«The International Transmission of Bank Liquidity Shocks: Evidence from an Emerging Market PHILIPP SCHNABL∗ ABSTRACT I exploit the 1998 Russian ...»
THE JOURNAL OF FINANCE • VOL. LXVII, NO. 3 • JUNE 2012
The International Transmission of Bank
Liquidity Shocks: Evidence from an Emerging
I exploit the 1998 Russian default as a negative liquidity shock to international banks
and analyze its transmission to Peru. I ﬁnd that after the shock international banks
reduce bank-to-bank lending to Peruvian banks and Peruvian banks reduce lending to Peruvian ﬁrms. The effect is strongest for domestically owned banks that bor- row internationally, intermediate for foreign-owned banks, and weakest for locally funded banks. I control for credit demand by examining ﬁrms that borrow from sev- eral banks. These results suggest that international banks transmit liquidity shocks across countries and that negative liquidity shocks reduce bank lending in affected countries.
AN IMPORTANT QUESTION IN FINANCE is whether ﬁnancial institutions transmit liquidity shocks across markets, and if so, whether such shocks impact real economic activity. On the one hand, efﬁcient market theory suggests that, as long as investment opportunities are constant, shocks to ﬁnancial institutions in one market should not affect lending in other markets. On the other hand, if ﬁnancing frictions prevent ﬁnancial institutions from accessing alternative ﬁnancing sources to cover shortfalls, then liquidity shocks in one market may affect lending in other markets.
∗ Philipp Schnabl is at the Stern School of Business, New York University. I am deeply grateful to my advisors Jeremy Stein, Andrei Shleifer, Sendhil Mullainathan, and Edward Glaeser for their support and guidance. I am deeply indebted to an anonymous referee for numerous comments that signiﬁcantly improved the paper, as well as an Associate Editor and the Editor, Cam Harvey. I thank Viral Acharya, Ashwini Agrawal, Marianne Bertrand, Edward Glaeser, Luigi Guiso, Jens Hilscher, Victoria Ivashina, Marcin Kacperczyk, Asim Khwaja, Alexander Ljungqvist, Deborah Lucas, Thomas Mertens, Atif Mian, Holger Mueller, Anthony Saunders, David Scharfstein, Antoinette Schoar, and Eric Werker for helpful comments and suggestions. This paper also bene- ﬁted greatly from comments of seminar participants at Chicago Booth School of Business, Columbia Business School, Duke University (Fuqua), European University Institute, Federal Reserve Board, Harvard University, Inter-American Development Bank, International Monetary Fund, London Business School, London School of Economics, New York University (Stern), Northwestern Uni- versity (Kellogg), MIT (Sloan), Stanford University, University of Pennsylvania (Wharton), and the 2008 Conference on Bank Structure. I am grateful to the Superintendencia de Banca, Seguros y AFP (SBS) for providing the data in this paper and I thank Diego Cisneros, Jorge Mogrovejo, Javier Poggi, Jorge Olcese, Mitchell Canta, and especially Adriana Valenzuela for clarifying many data questions. The results in this paper do not necessarily reﬂect the views of SBS. All errors are my own. The paper was previously circulated as “Financial Globalization and the Transmission of Credit Supply Shocks: Evidence from an Emerging Market.”
897898 The Journal of Finance R
To provide compelling evidence that ﬁnancial institutions transmit liquidity shocks across markets and that liquidity shocks have real effects, one has to address several challenges. First, liquidity shocks are typically systemic (or aggregate) shocks that affect many ﬁnancial institutions at the same time.
Hence, one needs to identify a setting in which similar ﬁnancial institutions differ in their exposure to a systemic shock. Second, the same economic forces that trigger liquidity shocks may also directly affect ﬁrms’ investment opportunities. One therefore needs to control for changes in investment opportunities, ideally at the ﬁrm level. Third, ﬁrms can offset bank liquidity shocks by borrowing elsewhere. Hence, one needs to observe real ﬁrm outcomes to evaluate the impact of liquidity shocks on real economic activity.
In this paper, I address these challenges by interpreting the 1998 Russian debt default as a negative liquidity shock to international banks and by analyzing its impact on bank lending in Peru. This setting is ideal for three reasons.
First, the impact of the liquidity shock varies across Peruvian banks because Peruvian banks vary in their exposure to international bank-to-bank lending.
I use this variation to construct bank-speciﬁc exposure to the liquidity shock.
