The Second Wave of Global Liquidity: Why Are Firms Acting Like Financial Intermediaries?

Caballero, Julián; Panizza, Ugo; Powell, Andrew Working Paper The Second Wave of Global Liquidity: Why Are Firms Acting Like Financial Intermediaries? IDB Working Paper Series, No. IDB-WP-641 Provided in Cooperation with: Inter-American Development Bank (IDB), Washington, DC Suggested Citation: Caballero, Julián; Panizza, Ugo; Powell, Andrew (2016) : The Second Wave of Global Liquidity: Why Are Firms Acting Like Financial Intermediaries?, IDB Working Paper Series, No. IDB-WP-641, Inter-American Development Bank (IDB), Washington, DC, https://hdl.handle.net/11319/7638 This Version is available at: https://hdl.handle.net/10419/146447 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc-nd/3.0/igo/legalcode The Second Wave of Global Liquidity: Why Are Firms Acting Like Financial Intermediaries? Julián Caballero Ugo Panizza A ndrew Powell IDB WORKING PAPER SERIES Nº IDB-WP-641 A pril 2016 Department of Research and Chief Economist Inter-American Development Bank A pril 2016 The Second Wave of Global Liquidity: Why Are Firms Acting Like Financial Intermediaries? Julián Caballero* Ugo Panizza** A ndrew Powell* * Inter-American Development Bank ** The Graduate Institute, Geneva, and Centre for Economic Policy Research (CEPR) Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library Caballero, Julián The second wave of global liquidity: why are firms acting like financial intermediaries? / Julián Caballero, Ugo Panizza, Andrew Powell. p. cm. — (IDB Working Paper ; 641) Includes bibliographic references. 1.Corporations–Finance . 2. Bonds. I. Panizza, Ugo. II. Powell, Andrew (Andrew Philip). III. Inter-American Development Bank. Department of Research and Chief Economist. IV. Title. V. Series. IDB-WP-641 Copyright © Inter-American Development Bank. This work is licensed under a Creative Commons IGO 3.0 Attribution- NonCommercial-NoDerivatives (CC-IGO BY-NC-ND 3.0 IGO) license (http://creativecommons.org/licenses/by-nc-nd/3.0/igo/ legalcode) and may be reproduced with attribution to the IDB and for any non-commercial purpose, as provided below. No derivative work is allowed. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the UNCITRAL rules. The use of the IDB's name for any purpose other than for attribution, and the use of IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC-IGO license. Following a peer review process, and with previous written consent by the Inter-American Development Bank (IDB), a revised version of this work may also be reproduced in any academic journal, including those indexed by the American Economic Association's EconLit, provided that the IDB is credited and that the author(s) receive no income from the publication. Therefore, the restriction to receive income from such publication shall only extend to the publication's author(s). With regard to such restriction, in case of any inconsistency between the Creative Commons IGO 3.0 Attribution-NonCommercial-NoDerivatives license and these statements, the latter shall prevail. Note that link provided above includes additional terms and conditions of the license. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. http://www.iadb.org 2016 1 Abstract* Recent work suggests non-financial firms have acted like financial intermediaries particularly in emerging economies. This paper corroborates these findings but then asks “why?” The results indicate evidence for carry-trade activities, but they are focused on countries with higher levels of capital controls, particular controls on inflows. There is little evidence for such activities given other potential motives. It is posited that this phenomenon is due more to the reaction of countries in the face of low global interest rates, quantitative easing and strong capital inflows than incomplete markets or the retreat of global banks due to impaired balance sheets or tighter regulations. JEL classifications: E51, F30, F33 Keywords: Corporate finance, Bond issuance, Currency mismatch, Carry trade, Capital controls * We would like to thank Jorge Carrera, Claudio Raddatz, Dave Seerattan and other participants at the IDB’s Network of Central Banks and Finance Ministries of Latin America in Lima, October 2015, participants at seminars at FGV Sao Paulo, University of St. Gallen and IDB for comments. We would also like to thank Jaime Ramírez and Daniela Sánchez for excellent research assistance. The views expressed are strictly those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors or the countries they represent. Contact author Andrew Powell at [email protected]. 2 1. Introduction The recent fall in international bank lending and the rise of dollar-denominated international bond issuance, particularly from non-financial corporations based in emerging economies, has been labelled “the second phase of global liquidity” (Shin and Zhao, 2013). Since 2010 international bond issuances by non-financial corporates based in emerging economies has nearly doubled, reaching $400 billion by the end of 2014 (Acharya et al., 2015; see also Turner, 2013, and IMF, 2015). What did firms do with the proceeds of these bond issues? Bruno and Shin (2015) show that bond issuance has not been used solely for real investment but also to increase cash holdings or other liquid assets. Powell (2014) documents a positive correlation among U.S. dollar issuances in Latin America, corporate deposits in Latin American and Caribbean (LAC) financial systems and domestic credit. This behavior is consistent with the idea that, by acting as financial intermediaries, non-financial firms have replaced banks as the conduit through which international financial conditions affect domestic liquidity and credit growth in emerging economies. In this paper we ask why non-financial firms have taken on this role. We suggest that non-financial firms are more likely to act like financial intermediaries in countries with tighter capital controls because non-financial corporates have a comparative advantage in arbitraging capital controls or other regulations that have prevented banks from pursuing what appear to be profitable opportunities. Low interest rates in advanced economies fueled the fear in emerging economies