Life Below Zero: Negative Policy Rates and Bank Risk Taking

Transcription

1 Life Below Zero: Negative Policy Rates and Bank Risk Taking Florian Heider ECB & CEPR Farzad Saidi University of Cambridge June 2, 2016 Preliminary and incomplete Glenn Schepens ECB Abstract This paper investigates the impact of negative policy rates on banks risk taking in the syndicated-loans market. Existing work on the bank risk-taking channel focuses on lower but positive interest rates and, thus, offers little guidance for environments in which short-term rates can become negative. For identification, we exploit the inability of banks to pass on negative rates to depositors, which creates differences in the pass-through of negative policy rates across banks with differential reliance on deposit funding. Combining bank and firm balance-sheet data with transaction-level loan data allows us to assess the riskiness of firms that receive new loans from banks with different deposit ratios before and after policy rates become negative. For a tenpercentage-point increase in the lender s deposit ratio, a reduction of the policy rate governing central-bank deposits from 0 to -0.2% in our sample period leads to an increase of at least 12% in the standard deviation of the return on assets of borrower firms. The treatment effect increases when borrowers are isolated from a change in the policy rate because they are located in a different currency zone. Finally, a placebo at the time when policy rates fall but are still positive shows no effect. JEL classification: E44, E52, E58, G20, G21 Keywords: monetary policy, zero lower bound, negative interest rates, bank risk taking We thank Luc Laeven, Anthony Saunders, and Sascha Steffen for their comments and suggestions. We also thank Valentin Klotzbücher and Francesca Barbiero for excellent research assistance. The views expressed do not necessarily reflect those of the European Central Bank or the Eurosystem. European Central Bank, Financial Research Division, Sonnemannstr. 22, Frankfurt am Main, Germany. florian.heider@ecb.int University of Cambridge, Judge Business School, Trumpington Street, Cambridge CB2 1AG, United Kingdom. f.saidi@jbs.cam.ac.uk European Central Bank, Financial Research Division, Sonnemannstr. 22, Frankfurt am Main, Germany. glenn.schepens@ecb.int

2 1 Introduction The global financial crisis, and further ramifications such as the European sovereign debt crisis, have thrown a negative light on the role and effectiveness of monetary policy. In the pre-crisis period, accommodative monetary policy is thought to have contributed to the emergence of the financial crisis (see, among others, Borio and Zhu (2008), Adrian and Shin (2009), Taylor (2009)). In the post-crisis period, accommodative monetary policy allegedly had only a limited impact on the real economy while, at the same time, undermining the stability of the financial system and, possibly, sowing the seeds for the next financial crisis. Adding complexity to the negative light, monetary policy in the post-crisis period occurs mostly in unchartered territory. Facing ultra-low interest rates, central banks around the world and perhaps most notably the U.S. Federal Reserve Board turned to non-standard measures in the form of large-scale asset purchases (Chakraborty, Goldstein, and MacKinlay (2016); Kandrac and Schlusche (2016)). Even less explored is standard interest-rate setting beyond the zero lower bound. The European Central Bank (ECB), but also the central banks of Denmark, Switzerland, and Sweden, have recently implemented negative policy rates. 1 This paper studies financial stability, and more precisely the bank risk-taking channel, in an environment with negative policy rates. Ever since the 2008/9 financial crisis, there has been a growing interest in the impact of prolonged periods of low interest rates on bank risk taking. Existing theoretical models reveal that there are two opposing forces that steer the relation between interest rates and bank risk taking (Dell Ariccia, Laeven, and Marquez (2014)). On the one hand, a portfolio-reallocation channel causes a shift towards more risky assets when interest rates decrease. One way to see this is that expected profits for banks decrease when interest rates fall, which reduces their monitoring effort and leads to higher risk taking on the asset side. 2 On the other hand, the classic risk-shifting channel implies that lower 1 Before the introduction of negative policy rates in Europe, Saunders (2000) laid out potential implications for bank behavior by considering the case of Japan in the late 1990s. 2 This effect is similar to a search-for-yield effect, which predicts that when rates on safe assets decline, banks will increase their demand for risky assets. 1

3 rates lead to a decrease in funding costs, which increases a banks net worth and, thus, provides greater incentives for prudent behavior. Recent empirical papers document that low rates tend to lead to more risky bank behavior, therefore indicating that the portfolio-reallocation channel tends to outweigh the risk-shifting channel (Jiménez, Ongena, Peydró, and Saurina (2014); Ioannidou, Ongena, and Peydró (2015); Dell Ariccia, Laeven, and Suarez (2016)). Yet, a great degree of uncertainty remains about the exact role of leverage and the relative strength of both channels. 3 Far less is known, however, about how negative rates impact bank risk-taking behavior. The fact that a number of major central banks in Europe have set key policy rates below zero has made this question the center of recent debates among both practitioners and policy makers (see, e.g., Bech and Malkhozov (2016)). At the same time, existing theoretical and empirical models remain silent on the extent to which the effect of negative rates is distinct compared to lower but positive interest rates. To the best of our knowledge, this paper is the first to identify the impact of negative policy rates on bank risk-taking behavior. We document that in contrast to rate reductions in normal times, the liability structure of a bank, and more precisely the amount of deposit funding, plays a decisive role in determining whether lower rates lead to more risk taking by banks. The main reason for this is that the liability side of the risk-shifting channel is shut down for banks that mainly fund themselves through customer deposits, as there is effectively a zero lower bound for the rates on these deposits. Thus, banks funding models that are typically seen as more stable in normal times, i.e. the ones with a strong deposit base, could be more vulnerable during times of negative policy rates. Our main results illustrate that banks with high deposit ratios lend to riskier borrowers when policy rates become negative compared to banks with low deposit ratios. We employ transaction-level data on syndicated loans granted by Eurozone banks. In doing so, we 3 While Dell Ariccia, Laeven, and Suarez (2016) show that the effect of interest rates on risk taking is less pronounced for poorly capitalized banks, which is consistent with the risk-shifting channel, Jiménez, Ongena, Peydró, and Saurina (2014) find that banks with low capitalization commit larger loan volumes to ex-ante risky firms. 2

