Maturity Puzzle

Maturity Clientele Effects in the Corporate Bond Market Alexander W. Butler, Xiang Gao, and Cihan Uzmanoglu*

July 17, 2018

Abstract

Traditional determinants of debt maturity fail to explain the declining trend in U.S. corporate bond maturities observed since 1970s. This paper shows that accounting for the declining share of insurance companies, who invest in long-term bonds, in the corporate bond market explains this maturity puzzle. The growth of mutual funds, who invest in shorter-term bonds, during the same period exacerbates the maturity contraction. The results from a Granger causality test, an instrumental variable approach, and a quasi-natural experiment establish a causal link between insurance company ownership and bond maturity. These findings illustrate how developments in financial institutions can affect the real economy.

Keywords: Bond Maturity; Supply of Credit; Demand for Bonds; Insurance Company Ownership; Mutual Fund Ownership. JEL Code: G20; G22; G23; G30; G32.

*

Alexander W. Butler, Rice University, [email protected]; Xiang Gao, Binghamton University, [email protected]; Cihan Uzmanoglu, Binghamton University, [email protected].

Maturity Clientele Effects in the Corporate Bond Market

Abstract

Traditional determinants of debt maturity fail to explain the declining trend in U.S. corporate bond maturities observed since 1970s. This paper shows that accounting for the declining share of insurance companies, who invest in long-term bonds, in the corporate bond market explains this maturity puzzle. The growth of mutual funds, who invest in shorter-term bonds, during the same period exacerbates the maturity contraction. The results from a Granger causality test, an instrumental variable approach, and a quasi-natural experiment establish a causal link between insurance company ownership and bond maturity. These findings illustrate how developments in financial institutions can affect the real economy.

1. Introduction From 1970 to 2015, the average maturity of new corporate bond issues in the U.S. has gradually declined from 20 years to 10 years, and the macroeconomic, firm, or bond level determinants of maturity fail to explain this declining trend (Custódio, Ferreira, and Laureano, 2013). Understanding the reasons for such maturity shortening is important as it can affect firms’ financing and investment decisions. For instance, Harford, Klasa, and Maxwell (2014) provide evidence suggesting that firms increase their precautionary cash holdings to mitigate the refinancing risk associated with this decrease in debt maturity. In this paper, we investigate whether accounting for maturity clientele effects in the corporate bond market reconciles the maturity puzzle. Insurance companies are the primary investors in the U.S. corporate bond market, yet their market share has declined from 40% in 1970s to its current level of 25%. 1 This decline is partly driven by the tax and retirement policies implemented in 1980s that led to the rise of mutual fund industry (Rydqvist, Spizman, and Strebulaev, 2014). Open-end mutual fund (hereafter, the term mutual fund refers to an open-end mutual fund) ownership in U.S. corporate bonds was less than 2% in 1970s, but, today, it is 15%. Thus, the composition of institutions supplying bond capital has evolved through time, which may have implications for the maturity of new bond issues. Insurance companies and mutual funds are heterogeneous with respect to their liability structures. In general, insurance companies have longer term liabilities than mutual funds who are subject to redemption risk. For this reason, insurance companies also invest in assets with longer maturities when compared to mutual funds as an asset-liability management strategy (e.g., Mayer 1977; Diamond, 1991; Stohs and Mauer, 1996). With the declining market share of insurance

1

According to the U.S. flow of funds data published by Federal Reserve.

2

companies and the increasing market share of mutual funds in the U.S. corporate bond market, therefore, the demand for long term bonds (or the supply of long term capital) has also decreased since 1970s. We investigate whether this diminishing appetite for longer term bonds accounts for the unexplained declining trend in bond maturities. We study a comprehensive sample of over 19,000 new bonds issued by public U.S. industrial firms between 1975 and 2015, and observe that, after controlling for its known determinants, bond maturity declines by an average of 1.34 months each year, totaling 4.58 years during the sample period. We then construct a measure of insurance company share in the corporate bond market using the U.S. flow of funds data and find that it is significantly positively related to bond maturity. More importantly, the declining trend becomes insignificant after controlling for the insurer market share in bond maturity regressions. We also find that the growth of mutual fund investments in corporate bonds exacerbates the maturity contraction, but it does not explain the trend. These findings, which are robust to alternative specifications and controlling for firm fixed effects, suggest that the maturity puzzle traces to the diminishing share of insurance companies in the corporate bond market. We next construct bond-level ownership data between 1999 and 2015 to further examine the maturity clientele effects in the bond market. We find that insurance company ownership in bonds is associated with longer bond maturities while mutual fund ownership in bonds is associated with shorter bond maturities. This relation between bond ownership structure and maturity is economically meaningful: a 10-percentage point increase in insurance company (mutual fund) ownership translates into a 10% (–6%) change in the maturity of bonds. These findings suggest that the evolution of maturity preferences in the bond market may indeed influence the maturity of bonds.

