The Underwriter Relationship and Corporate Debt Maturity

Transcription

1 The Underwriter Relationship and Corporate Debt Maturity Indraneel Chakraborty Andrew MacKinlay May 11, 2018 Abstract Supply-side frictions impact corporate debt maturity choices. Similar to bank loan markets, corporate debt markets exhibit repeated issuance relationships between underwriters and issuers. Using microlevel data on investor-underwriter-issuer relationships, we uncover significant frictions in the U.S. public corporate debt markets, even though firms issuing corporate debt often have relatively easier access to capital. Firm debt maturity is more sensitive to underwriter preferences than to a firm s past debt maturity. Firm issuance amounts decrease with an increase in the maturity preference mismatch between firms and underwriters. Supply-side frictions in the corporate debt market lead to slower lending and firm asset growth. JEL Code: G24, G32. Keywords: Debt maturity, Insurance companies, Underwriters, Maturity mismatch. For helpful comments and discussions, we thank Elena Loutskina, René Stulz; and seminar participants at the Darden School, University of Virginia and Virginia Tech. Indraneel Chakraborty: School of Business Administration, University of Miami, Coral Gables, FL i.chakraborty@miami.edu. Andrew MacKinlay: Pamplin College of Business, Virginia Tech, Blacksburg, VA acmackin@vt.edu.

2 Do financing relationships matter in determining contract terms in public debt markets? The answer to this question depends on the extent of financing frictions in public debt markets. A reasonable prior is that public debt markets in the U.S., given their size and scope, do not suffer from meaningful supply-side frictions. This perhaps explains why the literature has generally focused on bank-debt markets to investigate the importance of relationships in determining loan amounts and contract terms. 1,2 In this paper, we use micro-level data to test whether supply-side frictions affect firm corporate public debt decisions and if such frictions translate into real effects. Focusing on the maturity of newly issued debt, we explicitly link firm decisions to the underwriter helping originate the new debt. Further, we link underwriters to the investors which purchase these debt securities. We focus on the maturity of corporate debt for two reasons. One, maturity is a contract term for which investors have preferences that are unrelated to a particular firm. Insurance companies, which we consider, have significant liabilities of different durations stemming from their lines of business. They will therefore have preferences over which maturities to include in their fixed income portfolios. Two, we are able to observe the maturities of the firm s existing debt structure. This data allows us better account for the firm s own preferences when it comes to issuing new corporate debt. Our analysis allows us to make three contributions. First, we document the importance of the relationship between the firm and the underwriter. In our sample, we find that a firm has a 28% probability of working with the same underwriter on its next bond issue. This number compares to a 36% probability of a firm working with the same bank on its next loan in the syndicated loan market. Presumably for similar reasons as in the syndicated loan market, firms appear to reuse underwriters at similar rates for the public debt market. We also document that firms which raise public debt do so frequently: we find that the median firm in our sample issues public debt every 0.9 years. Second, we measure the extent to which underwriters preferences regarding debt maturity affect firms maturity choices when new debt is issued. Given the likelihood of using the same underwriter and the 1 Almeida, Campello, Laranjeira, and Weisbenner (2012); Custódio, Ferreira, and Laureano (2013); Harford, Klasa, and Maxwell (2014) are important exceptions in recent literature, who have addressed corporate debt maturity and its implications on firm outcomes. 2 Petersen and Rajan (1994) discuss the benefits of bank lending relationships regarding the availability and the cost of funds to a firm. Petersen and Rajan (1995) show that creditors are willing to finance constrained firms in concentrated credit markets. Berger, Miller, Petersen, Rajan, and Stein (2005) show that small banks are better able to collect and act on soft information. Chava and Purnanandam (2011) show that firms that borrow from banks facing adverse shocks receive less credit at higher loan rates; Murfin (2012) shows that lender-specific shocks affect contract strictness. 2

3 frequency of issuing public debt, if public debt market supply-side frictions are important, firms should be influenced by their underwriter relationship. We find that a firm s debt maturity is more than twice as sensitive to debt underwriter preferences than the firm s own existing maturity structure. We find this result despite the fact that firms in our sample are larger, with access to corporate debt and equity markets, and hence should be less capital constrained. Third, we show that supply-side frictions in the U.S. public debt market have real effects. Absent financial frictions, firms will choose debt maturity based on their firm-level determinants. 3 However, in the presence of financial frictions, firms make financing decisions that trade off better contract terms (e.g. larger amounts or lower yields) for their preferred maturity choice. To capture this trade-off, we use the difference in maturity between the firm s issuance and the average maturity of other contemporaneous bonds issued by the same underwriter. The larger the mismatch between the firm s maturity choice from the peak demand among the underwriter s investor base, the more likely the firm will be sacrificing other beneficial contract terms. Our results show that larger deviations in maturity are associated with lower loan amounts. In particular, a 1% increase in maturity deviation is associated with a 0.92% 1.36% decrease in the amount raised, depending on the specification. 4 Further, we do not find that firms are able to replace this capital from other sources: at the firm-level, we find that firms which deviate from the underwriter s average maturity have smaller overall debt increases. For a 1% increase in the mismatch between underwriter and firm preferences on debt maturity, firm debt growth slows by 3.4 basis points. We also find that these firms asset growth suffers as a result. In the case of asset growth, the sensitivity is 2.3 basis points slower growth for a 1% increase in mismatch. 5 In the cross-section of firms, these effects are concentrated in firms which have only one underwriter relationship. In addition, we do not find evidence that firms with a AAA credit rating face these supply-side frictions, pointing to a significant investor preference for the highest rating category. However, we find that for firms with an investment-grade rating of A rating or lower, there are significant supply-side frictions for firms in the U.S. corporate debt market. 3 Firm-level determinants include, but are not limited to, liquidity risk (Diamond, 1991a), growth options, information asymmetry (Barclay and Smith, 1995), credit ratings (Guedes and Opler, 1996), firm size and risk (Stohs and Mauer, 1996), and asset salability (Benmelech, 2008). 4 In unreported analysis, we also consider whether there are similar effects for the maturity deviation on the bond s yield at the time of offering. Although we find weak evidence that greater maturity deviations lead to higher yields, the primary dimension affected appears to be the amount of capital raised. 5 The estimated sensitivities for debt growth and asset growth come from column 5 of Table V and Table VI, respectively. 3