that strong capital inflows, including carry-trade type activities, would led to credit booms and currency appreciation. Some emerging economies responded to this situation with tighter regulations on capital movements. However, non-financial firms may have ways of escaping such controls as they can issue bonds in offshore financial centers and then bring the proceeds of that issuance into the home country via an inter-company loan which in the balance of payments is normally counted as FDI and may thus elude capital controls or taxes levied on portfolio flows (see McCauley, Upper, and Villar, 2013). We test if the presence of capital controls increases the likelihood that non-firms act like financial intermediaries by using data for the period 2000-14 covering 766 non-financial firms located in 18 emerging market countries. We show that these corporations are more likely to hold the resources obtained from foreign currency bond issuances in liquid assets when potential 3 returns from carry trade are high and there are capital account restrictions on inflows. We conjecture that in countries with no or few capital controls, banks remain the main conduit for transmitting global financial conditions to domestic markets but in those countries that have adopted tighter capital controls (especially controls on inflows) this role is at least to some degree being played by corporates. We also show that our main result is not driven by the fact that global banks have been retreating from lending to emerging market countries. Our paper is related to several strands of the literature spanning financial depth and corporate financial structure, the role of international banks, and the credit cycle and systemic macroeconomic financial risks. A useful starting point is the corporate finance literature that discusses a “pecking order” for firm financing. This implies that a firm would normally use internal sources to finance projects or operations and only seek outside funds when those are exhausted (Myers, 2003). An implication is that, unlike financial intermediaries, non-financial corporations’ liabilities and liquid financial assets should be negatively correlated (Shin and Zhao 2013). While this is the case for US firms, in emerging economies there is a positive correlation between debt and liquid assets (Shin and Zhao, 2013, and Bruno and Shin, 2015). Bruno and Shin (2015) is perhaps the paper which is closest to ours. This paper also considers the rise in issuance of non-financial corporates and, in analyzing the determinants of issuance, finds evidence in favor of carry trade activities. Our data and methodology in identifying carry trade activities are somewhat different but still we corroborate their findings in this regard. More importantly, we highlight the importance of capital controls and conclude that carry trade activities by non-financial firms are consistent with the presence of regulatory arbitrage. We also test for alternative hypotheses focusing on the retreat of global banks or credit market imperfections but find no evidence in these directions. A related strand of literature focuses on recent trends in international credit flows. Turner (2013) highlights the shift from bank financing to bond financing, particularly for emerging economies. Chung et al. (2014) document the importance of this trend in terms of overall global liquidity and discuss the potential ramifications for financial stability. Powell (2014) considers the case of four large Latin American economies (Brazil, Chile, Colombia and Mexico) and documents a strong increase in issuance from non-financial firms, particularly in US dollars (Rodrigues-Bastos, Kamil, and Sutton 2015 show that this pattern also holds for a larger sample of Latin American economies). He also presents evidence of an increase in local currency 4 denominated domestic credit which appears to be financed by corporate deposits and documents a deterioration of firms’ balance sheets due to a combination of rising dollar amortization schedule and falling earnings ratios. In our paper, we move beyond cross-country correlations and use firm-level data to document when and why non-financial firms act as financial intermediaries. Our paper also relates to the literature on the links between offshore bond issuances and capital controls Shin (2013) and McCauley, Upper, and Villar (2013) document the recent increase in issuance of Brazilians and Chinese non-financial firms through subsidiaries in offshore financial centers and suggest that issuances through foreign subsidiaries may enable firms to evade capital controls or taxes on certain inflows. Powell (2014) considers this issue in the case of Latin America and shows that while in the case of Brazil issuance on a nationality basis exceeds issuance on a residency basis the opposite is true for Chile. As Chile does not have capital controls, while Brazil does, this difference provides prima facie evidence for the potential importance of such controls. In this paper, we test this hypothesis and find that capital controls do indeed increase financial firms’ incentives to act as financial intermediaries. Finally, our paper relates to the recent literature attempting to explain relatively high corporate cash holdings in the United States. Bates, Kahle and Stulz (2009), for example, argue that precautionary motives may play an important role in explaining U.S. firms’ high cash-to- assets ratios. As corporates in emerging economies are operating in an environment of incomplete financial markets, their actions may well be different than those of corporates in advanced economies. Large corporates, for instance, may have better access to capital markets than smaller firms that they have relationships with, such as suppliers, and hence might borrow more to be able to pass the proceeds on in the form of direct loans to these firms exploiting the business relationships. In this manner larger firms may attempt to complete financial markets in environments where financial depth is limited (Petersen and Rajan, 1997; Demirgüç-Kunt and Maksimovic, 2001; Fisman and Love, 2003; Levine, 2005). This line of argument suggests that there might then be a link between the financial structure of large corporates