3

We also investigate whether the declining trend in maturities is more pronounced for corporate bonds that insurance companies prefer. We model such preferences by estimating the determinants of insurance company ownership. The regression estimates suggest that insurance company ownership is higher for non-convertible and fixed coupon bonds issued by higher-quality firms. Using these estimates, we then predict insurance company ownership for the entire sample of bonds, and then classify bonds into high and low predicted insurance company ownership groups. We find that the declining trend in maturities is only visible among the subsample of bonds that face high demand from insurance companies. Hence, insurance company investments appear to be driving the trend. As an additional test, we examine the relation between the types of insurance companies and bond maturity. If the maturity preferences of insurance companies are driving our findings, then, within the insurance sector, the declining trend in bond maturities should be more closely associated with the market share of life insurance companies, which tend to have longer term liabilities compared to property & casualty insurance companies (e.g., Fan, Titman, and Twite, 2012; Becker and Ivashina, 2015). We find results consistent with this prediction. The relation we observe between ownership structure and bond maturity is jointly determined based on the preferences of both issuers and investors. So, a primary concern in our study is the reverse causality of debt maturity and the insurance company presence, in that, firms’ decision to issue shorter term bonds could lead to the declining insurance company ownership. We implement three tests to address this issue: a Granger causality test, an instrumental variable regression, and a natural experiment. To implement a Granger causality test, we first compute the average maturity of new bond issues in each quarter between 1975 and 2005. We then match this quarterly series of average

4

maturities with quarterly market share of insurance companies. Using this dataset, we find that lagged insurer market share is a positive and significant determinant of maturity in the subsequent quarter, but the lagged maturity is insignificant in predicting insurer market share. These findings suggest that insurer market share Granger-causes bond maturity. Next, we use the physical distance between bond issuers and insurance companies as an instrument for insurance company ownership when estimating the maturity of new bond issues. We find that, conditional on investing, insurance companies that are in closer distance to an issuer invest significantly more in its bonds, and we argue that this distance is unlikely to be related to bond maturity except through its correlation with insurance company ownership. Hence, the physical distance variable is likely to be a valid instrument for insurance company ownership in bonds. The results from the instrumental variable regression confirms that insurance company ownership is positively associated with bond maturity. The coefficient estimates from this regression indicate that a 10-percentage point decline in insurance company ownership leads to approximately a 30% decline in bond maturity. Therefore, it is plausible that the decline in insurance company share from 40% in 1970s to 25% in 2000s may have caused the 50% decline in bond maturities observed during the same period. Finally, we utilize natural disasters as a quasi-natural experiment to examine the direction of the relation between insurance company ownership and bond maturity. Massa and Zhang (2011) show that insurance companies exposed to Hurricane Katrina liquidated their positions in corporate bonds, and that the effects of this demand shock lasted for several months. We capitalize on these findings and use natural disasters that result in large insurable losses as a shock to demand for bonds from insurance companies to examine its impact on bond maturity. We find that, on average, insurance company ownership is significantly lower following natural disasters,

5

validating the demand shock argument. We then test whether bonds issued in the quarter after natural disasters have shorter maturities, and find that they do. We also find that firms offer a higher yield to issue longer term bonds during the natural disaster periods, further providing evidence of the maturity clientele effects in the bond market. Overall, the results from both this natural experiment and the instrumental variable regression show that higher insurance company ownership in bonds leads to longer bond maturities. These findings provide evidence that a declining presence of insurance companies in the corporate bond market can result in shorter bond maturities. Our findings contribute to the literature in three ways. First, they resolve the maturity puzzle by explaining the declining trend in the maturity of U.S. corporate bonds. Second, they show that the institutional environment can influence the maturity of corporate debt, and hence, complements the earlier findings that the development of financial and legal institutions affects the capital structure of companies (e.g., Demirguc-Kunt and Maksimovic, 1998, 1999; Rajan and Zingales, 1998; Giannetti, 2003). Third, the literature shows that the supply and uncertainty of credit can have real consequences for corporations (e.g., Ellul, Jotikasthira, and Lundblad, 2011; Massa, Yasuda, and Zhang, 2013). Our evidence adds to this stream of literature evidence that the type of institutions supplying public debt is an important determinant of corporate debt maturity. The rest of the paper is organized as follows. Section 2 describes the sample and variable definitions. Section 3 explains the empirical setting and presents the results. Finally, Section 4 concludes with a summary of findings. 2. Data, Sample Selection, and Variables We construct our sample by combining all U.S. dollar denominated corporate bonds issued by U.S. industrial firms from Mergent FISD and SDC New Issues databases. We obtain the list of