4 Finally, while our identification is based on supply-side frictions at the micro-level, we also estimate aggregate implications of the maturity preference mismatch between investors and firms on credit supply. We estimate that the average mismatch in duration preferences between credit suppliers and demanders in this market leads to a decline of 1.96% in the growth rate of corporate debt. In dollar terms, maturity mismatch reduced corporate debt growth by $112 billion based on the total outstanding amount in 2016, which is the latest year for which we have data. Note that this cost is derived by comparing firms which issue public debt to counterfactual firms without any underwriter-related maturity mismatches. We do not estimate the costs to these firms of having no access to the corporate public debt market, which are substantial (Faulkender and Petersen, 2006). In such a framework, the presence of a firm-underwriter relationship provides clear benefits. Our work builds upon the recent literature that investigates the relationship between supply-side frictions, firm maturity choices, and firm outcomes. Almeida, Campello, Laranjeira, and Weisbenner (2012) use the crisis to investigate the causal effect of debt maturity on real firm behavior. The authors find that firms with significant long-term debt maturing during the crisis cut investment more than firms with long-term debt maturing in later years. Custódio, Ferreira, and Laureano (2013) show that corporate use of long-term debt has decreased in the U.S. The authors investigate this trend and find that changes in the composition of firms over time, changes in firm characteristics, and supply-side effects such as investor demand help explain the increased use of short-term debt by firms. Gopalan, Song, and Yerramilli (2014) show that short-term debt exposes a firm to rollover risk, increasing the firm s overall credit risk. Harford, Klasa, and Maxwell (2014) show that firm cash holdings can be explained by debt maturity. Our paper also complements a recent literature considering the aggregate supply-side effects on the corporate debt market. Greenwood, Hanson, and Stein (2010) show that maturity choices are affected by aggregate market conditions such as Treasury issuances. Graham, Leary, and Roberts (2014) consider the effect of changes in U.S. government borrowing on corporate financing and investment. Our paper, in contrast, investigates the impact of supply-side frictions by using micro-level data between firms and their underwriters. We use investor demand to isolate credit-supply frictions and estimate the impact of these supply-side frictions on firm debt maturity choices, credit supply and real outcomes. Our paper suggests that the corporate debt market has significant supply-side frictions related to these relationships in addition 4

5 to the aggregate effects uncovered in other work. A well-established literature has investigated the impact of bank credit-supply frictions on firm outcomes. In addition to the papers mentioned earlier, the literature has studied the impact of aggregate shocks on credit supply. Peek and Rosengren (2000) show that due to the Japanese crisis in the early 1990s, Japanese banks reduced credit supply in the U.S. Leary (2009) shows that bank credit in the first half of 1960s expanded following the introduction of the certificates of deposits. In the cross-section of banks, Kashyap and Stein (2000) find that the impact of monetary policy on lending is stronger for banks with less liquid balance sheets. Ivashina and Scharfstein (2010) show that banks cut their lending less if they had better access to deposit financing. In this paper, we investigate whether credit-supply frictions in the corporate debt market have significant firm-level effects. Our work is also related to the securities issuance literature which finds that relationships with underwriters are beneficial for issuing firms. Schenone (2004) finds firms with a pre-ipo banking relationship with a prospective underwriter face significantly lower IPO underpricing than firms without such banking relationships. James (1992) and Burch, Nanda, and Warther (2005) show that a repeated relationship between underwriting banks and firms reduces the amount of fees charged for equity issuances. Drucker and Puri (2005) show that when banks lend to issuers while also underwriting their public securities offerings, issuers benefit through lower underwriter fees and discounted loan yield spreads. These papers point toward significant supply-side frictions in the equity IPO market. In the corporate debt market, Gande, Puri, and Saunders (1999) find that underwriter spreads and ex-ante yields have declined significantly with bank entry in the 1990s, consistent with the market becoming more competitive. Yasuda (2005) finds that bank relationships have positive and significant effects on a firm s underwriter choice. Thus, relationships are important for public debt issuances as well. Our paper shows that debt underwriter relationships have significant effects on firm debt maturity decisions. The rest of the paper is structured as follows. Section I discusses the empirical framework and the sources and characteristics of the datasets utilized. Section II discusses the presence of underwriter relationships and their effects on debt maturity and the amount of capital raised. Section III considers how mismatches in maturity preferences affects firms debt and asset growth. Section IV provides further tests of the underlying mechanisms behind the frictions and some aggregate implications. Section V concludes. 5

6 I I.A Empirical framework and data Empirical framework A firm s choice of debt maturity depends on many considerations. Theoretical work by Flannery (1986) shows that firms can convey a signal about their quality through debt maturity choices. Better-quality firms use shorter-maturity debt to signal to the markets that they do not anticipate a deterioration in credit quality, and hence are willing to approach the debt market more frequently. Diamond (1991a) analyzes debt maturity structure when borrowing firms choose an optimal maturity structure which trades off a preference for shorter debt maturity due to the expectation of an improvement in their credit quality with the risk of excessive liquidation. The author suggests that the best and worst rated firms prefer short-term debt, and the intermediate credit quality firms prefer longer maturities. Further, since debt maturity can affect real investment decisions (Myers, 1977; Diamond and He, 2014) and market leverage (Gertner and Scharfstein, 1991), firms should optimally choose investment, leverage, and debt maturity together. Finally, the source of financing, i.e. public or bank debt (Diamond, 1991b; Rajan, 1992), and the ease with which firms can obtain capital from one source (equity or bank debt) compared to another source (public debt) can also potentially affect maturity choices (classic work includes Ross, 1977; Myers and Majluf, 1984; Petersen and Rajan, 1994, 1995). Our empirical framework needs to ensure that demand-side characteristics, which include the set of determinants discussed above and more, can be separated from supply-side determinants. This separation is important because our paper seeks to identify the impact of supply-side frictions on debt issuance. As observed debt issuance characteristics are determined in equilibrium by credit suppliers (underwriters) and firms that demand capital, regressing issuance characteristics on supply characteristics is not meaningful (see Figure 1). A comparable situation arises when regressing price against quantity in a supply-and-demand analysis, which is also not appropriate. To identify the effect of supply-side frictions, we require an instrument that influences an underwriter s willingness to supply debt of certain maturities without directly affecting the firm s demand for that capital. This effectively means we shift the supply of the underwriter around the equilibrium point while keeping the firm s demand for capital fixed. Our approach is similar to recent work by Ivashina (2009), who estimates the costs of information asymmetry between banks in a syndicate. 6

7 To shift the preferences of the underwriter regarding a certain maturity in a manner exogenous to the firm s demand for debt at that maturity, we instrument underwriter preferences with investor preferences. The instrument is the maturity preferences of a group of investors that purchase debt securities from an underwriter. Specifically, we observe the purchase activity of insurance companies from a given underwriter and use those purchases to calculate an average maturity preference. The debt issuance of the firm in question is removed from the calculation. The argument is that this instrument, in combination with time fixed effects and other pertinent controls, captures shifts in investor demand unrelated to the particular issuance in question. Insurers make a good investor base for an instrument for two reasons: one, they are a meaningful fraction of suppliers of corporate debt capital. Figure 2 shows that the fraction of outstanding corporate bond debt held by insurers is substantial. From , life and property-casualty insurers held between 18% and 38% of outstanding corporate debt. Two, their asset purchase decisions are driven by their specific liability structures related to their own lines of business. Their security purchase decisions, and maturity decisions in particular, are primarily a response to the duration of their own liabilities, and not any factors of the firm itself. The instrumental variables strategy isolates the impact of supply-side preferences from firm demand. However, an additional concern is that firms may have chosen a specific underwriter due to that underwriter specializing in debt of a certain maturity. Thus, in addition to the maturity being determined in equilibrium, the choice of the underwriter by a firm is also driven by the firm s characteristics. It is important to note that given our effects are negative, an endogenous selection by a firm of an underwriter creates a bias against us finding any impact of supply-side frictions. This is because firms should attempt to choose underwriters that diminish such frictions. Nevertheless, to address this concern of selection bias, we include specifications with underwriter fixed effects, firm-underwriter fixed effects, and firm-year fixed effects in our analysis. Inclusion of underwriter fixed effects eliminates static underwriter preferences. If there are persistent preferences related to a specific relationship between an underwriter and firm, this is captured by firm-underwriter fixed effects. Further, if firm characteristics that change over time lead to a choice of a specific underwriter, the inclusion of firmyear fixed effects control for such characteristics. Identification in this case is obtained from variation in 7