in emerging economies and financial depth. The lower is financial depth the more we may expect to see larger corporates borrowing to be able to correct such market failures in financial markets. We test whether the incentives of non-financial firms to act as financial intermediaries depend on 5 credit market imperfections (as proxied by financial depth or creditors’ rights) and do not find any evidence supporting this hypothesis. Another literature focuses on the role of international banks. International banks expanded during the 1990s and early 2000s through increased direct lending to clients in other countries, through establishing branches and subsidiaries in host nations, and hence collecting deposits and lending, and through the purchase of securities and structured products issued by foreign entities.1 However, the global financial crisis severely hit banks’ balance sheets, provoking a reduction in leverage and a retreat from international activities such that banks focused more on core assets and provoked a fundamental rethink of regulation. García-Luna and Van Rixtel (2014) provide a description of the retreat of global banks and discuss motivations including impaired balance sheets and regulatory developments, and Karam et al. (2014) consider changes in country ratings and their impacts on bank funding. If the rise of corporates as financial intermediaries is related to the retreat of global banks, then we would expect to see a relation with risk. In other words, if banks retreated more quickly from countries with lower ratings, then perhaps it is in those countries where corporates are now acting more like financial intermediaries. In what follows we test this proposition and we also test directly if corporates are behaving like financial intermediaries where international bank claims have fallen the most. 2. Data We collected annual data for the period 2000-2014 on firms’ balances sheets and bond issuances from two different sources. We focus on a sample composed of the 50 largest listed non-financial and non-foreign owned firms in each of 18 emerging markets. The baseline analysis includes a total of 766 firms.2 Bond issuance by our sample of firms was on a rising trend before the global financial crisis hit; it contracted in 2008 and then boomed over 2009-2013 (Figure 1). We obtained annual data on firms’ liquid financial assets and other balance sheet variables from the Thomson-Reuters Worldscope database and sourced data on bond issuances 1 On the expansion and role of foreign banks in emerging economies see for example Goldberg (2002), Martínez- Peria, Powell and Vladkova (2005) and Galindo, Micco and Powell (2005). 2 We have fewer than 50 firms in countries where there are less than 50 listed domestically owned non-financial corporations. 12 In Table 4, we measured capital account openness using the updated version of the Chinn and Ito (2006) aggregate index. This data source does not contain separate indicators for controls on inflows as opposed to controls on outflows. It is, however, plausible that controls on inflows are more relevant for non-financial firms that are trying to elude capital controls to exploit carry trade opportunities. To test this hypothesis, we use the Fernández et al. (2015) database on capital controls, which does contain separate measures for controls on outflows and on inflows. Table 5 reports the results using our baseline measure of FXB (the results are robust to using the other definitions). In the first column of Table 5, we estimate the same model of column 2, Table 4 by replacing the Chinn and Ito index with the overall measure (inflows and outflows) of capital account openness of Fernández et al. (2015).9 The results are similar to those of Table 4 (Panel B of Figure 3, plots the results), but the coefficients are not as precisely estimated as when we use the Chinn and Ito index (𝛿 and 𝜙 are statistically significant at the 10 percent confidence level while they were significant at the one percent confidence level in Table 4). Next, we use the Fernández et al. (2015) measures of openness to inflows (KI, Column 2) and outflows (KO, column 3). We find that the regression that uses controls on inflows yields results which are similar to those obtained for the overall index, but the coefficient are more precisely estimated (they are statistically significant at the 5 percent confidence level). The regression that uses controls on outflows, instead, yields results that are qualitatively similar to the regression that uses the overall index but with statistically insignificant coefficients for 𝛿 and 𝜙. In fact, in this case we find that capital account openness affects the elasticity of holdings of liquid financial assets to bond issuances but that the spread does not matter. The fact that openness to inflows and openness to outflows yield different results is particularly telling as the two components of the index are highly correlated (the correlation coefficient is 85 percent and a regression of one KI over KO yields a coefficient of 0.7, with a tstatistics of 70 and an R-squared of 0.7). If we include both components in a horserace regression, we still find that openness to inflows decreases the sensitivity of holdings of liquid financial assets to spreads but openness to outflows does not (column 4). In fact, the two bottom panels of Figure 4 suggest that the effect of openness to outflows goes in the opposite direction. 9 Note that the original Fernández et al. (2015) index gives higher values for countries with a closed capital account. We rescaled the index such that 1 means open capital account and 0 means closed capital account. 