6

unique bond issues from these databases using the CUSIP number of bonds. When a CUSIP is not available, we identify unique bonds based on their offering date, maturity date, offering amount, and coupon rate information. Our final sample includes 19,101 bonds issued by 2,988 firms between 1975 and 2015 with available information in CRSP and Compustat databases. Although the coverage in the merged bond database starts in 1970, our sample period starts in 1975 because the small number of bond issues (fewer than 10 bonds in each year) that satisfy the screening criteria during the beginning years prevents a meaningful comparison of average bond maturities through time. We compute Bond Maturity as the number of years between a bond’s maturity and issue dates. Table 1 reports the summary statistics of sample bond maturities by years. Each year, on average, there are 466 new bonds issued by 237 firms in our sample. The mean (median) of bond maturity is 21 years (25 years) in 1975, but it declines to 10 years (7 years) in 2015. The mean and median maturities during the entire sample period are 11 years and 10 years, respectively. Figure 1 plots the average Log(Bond Maturity) over time, and illustrates that the declining trend is visually evident during the analysis period and more pronounced during the 1975–1995 period when compared to the post-1995 period. To estimate the aggregate shares of insurance companies and mutual funds in the U.S. corporate bond market, we use the quarterly data from the U.S. flow of funds accounts published by the Federal Reserve. Accordingly, we define Insurer Market Share (Mutual Fund Market Share) as the amount of insurance company (mutual fund) ownership in U.S. corporate bonds and foreign entity bonds—issued through U.S. dealers and purchased by U.S. residents—divided by their total outstanding amount. The ownership data include both U.S. corporate and foreign entity bonds as they are reported jointly. Nevertheless, foreign entity bonds make up a small portion of the

7

institutional investor holdings (Becker and Ivashina, 2015), and hence, we interpret the ownership ratios as the shares of institutions in the U.S. corporate bond market. Appendix A provides more details on these variables, including the source codes of the items used. Table 1 reports the summary statistics of the insurance company and mutual fund market shares by years. Reported in each year are the mean of their quarterly values in that given year. The market share of insurance companies varies substantially through time. The average insurance company market share during the sample period is 32% with its highest value of 41% observed in 1979 and the lowest value of 19% observed in 2008 and 2009. As Rydqvist, Spizman, and Strebulaev (2014) show, this decline in insurance company presence is partly driven by tax and retirement policies adopted in early 1980s that resulted in the growth of mutual fund industry. Although the mutual fund market share is less than 2% in 1975, it gradually rises and reaches to 15% in 2015. The average market share of mutual funds during the analysis period is 6%. Figure 1 graphs the average insurance company and mutual fund market share by years along with the average Log(Bond Maturity). It shows that, in the absence of any control variables, insurance company (mutual fund) presence is positively (negatively) associated with bond maturity. Following the existing literature (e.g., Barclay and Smith, 1995; Guedes and Opler, 1996; Greenwood, Hanson, and Stein, 2010; Custódio, Ferreira, and Laureano, 2013), we construct an extensive list of firm, bond, and macro level variables as determinants of bond maturity. Firm level variables are Market Value of Equity, Total Debt/Total Assets, Net Income/Total Assets, Tangibility, Market-to-Book Value of Assets, Stock Return, Industry Dummies, and Firm IPO Decade Dummies, and bond level variables are Callable Dummy, Floating Dummy, Convertible Dummy, Puttable Dummy, Sinking Fund Dummy, Global Dummy, and Credit Rating Dummies. We consider four macro-level control variables in our analyses: Term Spread, Default Spread, Real Short-Term Rate,

8

and Share of Long-Term Government Debt. Appendix A provides a detailed description of these variables. Table 2 presents the summary statistics of the control variables in our study. The average issuer in our sample has a market cap of $17 billion, Market-to-Book ratio of 1.9, net income to assets ratio of 0.3%, leverage ratio of 34.8%, asset tangibility ratio of 0.45, and an average monthly stock return of 6.4% that is measured during the 3-month period before the bond offering date. About one half of the issuers in our sample went public before 1970, and less than 10% of them went public during the post-2000 period, and they are evenly distributed across industries. In our sample, we observe bonds that are callable (62%), convertible (17%), global issues (11%), puttable (7%), and floating rate (6%). Bonds with a sinking fund provision account for 5% of the sample and the average bond offering amount is $0.3 billion. About 43% of our sample bonds are investment-grade, 26% are speculative-grade, and 32% are not rated by Moody’s, S&P, or Fitch at issuance. During the analysis period, the mean values of the share of long-term government debt, real short-term rate, term spread, and default spread are 7.4%, 2.9%, 1.5%, and 1%, respectively.2 We also construct bond-level ownerships of insurance companies and mutual funds at the time of bond issuance. Bloomberg compiles the bond holding information of institutional investors from the 13F, Schedule D, 10-K, Form 990, and Form 5500 filings since 1998 at a quarterly frequency.3 Using this database, we define Insurer Ownership (Mutual Fund Ownership) as the

2

Figure 2 reports the distribution of bond characteristics through years, and shows that bond features and maturity do not appear to follow the same trend. 3

SEC requires all institutional investment managers that exercise investment discretion over $100 million to report its holdings on Form 13f. National Association of Insurance Commission (NAIC) requires all U.S. insurance companies to file Schedule D to reveal their holdings. Form 990 is a document filled with IRS by certain federally tax-exempt organizations. Form 5500 is filed with the Department of Labor by the sponsor of any employee benefit plans subject to Employee Retirement Income Security Act (EIRSA). Appendix B provides an example of the ownership information for a sample bond.

