*** 0.164*** 0.185*** 0.165*** (0.0485) (0.0484) (0.0487) (0.0498) (0.0492) (0.0471) (0.0510) (0.0498) (0.0523) (0.0477) (0.0481) (0.0432) (0.0484) (0.0459) ∆dd_Host 0.136*** 0.127*** 0.139*** 0.138*** 0.131*** 0.134*** 0.132*** 0.151*** 0.147*** 0.136*** 0.136*** 0.129*** 0.132*** 0.135*** (0.0275) (0.0262) (0.0263) (0.0265) (0.0269) (0.0265) (0.0267) (0.0310) (0.0262) (0.0273) (0.0268) (0.0265) (0.0266) (0.0278) ∆DEF -0.0101*** -0.0127*** -0.0101** -0.0109*** -0.0114*** -0.0111*** -0.0111*** -0.0122*** -0.0125*** -0.0107*** -0.0114*** -0.0102** -0.0113*** -0.0125*** (0.00369) (0.00381) (0.00374) (0.00384) (0.00370) (0.00364) (0.00377) (0.00371) (0.00340) (0.00364) (0.00370) (0.00375) (0.00370) (0.00397) ∆VIX -0.0233* -0.0288* -0.0331** -0.0260 -0.0265* -0.0270* -0.0289* -0.0297* -0.0290* -0.0253* -0.0272* -0.0261* -0.0271* -0.0285* (0.0137) (0.0154) (0.0141) (0.0154) (0.0143) (0.0147) (0.0148) (0.0162) (0.0154) (0.0145) (0.0150) (0.0153) (0.0146) (0.0145) Obs. 3,526 3,367 3,423 3,409 3,345 3,575 3,369 3,198 3,354 3,328 3,526 3,470 3,497 3,395 Subsidiary Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes fixed effects R-squared 0.291 0.292 0.293 0.289 0.285 0.288 0.283 0.283 0.308 0.278 0.288 0.278 0.289 0.291 Number of 88 87 89 90 87 90 89 84 88 88 92 91 91 89 Subsidiaries 32 Table 6: The association between parent banks’ and subsidiaries’ simplified Merton’s distance to default Regression results for the model ∆ _, = 0 + 1 ∆ _, + 2 ∆_ + 3 ∆ _ + 4 ∆ + 5 ∆ + + , are reported in this table. ∆ _, is the weekly change in distance to default of the subsidiary i in week t; ∆ _, is the weekly change in distance to default of the parent of subsidiary i; ∆ _ and ∆ _ 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. Distance to default is computed using the simplified approach outlined in the text in section 3. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. Regressions also include subsidiary fixed effects, . 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. Variables (6.1) (6.2) (6.3) (6.4) (6.5) ∆dd_Parent 0.335*** 0.191*** 0.196*** 0.190*** 0.190*** (0.0271) (0.0244) (0.0247) (0.0242) (0.0243) ∆dd_Home 0.184*** 0.187*** 0.199*** 0.202*** (0.0283) (0.0285) (0.0273) (0.0286) ∆dd_Host 0.142*** 0.145*** 0.148*** 0.148*** (0.0351) (0.0352) (0.0368) (0.0363) ∆DEF 0.00123 0.00116 (0.00339) (0.00337) ∆VIX 0.0149* 0.0148* (0.00814) (0.00812) Observations 4,326 4,172 4,008 4,008 4,008 Subsidiary fixed Yes Yes Yes Yes Yes effects R-squared 0.134 0.180 0.187 0.188 0.188 Number of 87 87 87 87 87 Subsidiaries 33 Table 7: The association between parent banks’ and subsidiaries’ synthetic distance to default Regression results for the model ∆ _, = 0 + 1 ∆ _, + 2 ∆_ + 3 ∆ _ + 4 ∆ + 5 ∆ + + , are reported in this table. ∆ _, is the weekly change in distance to default of the subsidiary i in week t; ∆ _, is the weekly change in distance to default of the parent of subsidiary i; ∆ _ and ∆ _ 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. Distance to default is computed using the ‘market comparable’ approach of Falkenheim and Pennachi (2003) outlined in the text in section 3. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. Regressions also include subsidiary fixed effects, . 