4 characterize bank risk taking based on the volatility of both publicly listed and private firms profits as well as public firms stock returns, which captures also the risk realized through loan financing in the real economy. Using a difference-in-differences setup around the period when policy rates became negative in the Eurozone (June 2014), our baseline results indicate that a one-standard-deviation increase in a bank s deposit ratio translates into a 12-percent increase in ex-ante ROA volatility of the firms associated with new loans when policy rates turn negative. Importantly, we do not find similar results during periods of decreasing, but still positive, interest rates, confirming the distinct nature of the risk-taking channel in times of negative rates. Another contribution of this paper lies in the precise identification of the mechanism through which bank risk taking operates when policy rates are decreasing. The fact that the risk-shifting channel is shut down for some banks in our sample allows us to cleanly identify the importance of the portfolio-reallocation channel for bank risk taking, without having to rely on the heterogeneity in bank equity ratios to disentangle both channels, which leads to contradictory results in previous research (Jiménez, Ongena, Peydró, and Saurina (2014); Dell Ariccia, Laeven, and Suarez (2016)). In general, our results confirm the existence of a portfolio-reallocation channel. Finally, our empirical setup and the data that we use allow us to control better than most existing studies for the concern that monetary policy is endogeneous to bank risk taking. Most importantly, we analyze a subsample of borrowers that are not in the Eurozone area. While papers that focus on monetary policy and risk taking in one country potentially suffer from the possibility that economic conditions drive both the riskiness of borrowers and monetary-policy decisions, it is unlikely that the economic conditions faced by borrowers outside the Eurozone area govern monetary-policy decisions in the Eurozone area. The remainder of the paper is organized as follows. In Section 2, we discuss our identification strategy, and describe the data. Our results are in Section 3, and Section 4 concludes. 3

5 2 Empirical Strategy and Data In this section, we start by providing background information on the introduction of negative policy rates, on the basis of which we develop our hypotheses. We then lay out our identification strategy for estimating the effect of negative policy rates on bank risk taking. Finally, we describe the empirical implementation and the data. 2.1 Background and Hypothesis Development On June 5, 2014, the European Central Bank (ECB) Governing Council lowered the Main Refinancing Operation (MRO) rate to 0.15% and the Deposit Facility (DF) rate to -0.10%. Shortly after, on September 4, 2014, the rates were lowered again: the MRO rate to 0.05% and the DF rate to -0.20%. With these actions, the ECB ventured into negative territory for some policy rates for the first time in its history. Ever since, the DF rate has continued to drop, to -0.40% on March 10, The main goal of lowering the rates was to provide monetary policy accommodation (in accordance with the ECB s forward guidance). In order to preserve the difference between the cost of borrowing from the ECB (at the MRO rate) and the benefit of depositing with the ECB (at the DF rate), which preserves banks incentive to lend in the interbank market, the DF rate became negative. Because banks hold significant amounts of excess liquidity and short-term market rates closely track the DF rate, these rates became negative, even though the MRO rate remained positive. Within Europe, Eurozone banks are not the only ones exposed to negative policy rates. Almost simultaneously, the Swedish Riksbank reduced its deposit rate from 0% to -0.50% on July 9, 2014, and since then gradually reduced it further to -1.10%. The Swedish experience is furthermore preceded by the Danish central bank, Nationalbanken, lowering the deposit rate to -0.20% on July 5, While the Danish deposit rate was raised in 2013, it was brought back to negative territory, -0.05%, on September 5, Most recently, the Swiss National Bank went negative on December 18, 2014, by imposing a negative interest rate of 4

6 -0.25% on sight deposits exceeding a given exemption threshold. We next discuss the relationship between lower short-term interest rates and bank risk taking. In particular, we focus on the extent to which rate decreases when rates are negative, rather than positive, enables the empirical identification of a risk-taking channel and how the inner workings of the risk taking channel change in an environment with negative policy rates. The literature identifies two channels through which a decrease in central-bank policy rates impacts bank behavior: the bank lending channel and the bank risk-taking channel. The two channels differ not just in outcomes but also in terms of their predictions under negative policy rates. The bank lending channel focuses on the volume of lending after an interest-rate change (Bernanke and Gertler (1989)). According to the bank lending channel, a decrease in the short-term interest rate lowers the funding cost of a financially constrained bank. Lower rates increase a bank s net worth and allow for raising more external financing, which leads to more lending. While the bank lending channel has empirical support (see, for example, Bernanke and Blinder (1992), Jayaratne and Morgan (2000), Stein and Kashyap (2000), Kishan and Opiela (2000), as well as Jiménez, Ongena, Peydró, and Saurina (2012)), its implications for bank risk taking are ambiguous. This is because more lending is not necessarily synonymous with greater risk taking, unless the marginal new loan is actually riskier. The lack of predictions for risk in the bank lending channel, combined with extant evidence of risky behavior by various financial institutions during periods of low interest rates (Maddaloni and Peydró (2011); Jiménez, Ongena, Peydró, and Saurina (2014); Ioannidou, Ongena, and Peydró (2015); Dell Ariccia, Laeven, and Suarez (2016); and Kacperczyk and Di Maggio (2016)), led to the development of a bank risk-taking channel. We base our discussion of the bank risk-taking channel on Dell Ariccia, Laeven, and Marquez (2014) and Dell Ariccia, Laeven, and Suarez (2016). Crucially, a decrease in shortterm interest rates affects both the assets and liabilities of banks. This, in turn, induces 5

7 them to adjust their behavior, potentially along three dimensions. The first potential adjustment is a portfolio rebalancing on the asset side. A lower shortterm interest rate makes safe, short-term assets less attractive relative to riskier, long-term assets. Holding the liability side constant, this will lead to a shift away from safe, short-term assets and towards risky, long-term assets. The extent of portfolio rebalancing depends on the relative rate of return on safe and risky assets, and does not depend on whether the level of interest rates is positive or negative. The second adjustment could stem from a larger franchise value. Holding bank leverage constant, a lower interest rate reduces the cost of debt/deposits, and increases the net worth of banks. 4 A higher net worth should reduce risk-taking incentives, as banks desire to maintain their franchise value. The third potential adjustment is a change of bank leverage. An increase in net worth after a decrease in the interest rate allows financially constrained banks to raise more external financing. As the marginal source of funds for banks is debt/deposits, this increases leverage and risk-taking incentives, and potentially exacerbates debt overhang. These three margins for adjustment in bank-level behavior have opposite implications for risk taking. This renders the overall effect of a decrease in the short-term interest rate on bank risk taking is ambiguous, putting the bank risk-taking channel on equal footing with the above-discussed bank lending channel. We argue, and later also provide evidence, that when short-term rates become negative, this ambiguity is resolved for the bank risk-taking channel, if there exists a zero lower bound for bank deposit rates. A zero lower bound on deposits undermines the second and the third adjustment effect and only the first margin for adjustment portfolio rebalancing remains. Thus, banks incentives to invest in riskier assets increase when policy rates become negative. Importantly, this also implies that the risk-taking channel works differently during periods of negative interest rates, as the amount of deposit funding becomes a crucial factor 4 For evidence that a decrease in the policy rate increases bank profitability, see English, Van den Heuvel, and Zakrajšek (2014). 6