13 Panel C plots the coefficient for openness to inflows. We find the usual negative relationship of Panels A and B, but the curve is steeper, and the point at which the coefficient becomes insignificant is higher than in the regressions that use total openness. Panel D, instead, shows that openness to outflows is positively correlated with our measure of carry trade activity, but the coefficient is never statistically significant. To probe further, we regress the inflow and outflows measures on the overall index of capital controls and use the errors of this regression as measures of controls on inflows and outflows that are orthogonal to overall capital controls (again, we rescale these two measures to range between 0 and 1, with 1 indicating maximum openness). In column 5 of Table 5, we control for both overall capital account openness and for openness to inflows that is orthogonal to overall openness (KI_R), we find that what matters is openness to inflows. In column 6, we repeat the experiment but now we include openness to outflows (KO_R). We find that the effect goes in the opposite direction, indicating that openness to outflows actually amplifies carry trade activities (this finding is consistent with what we showed in panel D of Figure 4). Our results are consistent with the idea that firms are unlikely to engage in carry trade activities if they have doubts on their ability to repatriate profits. It may also mean that controls on outflows are more tightly enforced (or enforceable) than controls on inflows. Next, we substitute our spread variable with a measure of potential carry trade returns. We use the Bloomberg Carry Return Index (CTI). CTI is a proxy of the carry-to-risk ratio obtained by summing the returns from interest rate differentials and exchange rate movements. The CTI is often interpreted as an ex ante measure of the attractiveness of carry trade. Table 6 shows that high carry trade returns increase the correlation between foreign bond issuances and holdings of liquid financial assets in country-years with a closed capital account but have no effect on the elasticity of holdings of liquid financial assets in country-years with an open capital account (column 1). We also find that the result holds for openness to inflows (column 2), but does not hold for openness to outflows (column 3) and that the result is robust to running a horserace that includes openness to both inflows and outflows (column 4). 14 5. Alternative Explanations: Incomplete Capital Markets and the Retreat of Banks The presence of capital controls is just one of several potential explanations for non-financial firms to act as financial intermediaries. Non-financial corporations may also be playing this role because emerging countries have under-developed capital markets or because international banks have retreated. We posit that financial depth and creditor rights are reasonable proxies for the lack of complete financial markets and that banks suffering from either impaired balance sheets or increased regulation are likely to retreat more from countries with lower credit ratings on long-term foreign currency bonds.10 Hence we test for these alternative views by estimating equation (2) replacing capital account openness with a measure of creditor rights, a measure of financial depth, and two indicators for the retreat of global banks, country ratings (risk) and international bank claims. In column 1 of Table 7, we use the index of creditors’ rights compiled by the Doing Business report. We rescale the variable to range between 0 and 1 (1 meaning stronger creditor rights).11 We find that creditor rights do not affect the correlation between foreign bond issuance and holdings of liquid financial assets of non-financial corporations (column 1). Next, we use a standard measure of financial depth (credit to the private sector as a percent of GDP) as a proxy of financial development.12 We find that financial depth does not affect the correlation between foreign bond issuances and holdings of liquid financial assets (Column 2). Finally, in columns 3 and 4 we run two horserace regressions that include financial development (creditors’ rights in column 3 and financial depth in column 4) and capital account openness (we use controls on inflows, but the results are robust to using overall capital account openness). We find that the effect of capital account openness is robust to controlling for financial depth, while the 10 Powell and Martínez (2008) in particular argue that ratings are actually fairly easy to model as rating agencies give considerable information as to what factors drive their ratings and hence suggest that ratings may be considered a convenient summary of those macroeconomic fundamentals and judgements regarding political and other less quantifiable risks. Cavallo, Powell and Rigobón (2013), within an errors in measurement-type methodology, show that sovereign ratings do add value in the sense that market variables are found to respond on average to changes in ratings. These results indicate that ratings may indeed be considered on average as at least a useful summary of fundamentals that drive more market measures of country risk. 11 As Doing Business data for creditors’ rights start in 2005, we use 2005 values for the 2000-2004 period. The results are robust to dropping the 2000-2004 period. 12 As cash deposits of corporations that borrow abroad may have an impact on the provision of domestic credit, we set FD to be equal to credit to the private sector in the year 2000. 15 coefficients have similar magnitude but are not statistically significant when we control for creditors’ rights (column 3). In Table 8 we look at the role of sovereign risk using both Standard and Poor’s and Moody’s credit ratings (again, we rescale the index to range between 0 and 1). Column 1 of Table 8 uses S&P ratings and shows that credit ratings are not statistically significant in our model. Column 2 includes both credit ratings and capital controls and shows that our baseline results are robust to controlling for credit ratings interactions. Columns 3 and 4 repeat the experiment with Moody’s ratings and find identical results. We also check to see whether our results are driven more directly by changes in the behavior of global banks. Specifically, we augment our model with a variable (which we call BIS) which measures the change in a country’s liabilities vis-à-vis BIS reporting banks divided by GDP. We try several different measures of liabilities employing the BIS locational banking statistics, namely: i) total liabilities; ii) liabilities of non-banks; iii) total loans; and iv) loans of the non-bank sector. We find that none of the BIS variables are statistically significant in our regressions and that there are no qualitative changes in our baseline results (Table 9). 