9

amount held by insurance companies (mutual funds) measured at the end of the issuance quarter divided by the bond’s issue amount. We construct these ownership variables since 1999 to eliminate a potential coverage bias in the database inception year, and also perform several quality control checks (e.g., total reported institutional holdings should be less than or equal to the offering amount) to ensure data integrity. Our final sample with bond-level ownership data includes 7,337 bonds with non-missing information. The aggregate amount of insurance company and mutual fund holdings at bond issuance is $1.08 trillion, which accounts for about 30% of the total offering amount ($3.70 trillion). This level of ownership coverage in bonds is comparable to that reported in the literature.4 Table 2 reports that the mean of Insurer Ownership is 15.1% and that of Mutual Fund Ownership is 12.5% in new bonds issued during the 1999-2015 period. The percentage of insurance company ownership in bonds is less than that reported by the U.S. flow of funds. This is partly because the flow of funds data covers all outstanding bonds, but the ownership data is only for the new bond issues. In order for the average ownership of insurance companies in outstanding bonds to decline through time, their ownership in new bond issues should decline even more. The fact that this effect is more pronounced for insurance companies than for mutual funds is consistent with insurance companies investing in bonds with longer maturities when compared to mutual funds. 3. Analyses of Bond Maturity

4

For instance, Becker and Ivashina (2015) collect the bond ownership information from Lipper eMaxx database and report that the total ownership of insurance companies, mutual funds, and pension funds accounts for 33.7% of the total face value of bonds in 2010. Massa, Yasuda, and Zhang (2013) and Dass and Massa (2014) also report similar ownership statistics using the same database. Lipper eMaxx compiles bond ownership information based on regulatory disclosure to the National Association of Insurance Commissioners (NAIC) for insurance companies and to the Securities and Exchange Commission (SEC) for mutual funds, which are similar to the information sources of Bloomberg.

10

In this section, we investigate whether the declining trend in bond maturities is associated with maturity clientele effects driven by changes in the composition of investors in the corporate bond market. 3.1. Bond Maturity and the Market Shares of Insurance Companies and Mutual Funds We start our analysis by running the following regression that estimates the declining trend in bond maturities: Log ( Bond Maturityij )    Wjt'   Zi'  X t'  Trendt   ij ,

(1)

where Bond Maturityij is the maturity of bond i issued by firm j, α is the intercept, Wjt, Zi, and Xt represent firm, bond, and macro level control variables, respectively, Trend is the difference between the year of bond issuance and the year when our sample period starts (1975), and εij is the error term. The earlier section and Appendix A provide the list and definitions of these control variables. In this regression framework, the coefficient on Trend (λ) provides an estimate of the time-series trend in bond maturities. We also introduce firm fixed effects to this regression model to examine the within firm trend in bond maturities. Column (1) in Table 3 reports the coefficient estimates from the bond maturity regression. The sign and significance of the coefficient estimates on the control variables are broadly consistent with their economic meanings. Larger firms and firms with better stock performance, higher tangibility, lower leverage, and lower book-to-market ratios issue longer maturity bonds. Bond characteristics are also important determinants of maturity: callable, fixed coupon rate, nonconvertible, puttable, and sinking fund bonds have longer maturity. At the macro level, bond maturity increases when the Treasury issues fewer long term bonds and the default spread declines. Controlling for the known determinants of bond maturity, we find that the coefficient estimate on Trend (multiplied by 100) is –1.12 and significant, indicating a 1.12% decline in bond

11

maturity during each calendar year. This is equivalent to an annual decline in maturity of 1.5 months for the average bond in our sample, and a total unexplained decline of 6 years in bond maturity during the sample period. Therefore, our comprehensive set of firm, bond, and macro level covariates are only able to explain about half of the 10-year decline in bond maturities observed since 1970s. Column (2) reports that the coefficient estimate on Trend maintains its sign and significance controlling for firm fixed effects, showing that the declining trend in bond maturities also persists within firms. Next, we introduce Insurer Market Share as an additional control variable to the maturity regressions and report the results from the OLS and fixed effects specifications in Columns (3) and (4) of Table 3, respectively. The coefficient estimate on Insurer Market Share is 1.9 and significant in both specifications, indicating that a 10-percentage point increase in Insurer Market Share is associated with a 21% increase in bond maturity. Hence, as predicted, insurance company presence in the bond market is positively related to bond maturity. More importantly, the coefficient estimate on Trend becomes nearly zero and insignificant after controlling for Insurer Market Share. Therefore, accounting for the share of insurance companies in the corporate bond market is sufficient to eliminate the declining trend in bond maturities observed since 1970s. The literature shows that this decline in the market share of insurance companies was partly driven by the growth of mutual fund investments in the corporate bond market (Rydqvist, Spizman, and Strebulaev, 2014). As mutual funds prefer investing in shorter maturity bonds when compared to insurance companies, we next investigate whether the increase in the mutual fund share in the bond market can explain the declining trend in bond maturities. Columns (5) and (6) in Table 3 report the results from OLS and fixed effects regressions of bond maturity, respectively, controlling for Mutual Fund Market Share as an additional right-hand side variable. We find that

