بازارهای سرمایه داخلی و وام دهی شعبه های فرعی بانک چندملیتی

We use new panel data on the intra-group ownership structure and the balance sheets of 45 of the largest multinational bank holdings to analyze what determines the credit growth of their subsidiaries. We find evidence for the existence of internal capital markets through which multinational banks manage the credit growth of their subsidiaries. Multinational bank subsidiaries with financially strong parent banks are able to expand their lending faster. As a result of parental support, foreign bank subsidiaries also do not need to rein in their credit supply during a financial crisis, while domestic banks need to do so.

In this paper, we analyze two interrelated questions. First, we ask whether multinational banks operate an internal capital market across national boundaries. By an internal capital market we mean that parent banks allocate scarce capital to their subsidiaries. A multinational bank would not operate such a market in the absence of capital market frictions (Stein, 1997). Rather, subsidiaries would attract sufficient liabilities to finance profitable investment projects themselves. However, if capital markets are not functioning perfectly, it can be advantageous for parent banks with better access to external funding to internally allocate capital to subsidiaries in order to manage their lending growth. Whether a bank operates an internal capital market or not is important because such a market creates financial linkages between subsidiaries. The second and related issue we analyze concerns the consequences that the operation of an internal capital market by multinational banks may have for the countries involved. We ask whether the presence of multinational bank subsidiaries means that economic shocks are more easily transmitted across borders. In addition we want to know whether multinational bank subsidiaries ‘cut and run’ during a financial crisis or whether they, on the contrary, provide a stable source of credit (in particular when compared to domestic banks). We contribute to the research on the existence and the ramifications of internal capital markets in several ways. As regards the existence of internal capital markets, we make two contributions. First, we examine whether banks operate internal capital markets across national boundaries. Empirical evidence on internal capital markets within banks only exists for the United States. Houston et al. (1997) show that the credit growth of a subsidiary is negatively correlated with the loan growth in other US subsidiaries of the bank holding. Dahl et al. (2002) show that such correlated credit growth patterns are due to equity flows between the parent bank and its subsidiaries. Ashcraft (2004) demonstrates that banks that are affiliated with a multi-bank holding company are less likely to experience financial distress and recover more quickly in case of such distress because of capital injections by the parent company. We extend this line of research to multinational banks.1 Second, our detailed information on intra-bank ownership allows us to analyze whether particular types of multinational bank subsidiaries are more closely integrated into internal capital markets than others. We distinguish between greenfield subsidiaries and takeover subsidiaries as well as between subsidiaries that are geographically close to their parent bank and those that are further away. Earlier empirical research treated bank subsidiaries as a homogenous group and ignored the potential differences between subsidiary types. We also contribute to the research on the effects of multinational banking. Earlier research shows that lending by multinational banks tends to transmit home country financial shocks (Peek and Rosengren, 1997, Peek and Rosengren, 2000a, Van Rijckeghem and Weder, 2000 and Van Rijckeghem and Weder, 2001) but to dampen host country financial shocks (Peek and Rosengren, 2000b and De Haas and Van Lelyveld, 2006). Multinational bank lending also tends to be influenced by the home country business cycle (Martinez Peria et al., 2002 and Morgan and Strahan, 2004). We improve on this work in two ways. First, most studies limit themselves to multinational bank linkages between one specific home region (United States, Japan or Western Europe) and one specific host region (Latin America or central and eastern Europe). Our approach differs in that we do not make an ex ante distinction between home and host regions but acknowledge that countries can be home and host country at the same time. We also use a more comprehensive country sample that better reflects the current state of bank globalization: • Our dataset includes 45 multinational banks from 18 home countries with 194 subsidiaries across 46 countries. Most parent banks (83 per cent) and subsidiaries (73 per cent) are based in Europe, partly reflecting the eastward expansion of many European banks after the fall of the Berlin Wall. Only about 14 per cent of all parent banks and subsidiaries are based in North America. North American banks are relatively domestically oriented, whereas European banks are on average more internationalized (IMF, 2007). • The number of multinational bank subsidiaries in Africa and Asia is limited as many countries in these regions still have limitations on majority foreign bank ownership. Latin America is host to some 5 per cent of all subsidiaries in our dataset, mainly of Spanish origin. • The time dimension (1991–2004) of our dataset reveals that banking systems have become increasingly globalized over time. Not only did the number of multinational banks increase, individual banks also became more globalized as measured by the number of foreign subsidiaries, especially through foreign takeovers. A second improvement of our empirical approach compared to existing studies is that we use bank-level information on intra-bank linkages rather than aggregate bank lending data. Van Rijckeghem and Weder, 2000 and Van Rijckeghem and Weder, 2001 find that multinational banks adjust credit lines to third countries in reaction to losses in a crisis country. However, the authors use aggregate Bank for International Settlements (BIS) data and therefore cannot measure intra-bank linkages at the bank level. Peek and Rosengren (2000a), in their study of multinational bank lending in Latin America, rely on aggregated BIS data as well. Our bank-level data allow us to more precisely attribute empirical findings to banks’ internal capital markets rather than to more general cross-country correlations. We find that subsidiaries of stronger parent banks grow faster and that parent banks trade off lending across countries. As a result of parental support, foreign bank subsidiaries do not typically rein in their lending during a financial crisis, while domestic banks are forced to do so. These findings are in line with earlier empirical results for specific regions, but are based on a much broader sample of home and host countries and on bank-level rather than aggregated data. Our findings are also the first to show that banks not only operate internal capital markets at the national but also at the international level. Greenfield subsidiaries and remote subsidiaries are most closely integrated into such internal capital markets. A limitation of our empirical approach is that we cannot track the actual transactions within internal capital markets and therefore cannot fully ascertain that the bank lending patterns we find are caused by equity flows between parent banks and their subsidiaries (as Dahl et al. (2002) showed for bank holdings in the USA). However, on the basis of a number of innovative robustness tests we are able to rule out the most likely alternative explanations for the lending patterns we observe. The remainder of this paper is structured as follows. In Section 2 we develop our theoretical predictions. Section 3 discusses the data we use, after which Section 4 explains our estimation methodology. Sections 5 and 6 present the empirical results and Section 7 concludes.