8 in determining the strength of the impact of a reduction in policy rates on bank risk taking. The above argument strongly relies on the existence of a zero lower bound for deposit rates, together with a pass-through of lower policy rates to loan rates. First, Figure 1 documents the lower bound for overnight household and non-financial-corporation (NFC) deposits, which are the two most important customer-deposit categories for Eurozone banks. 5 From mid-2014 onwards, rates on these deposits were very stable and consistently above the declining market rate (proxied by the three-month Euribor). As such, the gap between the cost of deposit funding and the cost of alternative funding sources steadily increased. Figure 2 depicts median rates on deposits with an agreed maturity below one year, which is another important deposit-funding source for Eurozone banks (about 20 percent of total household and NFC deposits). These rates also flatten out, the sole difference compared to the overnight deposits is that the disconnect between the market rate and the deposit rate is visible only from mid-2015 onwards. Second, Figure 3 documents the pass-through of policy rates to loan rates. The figure plots the total cost of credit for syndicated loans originated by Eurozone banks during our sample periods. In accordance with the decreasing policy rates in the Eurozone, the cost of credit decreases for Eurozone borrowers (with direct exposure to ECB policy rates) as well as for non-eurozone borrowers (without direct exposure). Overall, these figures illustrate that there is effectively a zero lower bound for deposit rates, while lower policy rates lead to lower loan rates. Additionally, the summary statistics in Table 1 show that there are large differences in the funding structure between banks, with deposit ratios ranging between 0.5 and 68 percent. These facts allow us to exploit the heterogeneity in banks funding structure to identify the impact of negative policy rates on bank risk taking. 6 We summarize our argument in the following testable hypothesis: 5 For the average Eurozone bank, overnight deposits make up 55 to 60 percent of total customer (households and NFCs) deposits during our sample period. A more detailed breakdown of the deposit composition of Eurozone banks is shown in Figure B.1 in the Online Appendix. All data on deposit rates and on the composition of deposits are from the IBSI and IMIR database at the ECB. 6 A potential threat to our identification occurs when banks with different deposit ratios can adjust their cost of funding differentially by levying fees on depositors. We assume the capacity to adjust the cost of funding via fees does not depend on a bank s deposit-to-assets ratio. 7

9 Hypothesis: Owing to a lower bound on bank deposit rates, negative policy rates lead to greater bank risk taking. In particular, banks willingness to lend to riskier firms increases in their ratio of deposits over total assets. The presence of a zero lower bound for bank deposit rates also allows us to examine the working of the bank lending channel when policy become negative. Namely, funding costs of high-deposit banks decrease less than those of low-deposit banks. This, in turn, reduces the net worth of high-deposit banks relatively more when their deposit rates remain fixed at the zero lower bound. We therefore expect the volume of lending of high-deposit banks to decrease, relative to that of low-deposit banks. Consistent with this intuition, Figure B.2 shows that indeed, this is the case. 2.2 Identification Strategy The setting at hand lends itself to a difference-in-differences strategy, which we implement by comparing risk taking by Eurozone banks with different deposit ratios around the ECB s introduction of negative policy rates in June We characterize bank risk taking by means of the ex-ante firm-level volatility of borrower firms, thereby capturing the amount of risk realized in the real economy. In this manner, we capture the observable riskiness of firms that were granted loans by differentially treated banks. To test the impact of negative policy rates on the level of risk of loan-financed firms, we estimate the following difference-in-differences specification at the level of loans granted to firm i by Eurozone lead arrangers j at date t: y ijt = β 1 Deposit ratio j After(06/2014) t + β 2 X it + δ t + η j + ɛ ijt, (1) where y ijt is an outcome variable reflecting firm-level risk, Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j in 2013, After(06/2014) t is a dummy variable for the period from June 2014 onwards, X it denotes firm-level control variables, namely industry(-year) and country(-year) fixed effects, and δ t 8

10 and η j denote month-year and bank fixed effects, respectively, where bank fixed effects are included for all Eurozone lead arrangers. Standard errors are clustered at the bank level, using a vector of all banks j that acted as lead arrangers to firm i for a given loan. We hypothesize the difference-in-differences estimate, β 1, to be positive, indicating that banks with higher deposit ratios financed riskier firms following the introduction of negative policy rates. For identification, we use a relatively short window around the June-2014 event, from January 2013 to October In this manner, we ensure that our differencein-differences estimate, at the time-varying bank level jt, is not contaminated by any other major bank-level shocks. More than that, to control for within-year time trends and time-invariant unobserved bank heterogeneity, we always include month-year and bank fixed effects. Bank fixed effects are included for all Eurozone lead arrangers of a given loan, which underlie the the calculation of the average Deposit ratio j in Thus, we effectively estimate the average risk associated with loans granted by banks with different deposit ratios before and after June In this setting, a potential concern regarding the identification of a causal chain from negative policy rates to bank risk taking may be centered on bank-firm matching. Given the relatively short time window around the June-2014 event, most firms are observed to have received only one loan, which eradicates the possibility of including (bank-)firm fixed effects. This is, however, crucial insofar as central banks lower interest rates when the economy is doing badly, which is also when lending tends to be riskier because of riskier borrowers. This renders it difficult to distinguish between our supply-side explanation, i.e., banks picking riskier borrowers, and an alternative demand-side explanation, i.e., risky borrowers demanding relatively more credit from high-deposit banks in times of negative policy rates. We take two steps to control for this possibility. First, we include industry-year and country-year fixed effects to capture time-varying unobserved heterogeneity of borrowers 7 Deposit ratios tend to be stable over time. As we show in the Online Appendix, our results are robust to using other years for the determination of a bank s ratio of deposits over total assets. 9