12

the coefficient on Mutual Fund Market Share is positive and insignificant, and the coefficient on Trend is negative and significant.5 Therefore, controlling for the market share of mutual funds is not able to explain the declining trend. We then include both Insurer Market Share and Mutual Fund Market Share as additional control variables to the maturity regressions and report the results in Columns (7) and (8). We find that the coefficient estimate on Mutual Fund Market Share is now negative, consistent with mutual funds preferring shorter term bonds, albeit it is still insignificant. The coefficient on Trend is also insignificant, but that on Insurer Market Share is positive and significant. These findings suggest that, although the rise of the mutual fund industry appears to exacerbate the maturity contraction, the maturity puzzle traces to the declining share of insurance companies in the corporate bond market. Overall, the results of this section show that accounting for the share of insurance companies in the corporate bond market explains the declining trend in bond maturities observed since 1970s. In the next section, we explore the relation between bond maturity and investor preferences by using bond level ownership data. 3.2. Ownership Structure and Bond Maturity We find in the previous section that changes in the market shares of investors in the corporate bond market can influence bond maturity. To better understand such maturity clientele effects, we study the relation between investor preferences and bond maturity using bond level ownership data. As described in Section 2, we are able to reliably construct this ownership data in the quarter of bond issuance since 1999 for 7,337 bonds in our sample. Using this subsample, we

5

We later find in Section 3.5 that, consistent with mutual funds demanding shorter maturity bonds, the coefficient estimate on Mutual Fund Market Share is negative and significant when Trend is excluded as a control variable. This finding suggests that a possible multicollinearity problem biases the coefficient estimate on Mutual Fund Market Share in the baseline model.

13

estimate the baseline regression of Log(Bond Maturity) by controlling for bond level ownerships of insurance companies and mutual funds. Columns (1) and (2) in Table 4 report that the coefficient estimate on Insurer Ownership is positive and significant in both OLS and firm fixed effects specifications, indicating that bond maturity increases with insurance company ownership. The estimates from Columns (3) and (4) show that this relation is non-linear: the coefficient on Insurer Ownership is positive and that on its square term is negative, suggesting that bond maturity increases with insurance company ownership at a decreasing rate. These coefficient estimates indicate that a 10-percentage point increase in insurance company ownership results in a 10% increase in bond maturity, which is equivalent to 1.1 years for the average bond in our sample. Similar to Columns (1)–(4), Columns (5)–(8) report the results from regressions that study the influence of mutual fund ownership on bond maturity. We find that bond maturity is significantly negatively correlated with Mutual Fund Ownership and it decreases with Mutual Fund Ownership at a diminishing rate. The coefficient estimates reported in Column (8) indicate that a 10-percentage point increase in mutual fund ownership corresponds to a 6% decline in bond maturity, which translates into a 0.66-year decline in maturity for the average bond in our sample. Using bond level ownership data, our analyses in this section show that bond maturity is positively correlated with insurance company ownership and negatively correlated with mutual fund ownership. These findings strengthen our baseline finding that, in the aggregate, the declining share of insurance companies—together with the increasing share of mutual funds—in the corporate bond market may result in the declining trend in bond maturities. In the next section, we further explore how bond maturity varies with investor preferences. 3.3. Insurance Company Preferences and the Declining Trend in Bond Maturity

14

In this section, we investigate whether the declining trend in bond maturities varies predictably based on bond and insurance company characteristics that are related to the maturity preferences of insurance companies. If the decline in insurance company ownership drives the declining trend in bond maturities, this effect should be more pronounced among bonds whose primary investors are insurance companies. For instance, because they are subject to risk-based capital regulations, insurance companies invest in safer and non-convertible bonds, and hence, their declining market share should lead to a greater decline in maturity for these types of bonds. To investigate whether this is the case, we first predict bond-level insurance company ownership using the subsample of 7,337 bonds with available ownership information (see Appendix C for the regression results). Regression results show that insurance company ownership is higher for less risky, nonconvertible, and fixed coupon bonds issued by profitable and low leveraged firms. Using the coefficients from this regression, we then predict the level of insurance company ownership for the entire sample of 19,101 bonds. Finally, we classify each bond into a high or low insurance company ownership subsample based on whether its predicted insurance company ownership is above or below the sample’s median predicted insurance company ownership (11.98%). Columns (1)–(2) and (3)–(4) of Table 5 report the coefficient estimates from the regression of Log(Bond Maturity) for the high and low expected insurance company ownership subsamples, respectively. The coefficient estimate on Trend is –1.62 and significant for the high expected insurer ownership subsample, and it is close to zero and insignificant for the low expected insurer ownership subsample. The difference in these coefficient estimates is statistically significant, indicating that, as predicted, the declining trend in bond maturity is more pronounced for the subsample of bonds that face a higher demand from insurance companies.