We examine whether multinational banks operate an internal capital market and, if so, whether the presence of financially integrated banks exposes countries to foreign economic developments. We analyze how multinational bank lending is influenced by the macroeconomic situation in the host country and in the home country, and by the financial characteristics of the subsidiary itself, those of the parent bank, and those of other subsidiaries. This is the first paper to analyze these micro and macro determinants of multinational bank lending within an integrated empirical framework. Our findings indicate that multinational banks manage the credit growth of their subsidiaries. In particular, we find strong proof of support effects. Subsidiaries of strong parent banks—that have high net interest margins or low loan loss provisioning to net interest revenue—grow faster. Subsidiaries of parent banks that keep less liquid assets on their balance sheets are able to grow faster as well. If other subsidiaries in the same banking group are relatively profitable this also positively influences subsidi- ary lending. Finally, we find that during systemic banking crises, multinational bank subsidiaries keep lending, whereas domestic banks are forced to sharply restrict their credit supply. Robustness tests add further credibility to the claim that our findings do not reflect macroeconomic linkages but in- tra-bank capital management. As Houston et al. (1997) did for national bank holdings in the US, we provide evidence for the operation of internal capital markets at the international level. We find weaker empirical support for substitution effects. We show that, in line with substitution effects, multinational bank subsidiaries expand lending faster when economic growth in their home country decreases. Likewise, we find that high host country growth stimulates credit growth, although this effect is limited to greenfield and remote subsidiaries. However, our robustness tests show that domestic bank lending is also procyclical (though somewhat less so) and also negatively related to economic growth in the home countries of foreign competitor banks. So while our findings are in line with substitution effects, they may also partly reflect macroeconomic linkages between countries. Our findings imply that openness to multinational bank subsidiaries can benefit host countries. Since the pace of multinational bank lending is partly determined abroad, the aggregate credit supply in the host country becomes more stable and less strongly correlated with the local business cycle. We find that multinational banks provide a stabilizing factor during local financial turmoil in particular. Diversity in bank ownership therefore contributes to credit stability, allowing firms with binding cred- it constraints to optimize investments and households to better smooth consumption over time. An important caveat is that the above interpretation of our results presumes that parent banks operate an internal capital market because they are better able to attract liquidity and raise capital than individual subsidiaries. If parent banks themselves experience problems with raising funds, they may no longer be able to support subsidiaries. Lending by foreign subsidiaries may even be scaled back in order to free up capital for the parent bank, leading to contagion from home to host countries. Indeed, the global financial crisis is currently testing the resilience of the support effects docu- mented in this paper. Multinational banks have so far continued to support their foreign subsidiaries. For instance, in Kazakhstan, cross-border foreign bank credit to domestic banks has dried up and the latter consequently had to rein in their own lending. Multinational bank subsidiaries were much less affected. ATF Bank, a mid-sized Kazakh bank owned by the Italian UniCredit group, for instance ob- tained credit lines and capital injections from its parent bank. In Latvia, multinational bank subsidiar- ies continued to receive funding from their parent banks, while Parex Bank, the second-largest bank and the only domestic bank left in the country, suffered a run on its deposits, forcing the government to step in. Both examples illustrate that lending by multinational bank subsidiaries tends to be more stable than cross-border foreign bank lending (see also Peek and Rosengren, 2000a ). For countries that plan to open up their banking system, it may thus make sense to not only permit cross-border foreign bank lending, but to also allow committed multinational banks to establish or buy local subsidiaries. When doing so, countries should ideally encourage entry from a variety ofhome countries. Diversified multinational bank ownership is preferred to highly concentrated foreign ownership as the latter implies an overreliance on intra-bank support from one or a very small num- ber of foreign parent banks.