6. Further Robustness Checks We further check if our results are robust to a series of alternatives specifications and subsamples. First, we split our sample into three different regions: Latin America (six countries, 261 firms and 2,940 observations), Asia (five countries, 237 firms and 2,512 observations), and Europe (5 countries, 158 firms and 1,404 observations).13 Table 9 shows that our results are robust in all sub-regions, but that they are weaker in Europe. This might be due to the fact that in our regressions we use U.S. dollar borrowing rates, but for European emerging markets the relevant currency is likely to be the Euro.14 Moreover, three of the countries included in the European subsample (Czech Republic, Hungary, and Poland) are part of the European Union, and this may cause differences with other emerging markets. To check whether our results are driven by influential observations, we estimate our benchmark regression by dropping one country at a time. Table A4 reports the results for the 13 We exclude South Africa and Israel, which do not belong to any of the geographical regions of Table 7. 14 In Table 8, we use openness to inflows, but the results are robust to using overall openness. 16 FXB*SP*KI coefficient. It shows that the coefficient is always negative (ranging between - 0.01 and -0.03) and statistically significant. We also run a set of placebo regressions (we run 500 regressions that randomly allocate capital controls across country-periods) and find that the average placebo coefficient is centered at zero and that only 5 percent of the placebo regressions (4.7 percent to be precise) are statistically significant at the 5 percent confidence level. This is exactly what one would expect to find if the coefficient is not statistically significant. Next, we split the sample in two sub-periods, 2007-2014 and 2000-2006, and find that the results for inflows hold for both sub-periods, but the results for overall capital account openness only hold for 2000-06 (Table A5). One possible issue with the carry trade interpretation of our results is that, rather than engaging in carry trade activities, non-financial corporations hold the proceedings of bond issuances in liquid financial assets because it takes some time between the moment in which they borrow and the moment in which they need the funds to finance an investment project (of course, it is not obvious why this lag should depend on the spread or why this result should only hold for foreign bond issuances). To check if our results are driven by this possibility, we look at holdings of liquid financial assets one year after bond issuances. We start by showing that our results are robust to regressing holdings of liquid financial assets at time t on all controls at time t-1. The first two columns of Table A6 estimate the model of columns 1-2 of Table 5 but with lagged explanatory variables and find results which are essentially identical to those of Table 5. The last two columns of Table A6 measure all explanatory variables at time t with the exception of bond issuances, which are measured at time t-1. Again, the results are robust to this specification. It is possible that countries introduce capital controls exactly to limit the type of carry trade activities that we describe in this paper. If this were the case, our estimates would be upward biased. While the use of country-year fixed effects should allay most concerns of reverse causality, we also use two strategies to assess whether our results are robust to controlling for the endogeneity of capital controls. First, we restrict our sample to 10 years of data and estimate our model for the period 2004-14 by using the level of capital controls in 2003. If changes in capital controls were a reaction to the massive inflows that followed the global financial crisis, using their 2004 values 17 should address any endogeneity concern. The first two columns of Table 11 show that our results are robust to this specification. Second, we use the results of Table 5, which show that, while controls on inflows are highly correlated with controls on outflows, controls on outflows do not matter when our regressions include controls on inflows. Therefore, we instrument controls on inflows with control on outflows (specifically we instrument FXB*KI and FXB*KI*SP with FXB*KO and FXB*KO*SP). Column 3 of Table 11 shows that in the IV regression the triple interaction coefficient has the right sign but it is no longer statistically significant. This may be due to the loss of efficiency of the IV estimator. In fact, if we limit our sample to Asia and Latin America (the regions for which our results are stronger, see Table 10), the results are also statistically significant in the IV regression (Column 4 Table 11).15 7. Conclusions This paper adds to the growing literature considering the increase in issuance of non-financial firms, particularly in dollars and especially from emerging economies. There are several potential implications of this phenomenon which is seen as central to the characteristics of the second phase of global liquidity. However, in order to draw the right conclusions and to assess potential risks it is important to understand firms’ motives and heterogeneous behaviors. We first corroborate two results already found in the literature, namely i) that firms are issuing and some are clearly not using the proceeds solely for real investment but also to maintain liquid financial assets and ii) that they appear to be doing so when the conditions for pursuing carry trade activities are more attractive. This suggests, as others before us have also indicated, that these firms are behaving like financial intermediaries. However, non-financial firms may be behaving like financial intermediaries for various motives. A benign view would be that they are attempting to correct market failures and hence serving a role in trying to complete incomplete financial markets. However, when we attempt to test such a view considering how such behavior varies with financial depth, or by creditor rights, we do not find any statistically significant results. Alternatively, non-financial firms may be taking the place of global banks that have been retreating due to impaired balance sheets or 15 In the IV regressions we end the sample in 2013 because the capital controls data for 2014 are imputed to be equal to the 2013 values. 