Second, I collect a novel data set that covers all bank loans in Peru. This data set allows me to control for changes in ﬁrm-level investment opportunities by analyzing ﬁrms that borrow from several banks. Third, I observe total borrowing, loan default, and survival of Peruvian ﬁrms. I can therefore estimate the impact on real economic activity for the universe of Peruvian ﬁrms.
The following example illustrates the transmission channel under investigation. Suppose Citibank-US and UBS are both international banks that provide loans to Peruvian banks. Assume Citibank-US has a subsidiary in Peru (e.g., Citibank-Peru), but UBS does not. I refer to banks such as Citibank-US as owners and to banks such as UBS as arm’s-length lenders. I show that owners are less likely to transmit liquidity shocks to their subsidiaries than arm’s-length lenders to their borrowers. The intuition is straightforward: An owner can directly monitor its subsidiary’s lending and can therefore prevent risk shifting in response to higher interest rates but an arm’s-length lender cannot. Hence, among banks that borrow internationally, the transmission of liquidity shocks by arm’s-length lenders is stronger than the transmission by owners.
To analyze the transmission of the Russian default empirically, I proceed in several steps. First, I document the impact of the Russian default on international banks. I ﬁnd that, after the Russian default, both owners (e.g., Citibank-US) and arm’s-length lenders (e.g., UBS) experience a sharp drop in their share prices. The drop is similar in magnitude for both groups of banks.
This ﬁnding suggests that the Russian default represents a systemic liquidity shock that negatively affects all international banks.
Second, I examine the impact of the liquidity shock on bank-to-bank lending to Peruvian banks. Among Peruvian banks that borrow internationally, I distinguish between foreign-owned banks (e.g., Citibank-Peru) and domestically owned banks (e.g., Banco Wiese). I ﬁnd that owners lend more to their Peruvian subsidiaries and less to other Peruvian banks (e.g., Citibank-US lending more to Citibank-Peru and less to Banco Wiese), but arm’s-length lenders reduce International Transmission of Bank Liquidity Shocks 899 loans to all Peruvian banks (e.g., UBS lending less to Citibank-Peru and less to Banco Wiese). As a result, after the Russian default, bank-to-bank loans to domestically owned banks decrease by 61% but bank-to-bank loans to foreignowned banks decrease by only 26%.
I next analyze the impact of the liquidity shock on lending by Peruvian banks.
The liquidity shock affects lending under two conditions. The ﬁrst condition is that banks cannot offset the shock by accessing other sources of ﬁnancing. The second condition is that ﬁrms cannot offset the shock by switching across banks or borrowing elsewhere.
To evaluate whether banks can offset the shock, I use loan-level data and analyze lending by Peruvian banks before and after the liquidity shock. I control for changes in investment opportunities at the ﬁrm level by focusing on ﬁrms that borrow from more than one bank and compare changes in borrowing from different banks within ﬁrms (e.g., for the same ﬁrm, I compare the change in borrowing from Citibank-Peru and Banco Wiese). The analysis shows that domestically owned banks with international bank-to-bank loans (e.g., Banco Wiese) reduce lending by 8.2% relative to foreign-owned banks (e.g., CitibankPeru). As an additional control group, I analyze locally funded banks that did not borrow from international banks prior to the Russian default. I ﬁnd that domestically owned banks with international borrowing reduce lending by 13.1% relative to locally funded banks. These results suggest that Peruvian banks transmit the liquidity shock to Peruvian ﬁrms and that the transmission is larger for banks that are more exposed to the liquidity shock.
I conduct several robustness tests to ensure that my results reﬂect the impact of the liquidity shock. First, I control for a large number of loan, ﬁrm, and bank characteristics. The results are robust to including these variables.
Second, I analyze changes in lending by ﬁrm age, ﬁrm size, and export status.
I ﬁnd that the results are similar across groups, which suggests that the results are not driven by a single group that changes its bank preferences after the Russian default. Third, I estimate the transmission of the liquidity shock without controlling for ﬁrm-level investment opportunities. The analysis yields qualitatively similar results, which suggests that the change in lending can be explained solely by the liquidity shock. Fourth, I estimate the impact of the liquidity shock using the change in international bank-to-bank borrowing as the main explanatory variable. I ﬁnd qualitatively similar results, which provides direct evidence on the transmission channel under investigation. Fifth, I estimate a placebo regression 1 year before the liquidity shock and ﬁnd no statistically signiﬁcant effect of bank exposure on bank lending.