15

Next, we examine the relation between the market shares of different types of insurance companies and bond maturities. Within the insurance sector, life insurance companies tend to prefer longer term assets (have longer term liabilities) when compared to property & casualty insurance companies. Therefore, if the maturity preferences of insurance companies drive the declining trend in bond maturities, this trend should be more closely associated with the market share of life insurance companies than that of property & casualty insurance companies. To test this prediction, we construct quarterly market shares of life and property & casualty insurance companies in the corporate bond market using the U.S. flow of funds data, and then estimate their influence on bond maturity. Columns (5)–(6) and (7)–(8) in Table 5 report the results from the baseline regressions of bond maturity that control for market shares of life and property & casualty insurance companies, respectively, as an additional control variable. The coefficient estimates on Life Insurer Market Share and Property & Casualty Insurer Market Share variables are positive and significant in both OLS and fixed effects specifications. Hence, market shares of both insurer types are positively associated with bond maturity. Controlling for Life Insurer Market Share, the coefficient on Trend is close to zero and insignificant in both regression specifications. On the other hand, the coefficient on Trend is –0.45 and significant when controlling for Property & Casualty Insurer Market Share in the OLS model, and it is negative and insignificant in the fixed effects model. The coefficients on Trend variable estimated from the OLS models for different insurance company types are significantly different from each other. Hence, as predicted, the declining trend in bond maturities is more closely related to the market share of life insurance companies when compared to that of property & casualty insurance companies.

16

Overall, this section shows that the intensity of the insurance company effect on bond maturity varies predictably across bond and insurance company types that are associated with insurance company preferences for maturity. 3.4. Alternative Identification Strategies to Estimate the Maturity Clientele Effects We have so far established that higher insurance company ownership in bonds is associated with longer bond maturities, and that the declining trend in insurance company presence in the bond market coincides with the declining trend in bond maturities observed during the same period. However, our results do not shed light on whether these maturity clientele effects can cause the decline in bond maturities. A primary difficulty in drawing such conclusions arises from the joint determination of the firms’ maturity choice and the investors’ preferences. To examine whether changes in the demand for bonds from insurance companies can have a causal effect on bond maturity, in this section, we implement a Granger causality test, an instrumental regression approach, and a natural experiment. 3.4.1. Granger Causality Test We implement a Granger causality test to examine the direction of the relation between Insurer Market Share and Bond Maturity. To construct the data for this test, we first calculate the average Log(Bond Maturity) for our sample bonds in each quarter from 1975 to 2015. Next, we match this quarterly maturity series with Insurer Market Share observed at the beginning of each quarter, assuming that issuers and underwriters observe the demand for bonds from insurance companies at the beginning of the quarter and determine the maturity of new bond issues accordingly. Using these two time-series, we then implement the Granger causality test. Column (1) in Table 6 reports the results from a regression where the dependent variable is the quarterly average Log(Bond Maturity), and the independent variables are the one-quarter

17

lagged values of average Log(Bond Maturity) and Insurer Market Share. The coefficient estimate on both variables are positive and significant, indicating that lagged insurer market share and maturity are significant determinants of maturity in the subsequent quarter. Column (2) in Table 6 reports the results from an identical regression specification with the exception that the dependent variable is Insurer Market Share. In this regression, the coefficient estimate on lagged Insurer Market Share is positive and significant, but that on lagged average Log(Bond Maturity) is insignificant. These findings show that insurer market share appears to Granger-cause bond maturity. Table 6 also reports the results from a Wald test that further confirms this conclusion. 3.4.2. Instrumental Variable Approach Prior literature documents that investors prefer securities issued by geographically proximate companies (e.g., Coval and Moskowitz, 1999; Ivković and Weisbenner, 2005; Massa, Yasuda, and Zhang, 2013). In this sense, the geographical distance between an insurance company and an issuer can be negatively related to the amount it invests in the issuer’s bonds, but this distance is unlikely to be correlated with bond maturity except through its correlation with insurance company ownership. Accordingly, we argue that such a geographical distance variable can be a valid instrument for insurance company ownership in our bond maturity regressions. We construct this geographical distance variable for each bond as the average distance between the bond’s issuer and the bond’s insurance company investors. This bond-level distance can vary within a firm’s bonds, thereby allowing for controlling firm fixed effects in regressions. In addition, because it is constructed using insurance companies that hold the bond, this distance measure implicitly accounts for firm and bond characteristics that may make a certain bond undesirable for certain insurance companies. Alternatively, using all insurance companies to construct the distance measure would require us to control for their investment opportunity sets.