8 instrumented underwriter preferences for a firm which undertakes issuances from different underwriters in a year. This approach is similar to those of Khwaja and Mian (2008) and Lin and Paravisini (2013), among others. The set of firms that obtain financing from multiple underwriters in the same year likely have more access to external debt markets. Thus, any observed impact of financial frictions in this subset of firms should understate the impact on the average firm in the sample. I.B Data and overview of main variables We combine four main datasets for our analysis. We use Compustat and Capital IQ for firm-level information. We use Mergent FISD for debt issuances and NAICS for insurance company bond purchases. Firm size, book leverage, market-to-book ratio, and profitability are from Compustat. Average debt maturity is from Capital IQ. For our main dataset, we require the firm s detailed capital structure (from Capital IQ) in addition to standard Compustat information. Table I provides variable definitions and reports summary statistics on firm and issuance characteristics in the final dataset. The sample period contains annual observations from However, because of the coverage of the Capital IQ data, the vast majority of observations are from The median firm in our sample is sizable, with sales of approximately $1.6 billion U.S. dollars. The median book leverage ratio is 33% and median market-to-book ratio is 1.29, which is in line with large firms during the sample period. The median firm has an annual profitability ratio of 5.6%, debt growth of 16%, and asset growth of 10%. Because we focus on the impact of the firm s debt maturity structure on future performance, we introduce some additional variables in the analysis. The first is the firm s average debt maturity in years. This variable is calculated as the dollar-weighted average of the firm s outstanding liabilities and uses the additional granularity provided by the Capital IQ data. 6 The median firm has a debt maturity of 6.6 years and the average debt maturity of firms in the sample is 8.71 years. Following Guedes and Opler (1996), we calculate the maturity of the firm s assets. The median asset maturity is 8.38 years and the average asset maturity is 12.9 years. Table I also provides summary details for the debt issuances matched to the firms in our sample. These 6 This granularity is helpful since the traditional Compustat data does not separately detail the maturity of debt beyond classifying the dollar amount due within one year, two years, three years, four years, or five years and later. 8

9 debt issuances are generally sizable with a longer maturity. We are able to match 470 firms to a total of 39,892 issuances with 67 lead underwriters. 7 The average bond issuance size in our sample is $123.2 million, with a median issuance of $35 million. The maturity of the average debt issuance is approximately 6.11 years, at an average yield of 4.9%. The average dollar-weighted maturity for an underwriter is 7.19 years, suggesting that the larger issuances tend to have longer maturities. Considering the dollar-weighted purchases of life insurance companies with each underwriter, they tend to purchase these longer maturity bonds. Their average purchase maturity is 12.3 years, compared to only 8.43 years for property-casualty insurance companies. In later analysis, we focus on how much a firm s specific issuance maturity deviates from the underwriter s average maturity at that point in time. As such, we construct Maturity Mismatch, defined as the absolute value of the log difference between a firm s given issuance maturity and the underwriter s average maturity in that year. Similarly, we construct analogues to the mismatch variable but use the average maturity of insurance company purchases from that specific underwriter instead of the underwriter s overall average maturity. These variables are Maturity Mismatch, Life Insurance Cos. and Maturity Mismatch, P/C Insurance Cos. and are used as instruments in some specifications. II Relationships, maturity, and credit supply Our argument that supply-side frictions exist and their presence has real effects requires two elements: (i) supply-side preferences affect firms debt maturity, and (ii) firms debt maturity decisions affect their credit supply. If changes in the firms debt maturity do not have effects on their credit supply, then firms can adjust maturity without suffering any adverse consequences. In such a case, we cannot argue that supply-side determinants are evidence of supply-side frictions. Hence, the first part of our analysis investigates the effect of underwriters on debt maturity choices at the issuance level. The second part of the analysis considers the effects of mismatch in maturity preferences between firms and underwriters on firm credit supply. 7 For our analysis, we focus on the lead underwriter of a corporate debt offering. We also exclude those issuances with multiple lead underwriters. We have run our analysis including these observations, and although the results are qualitatively similar, the meaning of a maturity mismatch is less clear. 9

10 II.A Underwriter relationships and maturity preferences Even though corporate debt is relatively arm s length compared to bank loans (Diamond, 1991b; Rajan, 1992), issuing firms have frequent interaction with financial intermediaries. Figure 3 investigates the frequency of interactions between firms and corporate debt markets with a specific underwriter and in general. The top figure presents the distribution of the time interval between long-term debt issuances for each firm with all underwriters. The bottom figure presents the distribution of the time interval between long-term debt issuances for a specific underwriter and firm. For the purposes of calculation, we treat multiple issuances on the same day as one issuance. The figures show that the majority of issuances in our sample occur with the same quarter, and 96% of issuances by firms in our sample are followed by another issuance within one year. While these numbers are influenced by some particularly frequent issuers, even averages at the firm-level show frequent issuance: the median firm in our sample issues every 326 days. If considering the average issuance time for firm-underwriter pairs, the median pair undertakes an issuance every 135 days. It is clear that firms are frequently raising money in the corporate debt market. To gain a better sense of the persistence of a given relationship between an underwriter and firm, in Figure 4 we calculate the hazard function for firms that issue debt. Specifically, for each issuance between the firm and its lead underwriter, we calculate the conditional probability that the next issuance by the firm uses the same lead underwriter. The firm-underwriter issuances need not be consecutive: provided they undertake multiple issuances over our sample, they will be included in the calculation for the appropriate number of issuances. Overall, there is a 28% probability that the firm will have two consecutive issuances that use the same underwriter. The relationship appears stronger at the beginning of a relationship. The probability of the firm follows its first issuance with a second from the same underwriter is about 48%. As seen in Figure 4, this probability decays over time, and after about 10 issuances settles at a probability of around 25%. As a point of reference, in the same figure we plot the conditional probability that the next bank loan for a firm uses the same lead bank. 8 We find that while banks have higher probability of a continued relationship, the differences are not large. On average a firm is likely to use the same bank for its next loan 36% of the time. This compares to 28% for the same debt underwriter. The frequency of issuances and the repeated interactions between firms and underwriters suggests that 8 For this analysis we use a sample of Compustat borrowers matched to banks in DealScan bank loan data. 10