11 that could be explained by their industry or country dynamics. Second, we limit out sample to non-eurozone borrowers with syndicated loans granted by Eurozone lead arrangers to filter out any effect of an environment with negative policy rates on the composition of borrowers. In order to buttress the supply-side channel, we provide evidence for the specificity of the treatment effect to rate decreases into negative, rather than non-negative, territory. For this purpose, we use the reduction of the DF rate to what was believed to be the zero lower bound in July 2012, and show that high-deposit and low-deposit banks were not differentially affected in their risk taking. To test this, we extend our sample to the period from January 2011 to October 2015, and include the interaction Deposit ratio j After(07/2012) t, where After(07/2012) t is a dummy variable for the period from July 2012 onwards, in (1). This lends support to the idea that the bank risk-taking channel is identified only when the passthrough of loan rates and deposit rates is asymmetric, which is the case when short-term rates become negative, rather than when they decrease but remain positive. Crucially, if firm-level demand was driving our findings, we should find similar effects after both rate decreases in July 2012 and June Lastly, we show our results to be robust to the inclusion of Danish, Swedish, and Swiss lenders by exploiting the staggered timing of negative policy rates across these countries and the Eurozone. To this end, we modify (1) as follows: y ijt = β 1 Deposit ratio j After jt + β 2 X it + δ t + η j + ɛ ijt, (2) where Deposit ratio j is now the average ratio (in %) of deposits over total assets across all Eurozone or Danish, Swedish, or Swiss lead arrangers j in 2013, After jt is a dummy variable for the period from June 2014 onwards for all loans with any Eurozone lead arrangers, or from July 2012, July 2014, or January 2015 for all loans with only Danish, Swedish, or Swiss lead arrangers, respectively, and η j denotes bank fixed effects, which are included for all Eurozone, Danish, Swedish, and Swiss lead arrangers. 10

12 2.3 Data Description To measure bank risk taking, we use the riskiness associated with syndicated loans for which banks acted as lead arrangers. For our loans sample, we use DealScan data, which we match with Bureau van Dijk s Amadeus data on European firms. We consider the lead arrangers when identifying the types of banks that granted the loan. We determine their ratio of deposits over total assets as our treatment-intensity measure by hand-matching the respective lead arrangers with balance-sheet and P&L data at the bank-group level from SNL. In our baseline sample, we use syndicated loans with any Eurozone lead arrangers from January 2013 to October When we include Danish, Swedish, and Swiss lenders, we limit the sample of additional loans to those with only Danish, Swedish, or Swiss lead arrangers, as Sweden and Switzerland introduced negative policy rates, and Denmark re-introduced them, only after the Eurozone did. To accommodate these dates, we use the sample period from January 2011 to October For each loan granted to firm i by lead arranger(s) j at date t, we define the associated level of ex-ante observable firm risk by means of two measures. First, for both private and publicly listed firms, σ(roa i ) 5y is the five-year standard deviation of firm i s return on assets (ROA, using P&L before tax) from year t 5 to t 1, Second, for public firms only, which make for roughly one-third of our sample, σ(return i ) 36m is the standard deviation of firm i s stock returns in the 36 months before t. In Table 1, we present summary statistics for all key variables in our analysis. An interesting feature about European syndicated loans is their relatively long maturity, five years on average. Note, furthermore, that all loans in our sample are floating-rate loans. Importantly, while roughly half of the loans in our sample have a unique lead arranger, the average number of lead arrangers is 3.8. This set of lead arrangers serves as the basis for Deposit ratio j, which is the average ratio (in %) of deposits over total assets across all applicable lead arrangers j in Accordingly, in regression specification (1), we include 11

13 bank fixed effects η j for all such lead arrangers of a given loan. Hence, a convex combination of these bank fixed effects captures the level effect of Deposit ratio j, leaving the coefficient on Deposit ratio j After(06/2014) t as our difference-in-differences estimate. 3 Results We start our empirical analysis by visualizing the main finding in Figure 4, namely that high-deposit Eurozone banks financed significantly riskier firms following the introduction of negative policy rates in June In the top panel, we plot the four-month average 8 of ROA volatility of all firms that received loans from Eurozone lead arrangers that were in the top vs. bottom tercile of the distribution of Deposit ratio j. In the bottom panel, we use the top and bottom quintiles. That is, we yield three data points per year. In the period leading up to the introduction of negative policy rates, comprising three data points, we observe parallel trends between treated high-deposit and control low-deposit banks in terms of risk taking, and high-deposit banks financed less risky firms than low-deposit banks. This gap closes when policy rates became negative (the June-2014 data point uses data from June to September 2014), and the previous trend is eventually reversed, implying significantly greater risk taking by highdeposit banks after June This effect is even more pronounced when comparing the top vs. bottom quintiles, rather than terciles, of Deposit ratio j in the bottom panel of Figure 4, suggesting a positive difference-in-differences estimate. In Table 2, we show that this is indeed the case by estimating (1). In the first column, we find that a positive and significant treatment effect. As Deposit ratio j is expressed in %, one can infer the percent change in ROA volatility by multiplying the difference-in-differences estimate with According to Table 1, Deposit ratio j exhibits a standard deviation of approximately 10.9%. Thus, a one-standard-deviation increase in Deposit ratio j translates into a 12-percent increase in ROA volatility ( = 0.12), which is substantial. 8 This is to ensure that we yield enough observations for the calculation of the mean. 12

14 Our difference-in-differences estimate further increases from to after including industry-year and country-year fixed effects in the fourth column. In the fifth column, we extend the sample to the period from January 2011 to October 2015, and include the interaction Deposit ratio j After(07/2012) t to test the impact of reducing policy rates to zero in July Not only is the respective estimate close to zero and insignificant, it is also significantly different (at the 2% level) from that on Deposit ratio j After(06/2014) t. Besides reaffirming the parallel-trends assumption, this lends support to the idea that differential risk taking by high-deposit vs. low-deposit banks is specific to rate decreases when the policy rate is negative, rather than positive. Finally, in the last column of Table 2, we reduce the sample from the fifth column to non-eurozone borrowers so as to filter out any correlation between negative interest-rate environments and firm characteristics. In this subsample, firms should not be affected by economic policies associated with negative interest rates, other than through trade and other connections to Eurozone firms. The difference-in-differences estimate on Deposit ratio j After(06/2014) t is even stronger in this subsample, and still significantly different (at the 5% level) from the coefficient on Deposit ratio j After(07/2012) t. Thus far, we have considered both privately held and publicly listed borrower firms. For the subsample of public firms, our results are robust to using as dependent variable borrower firms stock-return volatility, based on monthly returns, in Table 3. Note that statistical significance may suffer, but survives, at times due to the drop in sample size in the already quite short sample period. One concern regarding these results for stock-return volatility may be that public firms potentially differ along specific characteristics that are relevant for our outcome. To this end, in Table 3, we re-run the regressions from Table 2, for the sample of both private and public firms with ROA volatility as dependent variable, on the subsample of public firms only. All insights carry over, and the difference-in-differences estimates are even larger. In the last column, where we limit the sample to non-eurozone borrowers, the respective coefficient is imprecisely estimated, which is due to the drop in sample size (327 observations are left). 13