18

To measure geographical distance, we obtain county level ZIP codes for the headquarters of issuers and insurance companies in our sample from COMPUSTAT and A.M. Best Rating Service, respectively. Eliminating issuers and insurers that are located outside the continental U.S. (i.e., institutions from Hawaii, Alaska, and Puerto Rico) results in a sample of 6,235 bonds with non-missing information. We then identify the latitude and longitude for each county from the U.S. Census Bureau’s Gazetteer Place and Zip Code database, and use the standard formula to compute the geographic distance between any two counties.7 Next, we construct the distance variable for each bond by averaging the distances between the bond’s issuer and the bond’s insurance company investors. Finally, we define a Closer Dummy variable that equals one for bonds with the average distance less than the median distance (805.79 miles) in our sample, and zero otherwise, and use it as the instrument for insurance company ownership in bond maturity regressions.8 Columns (1) and (2) in Table 7 report the first-stage estimation results from OLS and fixed effects regressions of insurance company ownership, respectively, using Closer Dummy as an instrument. The coefficient estimate on Closer Dummy is positive and significant in both specifications, verifying that, conditional on investing, insurance companies located in closer distance to issuers invest more in their bonds. Closer Dummy is also unlikely to be a weak instrument as its F-statistic is greater than 10 in both regression specifications. Columns (3) and (4) in Table 7 report the estimates from the second-stage regression of Log(Bond Maturity) using OLS and fixed effects models, respectively. The coefficient estimate on

Following Ivković and Weisbenner (2005), we calculate the distance between county a and b as: d(a, b) = arccos{cos(a1) cos(a2) cos(b1) cos(b2) + cos(a1) sin(a2) cos(b1) sin(b2) + sin(a1) sin(b1)}r, where a1 and b1 (a2 and b2) are the latitudes (longitudes) of the two counties (expressed in radians), respectively, and r is the radius of the Earth (3,963 miles). 7

8

Our instrumental variable is Closer Dummy because it is a stronger predictor of insurance company ownership compared to the actual bond-level distance. This finding suggests that the relation between the distance and insurance company ownership is non-linear.

19

Instrumented Insurer Ownership variable is about 4 and significant in both regression models, indicating that a 10-percentage point decline in insurance company ownership leads to about a 30% decline in bond maturity. Therefore, it is plausible that the decline in insurance company share from 40% in 1970s to 25% in 2000s may cause the 50% decline in bond maturities observed during the same period. 3.4.3. Natural Experiment Massa and Zhang (2011) show that insurance companies affected by Hurricane Katrina liquidated their corporate bond holdings and that the effects of this demand shock lasted for several months. We extend these findings and utilize the ten natural disasters that led to the largest insurable losses during our sample period as an exogenous shock to insurance company ownership, and investigate whether the maturities of new bonds issued during the post disaster periods are shorter than the average.9 We define a Disaster Dummy variable that equals one for 627 bonds issued during the disaster quarters, and zero otherwise. Columns (1) and (2) in Table 8 report the results from OLS and fixed effects regressions of Insurer Ownership, respectively. The coefficient estimate on Disaster Dummy is –0.02 and significant in both specifications, indicating a 2-percentage point decline in insurance company ownership during the disaster quarters. In contrast, Columns (3) and (4) show that mutual fund ownership does not change significantly during the same period. Hence, natural disasters appear to serve as a shock to specifically insurance company ownership.

9

We identify the ten largest natural disasters based on the insurable losses provided by Swiss Re and Insurance Information Institute. Here is the list of the disasters with event quarter and their insured losses (in billions) in parentheses, are Charley (2004Q3; $10), Frances (2004Q3; $6), Ivan (2004Q3; $16), Katrina (2005Q3; $81), Rita (2005Q3; $13), Wilma (2005Q4; $15), Ike (2008Q3; $23), Super Outbreak (2011Q2; $8), Irene (2011Q3; $6), and Sandy (2012Q4; $30).

20

Next, we test whether bonds issued in the natural disaster quarters have shorter maturities.10 Column (5) in Table 8 reports the estimates from the regression of Log(Bond Maturity) and shows that the coefficient on Disaster Dummy is –0.07 and significant, indicating an average of 7% decline in maturities of bonds issued during the natural disaster periods. Column (6) shows that the results are similar controlling for firm fixed effects. These findings suggest that a 2 (10) percentage point exogenous decline in insurance company ownership is associated with a 7% (35%) decline in bond maturity, which is consistent with the economic magnitude of the insurance company effect estimated from the instrumental variable approach. As an additional test, we investigate in Columns (7) and (8) of Table 8 the price impact of the exogenous decline in insurance company ownership. If insurance company demand declines, firms may also have to offer a higher yield to issue long term bonds. To test this prediction, we run a regression of bond offering yields controlling for Disaster Dummy, Log(Bond Maturity), and their interaction term in addition to controlling for the variables in maturity regressions. In both OLS and fixed effects regression models, the coefficient on the interaction of Disaster Dummy and Log(Bond Maturity) is positive and significant, indicating that the cost to issue longer maturity bonds increases when the demand from insurance companies declines. This finding provides further evidence of the maturity clientele effects in the bond market. Overall, the results from both this natural experiment and the earlier instrumental variable regression show that changes in insurance company ownership in bonds can have a causal effect on their maturities. These findings suggest that the declining presence of insurance companies in the corporate bond market can indeed result in shorter bond maturities in the aggregate.

10

Untabulated results show that the natural disaster event is a strong instrument for insurer ownership in a two-stage least squares model and that, in the second stage, the instrumented insurer ownership is significantly positively associated with bond maturity.