11 these relationships in the corporate debt market are important. Insofar as firms do not appear to be issuing infrequently or using different underwriters with each issuance, it suggests that there are reasons for such relationships to exist. Such relationships are a necessary precondition for underwriters preferences to be transmitted to the firm. Next, we investigate why underwriters have preferences in terms of maturity structure. Corporate debt investors who make portfolio allocation decisions prefer certain maturities over others, along with other security characteristics such as credit quality, coupon rate, etc. We therefore consider the relationship between the average maturity of bond issues sold by underwriters and the average maturity of purchases by a subset of investors. We focus on insurance companies as the investor group because of their relatively significant presence in the market and data availability. Over our full sample period of , insurance companies held on average 29% of outstanding corporate and foreign bonds. 9 In the next section, we investigate the effect of the average maturity of an underwriter on the maturity decision of the firm, using insurance company bond purchases as an instrument. To show the relevance of our instrument, we present the associated first-stage regression: Log(Avg. Maturity), UW i jt = γ t + δ i + α j + κ 1 Log(Avg. Maturity), Investor i jt + κ 2 Log(Avg. Maturity), Firm i jt + κ 3 Firm Characteristics it + ε i jt. (1) Log(Avg. Maturity), UW is the log-transform of the dollar-weighted average maturity of all bond issuances by a specific underwriter j in a given calendar year t. It also varies by borrowing firm i as any of the firm s own issuances with the underwriter are purposefully excluded from the calculation. Log(Avg. Maturity), Investor is calculated using the dollar-weighted average maturity of bond purchases from that underwriter by life insurance companies and property-casualty insurance companies. In this first-stage, we also exclude any insurer purchases of the firm s bond issuance from the calculation of the investor s average maturity. To focus on insurance purchases that are more likely purchased in or shortly after the initial offer and therefore indicative of the relation between the underwriter and investor, we restrict the set of purchases to bonds issued by the underwriter in the past year. We include year fixed effects (or firm-year fixed effects) in all 9 See Figure 2. This amount is calculated by combining the holdings of property/casualty and life insurance companies in the Fed Flow of Funds data. 11

12 specifications and underwriter fixed effects in some specifications to remove any macroeconomic trends in average debt maturities and underwriter-specific differences in maturities, respectively. Column 1 of Table II presents the relation between life insurance and property-casualty insurance companies purchases on the underwriter s average maturities. We find that for a 1% increase in the average maturity of purchases by life insurance companies, there is a 0.3% increase in the average maturity of new issuances by that underwriter. Similarly, a 1% increase in the average maturity of purchases by propertycasualty insurance companies is associated with a 0.3% increase in the average maturity of new issuances by that underwriter. This specification includes year fixed effects, which capture any aggregate changes in bond maturities at that time. The remaining columns of Table II runs a similar specifications, but with different applications of fixed effects. Across all specifications, the instruments load positively and are typically significant. As these columns include versions of firm, underwriter, and year fixed effects, this identification is driven by groups of insurers buying specific types of bonds from underwriters at a particular time, and not because of any particular underwriter maturity specialization. Overall, insurers bond purchase activity is a relevant factor in the average maturity that underwriters issue. II.B Supply-side effects on issuance maturity This section attempts to isolate the impact of the maturity preferences of capital suppliers on the maturity of new debt issuances. Given our evidence of relationships between firms and underwriters, we expect that the maturity preferences of underwriters (and bond investors behind them) should be reflected in new issuances. However, as discussed in Section I.A, we in general observe the equilibrium outcome of a firm s and underwriter s different preferences. Separating supply-level preferences from a firm s demand for a certain maturity structure is therefore important. Our identification strategy is based on exogenous changes in the firm s debt maturity structure. Any maturity changes driven by the firm s characteristics are by definition endogenous. Changes in investor preferences, and the impact of such changes on the debt maturity of firms, is the identifying dynamic in our setup. We first consider whether observed investor and underwriter preferences affect the maturity choice for 12

13 individual issues. We run the following specification: Log(Maturity), Issuance i jt = γ t + δ i + α j + β 1 Log(Avg. Maturity), UW i jt + β 2 Log(Avg. Maturity), Firm it 1 + β 3 Firm Characteristics it 1 + ε i jt. (2) Log(Maturity), Issuance is the log of the maturity (in years) for the debt issue of firm i using underwriter j in year t. The unit of observation in this panel is a firm s specific debt issuance. Log(Avg Maturity), UW is the average maturity for the underwriter in that year. The average underwriter maturity technically varies by firm as we exclude any issuances from the firm with that underwriter from the calculation. Log(Avg Maturity), Firm is the average maturity of the firm s existing debt structure in the most recent year before the issuance. We also include the firm s leverage, market-to-book ratio, profitability, and size in the most recent year before the issuance as additional controls. These controls are meant to capture demand factors that are separate from the firm s existing debt structure but may still affect the firm s maturity choice. All specifications include year (or firm-year) fixed effects. We cluster errors by both firm and underwriter to ensure we account for correlation across error terms along these two dimensions. Column 1 of Table III provides an ordinary least squares estimate of the impact of underwriter and firm maturity preferences on new issuances. In columns 2 6, underwriter preferences are instrumented using the maturity preferences of insurance companies. The first stage of the instrumental variables approach for underwriter maturity preferences is reported in Table II. This instrumentation, under the assumption that insurance purchases not involving the firm in question only affect the firm s issuance maturity through the underwriters preferences, removes variation in underwriter maturity coming from the firm s own demand. Column 1 reports that firms that borrow from underwriters with a 1% higher average maturity issue debt at 0.2% higher maturity. The coefficient estimate for the firm s existing debt maturity also suggests that firms with higher average debt maturities, compared to other firms, continue to issue debt at higher maturities. It appears that firms attempt to maintain a relatively stable maturity structure. Column 2 conducts an IV analysis, the framework of which is discussed in Section I.A. We find that when we instrument the underwriter s maturity with those of its investors, the impact of the underwriter s maturity on the firm s debt maturity is stronger. Columns 3 and 4 include firm fixed effects. Within firm, we find that firms issue shorter maturity debt 13

14 if their current overall debt maturity is longer than their sample average. This finding again suggests that firms attempt to maintain a stable debt maturity structure. Nevertheless, column 3 still finds a significant impact of the underwriter s average debt maturity on the firm s new issuance maturity. One concern is that different underwriters have different debt maturity ranges in which they specialize. If firms then switch between these underwriters as needed, we would still find that the underwriter s average maturity matters, but for a different reason. To check if this is the case, in column 4 we include underwriter fixed effects which absorb this type of heterogeneity across underwriters. We find that the impact of the underwriter s maturity on the firm s debt maturity now falls by about half. Yet, column 4 shows that for a 1% increase in the average issuance maturity of an underwriter, a firm issues new debt with a statistically-significant 0.5% higher maturity. This supply-side effect is more than twice as strong as the firm s response to its existing debt structure (0.2%). Column 4 shows that while some of the underwriter s maturity effect is attributable to persistent differences across underwriters, a significant portion of the effect is driven by variation in maturity decisions for specific underwriters. A related concern may be that firm-underwriter relationship characteristics other than the underwriter s maturity preference are driving our results. For example, a firm s has a designated underwriter for issuances of a specific maturity or maturity range. Therefore, column 5 replaces the separate firm and underwriter fixed effects with firm-underwriter pair fixed effects. Now the identification is driven by variation in the underwriter s average debt maturity within the observations that relate to that firm-underwriter pair. We find the magnitude of the impact of the underwriter s maturity preference on the firm s debt issuance maturity remains similar. The firm s own past debt maturity has a similar impact in terms of magnitude, but loses statistical significance in this specification. Columns 3 5 include firm or firm-underwriter fixed effects and important firm-level controls. Despite this, a concern may be that unobservable or omitted firm-level characteristics that change over time are affecting the choice of the firm s debt maturity and possibly the choice of the underwriter for the issuance. Thus, it is not supply-side factors but omitted demand-side characteristics that are driving our findings. To address these concerns, we include firm-year fixed effects in column 6 (similar to Khwaja and Mian, 2008; Lin and Paravisini, 2013). If dynamic firm characteristics determine the choice of underwriters and debt maturity, when firm-year fixed effects are included, the firm-level determinants of such a choice are 14