18 increased regulatory pressure. If this were the case, then we would expect some relation to risk and hence we attempt to find a pattern using sovereign credit ratings, which are commonly used by banks and by bank regulators to proxy risk, but again we find no statistically significant results. Under this view, we would also expect to find a relation with the change in cross-border banking liabilities but again we find no significant relationship. A third view is that as a response to low global interest rates and quantitative easing in advanced economies, several emerging economies have imposed or tightened capital controls and that non-financial firms have mechanisms that are not available to banks to evade such controls. We argued that controls on inflows would be particularly relevant in this case. We find strong statistical evidence in favor of the view that non-financial firms are acting like financial intermediaries in countries with relatively high capital controls, particularly where there are controls on capital inflows. We leave an in-depth discussion of the policy implications of these results for future work (see Acharya et al., 2015, and IMF 2015). Suffice to summarize here however that while our results do not back the view that non-financial firms are attempting to complete incomplete markets nor take on a role left by global banks retreating, they do suggest that firms are attempting to gain from carry trade type activities where capital controls, particularly controls on inflows are prevalent. In turn, this suggests that any evaluation of the efficacy of capital controls should take into account the possibility that they may be evaded through such means. Indeed, to the extent that non-financial firms may issue abroad and are able to deposit the funds in the local financial system, evading any capital controls in place, then arguably macro-prudential policies applied on local financial systems may be more effective tools to limit capital inflows and reduce the risks of credit booms. In countries where non-financial firms are behaving in this fashion it may thus appear that corporates are highly liquid and hence may be expected to have the resources to repay those external liabilities. However, if the profitability of the carry trade diminishes and firms decide to withdraw their liquidity en masse then this might provoke problems for the domestic financial system. Moreover, the information on whether firms have hedged currency risks remains very partial. Finally, foreign currency bond issuance in countries with open capital accounts has also been strong, but in these cases our results indicate firms have been not been building financial reserves and hence that issuance has been more for real investment. In those countries the risks may then be different, such firms may not be as liquid 19 and risks may relate more to the underlying profitability of the investments. As commodity prices have plummeted and currencies depreciated, the risks created by such swings in relative prices may bring large profits or losses; such risks should also then be carefully monitored. 20 References Acharya, V. et al. 2015. “Corporate Debt in Emerging Economies: A Threat to Financial Stability?” Washington, DC, United States: Brookings Institution: Committee for International Policy Reform. Available at: http://www.brookings.edu/research/papers/2015/09/corporate-debt-emerging-economies Afonso, A., P. Gomes and P. Rother. 2007. “What ‘Hides’ behind Sovereign Debt Ratings?” Working Paper 711. Frankfurt, Germany: European Central Bank. Bates, T.W., K.M. Kahle and R.M. Stulz. 2009. “Why Do US Firms Hold So Much More Cash Than They Used To?” Journal of Finance 64(5): 1985-2021. Bruno, V., and H.S. Shin. 2015. “Global Dollar Credit and Carry Trades: A Firm-Level Analysis.” Washington, DC, United States: American University. Mimeographed document. Caballero, R., and A. Krishnamurthy. 2003. “Excessive Dollar Debt: Financial Development and Underinsurance.” Journal of Finance 58: 967–893. Cavallo, E., A. Powell and R. Rigobón. 2013. “Do Credit Rating Agencies Add Value? Evidence From The Sovereign Rating Business.” International Journal of Finance & Economics 18(3): 240-265. Chinn, M.D., and H. Ito. 2006. “What Matters for Financial Development? Capital Controls, Institutions, and Interactions.” Journal of Development Economics 81(1): 163-192. Chung, K. et al. 2014. “Global Liquidity through the Lens of Monetary Aggregates.” IMF Working Paper 14/9. Washington, DC, United States: International Monetary Fund. Claessens, S., and N. van Horen. 2012. “Foreign Banks: Trends, Impact and Financial Stability.” International Monetary Fund Working Paper 12/10. Washington, DC, United States: International Monetary Fund. De Haas, R., and N. van Horen. 2011. “Running for the Exit: International Banks and Crisis Transmission.” DNB Working Paper 279. Amsterdam, The Netherlands: De Nederlandsche Bank. Demirgüç-Kunt, A., and V. Maksimovic. 2001. “Firms As Financial Intermediaries: Evidence from Trade Credit Data.” Washington, DC, United States: World Bank. Mimeographed document. 21 Dittmar, A., J. Mahrt-Smith and H. Servaes. 2003. “International Corporate Governance and Corporate Cash Holdings. Journal of Financial and Quantitative Analysis 38(1): 111- 134. Fernández, A. et al. 2015. “Capital Controls Measures: A New Dataset.” NBER Working Paper 20970. Cambridge, United States: National Bureau of Economic Research. Fisman, R.J., and I. Love. 2003. “Trade Credit, Financial Intermediary Development, and Industry Growth.” Journal of Finance 58: 353–374. Galindo, A., A. Micco and A. Powell. 2005. “Loyal Lenders or Fickle Financiers: Foreign Banks in Latin America.” Research Department Working Paper 529. Washington, DC, United States: Inter-American Development Bank. García-Luna, P., and A. Van Rixtel. 