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. Variables (7.1) (7.2) (7.3) (7.4) (7.5) ∆dd_Parent 0.188*** 0.087*** 0.088*** 0.097*** 0.096*** (0.0194) (0.019) (0.019) (0.019) (0.019) ∆dd_Home 0.052** 0.055** 0.057** 0.059** (0.022) (0.022) (0.022) (0.022) ∆dd_Host 0.522*** 0.521*** 0.518*** 0.518*** (0.053) (0.053) (0.053) (0.052) ∆DEF -0.013*** -0.009*** (0.003) (0.002) -0.0378*** -0.0317*** ∆VIX (0.008) (0.008) Observations 17,170 16,702 16,702 16,702 16,702 Subsidiary fixed effects Yes Yes Yes Yes Yes R-squared 0.086 0.278 0.280 0.280 0.281 Number of Subsidiaries 310 310 310 310 310 34 Table 8: The impact of subsidiaries’ characteristics on the association between foreign subsidiaries and parents’ distance to default Regression results for the model ∆ _, = 0 + 1 ∆_, + 2 ∆ _ + 3 ∆ _ + 4 ∆ + 5 ∆ + 6 × ∆_, + + , are reported in this table. ∆_, is the weekly change in distance to default of the subsidiary i in week t; ∆ _ is the weekly change in distance to default of the parent of subsidiary i; ∆ _ and ∆_ are changes in average distance to defaults of all the publicly traded banks and companies in the host country and parent bank country, respectively. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. are foreign subsidiary characteristics computed as of December 2006. These variables are described in detail in Table 2. Regressions also include subsidiary fixed effects, . 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. Variables (8.1) (8.2) (8.3) (8.4) (8.5) (8.6) (8.7) (8.8) (8.9) (8.10) ∆dd_Parent 0.283*** 0.479*** 0.357*** 0.781*** 0.343*** 0.207*** 0.325*** 0.771*** 0.342*** 1.179*** (0.0421) (0.115) (0.0804) (0.221) (0.0476) (0.0684) (0.0399) (0.268) (0.0417) (0.294) ∆dd_Home 0.171*** 0.168*** 0.172*** 0.163*** 0.166*** 0.172*** 0.156*** 0.176*** 0.176*** 0.168*** (0.0484) (0.0577) (0.0476) (0.0491) (0.0506) (0.0484) (0.0482) (0.0473) (0.0465) (0.0583) ∆dd_Host 0.128*** 0.124*** 0.127*** 0.129*** 0.130*** 0.127*** 0.126*** 0.135*** 0.135*** 0.128*** (0.0266) (0.0321) (0.0267) (0.0260) (0.0262) (0.0266) (0.0266) (0.0263) (0.0256) (0.0305) ∆DEF -0.0127*** -0.0130*** -0.0124*** -0.0128*** -0.0128*** -0.0130*** -0.0116*** -0.0113*** -0.0116*** -0.0243 (0.00375) (0.00433) (0.00386) (0.00380) (0.00369) (0.00398) (0.00371) (0.00371) (0.00357) (0.0164) ∆VIX -0.0279* -0.0227 -0.0278* -0.0292* -0.0271* -0.0263* -0.0239 -0.0269* -0.0291* -0.0143*** (0.0150) (0.0154) (0.0150) (0.0156) (0.0149) (0.0152) (0.0141) (0.0148) (0.0147) (0.00438) Size×∆dd_Parent -0.197 (0.544) Capital ratio×∆dd_Parent -1.124* -0.859* (0.608) (0.454) Equity assets×∆dd_Parent -0.908 (0.702) Cust. Dep./Fund.× ∆dd_Parent -0.602** -0.579* (0.262) (0.316) Profitability×∆dd_Parent -3.597* -0.171 (2.011) (2.654) Liquidity×∆dd_Parent 0.294 (0.344) Provisions×∆dd_Parent -4.271 (2.912) Geo. distance×∆dd_Parent -0.0620* -0.0234 (0.0327) (0.0305) Cultural Prox.