15 More importantly, the coefficient is virtually unaltered compared to that in the fifth column, where we include all European borrowers. In addition, our findings are also robust to the definition of our treatment-intensity variable Deposit ratio j. We show this in Tables A.1 to A.3, where we re-run all regressions from Tables 2 to 4, but replace said variable by the average deposit ratio across all Eurozone lead arrangers from 2011 to 2013, rather than in We next turn to the estimation of (2), where we include Danish, Swedish, and Swiss banks to yield staggered timing of negative policy rates across these countries and the Eurozone. In Tables 5 to 7, we re-run the regressions from the first four columns of Tables 2 to 4, and define After jt as an indicator for the period characterized by negative policy rates that is specific to the Eurozone, Denmark, Sweden, and Switzerland. The results for ROA volatility are remarkably similar across the two sets of tables, and put further emphasis on the stronger treatment effect on public, rather than private, firms (which can be seen by comparing the first four columns of Table 2 and 4, as opposed to the estimates in Tables 5 and 7). The estimates for stock-return volatility in Table 6 are also virtually unaltered compared to those in Table 3. Thus far, our findings suggest that treated high-deposit banks financed significantly riskier firms following the introduction of negative policy rates. While we have shown this behavior to be motivated by a decrease in the net interest margin, it remains an open issue as to how, rather than why, these banks manage to rebalance their portfolio in the market for syndicated loans. In Section 2.1, we argue that banks with a high deposit ratio are more exposed to the drop in the net interest margin, which kicks in due to a zero lower bound on deposit rates despite negative policy rates. Thus, it becomes more attractive for treated banks to lend money than to deposit it with the central bank. In particular, loans that would previously not have been granted are now more attractive, implying that the marginal loan carries a lower expected return. Roughly speaking, a loan s expected return is the default-probability- 14

16 weighted average of the loan spread and the collateral value. As we have shown empirically, treated high-deposit banks lend to riskier firms, which plausibly increases the ex-ante probability of corporate default and, therefore, constitutes a relevant reason for why the marginal loan s expected return is lower. However, both the loan spread and the collateral value associated with a loan are negotiation outcomes between the borrower and the lender and, thus, subject to demand and supply considerations. According to the bank risk-taking channel, high-deposit banks extend the supply of loans for riskier firms, which should not all else equal lead to an increase in the loan spread. Similarly, if the supply-side argument implied by the bank risk-taking channel is at play, then treated high-deposit banks should be willing to finance riskier firms without adjusting collateral or any other loan terms that affect its expected return. To show that this is indeed the case for banks treated under negative policy rates, we re-estimate regression specification (1) for various loan-level (contractual) outcomes. In the first column of Table 8, we find that treated high-deposit banks charged significantly lower interest rates, which mirrors the findings in Paligorova and Santos (2016) and Ioannidou, Ongena, and Peydró (2015) for low-interest-rate environments. However, the effect loses statistical significance and decreases in size, as we add industry-year and country-year fixed effects. In the fifth column, there is no difference between the two difference-in-differences estimates around the two rate decreases in June 2014 and July In the last column, where we limit the sample to non-eurozone borrowers, the effect does not survive either. In Table 9, we re-run the same set of regressions, and use as dependent variable the size of the loan. We do not find any consistently significant effect on the latter, suggesting that riskier firms financed by treated banks did not receive smaller loans. Other loan terms at origination are not adjusted to reflect the higher risk of borrowers either: in Table 10, we fail to find any treatment effect on whether loans are secured, the (average) loan share retained by the lead arranger(s), the use of financial covenants, or loan maturity. 15

17 4 Conclusion When central banks charge negative policy rates to stimulate a post-crisis economy, they enter unchartered territory. We document the limits of such standard monetary policy below the zero bound. In particular, we identify negative policy rates to lead to greater risk taking by banks in the market for syndicated loans. Lowering policy rates into negative territory provides a suitable natural experiment to study the impact of central-bank decisions on bank behavior. Normally, it is difficult to disentangle the effect of lower policy rates on the asset side of banks balance sheets from the effect on the liability side. We exploit banks reluctance to pass on negative rates to their depositors. This effectively shuts down the effect on the liability side for high-deposit, rather than low-deposit, banks. We use transaction-level data on syndicated loans to examine bank lending behavior. While the market for syndicated loans represents only a fraction of overall bank lending, it offers two key advantages in our setting. First, it allows us to match banks with firms. We can therefore study the characteristics of firms that receive new loans most notably an ex-ante measure of risk from banks with differential exposure to lower policy rates (via their different reliance on deposit funding). Second, the market for syndicated loans is global. This enables us to study borrowers that are isolated from a change in the policy rate, because they are located in a different currency zone. This effectively shuts down the credit channel, i.e., the idea that monetary policy and economic environment are endogenous. Finally, deposit funding is both a blessing and a curse for the stability of banks. Typically, deposits are a stable source of funding, and insulate banks from the whims of financial markets. Yet in post-crisis times, when central banks engage in aggressive monetary policy, deposits prevent banks from benefitting from a lower cost of funding. That is, while negative policy rates are intended to deliver additional monetary stimulus, they operate through banks as suppliers of financing to the real economy. We show that this leads to unintended consequences, as these banks shift towards lending to significantly riskier firms. 16

18 References Adrian, T., and H. S. Shin (2009): Money, Liquidity, and Monetary Policy, American Economic Review, 99(2), Bech, M. L., and A. Malkhozov (2016): How Have Central Banks Implemented Negative Policy Rates?, BIS Quarterly Review No. 1. Bernanke, B., and M. Gertler (1989): Agency Costs, Net Worth, and Business Fluctuations, American Economic Review, 79(1), Bernanke, B. S., and A. S. Blinder (1992): The Federal Funds Rate and the Channels of Monetary Transmission, American Economic Review, 82(4), Borio, C., and H. Zhu (2008): Capital Regulation, Risk-Taking and Monetary Policy: A Missing Link in the Transmission Mechanism, BIS Working Paper No Chakraborty, I., I. Goldstein, and A. MacKinlay (2016): Monetary Stimulus and Bank Lending, University of Pennsylvania Working Paper. Dell Ariccia, G., L. Laeven, and R. Marquez (2014): Real Interest Rates, Leverage, and Bank Risk-Taking, Journal of Economic Theory, 149, Dell Ariccia, G., L. Laeven, and G. Suarez (2016): Bank Leverage and Monetary Policy s Risk-Taking Channel: Evidence from the United States, Journal of Finance, forthcoming. English, W. B., S. J. Van den Heuvel, and E. Zakrajšek (2014): Interest Rate Risk and Bank Equity Valuations, Wharton Financial Institutions Center Working Paper No Ioannidou, V., S. Ongena, and J.-L. Peydró (2015): Monetary Policy, Risk-Taking, and Pricing: Evidence from a Quasi-Natural Experiment, Review of Finance, 19(1),