To evaluate whether ﬁrms can offset the shock, I analyze the impact of the liquidity shock on real economic activity. I measure a ﬁrm’s exposure to the liquidity shock by computing borrowing from each type of bank as a share of total borrowing before the liquidity shock. I expect that ﬁrms with established relationships to less affected banks have a smaller reduction in total borrowing.
Indeed, I ﬁnd that a one-standard-deviation increase in the share of borrowing from foreign-owned banks increases total borrowing after the liquidity shock by 3.0%, reduces loan default by 1.1 percentage points, and increases ﬁrm survival 900 The Journal of Finance R by 1.4 percentage points relative to domestically owned banks that borrow internationally. I ﬁnd qualitatively similar results for locally funded banks but the quantitative impact is smaller and some results are not statistically signiﬁcant. These ﬁndings suggest that Peruvian ﬁrms cannot offset the negative liquidity shock and that the liquidity shock affects real ﬁrm outcomes.
Overall, my results establish the transmission of a bank liquidity shock originating in one country, Russia, to another country, Peru. The transmission channel is through international lending between banks. The transmission is strongest for domestically owned banks that borrow internationally, intermediate for foreign-owned banks, and weakest for locally funded banks. These results suggest that lending between international banks establishes a transmission channel for bank liquidity shocks across countries and that negative liquidity shocks reduce bank lending in affected countries.
The analysis of bank liquidity shocks connects to a large literature on the impact of ﬁnancial shocks on the real economy.1 Theoretical work by Bernanke and Blinder (1988), Bernanke and Gertler (1989), Holmstrom and Tirole (1997), and Stein (1998) shows that ﬁnancial shocks affect real ﬁrm outcomes only if there are credit market imperfections at both the bank and the ﬁrm level.
The early empirical literature by Bernanke (1983) and Bernanke and Blinder (1992) uses correlations between aggregate changes in liquidity and aggregate changes in output to show that ﬁnancial shocks affect real outcomes.
However, aggregate correlations may be driven by omitted variables that affect both bank credit supply and ﬁrm investment. More recent empirical work uses variation across banks and ﬁrms (e.g., Kashyap, Lamont, and Stein (1994), Kashyap and Stein (2000), Campello (2002), Ashcraft (2006), and Ashcraft and Campello (2007)) or natural experiments (e.g., Peek and Rosengren (1997), Peek and Rosengren (2000a), Ashcraft (2005), Gan (2007), Khwaja and Mian (2008), Paravisini (2008), Chava and Purnanandam (2011), Iyer and Peydro (2011)) to control for omitted variables. This paper is closest to that of Khwaja and Mian (2008), who also use a within-ﬁrm estimator to identify the bank lending channel. However, their paper studies a domestic shock to bank deposits rather than the transmission of liquidity shock across countries.
This paper relates to studies on the differences between foreign- and domestically owned banks in emerging markets. Empirical work using cross-sectional data on lending, such as Berger, Klapper, and Udell (2001), Mian (2006), and Detragiache, Tressel, and Gupta (2008), or using panel data on foreign bank entry, such as Gormley (2010), ﬁnds that foreign-owned banks tend to ﬁnance larger ﬁrms and domestically owned banks tend to ﬁnance smaller ﬁrms. Using bank-level data for Latin American and Asian countries, Diamond and Rajan (2001), Peek and Rosengren (2000b), and Detragiache and Gupta (2004), among others, ﬁnd that, after ﬁnancial crises, foreign-owned banks increase 1 I deﬁne a bank liquidity shock as a shock to a bank’s funding liquidity. Alternatively, some papers in the bank lending literature refer to such shocks as credit supply shocks. In the context of this paper, there is no substantive difference between bank liquidity shocks and credit supply shocks.
International Transmission of Bank Liquidity Shocks 901 lending relative to domestically owned banks. Other authors, such as Arena, Reinhart, and Vazquez (2007), ﬁnd only weak evidence that foreign- and domestically owned banks behave differently during ﬁnancial crises. My paper is different from previous studies because I distinguish between domestically owned banks that borrow internationally and domestically owned banks that do not. Moreover, I estimate the transmission of liquidity shocks using loanlevel data, which allows me to control for ﬁrm-level changes in investment opportunities.