×∆dd_Parent -0.194*** -0.190*** (0.0574) (0.0617) Observations 3,411 2,708 3,396 3,372 3,352 3,380 3,337 3,581 3,581 2,622 R-squared 0.287 0.304 0.288 0.294 0.293 0.295 0.292 0.298 0.292 0.326 Number of subsidiaries 88 66 87 87 86 87 86 93 93 64 35 Table 9: Exploring threshold effects in the subsidiary characteristics’ impact on the association between subsidiary and parent distance to default Regression results for the model ∆ _, = 0 + 1 ∆_, + 2 ∆ _ + 3 ∆ _ + 4 ∆ + 5 ∆ + 6 × ∆ _, + + , are reported in this table. ∆_, is the weekly change in distance to default of the subsidiary i in week t; ∆ _, is the weekly change in distance to default of the parent of subsidiary i; ∆ _ and ∆_ 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. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. are foreign subsidiary characteristics computed as of December 2006. These variables are described in detail in Table 2. Instead of using continuous variables, this table reports regression results using dummy variables which equal one for those affiliates whose characteristics rank above the median. Regressions also include subsidiary fixed effects, . 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.. Variables (9.1) (9.2) (9.3) (9.4) (9.5) (9.6) (9.7) (9.8) ∆dd_Parent 0.258*** 0.374*** 0.332*** 0.344*** 0.315*** 0.233*** 0.363*** 0.329*** (0.0355) (0.0519) (0.0496) (0.0448) (0.0411) (0.0318) (0.0470) (0.0491) ∆dd_Home 0.171*** 0.167*** 0.175*** 0.164*** 0.169*** 0.171*** 0.156*** 0.174*** (0.0481) (0.0584) (0.0479) (0.0498) (0.0501) (0.0487) (0.0474) (0.0477) ∆dd_Host 0.127*** 0.124*** 0.126*** 0.128*** 0.130*** 0.128*** 0.124*** 0.134*** (0.0264) (0.0319) (0.0264) (0.0258) (0.0265) (0.0265) (0.0266) (0.0262) ∆DEF -0.0125*** -0.0129*** -0.0120*** -0.0130*** -0.0126*** -0.0131*** -0.0115*** -0.0112*** (0.00376) (0.00424) (0.00383) (0.00376) (0.00376) (0.00394) (0.00377) (0.00368) ∆VIX -0.0272* -0.0227 -0.0277* -0.0297* -0.0270* -0.0266* -0.0235 -0.0270* (0.0151) (0.0158) (0.0149) (0.0158) (0.0149) (0.0152) (0.0142) (0.0149) H_Size×∆dd_Parent 0.0278 (0.0632) H_Capital×∆dd_Parent -0.125* (0.0659) H_Equity×∆dd_Parent -0.124* (0.0628) H_Funding×∆dd_Parent -0.133** (0.0600) H_Profitability×∆dd_Parent -0.0699 (0.0577) H_Liquidity×∆dd_Parent 0.0823 (0.0642) H_Provisioning×∆dd_Parent -0.142** (0.0678) H_Geo. Distance×∆dd_Parent -0.116* (0.0644) Observations 3,411 2,708 3,396 3,372 3,352 3,380 3,337 3,581 R-squared 0.287 0.305 0.291 0.294 0.291 0.296 0.295 0.292 Number of subsidiaries 88 66 87 87 86 87 86 93 36 Table 10: The impact of subsidiaries’ characteristics controlling for host country dummies-parent distance to default interaction Regression results for the model ∆ _, = 0 + 1 ∆_, + 2 ∆ _ + 3 ∆ _ + 4 ∆ + 5 ∆ + 6 × ∆ _, + 7 × ∆ _, + + , are reported in this table. ∆ _, is the weekly change in distance to default of the subsidiary i in week t; ∆ _, is the weekly change in distance to default of the parent of subsidiary i; ∆_ and ∆ _ are changes in average distance to defaults of all the publicly traded banks and companies in the host country and parent bank, respectively. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. 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, × ∆ _, , which are included but not reported. Regressions also include subsidiary fixed effects, . 