15 controlled for. In general, any omitted firm-level characteristics that affect the firm s debt maturity are removed. Identification in column 6 is therefore obtained from variation in underwriter preferences for a firm issuing through multiple underwriters within a year. The magnitude in this case declines by a third compared to column 5, but remains statistically significant. Here we find an average increase of 0.3% in issuance maturity for a 1% increase in the underwriter s maturity preference. The result suggests that underwriters have changing supply preferences beyond any aggregate effects and that we are capturing the effect of these supply-side factors on the firm s debt maturity. II.C Credit supply and the maturity preference mismatch The previous sections show that new issuances of firm debt depend on supply-side determinants. This section investigates how the firm s credit supply is affected if there is a mismatch between the firm s debt maturity preferences and those of the credit suppliers and namely the underwriters. If supply-side preferences for debt maturity constrain the firm s credit supply, then we should find that firms with a lesser mismatch will receive more credit supply. Figure 5 plots the relationship between the offering amount of a bond issuance (in mil. USD) as a function of the firm s deviation in debt maturity from the underwriter s average issuance maturity in a given year. In the figure, we find that the offering amount declines as the firm s maturity deviation increases. This effect occurs for both positive and negative maturity deviations, and is most pronounced for deviations between 5 and 10 years. Presumably there is a cluster of investors interested in maturities around the underwriter s average maturity, which mutes the effect at smaller deviations. While suggestive evidence, this local polynomial regression does not control for various other factors that could affect the offering amount for the firm. To more carefully analyze this effect, we estimate the following equation for firm i in calendar year t that uses underwriter j: Log(Amount) i jt = γ t + δ i + α j + κ 1 Maturity Mismatch i jt + κ 2 Log(Avg. Maturity), Firm it 1 + κ 3 Firm Characteristics it 1 + ε i jt, (3) where Maturity Mismatch is the absolute value of Log(Maturity), Issuance Log(Avg. Maturity), UW. Mo- 15

16 tivated by the evidence in Figure 5, we choose to focus on the absolute value of this difference. The unit of observation is again at the issuance level. Table IV reports the results of the estimation. Column 1 presents the baseline estimation with firm-level controls. Here we include yearly fixed effects for each possible maturity to ensure that our result is not driven by variation in the Treasury yield curve or other aggregate conditions may correlate with deviations from the underwriter s average issuance maturity. The estimated impact of the difference in the average maturity for the underwriter and the maturity of the firm s issuance is negative but statistically insignificant in the OLS specification. This result is not surprising as the firm s decision to deviate from the underwriter s average maturity is an equilibrium outcome. The firm is weighing the relative benefits and costs of deviation, accounting for the underwriter s supply preferences, and making its decision given these frictions. To better identify how the firm s deviation is determined by variation in the underwriter s supply preferences, we again use an instrumental variables approach. In this case, we construct analogous variables to our Maturity Mismatch variable but use the insurance companies average maturity purchases from that underwriter in place of the underwriter s overall average maturity. In the IV specification of column 2, we find that for a 1% increase in maturity mismatch between an underwriter s average issuance and a firm s debt issuance maturity, the amount financed by debt markets declines by a statistically and economically significant 1.36%. Column 3 includes firm fixed effects and finds similar results. To address concerns that static differences across underwriters maturity preferences are driving the results, column 4 includes underwriter fixed effects. The identification in this case is obtained from variation in a given underwriter s maturity preferences, controlling for firm-specific determinants and maturity-specific aggregate effects. Column 4 reports a similar elasticity between the maturity mismatch and firm financing as earlier columns: a 1% increase in the mismatch leads to a 1% decrease in the amount financed through the debt issuance. Column 5 includes firm-underwriter fixed effects to address concerns that firm-underwriter relationship characteristics other than the underwriter s general maturity preferences are driving our results. The estimates have the same economic and statistical significance as column 4. Column 6 includes firm-year fixed effects to most fully control for demand-side effects. Column 6 shows that for a 1% increase in the maturity mismatch between underwriters and firms, the issuance amount declines by 0.9%. This result provides 16

17 confidence that the reduction in the amount of debt raised is driven by the firm s choice to deviate from the underwriter s average maturity, and not some other firm-level determinant which is potentially correlated with the deviation. III Firm outcomes Even if individual issuances are affected by supply-side frictions, firms may obtain financing from multiple sources over time to ameliorate the impact of supply-side frictions on firm outcomes. To understand the net effect of these frictions, and potential actions taken by a firm as a response, this section investigates the impact of supply-side frictions on firm outcomes. The specific channel of firm outcomes we consider is how reductions in the offering amount of individual issuances affect the firm s overall liabilities. In particular, we first investigate if the firm s overall debt growth is affected. As a second related effect, we see if these reductions impact the firm s overall asset growth. III.A Debt issuance The top panel of Figure 6 plots the relationship between the firm s maturity deviation from the underwriter s average maturity and the subsequent debt growth. Similar to Figure 5, we find that both positive and negative deviations from the average result in lower average debt growth for the firm. This univariate evidence suggests that firms on average do not fully replace the lower issuance-specific capital with multiple issuances or other sources of debt, such as bank financing. To better control for the myriad of factors that could influence debt growth, we estimate the following equation: Debt Growth it = γ t + δ i + α j + θ 1 Maturity Mismatch i jt + θ 2 Log(Avg. Maturity), Firm it 1 + θ 3 Firm Characteristics it 1 + ε i jt. (4) Similar to the specifications in Section II.C, we focus on the absolute value of the difference in maturities between the firm s issuance and the average maturity of the underwriter. For our instrumental variable specifications, we use the same instrumental variables as in Section II.C. To be included in this panel, 17