19 Jayaratne, J., and D. P. Morgan (2000): Capital Market Frictions and Deposit Constraints at Banks, Journal of Money, Credit and Banking, 32(1), Jiménez, G., S. Ongena, J.-L. Peydró, and J. Saurina (2012): Credit Supply and Monetary Policy: Identifying the Bank Balance-Sheet Channel with Loan Applications, American Economic Review, 102(5), (2014): Hazardous Times for Monetary Policy: What Do Twenty-Three Million Bank Loans Say About the Effects of Monetary Policy on Credit Risk-Taking?, Econometrica, 82(2), Kacperczyk, M. T., and M. Di Maggio (2016): The Unintended Consequences of the Zero Lower Bound Policy, Journal of Financial Economics, forthcoming. Kandrac, J., and B. Schlusche (2016): Quantitative Easing and Bank Risk Taking: Evidence from Lending, Federal Reserve Board Working Paper. Kishan, R. P., and T. P. Opiela (2000): Bank Size, Bank Capital, and the Bank Lending Channel, Journal of Money, Credit and Banking, 32(1), Maddaloni, A., and J.-L. Peydró (2011): Bank Risk-taking, Securitization, Supervision, and Low Interest Rates: Evidence from the Euro-area and the U.S. Lending Standards, Review of Financial Studies, 24(6), Paligorova, T., and J. A. C. Santos (2016): Monetary Policy and Bank Risk-Taking: Evidence from the Corporate Loan Market, Federal Reserve Bank of New York Working Paper. Saunders, A. (2000): Low Inflation: The Behavior of Financial Markets and Institutions, Journal of Money, Credit and Banking, 32(4), Stein, J. C., and A. K. Kashyap (2000): What Do a Million Observations on Banks Say about the Transmission of Monetary Policy?, American Economic Review, 90(3),

20 Taylor, J. B. (2009): The Financial Crisis and the Policy Responses: An Empirical Analysis of What Went Wrong, NBER Working Paper No

21 5 Figures Figure 1: Deposit Rates on Overnight Deposits (Households and NFCs). This figure shows the evolution of overnight deposit rates at Eurozone banks between December 2011 and March The data are taken from the ECB IMIR interest rate statistics database, which provides monthly data on deposit rates for the median Eurozone bank at the monetary financial institution (MFI) level. 20

22 Figure 2: Deposit Rates on Short-term (<1y) Agreed Maturity Deposits (Households and NFCs). This figure shows the evolution of short term agreed maturity deposit rates at the median Eurozone bank between December 2011 and March The data are taken from the ECB IMIR interest rate statistics database, which provides monthly data on deposit rates for Eurozone banks at the monetary financial institution (MFI) level. 21

23 Figure 3: Evolution of Cost of Debt associated with Loans granted by Eurozone Banks. This figure plots the four-month (forward-looking) average of the total cost of credit charged by Eurozone lead arrangers, separately for Eurozone and non-eurozone borrowers. 22

24 Figure 4: ROA Volatility of Private and Public Firms associated with Loans granted by Eurozone Banks with High vs. Low Deposit Ratios. This figure plots the four-month (forward-looking) average of ROA volatility of all firms that received loans from Eurozone lead arrangers that were in the top vs. bottom tercile of the distribution of the average ratio of deposits over total assets in For a given loan at date t, the associated ROA volatility is measured as the five-year standard deviation of the borrower firm s return on assets (ROA, using P&L before tax) from year t 5 to t 1. 23

25 6 Tables Table 1: Summary Statistics Variable Mean Std. dev. Min Max N σ(roa i ) 5y ,327 σ(return i ) 36m Deposit ratio in % ,918 Eurozone firm {0, 1} ,918 All-in-drawn spread in bps Loan size in 2016 ebn ,080 Secured [0, 1] Average loan share of lead arrangers [0, 1] Financial covenants {0, 1} ,082 Maturity of loan in months ,018 No. of lead arrangers ,082 Notes: The baseline sample consists of all completed syndicated loans (package level) of both private and publicly listed firms i at date t granted by any Eurozone lead arranger(s) j from January 2013 to October σ(roa i ) 5y is the five-year standard deviation of firm i s return on assets (ROA, using P&L before tax) from year t 5 to t 1. σ(return i ) 36m is the standard deviation of firm i s stock returns in the 36 months before t. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j in Eurozone firm i is an indicator for whether firm i is headquartered in the Eurozone. The all-in-drawn spread is the sum of the spread over LIBOR and any annual fees paid to the lender syndicate. 24

26 Table 2: Impact of Negative Policy Rates on Firms ROA Volatility ln(σ(roa i ) 5y ) Sample , non-euro firms Deposit ratio After(06/2014) 0.010** 0.012** 0.012** 0.014** 0.015** 0.030** (0.005) (0.005) (0.005) (0.006) (0.006) (0.014) Deposit ratio After(07/2012) (0.004) (0.010) Bank FE Y Y Y Y Y Y Month-year FE Y Y Y Y Y Y Industry FE N Y Y N N N Industry-year FE N N N Y Y Y Country FE N Y N N N N Country-year FE N N Y Y Y Y N 1,327 1,327 1,327 1,327 2, Notes: The sample consists of all completed syndicated loans (package level) of both private and publicly listed firms i at date t granted by any Eurozone lead arranger(s) j, from January 2013 to October 2015 in the first four columns and from January 2011 to October 2015 in the last two columns. In the last column, we furthermore limit the sample to non-eurozone borrowers. The dependent variable is the logged five-year standard deviation of firm i s return on assets (ROA, using P&L before tax) from year t 5 to t 1. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j in After(06/2014) t is a dummy variable for the period from June 2014 onwards. After(07/2012) t is a dummy variable for the period from July 2012 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 25