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. Variables (10.1) (10.2) (10.3) (10.4) (10.5) (10.6) (10.7) (10.8) (10.9) ∆dd_Parent 0.393*** -0.139 0.553*** 0.959*** -0.0680 0.349*** 0.473*** 0.602 0.365*** (0.0402) (0.121) (0.198) (0.248) (0.0715) (0.112) (0.0627) (0.396) (0.0730) ∆dd_Home 0.169*** 0.166*** 0.171*** 0.166*** 0.166*** 0.168*** 0.155*** 0.171*** 0.172*** (0.0507) (0.0597) (0.0499) (0.0512) (0.0525) (0.0508) (0.0508) (0.0490) (0.0488) ∆dd_Host 0.116*** 0.110*** 0.115*** 0.116*** 0.115*** 0.116*** 0.113*** 0.123*** 0.123*** (0.0249) (0.0297) (0.0252) (0.0249) (0.0248) (0.0250) (0.0246) (0.0251) (0.0250) ∆DEF -0.0136*** -0.0148*** -0.0136*** -0.0134*** -0.0136*** -0.0141*** -0.0128*** -0.0128*** -0.0125*** (0.00364) (0.00392) (0.00370) (0.00369) (0.00365) (0.00387) (0.00362) (0.00352) (0.00350) ∆VIX -0.0319** -0.0270 -0.0318** -0.0329** -0.0308* -0.0313** -0.0287* -0.0325** -0.0319** (0.0153) (0.0161) (0.0153) (0.0155) (0.0153) (0.0153) (0.0143) (0.0147) (0.0148) Size×∆dd_Parent 0.464 (0.920) Capital ratio×∆dd_Parent 0.354 (0.638) Equity assets×∆dd_Parent -0.915 (1.208) Cust. deposits/Funding×∆ dd_Parent -0.658** (0.293) Profitability×∆dd_Parent -1.222 (1.857) Liquidity× ∆dd_Parent 0.115 (0.210) Provisions× ∆dd_Parent -2.203 (1.975) Geographic distance×∆dd_Parent -0.0423 (0.0497) Cultural Proximity× ∆dd_Parent -0.113* (0.0601) Host Country Dummy x∆ dd_Parent Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 3,411 2,708 3,396 3,372 3,352 3,380 3,337 3,581 3,581 R-squared 0.331 0.346 0.331 0.335 0.334 0.339 0.335 0.332 0.333 Number of subsidiaries 88 66 87 87 86 87 86 93 93 37 Table 11: The impact of host countries’ regulations on the association between foreign subsidiaries’ and parents’ distance to default Regression results for the model ∆ _, = 0 + 1 ∆_, + 2 ∆ _ + 3 ∆ _ + 4 ∆ + 5 ∆ + 6 × ∆ _, + + , are reported in this table. ∆_, is the weekly change in distance to default of the subsidiary i in week t; ∆ _ is the weekly change in distance to default of the parent of subsidiary i; ∆ _ and ∆_ 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. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. are subsidiary host country regulations measured as of December, 2006. These variables are described in detail in Table 2. Regressions also include subsidiary fixed effects, . 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. Variables (11.1) (11.2) (11.3) (11.4) (11.5) (11.6) (11.7) (11.8) (11.9) (11.10) 0.778*** 0.616*** 0.484*** 0.481*** 0.650*** 0.282*** 0.330 0.312*** 0.303*** 0.277*** ∆dd_Parent (0.239) (0.106) (0.1000) (0.155) (0.106) (0.0890) (0.208) (0.0499) (0.0612) (0.0792) 0.179*** 0.200*** 0.197*** 0.161** 0.174*** 0.175*** 0.186*** 0.178*** 0.230*** 0.176*** ∆dd_Home (0.0475) (0.0557) (0.0563) (0.0622) (0.0544) (0.0472) (0.0522) (0.0499) (0.0392) (0.0468) 0.133*** 0.125*** 0.129*** 0.127*** 0.125*** 0.134*** 0.138*** 0.127*** 0.108*** 0.134*** ∆dd_Host (0.0269) (0.0300) (0.0303) (0.0444) (0.0380) (0.0269) (0.0312) (0.0269) (0.0364) (0.0263) -0.011*** -0.0110*** -0.0106** -0.0123** -0.0124** -0.0118*** -0.0115*** -0.0111*** -0.0111** -0.0111*** ∆DEF (0.00362) (0.00400) (0.00411) (0.00504) (0.00447) (0.00358) (0.00387) (0.00378) (0.00464) (0.00364) -0.0278* -0.0180 -0.0196 -0.0392** -0.0258 -0.0281* -0.0265 -0.0218 -0.0393** -0.0270* ∆VIX (0.0140) (0.0157) (0.0148) (0.0179) (0.0177) (0.0148) (0.0160) (0.0141) (0.0182) (0.0147) -0.183** Disclosure×∆dd_Parent (0.0852) -0.0407*** Activity Restr.