18 firms need to have a debt issuance in the current year. Apart from the deviation in maturity, all other control variables are from the prior year. 10 Table V reports the results. Column 1 presents the baseline OLS estimation with firm-level controls and maturity-year fixed effects. Maturity-year fixed effects help address concerns that the yield-curve or other aggregate economic conditions are driving firms debt maturity choices. The estimated impact of the difference in the average maturity for an underwriter and the maturity of the specific issuance by the firm is negative and statistically significant in the OLS specification. However, as discussed in prior sections, we believe the IV approach allows us to better identify the effect of the maturity mismatch on firm s outcomes. In the IV specification of column 2, we find that for a 1% increase in the maturity mismatch between an underwriter s average issuance and a firm s debt issuance maturity, firm borrowing as represented by debt growth is lower by 0.1%, compared to other firms. Column 3 includes firm fixed effects to address concerns regarding heterogeneity across firms not captured by firm-level controls. In presence of firm fixed effects, sensitivity of firm level debt growth to supplierborrower maturity mismatch falls by half but still remains significant: a 1% increase in mismatch leads to a 0.05% decrease in the growth rate of debt. Column 4 includes underwriter fixed effects to address concerns that certain underwriters may systematically encourage different growth rates for debt. The estimated sensitivity of firm-level debt growth to maturity mismatch is similar in magnitude to the estimate in column 3. Finally, column 5 includes firm-underwriter fixed effects to eliminate static firm-underwriter relationship characteristics other than maturity preferences. We find that for a 1% increase in the maturity mismatch between underwriters and firms, debt growth declines by 0.034% in this specification. In sum, mismatches in maturity preferences have a significant effect on the firm s debt financing, suggesting that meaningful supply-side frictions exist in corporate debt markets. These firms are not able to fully recoup the capital lost when they choose to deviate from the underwriter s preferred maturity. 10 The unit of observation in this panel is at the issuance level. Although the dependent variable is at a firm level, we do not in general suffer from redundant observations. With the inclusion of firm-underwriter fixed effects, the dependent variable is interpreted as this year s deviation from the firm s average debt growth over the course of the firm s relationship with that specific underwriter. The same firm-year observation will be transformed differently because of differences in underwriter relationships. The application of maturity-year fixed effects further transforms the dependent variable by accounting for aggregate shocks to certain issuance maturities that could affect debt growth. 18

19 III.B Asset growth Even if firm debt financing is affected by supply-side frictions as documented above, it is possible that the firm is able to substitute away to other types of financing, such as equity, or even adjust payout policy. Therefore, this section investigates firm asset growth in response to the underwriter maturity mismatch. A firm should make optimal choices across the type of financing and specific intermediaries. Despite this, if we find that asset growth is negatively affected by the maturity mismatch, then one can conclude that the frictions are hard to eliminate by switching to other sources of financing. The bottom panel of Figure 6 reports the relationship between firm asset growth and the mismatch in maturity preferences of underwriters and firms. Rather than a flat relationship, we find the similar humpshaped relationship as in the case of firm debt growth. Table VI presents the results of the more formal estimation of the determinants of firm-level asset growth. Column 1 reports that a 1% increase in the maturity mismatch between underwriters and firms leads to a 0.038% lower growth rate for the firm s assets. Column 2, which instruments the maturity preference mismatch with investor preferences, finds a 0.056% reduction in asset growth. The results remain similar in columns 3 and 4 as well. Column 5, which is our most exhaustive specification, finds a 0.023% reduction in asset growth for a 1% increase in the maturity mismatch. To obtain some insights into the impact of supply-side frictions due to the underwriter-firm maturity preference mismatch on real outcomes, we compare the magnitude of the estimate in column 5 of Table VI with the estimate in column 5 of Table V. As the average book leverage in our sample is 38%, for a 1% increase in maturity mismatch, debt growth explains 0.013% slower asset growth ( ). If the firms were able to completely substitute to other sources of financing, then asset growth could even be insensitive to maturity preference mismatch. However, given that other types of financing are relatively costly compared to debt, it is reasonable that once the firm is unable to obtain optimal amounts of debt financing, other sources of financing are also reduced in some proportion. For example, if a firm wants to maintain a certain leverage ratio, the amount of debt growth would limit its equity growth. As 38% of the liabilities are debt, comparing the point estimates of column 5 in the two tables suggests that the remaining 62% of liabilities are growing at % slower rate ( )/0.62. Thus, for every unit reduction in debt financing, other 19

20 sources of financing also grow at a 50% (0.0168/0.0336) lower rate: suggesting a spill-over effect of the supply-side frictions. Alternative sources of financing such as equity, bank financing, or other short-term debt financing are thus unable to eliminate the supply-side frictions of the corporate debt market. IV IV.A Mechanism validation and aggregate implications Firm constraints In our sample, higher credit quality firms dominate in terms of the number of issuances. Despite the sample being mostly investment grade in terms of credit quality, it is notable that we find evidence of supply-side frictions. This suggests that the supply-side frictions must be even stronger in speculative-grade corporate debt. One validation test of our mechanism is that the supply-side frictions should be stronger in lower credit quality firms, even within investment-grade firms. As there is less overall demand for lower-rated debt, these firms are more beholden to the preferences of the investors who work with their particular underwriter. Focusing on the offering amounts (like in Section II.C), we divide our data sample into issuances by firms with a AAA credit rating and issuances by firms with an investment-grade credit rating of A+ or lower. Table VII reports the results for the two subsets of observations. 11 Issuances with a AA credit rating are excluded to provide an interval between the two groups. As the effect on AA-rated firms falls between the two groups, their inclusion does not change our results. It is notable that despite column 1 having a larger set of firms, it has fewer observations than column 2. This is because a small number of firms with AAA credit ratings are the most frequent issuers in our sample. Column 1 reports that for a 1% increase in the maturity preference mismatch between underwriters and firms, issuance amounts decline by 1.7% for investment-grade firms with ratings of A+ or lower. This comparison is within the group of A+ or lower-rated firms as the maturity mismatch changes. In contrast, column 2 reports no significant decline in issuance amounts for firms with a AAA credit rating because of a maturity preference mismatch. This class of firms is not affected by undewriter preferences, consistent with prior evidence of a significant demand of AAA-rated securities in general (Benmelech and 11 To include more firms, we use the specification similar to column 4 of Table IV, which has separate firm and underwriter fixed effects. Using firm-underwriter fixed effects (similar to column 5 of Table IV) generates similar results but uses fewer observations. 20

21 Dlugosz, 2009; He, Qian, and Strahan, 2011; Griffin and Tang, 2012). Our results suggest that there is no evidence of supply-side frictions as measured by the maturity preference mismatch for this small group of firms. However, we find that as soon as credit quality falls even to A+ or BBB ratings, which are still investment grade, firms face significant frictions in the corporate debt market. Indeed, we find that the difference between the coefficients for the maturity preference mismatch between the two subsamples is significant at the 1% level. IV.B Underwriter relationships To further test the proposed mechanism, we consider if firms with multiple underwriter relationships face relatively fewer supply-side frictions. This is similar to the literature that shows that firms that have relationships with multiple banks face fewer supply-side frictions (Bharath, Dahiya, Saunders, and Srinivasan, 2011). Table VIII again considers the amount of capital raised, splitting the sample into issuances by firms which have a relationship with one underwriter and issuances by firms which have relationships with three or more lead underwriters in the past 5 years. 12 If underwriter preferences, not firm-specific characteristics, are the source of the friction as we argue, then firms with more underwriter relationships should face fewer supply-side frictions. Firms in this situation have more bargaining power with a given underwriter and can pressure one of the underwriters to source more capital despite the deviation in maturity. Column 1 reports that firms with one underwriter relationship are the main source of our results: for a 1% increase in the maturity mismatch, the amount issued declines by 2.3%. Firms with multiple underwriter relationships do not face such supply-side frictions, as shown in column 2. Taking the results of this section and the previous section together, they provide evidence in favor the financing frictions we argue are present in U.S. corporate debt markets. Namely, firms that have more demand for their debt (due to a high credit rating) and have more bargaining power with underwriters (due to multiple relationships) are able to avoid the maturity-mismatch related frictions we document. 12 Here we run a specification analogous to column 4 of Table IV. 21