27 Table 3: Impact of Negative Policy Rates on Firms Stock-return Volatility ln(σ(return i ) 36m ) Sample , non-euro firms Deposit ratio After(06/2014) 0.006** 0.004* 0.006** 0.008** ( ) (0.003) (0.003) (0.003) (0.004) (0.004) (0.013) Deposit ratio After(07/2012) (0.003) (0.014) Bank FE Y Y Y Y Y Y Month-year FE Y Y Y Y Y Y Industry FE N Y Y N N N Industry-year FE N N N Y Y Y Country FE N Y N N N N Country-year FE N N Y Y Y Y N , Notes: The sample consists of all completed syndicated loans (package level) of publicly listed firms i at date t granted by any Eurozone lead arranger(s) j, from January 2013 to October 2015 in the first four columns and from January 2011 to October 2015 in the last two columns. In the last column, we furthermore limit the sample to non-eurozone borrowers. The dependent variable is the logged standard deviation of firm i s stock returns in the 36 months before t. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j in After(06/2014) t is a dummy variable for the period from June 2014 onwards. After(07/2012) t is a dummy variable for the period from July 2012 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 26

28 Table 4: Impact of Negative Policy Rates on Public Firms ROA Volatility ln(σ(roa i ) 5y ) Sample , non-euro firms Deposit ratio After(06/2014) 0.019*** 0.017*** 0.014* 0.019** 0.021** (0.007) (0.006) (0.007) (0.009) (0.010) (0.039) Deposit ratio After(07/2012) (0.007) (0.029) Bank FE Y Y Y Y Y Y Month-year FE Y Y Y Y Y Y Industry FE N Y Y N N N Industry-year FE N N N Y Y Y Country FE N Y N N N N Country-year FE N N Y Y Y Y N , Notes: The sample consists of all completed syndicated loans (package level) of publicly listed firms i at date t granted by any Eurozone lead arranger(s) j, from January 2013 to October 2015 in the first four columns and from January 2011 to October 2015 in the last two columns. In the last column, we furthermore limit the sample to non-eurozone borrowers. The dependent variable is the logged five-year standard deviation of firm i s return on assets (ROA, using P&L before tax) from year t 5 to t 1. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j in After(06/2014) t is a dummy variable for the period from June 2014 onwards. After(07/2012) t is a dummy variable for the period from July 2012 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 27

29 Table 5: Impact of Negative Policy Rates on Firms ROA Volatility Inclusion of Danish, Swedish, and Swiss Banks ln(σ(roa i ) 5y ) Deposit ratio After 0.007* 0.009** 0.010** 0.011** (0.004) (0.004) (0.004) (0.005) Bank FE Y Y Y Y Month-year FE Y Y Y Y Industry FE N Y Y N Industry-year FE N N N Y Country FE N Y N N Country-year FE N N Y Y N 2,262 2,262 2,262 2,262 Notes: The sample consists of all completed syndicated loans (package level) of both private and publicly listed firms i at date t granted by granted by any Eurozone or only by Danish, Swedish, or Swiss lead arranger(s) j from January 2011 to October The dependent variable is the logged five-year standard deviation of firm i s return on assets (ROA, using P&L before tax) from year t 5 to t 1. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone or Danish, Swedish, or Swiss lead arrangers j in After jt is a dummy variable for the period from June 2014 onwards for all loans with any Eurozone lead arrangers, or from July 2012, July 2014, or January 2015 for all loans with only Danish, Swedish, or Swiss lead arrangers, respectively. Bank fixed effects are included for all Eurozone, Danish, Swedish, and Swiss lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 28

30 Table 6: Impact of Negative Policy Rates on Firms Stock-return Volatility Inclusion of Danish, Swedish, and Swiss Banks ln(σ(return i ) 36m ) Deposit ratio After 0.006** 0.006*** 0.005** 0.006* (0.002) (0.002) (0.003) (0.003) Bank FE Y Y Y Y Month-year FE Y Y Y Y Industry FE N Y Y N Industry-year FE N N N Y Country FE N Y N N Country-year FE N N Y Y N 1,174 1,174 1,174 1,174 Notes: The sample consists of all completed syndicated loans (package level) of publicly listed firms i at date t granted by granted by any Eurozone or only by Danish, Swedish, or Swiss lead arranger(s) j from January 2011 to October The dependent variable is the logged standard deviation of firm i s stock returns in the 36 months before t. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone or Danish, Swedish, or Swiss lead arrangers j in After jt is a dummy variable for the period from June 2014 onwards for all loans with any Eurozone lead arrangers, or from July 2012, July 2014, or January 2015 for all loans with only Danish, Swedish, or Swiss lead arrangers, respectively. Bank fixed effects are included for all Eurozone, Danish, Swedish, and Swiss lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 29

31 Table 7: Impact of Negative Policy Rates on Public Firms ROA Volatility Inclusion of Danish, Swedish, and Swiss Banks ln(σ(roa i ) 5y ) Deposit ratio After 0.020*** 0.021*** 0.021*** 0.023*** (0.006) (0.006) (0.006) (0.008) Bank FE Y Y Y Y Month-year FE Y Y Y Y Industry FE N Y Y N Industry-year FE N N N Y Country FE N Y N N Country-year FE N N Y Y N 1,045 1,045 1,045 1,045 Notes: The sample consists of all completed syndicated loans (package level) of publicly listed firms i at date t granted by granted by any Eurozone or only by Danish, Swedish, or Swiss lead arranger(s) j from January 2011 to October The dependent variable is the logged five-year standard deviation of firm i s return on assets (ROA, using P&L before tax) from year t 5 to t 1. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone or Danish, Swedish, or Swiss lead arrangers j in After jt is a dummy variable for the period from June 2014 onwards for all loans with any Eurozone lead arrangers, or from July 2012, July 2014, or January 2015 for all loans with only Danish, Swedish, or Swiss lead arrangers, respectively. Bank fixed effects are included for all Eurozone, Danish, Swedish, and Swiss lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 30

32 Table 8: Impact of Negative Policy Rates on Loan Spreads ln(all-in-drawn spread) Sample , non-euro firms Deposit ratio After(06/2014) ** * (0.006) (0.006) (0.007) (0.008) (0.007) (0.016) Deposit ratio After(07/2012) (0.004) (0.016) Bank FE Y Y Y Y Y Y Month-year FE Y Y Y Y Y Y Industry FE N Y Y N N N Industry-year FE N N N Y Y Y Country FE N Y N N N N Country-year FE N N Y Y Y Y N , Notes: The sample consists of all completed syndicated loans (package level) of both private and publicly listed firms i at date t granted by any Eurozone lead arranger(s) j from January 2013 to October The dependent variable is the log of the all-in-drawn spread (in bps), which is the sum of the spread over LIBOR and any annual fees paid to the lender syndicate. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j in After(06/2014) t is a dummy variable for the period from June 2014 onwards. After(07/2012) t is a dummy variable for the period from July 2012 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 31