× ∆dd_Parent (0.0115) -0.0402** Capital Regulation× ∆dd_Parent (0.0162) -0.000432 Loan Classification×∆ dd_Parent (0.000306) -0.00236*** Provisioning × ∆dd_Parent (0.000666) -0.00545 Diversification× ∆dd_Parent (0.0579) -0.00530 Supervisory Powers× ∆dd_Parent (0.0157) -0.0125 Prompt Corrective×∆dd_Parent (0.0131) -0.00374* Reserves Req.× ∆ dd_Parent (0.00189) -0.00935 Financial Outflows×∆ dd_Parent (0.105) Obs. 3,576 3,073 3,073 2,158 2,512 3,555 3,319 3,309 2,414 3,581 R-squared 0.294 0.294 0.291 0.273 0.275 0.291 0.293 0.285 0.313 0.288 Number subsidiaries 92 77 77 55 62 91 84 84 62 93 38 Table 12: The impact of host countries’ regulations and subsidiaries' characteristics combined Regression results for the model ∆ _, = 0 + 1 ∆_, + 2 ∆ _ + 3 ∆ _ + 4 ∆ + 5 ∆ + 6 × ∆_, + 7 × ∆ _, + + , are reported in this table. ∆ _, is the weekly change in distance to default of the subsidiary i in week t; ∆ _, is the weekly change in distance to default of the parent of subsidiary i; ∆_ and ∆_ 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. ∆ is the change in interest rate spread between Bbb and Aaa rated companies. ∆ is the weekly change in the VIX volatility index. are foreign subsidiary characteristics (size, liquidity, funding structure, capital, etc.) computed as of December, 2006. are subsidiary host country regulations measured as of December, 2006. These variables are described in detail in Table 2. Regressions also include subsidiary fixed effects, . 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. Variables (12.1) (12.2) (12.3) (12.4) (12.5) (12.6) (12.7) (12.8) (12.9) (12.10) ∆dd_Parent 1.269*** 1.090*** 1.019*** 1.048** 1.275*** 0.711*** 1.125*** 0.891*** 0.865*** 0.871*** (0.366) (0.296) (0.281) (0.409) (0.275) (0.249) (0.308) (0.285) (0.293) (0.282) ∆dd_Home 0.171*** 0.189*** 0.182*** 0.134* 0.151** 0.165*** 0.170*** 0.172*** 0.214*** 0.167*** (0.0575) (0.0647) (0.0643) (0.0698) (0.0612) (0.0561) (0.0594) (0.0590) (0.0415) (0.0563) ∆dd_Host 0.124*** 0.114*** 0.117*** 0.117** 0.111** 0.126*** 0.123*** 0.119*** 0.0916** 0.124*** (0.0305) (0.0335) (0.0338) (0.0514) (0.0419) (0.0301) (0.0345) (0.0307) (0.0361) (0.0300) ∆DEF -0.014*** -0.0130** -0.0126** -0.0144** -0.0144** -0.0140*** -0.0131*** -0.0135*** -0.0138** -0.0137*** (0.00421) (0.00467) (0.00473) (0.00623) (0.00537) (0.00422) (0.00432) (0.00465) (0.00533) (0.00426) ∆VIX -0.0274* -0.0145 -0.0160 -0.0344* -0.0232 -0.0270* -0.0261 -0.0172 -0.0396* -0.0262 (0.0155) (0.0150) (0.0145) (0.0176) (0.0176) (0.0157) (0.0159) (0.0144) (0.0191) (0.0158) Disclosure ×∆dd_Parent -0.170* (0.0881) Activity Restrictions×∆dd_Parent -0.0349*** (0.00568) Capital Regulation×∆dd_Parent -0.0374** (0.0140) Loan Classification×∆dd_Parent -0.000215 (0.000224) Provisioning ×∆dd_Parent -0.00202*** (0.000681) Diversification×∆dd_Parent 0.0689 (0.0721) Supervisory Powers×∆dd_Parent -0.0179 (0.0105) Prompt Corrective×∆dd_Parent -0.0154 (0.0120) Reserves Req.