22 IV.C Aggregate implications For most of the analysis in the paper, we have taken the fact that the firm has a preferred debt maturity as a given. One fundamental determinant of the firm s preferred debt maturity structure is the maturity of its assets. Asset maturity itself depends on firm and industry-specific production technologies, changes in the firm s competitive landscape, and other factors that affect its business operations. As a result, asset maturities will vary over time. Likewise, the maturity preferences of insurance companies, which are the investor group we consider, are driven by changes in consumer demographic characteristics, demand for certain insurance products, and the evolution of their existing fixed income portfolios. Because these determinants of the firm s preferred maturity and investor s preferred maturity are driven by such different forces, the mismatch in maturity preferences varies over time. In the top panel of Figure 7, we calculate the dollar-weighted average asset maturity of Compustat firms over the past four quarters and calculate the dollar-weighted average maturity of insurer s bond purchases also over the past four quarters. 13 As evidenced from the figure, there are periods where the insurers debt purchase activity closely matches aggregate asset maturity. Other periods, such as in recent years, there is a meaningful separation in asset maturities and investor purchase preferences. To give a sense of how meaningful these maturity mismatches are in the aggregate, we consider the following: from , the average growth rate of total debt securities in the non-financial corporate sector for all maturities in the U.S. was approximately 6.3% (see the bottom panel of Figure 7). In 2016, the total amount of outstanding debt securities was approximately $5.76 trillion. The average asset maturity of firms in our sample is 12.9 years. Here we use this number instead of the average firm debt maturity to get an estimate of what maturities the firm would prefer absent any debt-market frictions. 14 To capture the preference of investors, we use the average issuance maturity of 7.19 years for underwriters in our sample. 15 Using these two values, the average maturity preference mismatch between 13 Here we combine life insurers and P/C insurers using their respective purchase amounts. We also only focus on bonds which were purchased within one year of issuance to make it more comparable to other parts of our analysis. We use the period of due to relatively sparse insurer purchase data before This choice assumes that firms want to perfectly match the maturity of assets and the maturity of their debt liabilities. This will not be strictly true for two reasons: (i) maturities are really only meaningful as they relate to duration, and (ii) the equity component of a firm s liabilities will have its own duration. For these reasons, the preferred debt maturity will not be exactly the asset maturity, but will still be closely related. 15 This value is clearly affected by the demands of the issuing firms. To the extent that this moves the average away from the investors pure maturity preferences, our back-of-the-envelope calculations will understate the aggregate effects. 22

23 underwriters and firms in our sample is approximately six years, which translates to a 58.5% mismatch in the manner we measure it in this paper. 16 Using the estimated sensitivity in column 5 of Table V (0.0336), we estimate that the average mismatch suggests an average decline of 1.96% in the growth rate of debt due to the debt maturity preference mismatch ( % = 1.96%). In dollar terms, for the total debt liabilities of U.S. non-financial corporate sector in 2016, this back-of-the-envelope calculation translates to a $112 billion reduction in debt issuance ($5.76 trillion 1.96% =$112 billion). It is important to note that this number does not take into account the extensive margin of supply-frictions: debt may not be issued by firms because there is insufficient aggregate investor demand. Thus, despite firms having relatively strong access to U.S. capital markets, we uncover significant supply-side frictions in the corporate debt market. When firms maturity choices deviate from underwriter preferences, they have to sacrifice capital. We should emphasize that this is an important comparison, but not the only possible counterfactual. For example, firms which do not have any access to public debt markets due to their smaller scale and inability to cultivate relationships with underwriters are even more impacted. These firms which have an underwriter relationship benefit in comparison to these excluded firms, even if they are still raising less capital than if there were no frictions related to mismatches in maturity preferences. V Conclusion This paper shows that supply-side frictions between underwriters and firms have important implications for firm financing and firm outcomes. These findings show that corporate debt markets are not so dissimilar from bank and equity markets, in which the importance of supply-side frictions and relationships with intermediaries are well documented. The median firm in our sample issues new debt every 0.9 years. Corporate debt market supply-side frictions lead to firms being strongly influenced by their underwriter relationship: a firm s debt maturity is two times more sensitive to debt underwriter preferences than to the firm s own past maturity. These maturity effects are not innocuous. As the maturity mismatch between the firm s debt issuance and the underwriter s preferences increases, issuance amounts fall with almost unit sensitivity. The firm s debt 16 The calculation is the log difference between the firm s asset maturity and the underwriter s average maturity log(12.9) log(7.19) =

24 growth decreases by approximately 1.96% for the average mismatch between asset maturity and underwriter maturity of approximately six years observed in our data. Firms seek to lessen these frictions by choosing underwriters and other sources and types of financing. Yet, we find that the asset growth of the firm, which is an outcome determined in part by the firm s financing choices, is less in response to the maturity preference mismatch between firms and credit suppliers. Our results suggest that U.S. corporate debt markets, despite being large in size and scope, have significant frictions, at least for all but the highest-rated firms. Overall, our results underscore that through the specific channel of maturity preferences and potential mismatches, corporate debt investors and underwriters play a meaningful role in firm outcomes. 24

25 References Almeida, Heitor, Murillo Campello, Bruno Laranjeira, and Scott Weisbenner, 2012, Corporate debt maturity and the real effects of the 2007 credit crisis, Critical Finance Review 1, Barclay, Michael J, and Clifford W Smith, 1995, The maturity structure of corporate debt, Journal of Finance 50, Benmelech, Efraim, 2008, Asset salability and debt maturity: Evidence from nineteenth-century American railroads, Review of Financial Studies 22, Benmelech, Efraim, and Jennifer Dlugosz, 2009, The alchemy of CDO credit ratings, Journal of Monetary Economics 56, Carnegie-Rochester Conference Series on Public Policy: Distress in Credit Markets: Theory, Empirics, and Policy November 14-15, Berger, Allen N, Nathan H Miller, Mitchell A Petersen, Raghuram G Rajan, and Jeremy C Stein, 2005, Does function follow organizational form? Evidence from the lending practices of large and small banks, Journal of Financial Economics 76, Bharath, Sreedhar T., Sandeep Dahiya, Anthony Saunders, and Anand Srinivasan, 2011, Lending relationships and loan contract terms, Review of Financial Studies 24, Burch, Timothy R, Vikram Nanda, and Vincent Warther, 2005, Does it pay to be loyal? An empirical analysis of underwriting relationships and fees, Journal of Financial Economics 77, Chava, Sudheer, and Amiyatosh Purnanandam, 2011, The effect of banking crisis on bank-dependent borrowers, Journal of Financial Economics 99, Custódio, Cláudia, Miguel A. Ferreira, and Luís Laureano, 2013, Why are US firms using more short-term debt?, Journal of Financial Economics 108, Diamond, Douglas W., 1991a, Debt maturity structure and liquidity risk, Quarterly Journal of Economics 106, Diamond, Douglas W., 1991b, Monitoring and reputation: The choice between bank loans and directly placed debt, Journal of Political Economy 99, Diamond, Douglas W., and Zhiguo He, 2014, A theory of debt maturity: The long and short of debt overhang, Journal of Finance 69, Drucker, Steven, and Manju Puri, 2005, On the benefits of concurrent lending and underwriting, Journal of Finance 60,