2014. “International Interbank Activity in Retreat.” BIS Quarterly Review March 2014: 14-15. Available at: http://www.bis.org/publ/qtrpdf/r_qt1403x.htm Goldberg, L. 2002. “When Is U.S. Bank Lending to Emerging Markets Volatile?” In: S. Edwards and J.A. Frankel, editors. Preventing Currency Crises in Emerging Markets Chicago, United States: University of Chicago Press. González-Miranda, M. 2012. “Nonfinancial Firms in Latin America: A Source of Volatility?” IMF Working Paper 12/279. Washington DC, United States: International Monetary Fund. International Monetary Fund (IMF). 2015. “Corporate Leverage in Emerging Markets: A Concern?” In: Global Financial Stability Report. Washington, DC, United States: IMF. Karam, P. et al. 2014. “The Transmission of Liquidity Shocks: The Role of Internal Capital Markets and Bank Funding Strategies.” IMF Working Paper 14/207. Washington DC, United States: International Monetary Fund. Levine, R. 2005 “Finance and Growth: Theory and Evidence.” In: P. Aghion and S. Durlauf, editors. Handbook of Economic Growth. Volume 1A. Amsterdam, The Netherlands: Elsevier. Martínez-Peria, S., A. Powell and I. Vladkova. 2005. “Banking on Foreigners: The Behavior of International Bank Lending to Latin America, 1985-2000.” IMF Staff Papers 52(3): 430- 461. 28 Table 6. Carry Trade Index Instead of Spread This table reports a set of firm-level regressions in which the dependent variable is the log of the ratio between holdings of liquid financial assets and sales, and the explanatory variables are foreign currency bond issuances (FXB, defined as ln(1+ issuances/sales)), the demeaned value of the Bloomberg index of carry trade return (CTI), the Fernández et al. index of capital account openness (K), the Fernández et al. index of capital account openness to inflows (KI), the Fernández et al. index of capital account openness to outflows (KO). All regressions control for the log of total debt over sales, the log of total sales, leverage, firm fixed effects and country-year fixed effects. (1) (2) (3) (4) FXB -0.250 0.548 0.173 0.617 (0.613) (0.548) (0.373) (0.516) FXB*CTI 1.173** 1.103** -0.0818 0.978** (0.589) (0.457) (0.308) (0.481) FXB*K 1.749 (1.092) FXB*CTI*K -3.402*** (1.290) FXB*KI 0.535 -0.641 (1.124) (1.230) FXB*CTI*KI -2.756** -2.673*** (1.083) (0.852) FXB*KO 1.941*** 1.719* (0.720) (0.946) FXB*CTI*KO -0.290 0.576 (1.139) (1.773) Observations 5,587 5,587 5,587 5,587 Number of firms 523 523 523 523 Firm Fixed effects Yes Yes Yes Yes Country year Fixed effects Yes Yes Yes Yes FXB is: 𝑙𝑙�1 + 𝐹𝐹𝐹 𝑆𝑆𝑙𝑆𝑆� Robust standard errors clustered at the firm level in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 29 Table 7. The Role of Creditors’ Rights and Financial Depth This table reports a set of firm-level regressions in which the dependent variable is the log of the ratio between holdings of liquid financial assets and sales, and the explanatory variables are foreign currency bond issuances (FXB, defined as ln(1+issuances/sales)), the demeaned spread between local deposit rate and borrowing costs in the United States (SP), The Doing Business index of creditor rights (CR, the index rescaled to range between 0 and 1, with 1 indicating stronger creditors’ rights), a measure of financial depth (FD is credit to the private sector over GDP in the year 2000), the Fernández et al. index of capital account openness to inflows (KI). All regressions control for the log of total debt over sales, the log of total sales, leverage, firm fixed effects and country-year fixed effects. (1) (2) (3) (4) FXB 0.230 0.622 0.420 0.646 (2.034) (0.824) (2.046) (0.932) FXB*SP 0.409 0.123 0.834 0.668** (0.543) (0.282) (0.569) (0.310) FXB*CR 0.170 0.049 (0.432) (0.528) FXB*CR*SP -0.040 -0.027 (0.129) (0.130) FXB*FD -0.089 0.308 (0.679) (0.710) FXB*FD*SP 0.008 0.039 (0.216) (0.225) FXB*KI 0.276 1.005 (2.411) (1.125) FXB*KI*SP -1.034 -0.786*** (0.608) (0.202) Observations 5,831 6,621 5,831 6,621 Number of firms 648 622 648 622 Firm Fixed effects Yes Yes Yes Yes Country year Fixed effects Yes Yes Yes Yes FXB is 𝑙𝑙�1 + 𝐹𝐹𝐹 𝑆𝑆𝑙𝑆𝑆� Robust standard errors clustered at the firm level in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 30 Table 8. Sovereign Risk This table reports a set of firm-level regressions in which the dependent variable is the log of the ratio between holdings of liquid financial assets and sales, and the explanatory variables are foreign currency bond issuances (FXB, defined as ln(1+ issuances/sales)), the demeaned spread between local deposit rate and borrowing costs in the United States (SP), numerical credit rating (RATING, the index is rescaled to range between 0 and 1, with 1 indicating AAA), the Fernández et al. index of capital account openness to inflows (KI). All regressions control for the log of total debt over sales, the log of total sales, leverage, firm fixed effects and country-year fixed effects. (1) (2) (3) (4) FXB -0.478 -0.199 -0.056 0.522 (1.120) (1.289) (1.145) (1.457) FXB*SP 0.371 0.901*** 0.373* 1.067*** (0.246) (0.309) (0.207) (0.335) FXB*RATING 2.438 1.378 1.726 0.437 (2.063) (2.239) (2.032) (2.397) FXB*RATING*SP -0.564 -0.743 -0.590 -0.964* (0.560) (0.534) (0.484) (0.515) FXB*KI 0.576 0.247 (0.878) (0.906) FXB*KI*SP -0.705*** -0.797*** (0.194) (0.241) Observations 7,622 7,622 7,310 7,310 Number of id 716 716 716 716 Firm Fixed effects Yes Yes Yes Yes Country year Fixed effects Yes Yes Yes Yes FXB is 𝑙𝑙�1 + 𝐹𝐹𝐹 𝑆𝑆𝑙𝑆𝑆� RATING is S&P Moody’s Robust standard errors clustered at the firm level in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 31 Table 9. Country Liabilities to BIS Reporting Banks This table reports a set of firm-level regressions in which the dependent variable is the log of the ratio between holdings of liquid financial assets and sales, and the explanatory variables are foreign currency bond issuances (FXB, defined as ln(1+ issuances/sales)), the demeaned spread between local deposit rate and borrowing costs in the United States (SP), the Fernández et al. (2015) index of capital account openness to inflows (KI), and the change in the ratio of liabilities versus BIS reporting banks and GDP (BIS). We use four measure for BIS: total liabilities versus BIS reporting banks (column 1); total liabilities of the non-bank sector versus BIS reporting banks (column 2); total loans with BIS reporting banks (column 3); total loans with BIS reporting bank of the non-bank sector (column 4). All regressions control for the log of total debt over sales, the log of total sales, leverage, firm fixed