× ∆dd_Parent -0.00309* (0.00153) Financial Outflows×∆dd_Parent -0.0403 (0.120) Capital ratio×∆dd_Parent -0.553 -0.755 -0.268 -1.294** -0.606 -1.034* -0.889* -0.790 -0.965** -1.071** (0.477) (0.520) (0.616) (0.533) (0.460) (0.510) (0.498) (0.487) (0.352) (0.459) Cust. deposits/Funding×∆dd_Parent -0.409 -0.341 -0.460 -0.405 -0.570* -0.315 -0.456 -0.405 -0.381 -0.365 (0.293) (0.377) (0.335) (0.465) (0.330) (0.313) (0.335) (0.352) (0.390) (0.321) Cultural Proximity×∆dd_Parent -0.207*** -0.200*** -0.206*** -0.241** -0.176** -0.222*** -0.242*** -0.221*** -0.215** -0.213*** (0.0637) (0.0545) (0.0512) (0.0934) (0.0838) (0.0639) (0.0602) (0.0588) (0.0930) (0.0623) Obs. 2,669 2,363 2,363 1,620 1,873 2,669 2,597 2,435 1,837 2,669 R-squared 0.326 0.325 0.322 0.296 0.297 0.322 0.330 0.312 0.354 0.322 Number of subsidiaries 65 56 56 39 44 65 62 59 44 65 39 Figure 1: Change in the mean distance to default across all multinational parent banks and across foreign subsidiaries 40 Appendix 1: The synthetic distance to default a la Falkenheim and Pennachi (2003) We model market leverage (mktlev) and standard deviation of asset returns (sigmaVA) as a linear function of accounting variables. Market leverage is total liabilities divided by the market value of assets computed from the Merton model (VA /X) described above. Similarly, the standard deviation of asset returns are computed from the Merton model (sA), using market equity data. For each weekly observation in our sample, we run a cross-sectional regression of mktlev and sigmaVA on a number of accounting variables. The explanatory variables we use are similar to those described in Falkenheim and Pennachi (2003). We make some modifications as our sample includes international banks operating in a number of different countries. We use book value of leverage (Asset / Liabilities) to proxy for market leverage. Given the significant impact earnings have on equity prices, we use three variables to capture their effects on market leverage and asset volatility. In particular, we use net income scaled by liabilities (Net Income / Liabilities), net income ratio squared (Net Income / Liabilities)2, and the quadratic term multiplied by a dummy variable set equal to one if net income is positive and zero if it is negative (Dummy*(Net Income / Liabilities)2). The quadratic terms allow for potential non- linearities in the relationship between net income and market leverage and asset volatility. We use standard deviation of earnings and standard deviation of liabilities growth to proxy for volatility of market value of assets. 18 Loan loss provisions (Loan loss provisions / Liabilities) capture differences in asset quality, net loans (Net loans / Liabilities) capture differences in business model, and deposits ratio (Deposits / Liabilities) capture differences in funding structure. We include bank size (log(Assets)) and a dummy variable set equal to one (BHC dummy) if the institution is a bank holding company to distinguish large holding companies from smaller more independent banks. Large banks and holding companies may have more 18 Liabilities growth for a given bank is measured as: log(Liabilitiest / Liabilitiest-1) 41 market power, greater brand recognition, and charter values. They may also be more complex, diversified and may have greater economies of scale. They are also more likely to engage in greater securitization and use of derivative products. Finally, we include eight region dummies to take into account geographical differences. 