26 Faulkender, Michael, and Mitchell A. Petersen, 2006, Does the source of capital affect capital structure?, Review of Financial Studies 19, Flannery, Mark J, 1986, Asymmetric information and risky debt maturity choice, Journal of Finance 41, Gande, Amar, Manju Puri, and Anthony Saunders, 1999, Bank entry, competition, and the market for corporate securities underwriting, Journal of Financial Economics 54, Gertner, Robert, and David Scharfstein, 1991, A theory of workouts and the effects of reorganization law, Journal of Finance 46, Gopalan, Radhakrishnan, Fenghua Song, and Vijay Yerramilli, 2014, Debt maturity structure and credit quality, Journal of Financial and Quantitative Analysis 49, Graham, John, Mark Leary, and Michael Roberts, 2014, How does government borrowing affect corporate financing and investment?, Working Paper National Bureau of Economic Research. Greenwood, Robin, Samuel Hanson, and Jeremy C Stein, 2010, A gap-filling theory of corporate debt maturity choice, Journal of Finance 65, Griffin, John M., and Dragon Yongjun Tang, 2012, Did subjectivity play a role in CDO credit ratings?, Journal of Finance 67, Guedes, Jose, and Tim Opler, 1996, The determinants of the maturity of corporate debt issues, Journal of Finance 51, Harford, Jarrad, Sandy Klasa, and William F. Maxwell, 2014, Refinancing risk and cash holdings, Journal of Finance 69, He, Jie, Jun Qian, and Philip E. Strahan, 2011, Credit ratings and the evolution of the mortgage-backed securities market, American Economic Review 101, Ivashina, Victoria, 2009, Asymmetric information effects on loan spreads, Journal of Financial Economics 92, Ivashina, Victoria, and David Scharfstein, 2010, Bank lending during the financial crisis of 2008, Journal of Financial Economics 97, The financial crisis: Lessons from corporate finance. James, Christopher, 1992, Relationship-specific assets and the pricing of underwriter services, Journal of Finance 47, Kashyap, Anil K., and Jeremy C. Stein, 2000, What do a million observations on banks say about the transmission of monetary policy?, American Economic Review 90,

27 Khwaja, Asim Ijaz, and Atif Mian, 2008, Tracing the impact of bank liquidity shocks: Evidence from an emerging market, American Economic Review 98, Leary, Mark T., 2009, Bank Loan Supply, Lender Choice, and Corporate Capital Structure, Journal of Finance 64, Lin, Huidan, and Daniel Paravisini, 2013, The effect of financing constraints on risk, Review of Finance 17, Murfin, Justin, 2012, The supply-side determinants of loan contract strictness, Journal of Finance 67, Myers, Stewart C., 1977, Determinants of corporate borrowing, Journal of Financial Economics 5, Myers, Stewart C., and Nicholas S. Majluf, 1984, Corporate financing and investment decisions when firms have information that investors do not have, Journal of Financial Economics 13, Peek, Joe, and Eric S. Rosengren, 2000, Collateral damage: Effects of the Japanese bank crisis on real activity in the United States, American Economic Review 90, Petersen, Mitchell A, and Raghuram G Rajan, 1994, The benefits of lending relationships: Evidence from small business data, Journal of Finance 49, Petersen, Mitchell A, and Raghuram G Rajan, 1995, The effect of credit market competition on lending relationships, Quarterly Journal of Economics 110, Rajan, Raghuram G., 1992, Insiders and outsiders: The choice between informed and arm s-length debt, Journal of Finance 47, Ross, Stephen A., 1977, The determination of financial structure: The incentive-signalling approach, Bell Journal of Economics 8, Schenone, Carola, 2004, The effect of banking relationships on the firm s IPO underpricing, Journal of Finance 59, Stohs, Mark Hoven, and David C. Mauer, 1996, The determinants of corporate debt maturity structure, Journal of Business 69, Yasuda, Ayako, 2005, Do bank relationships affect the firm s underwriter choice in the corporate-bond underwriting market?, Journal of Finance 60,

28 Amount (Firm) Amount (UW) Issuance amount E Firm preference Underwriter (UW) preference Figure 1: The issuance amount and debt maturity are determined together. Underwriters and firms agree on the issuance amount and maturity of issued debt based on their maturity preferences. The equilibrium outcome (point E) is observable in the data. To identify the impact of the maturity preference mismatch on the issuance amount, exogenous shifts in the maturity preferences of the underwriters are needed. 28

29 Holdings of Corporate Bonds Trillions USD Corporate Bonds Outstanding 1990q1 1995q1 2000q1 2005q1 2010q1 2015q1 Date Life Insurance P/C Insurance Figure 2: Amount of corporate bond debt outstanding and the amount of corporate bond debt held by life insurance and property-casualty (P/C) insurance companies. 29

30 Time between Issuances for Firm Percent Years Until Next Issuance Time between Issuances for Firm-Underwriter Pair Percent Years Until Next Issuance Figure 3: The top figure presents the distribution of the time interval between debt issuances for each firm with all underwriters. The bottom figure presents the distribution of the time interval between debt issuances for a specific underwriter and firm. Multiple issuances that occur on the same day are treated as a single issuance. 30

31 Conditional Probability of Next Issuance Firm-Underwriter Firm-Bank Number of Issuances for a Pair Figure 4: For corporate bond issuances, the conditional probability that a firm s next debt issuance will use the same lead underwriter as the current issuance as a function of the number of issuances between a firm and underwriter. For syndicated bank loans, the conditional probability that a firm s next loan facility will use the same lead bank as the current issuance as a function of the number of loans between a firm and bank. 95% confidence intervals presented for the conditional probability estimates. 31

32 Local polynomial smooth Offering Amount (Mil. USD) Deviation from Underwriter's Avg. Maturity (Years) 95% CI lpoly smooth kernel = epanechnikov, degree = 0, bandwidth = 1.9, pwidth = 3.8 Figure 5: Local polynomial regression of the offering amount of a bond issuance (in mil. USD) as a function of the firm s deviation in debt maturity from the underwriter s average issuance maturity in a given year. 32

33 Local polynomial smooth Debt Growth Deviation from Underwriter's Avg. Maturity (Years) 95% CI lpoly smooth kernel = epanechnikov, degree = 0, bandwidth = 1.9, pwidth = 3.8 Local polynomial smooth Asset Growth Deviation from Underwriter's Avg. Maturity (Years) 95% CI lpoly smooth kernel = epanechnikov, degree = 0, bandwidth = 1.9, pwidth = 3.8 Figure 6: Local polynomial regression of a firm s annual debt growth (top panel) and annual asset growth (bottom panel) as a function of the firm s deviation in debt maturity from the underwriter s average issuance maturity in a given year. 33

34 q1 2000q1 2005q1 2010q1 2015q1 Date Asset Maturity Average Bond Maturity Purchase Figure 7: The top panel plots the aggregate dollar-weighted firm asset maturity and dollar-weighted maturity of bonds purchased by insurance companies. The aggregate asset maturity and average bond maturity purchase are calculated using the average over the past four quarters. The calculation of the maturity for bond purchases focuses on life and property-casualty insurance companies and only includes bonds which are purchased within a year of initial offering. The bottom panel shows the annual growth rate of aggregate non-financial corporate debt in the U.S. 34