effects and country-year fixed effects. (1) (2) (3) (4) FXB 0.708 0.954* 0.609 0.515 (0.541) (0.554) (0.667) (0.581) FXB*SP 0.535*** 0.533*** 0.593*** 0.569*** (0.138) (0.142) (0.145) (0.139) FXB*BIS 28.03** 76.91* -0.169 59.32 (12.73) (40.66) (21.36) (39.01) FXB*SP*BIS -2.288 4.516 -0.958 -1.630 (2.882) (11.74) (4.866) (8.659) FXB*KI 0.490 0.408 0.362 0.935 (1.036) (1.004) (1.355) (1.135) FXB*SP*KI -0.638*** -0.614*** -0.676*** -0.700*** (0.198) (0.204) (0.245) (0.215) Observations 6,284 6,284 6,284 6,284 Number of firms 650 650 650 650 Firm FE Yes Yes Yes Yes Ctry-year FE Yes Yes Yes Yes FXB is BIS is All Liabilities All liabilities of non-banks All Loans All loans of nonbanks Robust standard errors clustered at the firm level in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 32 Table 10. Different Regions This table reports a set of firm-level regressions in which the dependent variable is the log of the ratio between holdings of liquid financial assets and sales, and the explanatory variables are foreign currency bond issuances (FXB, defined as ln(1+bond issuances/sales)), the demeaned spread between local deposit rate and borrowing costs in the United States (SP), the Fernández et al. index of capital account to inflows (KI). All regressions control for the log of total debt over sales, the log of total sales, leverage, firm fixed effects and country-year fixed effects. Column 1 focuses on Latin America, Column 2 on Asia and Column 3 on Emerging Europe. (1) (2) (3) FXB -0.061** 0.010*** 0.044 (0.027) (0.028) (0.068) FXB*SP 0.023*** 0.0120** 0.021 (0.004) (0.008) (0.015) FXB*KI 0.138*** -0.133** -0.019 (0.047) (0.064) (0.113) FXB*SP*KI -0.031*** -0.026* -0.056** (0.007) (0.016) (0.028) Observations 2,940 2,512 1,726 Number of firms 261 237 174 Firm Fixed effects Yes Yes Yes Country year Fixed effects Yes Yes Yes FXB is 𝑙𝑙�1 + 𝐹𝐹𝐹 𝑆𝑆𝑙𝑆𝑆� Region LAC ASIA EUROPE Robust standard errors clustered at the firm level in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 33 Table 11. Endogeneity This table reports a set of firm-level regressions in which the dependent variable is the log of the ratio between holdings of liquid financial assets and sales, and the explanatory variables are foreign currency bond issuances (FXB, defined as ln(1+ issuances/sales)), the demeaned spread between local deposit rate and borrowing costs in the United States (SP), the Fernández et al. index of capital account openness (K) in 2003 (column 1), the Fernández et al. index of capital account openness to inflows (K) in 2003 (column 2), the time-varying Fernández et al. index of capital account openness to inflows (K) instrumented with openness to outflows. All regressions control for the log of total debt over sales, the log of total sales, leverage, firm fixed effects and country-year fixed effects. (1) (2) (3) (4) FXB -0.099 0.046 0.556 0.469 (0.583) (0.616) (0.854) (0.800) FXB*SP 0.420** 0.497** 0.261* 0.384** (0.198) (0.197) (0.137) (0.196) FXB*K 0.342 0.206 -1.239 -0.835 (1.218) (1.093) (1.860) (1.836) FXB*K*SPR -0.732** -0.693*** -0.471 -0.670** (0.342) (0.238) (0.360) (0.320) Observations 6,685 6,685 6,851 4,531 Number of firms 750 750 735 474 Firm FE Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Estimation OLS OLS IV IV Capital account openness Capital account openness in 2003 Openness to inflows in 2003 Time-varying openness to inflows instrumented with time-varying openness to outflows Estimation period 2003-13 2003-13 2000-13 2000-13 Sample All Countries All Countries All Countries Asia and Latin America Robust standard errors clustered at the firm level in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 34 Figure 1. Domestic and Foreign Currency Bond Issuances 050 100 150 Billion USD 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Loc. Curr. For. Curr. 35 Figure 2. Evolution of Capital Account Openness This figure plots the evolution of different indexes of capital account openness for the sample of countries included in the regressions of this paper. In all graphs the solid line plots the median value of the index and the dashed lines plot the top and bottom 20th percentile of the index. Panel A uses the Chinn and Ito Index, Panel B the aggregate index of Fernández et al., Panel C the Fernández et al. index of openness to inflows, and Panel D the Fernández et al. index of openness to outflows. .2 .4 .6 .8 1 2000 2002 2004 2006 2008 2010 2012 2014 Year A. Chinn & Ito Index .2 .4 .6 .8 1 2000 2002 2004 2006 2008 2010 2012 2014 Year B. Fernandez et al. Index .2 .4 .6 .8 1 2000 2002 2004 2006 2008 2010 2012 2014 Year C. Inflows Index 0.2 .4 .6 .8 1 2000 2002 2004 2006 2008 2010 2012 2014 Year D. Outflows Index Index of Cap. Acc. Openn. 36 Figure 3. Distribution of Capital Account Openness across Countries in 2007 and in 2013 This figure plots the evolution of different indexes of capital account openness for the sample of countries included in the regressions of this paper. The box plots the interquartile range and the median and the whiskers the upper and lower adjacent values. Panel A uses the Chinn and Ito Index, Panel B the aggregate index of Fernández et al., Panel C the Fernández et al. index of openness to inflows, and Panel D the Fernández et al. index of openness to outflows. 0 .2 .4 .6 .8 1 Chin & Ito Index 0 .2 .4 .6 .8 1 Fernandez et al. Index 0 .2 .4 .6 .8 1 Inflows Index 0 .2 .4 .6 .8 1 Outflows Index 2007 2013 37 Figure 4. Marginal Effects This figure plots how the sensitivity of the relationship between foreign bond issuances and holding of liquid financial assets to our spread variable varies with capital account openness. The solid line plots the main effect and the dashed lines are 95 percent confidence intervals. Panel A is uses the model of column 2, Table 4; Panel B uses the model of column 1, Table 5; and panels C and D use the model of column 4, Table 5. -.5 0.5 1 d(dCash/dBond)/dSP 0.1 .2 .3 .4 .5 .6 .7 .8 .9 1 Cap. Acc. Openn. A. Chinn and Ito -.1 -.05 0.05 .1 .15 d(dCash/dBond)/dSP 0.1 .2 .3 .4 .5 .6 .7 .8 .9 1 Cap. Acc. Openn. B. Fernandez et al. -.2 -.1 0.1 .2 .3 d(dCash/dBond)/dSP 0.1 .2 .3 .4 .5 .6 .7 .8 .9 1 Cap. Acc. Openn. C. Inflows -.2 -.1 0.1 .2 .3 d(dCash/dBond)/dSP 0.1 .2 .3 .4 .5 .6 .7 .8 .9 1 Cap. Acc. Openn. D. Outflows