19 We run cross-sectional regressions each week over the years 2008 and 2009 using lagged values of accounting variables and region dummies to explain market leverage and asset volatility. We then use the weekly estimated coefficients on the lagged explanatory variables for non-publicly traded banks to create ‘synthetic’ market leverage and asset volatility measures for these banks. These ‘synthetic’ measures are then used to compute weekly distance to default measures. 20 Appendix Table A.1 reports Fama-Macbeth regression results for the time period used to estimate weekly credit risk measures. 21 The results are qualitatively similar to those reported in Falkenheim and Pennachi (2003). 19 The region coefficients capture relative effects for each of the regions with respect to the Africa region which is not included to avoid perfect multicollinearity. 20 We plug in the estimated values for market leverage and standard deviation of asset returns to compute weekly � 2 � �+�−− � � 2 distance-to-default: = � 2 √ 21 The standard errors for the coefficient estimates are based on the time series variability of the cross-sectional estimates, incorporating a Newey-West (1987) correction with three lags to account for possible autocorrelation in the estimates. 42 Appendix Table A.1. Synthetic credit risk scores This table reports results from weekly Fama and MacBeth (1973) regressions of market leverage (mktlev) and asset return volatility (sigmaVA) on lagged accounting variables. mktlev is total liabilities divided by the market value of assets computed from the Merton model sigmaVA is standard deviation of asset returns computed from the Merton model. The explanatory accounting variables are described in detail in the text. The time period is from September 2008 to December 2009. We report Fama and MacBeth regression coefficients as well as their corresponding Newey- West (1987) three lag corrected standard errors in parentheses. *** (**) (*) indicates significance at 1% (5%) (10%) level, respectively. VARIABLES mktlev sigmaVA log(Assets) -0.016*** -0.001*** (0.001) (0.000) Asset / Liabilities 0.138*** 0.074*** (0.053) (0.003) Net loans / Liabilities -0.084*** -0.011*** (0.007) (0.001) Loan loss provisions / Liabilities 1.950*** 0.074*** (0.133) (0.021) Deposits / Liabilities -0.020*** -0.002 (0.008) (0.001) Net Income / Liabilities 5.584*** 0.518*** (0.189) (0.028) (Net Income / Liabilities)2 0.659* -0.405*** (0.345) (0.036) Dummy×(Net Income / Liabilities)2 25.271*** 4.137*** (1.431) (0.238) Std Deviation (Net Income / Liabilities) 0.437*** 0.091*** (0.127) (0.012) Std Deviation (Liabilities growth) 0.127*** 0.020*** (0.019) (0.002) BHC dummy 0.007*** -0.000 (0.003) (0.000) Central Asia & Eastern Europe dummy -0.260*** -0.028*** (0.012) (0.002) East Asia and Pacific dummy -0.189*** -0.028*** (0.013) (0.002) Japan dummy -0.252*** -0.060*** (0.012) (0.002) Latin America & Caribbean dummy -0.214*** -0.021*** (0.013) (0.002) Middle East & North Africa dummy -0.121*** -0.004** (0.012) (0.002) North America dummy -0.255*** -0.033*** (0.012) (0.002) South Asia dummy -0.243*** -0.025*** (0.011) (0.002) Western Europe dummy -0.221*** -0.054*** (0.013) (0.002) Constant 1.650*** 0.022*** (0.067) (0.005) Observations 95,840 95,840 R-squared 0.638 0.462 43