33 Table 9: Impact of Negative Policy Rates on Loan Size ln(loan size) Sample , non-euro firms Deposit ratio After(06/2014) ** * (0.005) (0.004) (0.004) (0.005) (0.005) (0.008) Deposit ratio After(07/2012) (0.005) (0.008) Bank FE Y Y Y Y Y Y Month-year FE Y Y Y Y Y Y Industry FE N Y Y N N N Industry-year FE N N N Y Y Y Country FE N Y N N N N Country-year FE N N Y Y Y Y N 2,080 2,080 2,080 2,080 3, Notes: The sample consists of all completed syndicated loans (package level) of both private and publicly listed firms i at date t granted by any Eurozone lead arranger(s) j from January 2013 to October The dependent variable is the log of the loan size in 2016 e. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j in After(06/2014) t is a dummy variable for the period from June 2014 onwards. After(07/2012) t is a dummy variable for the period from July 2012 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 32

34 Table 10: Impact of Negative Policy Rates on Other Loan Terms Secured Lead share Covenants ln(maturity) Deposit ratio After(06/2014) (0.003) (0.002) (0.001) (0.002) Bank FE Y Y Y Y Month-year FE Y Y Y Y Industry-year FE Y Y Y Y Country-year FE Y Y Y Y N ,082 2,018 Notes: The sample consists of all completed syndicated loans (package level) of both private and publicly listed firms i at date t granted by any Eurozone lead arranger(s) j from January 2013 to October The dependent variable in the first column is the proportion, between 0 and 1, of facilities within the package that are secured, in the second column the average loan share, between 0 and 1, retained by all Eurozone lead arrangers, in the third column an indicator for whether the loan has at least one financial covenant, and in the last column the logged maturity. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j in After(06/2014) t is a dummy variable for the period from June 2014 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 33

35 Supplementary Appendix (Not for Publication) A Supplementary Tables Table A.1: Impact of Negative Policy Rates on Firms ROA Volatility Robustness to Definition of Deposit Ratio ln(σ(roa i ) 5y ) Sample , non-euro firms Deposit ratio After(06/2014) 0.010* 0.012** 0.012** 0.015** 0.015** 0.033** (0.006) (0.005) (0.006) (0.006) (0.007) (0.015) Deposit ratio After(07/2012) (0.005) (0.011) Bank FE Y Y Y Y Y Y Month-year FE Y Y Y Y Y Y Industry FE N Y Y N N N Industry-year FE N N N Y Y Y Country FE N Y N N N N Country-year FE N N Y Y Y Y N 1,327 1,327 1,327 1,327 2, Notes: The sample consists of all completed syndicated loans (package level) of both private and publicly listed firms i at date t granted by any Eurozone lead arranger(s) j, from January 2013 to October 2015 in the first four columns and from January 2011 to October 2015 in the last two columns. In the last column, we furthermore limit the sample to non-eurozone borrowers. The dependent variable is the logged five-year standard deviation of firm i s return on assets (ROA, using P&L before tax) from year t 5 to t 1. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j from 2011 to After(06/2014) t is a dummy variable for the period from June 2014 onwards. After(07/2012) t is a dummy variable for the period from July 2012 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 34

36 Table A.2: Impact of Negative Policy Rates on Firms Stock-return Volatility Robustness to Definition of Deposit Ratio ln(σ(return i ) 36m ) Sample , non-euro firms Deposit ratio After(06/2014) 0.005* * 0.008* ( ) (0.003) (0.003) (0.003) (0.004) (0.004) (0.015) Deposit ratio After(07/2012) (0.003) (0.016) Bank FE Y Y Y Y Y Y Month-year FE Y Y Y Y Y Y Industry FE N Y Y N N N Industry-year FE N N N Y Y Y Country FE N Y N N N N Country-year FE N N Y Y Y Y N , Notes: The sample consists of all completed syndicated loans (package level) of publicly listed firms i at date t granted by any Eurozone lead arranger(s) j, from January 2013 to October 2015 in the first four columns and from January 2011 to October 2015 in the last two columns. In the last column, we furthermore limit the sample to non-eurozone borrowers. The dependent variable is the logged standard deviation of firm i s stock returns in the 36 months before t. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j from 2011 to After(06/2014) t is a dummy variable for the period from June 2014 onwards. After(07/2012) t is a dummy variable for the period from July 2012 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 35

37 Table A.3: Impact of Negative Policy Rates on Public Firms ROA Volatility Robustness to Definition of Deposit Ratio ln(σ(roa i ) 5y ) Sample , non-euro firms Deposit ratio After(06/2014) 0.021*** 0.020*** 0.016* 0.023** 0.025** (0.008) (0.007) (0.008) (0.010) (0.012) (0.044) Deposit ratio After(07/2012) (0.008) (0.031) Bank FE Y Y Y Y Y Y Month-year FE Y Y Y Y Y Y Industry FE N Y Y N N N Industry-year FE N N N Y Y Y Country FE N Y N N N N Country-year FE N N Y Y Y Y N , Notes: The sample consists of all completed syndicated loans (package level) of publicly listed firms i at date t granted by any Eurozone lead arranger(s) j, from January 2013 to October 2015 in the first four columns and from January 2011 to October 2015 in the last two columns. In the last column, we furthermore limit the sample to non-eurozone borrowers. The dependent variable is the logged five-year standard deviation of firm i s return on assets (ROA, using P&L before tax) from year t 5 to t 1. Deposit ratio j is the average ratio (in %) of deposits over total assets across all Eurozone lead arrangers j from 2011 to After(06/2014) t is a dummy variable for the period from June 2014 onwards. After(07/2012) t is a dummy variable for the period from July 2012 onwards. Bank fixed effects are included for all Eurozone lead arrangers. Industry(-year) fixed effects are based on two-digit SIC codes. Country(-year) fixed effects are based on the firm s country of origin. Public-service, energy, and financial-services firms are dropped. Robust standard errors (clustered at the bank level) are in parentheses. 36

38 B Supplementary Figures Figure B.1: Composition of Household and NFC Deposits at Eurozone Banks. This figure shows the end-of-year composition of household and NFC deposits at Eurozone banks for 2013, 2014, and We differentiate between four types of deposits, namely overnight deposits, deposits with an agreed maturity below one year, deposits redeemable at notice (below three months), and other deposits. The data are taken from the ECB IBSI database, which provides monthly bank balance-sheet data for Eurozone banks at the monetary financial institution (MFI) level. We first calculate the different deposit shares for each bank, and then take the average over the different banks. 37