21

3.5. Other Trending Variables In this section, we investigate whether trending variables other than insurer market share can explain the declining trend in bond maturities. Based on the findings in the literature, the variables of interest are firms’ cash holdings, the share of pension funds in the bond market, and the volatility of interest rates. During our analysis period, firms in the U.S. have also increased their cash holdings (e.g., Foley, Hartzell, Titman, and Twite, 2007; Bates, Kahle, and Stulz, 2009; Harford, Klasa, and Maxwell, 2014). This may affect firms’ debt maturity as higher levels of cash reduces the cost of failing to rollover debt. Accordingly, higher levels of cash holdings may incentivize firms to issue shorter term debt to reduce their cost of borrowing. Our baseline regression of bond maturity [Regression (3), Table 3] controls for the effect of cash holdings indirectly through Tangibility variable. Alternatively, Regression (1) and (2) in Table 9 directly account for the influence of cash using the ratio of cash and equivalents to total assets as an additional control variable in OLS and fixed effects settings, respectively. The coefficient estimates on Cash/Assets variable is negative, consistent with higher cash holdings being associated with shorter debt maturity. However, the coefficient on Trend variable is negative and significant in both two regressions after controlling for Cash/Assets, indicating that variation in firms’ cash holdings cannot explain the declining trending in bond maturity. Similar to insurance companies, pension funds also prefer investing in long-term bonds and their share in the corporate bond market has declined from 30% to 11% between 1975 and 2015. To test whether the declining presence of pension funds can explain the maturity trend, we calculate Pension Fund Market Share using the U.S. flow of funds data. Regression (3) and (4) in Table 9 show that coefficient estimates on Trend is negative and significant after controlling for

22

Pension Fund Market Share in both OLS and fixed effects regressions. Hence, time-series variation in the pension fund market share cannot explain the declining trend in bond maturities. Finally, we investigate whether changes in interest rate volatility can explain the maturity trend. Given that longer maturity bonds have higher duration, the demand for long-term bonds may be lower when interest rate volatility is higher. Accordingly, firms may issue shorter maturity bonds as interest rate volatility increases, or vice versa. To test this prediction, we calculate Interest Volatility as the standard deviation of daily interest rate on the 10-year Treasury bond in the quarter right before bond issuance. Regressions (5) and (6) in Table 9 include Interest Rate Volatility as an additional control variable in the OLS and fixed effects specifications, respectively, and show that the coefficient on Trend maintains its negative sign and significance. Hence, changes in the volatility of interest rates cannot explain the decline in bond maturities. Overall, the tests in this section show that the variations in firms’ cash holdings, the share of pension funds in the corporate bond market, and interest volatility cannot explain the declining trend in bond maturities. 3.6. Robustness Tests In this section, we investigate the robustness of our findings and report the results in Table 10. For brevity, Table 10 reports the results from only the OLS regressions, but the fixed-effects regressions also produce similar results. We first examine the sensitivity of our findings to alternative regression specifications. Regression (1) adds Trend Square to the baseline regression of bond maturity [Regression (3), Table 3] to control for potential non-linearity of the declining trend in bond maturity. To examine whether our findings hold in subsample analyses, Regressions (2) and (3) estimate the baseline regression of bond maturity during the 1975-1994 and 1995-2015 periods, respectively. In

23

Regression (4), a continuous measure of trend that equals the number of years between the bond offering date and the starting date of our sample period (January 1st,1975) replaces the discrete trend variable in the baseline model. In Regression (5), all control variables measured in U.S. dollars are adjusted for inflation by converting them into 2015 U.S. dollars. In Regression (6), the bond offering amount is divided by the market value of equity to control for the relative size of the issue (Altinkilic and Hansen, 2000). Regression (7) controls for underwriter fixed effects to examine the influence of underwriter characteristics on our findings. We identify the lead underwriters for our bond sample from FISD and group them under their parent institutions. The results from these regressions show that the coefficient estimate on Trend is insignificant and that on Insurer Market Share is positive and significant, consistent with the findings from the baseline regressions. Trend, Insurer Market Share, and Mutual Fund Market Share variables are all correlated, thereby introducing a potential multicollinearity problem in our regression estimates. Hence, the positive (negative) relation observed between bond maturity and insurance company (mutual fund) market share may be driven by a multicollinearity problem arising from controlling Trend in the same regression. To address this concern, we estimate the baseline bond maturity regression [Regression (1), Table 3] without controlling for Trend but controlling for Insurer Market Share and Mutual Fund Market Share in Regressions (8) and (9), respectively. These regressions show that, in the absence of a potential multicollinearity problem, the coefficient estimate on Insurer Market Share is positive and significant, and that on Mutual Fund Market Share is negative and significant. Hence, consistent with the earlier findings, insurance companies (mutual funds) invest in longer (shorter) term bonds.

24

Finally, we investigate whether the maturity clientele effects are driven by an omitted liquidity preference variable. To do so, we construct a Bond Liquidity variable for each bond that equals the proportion of trading days with available prices measured during the post 3-month period following the bond offering date. We obtain the transaction level bond prices from TRACE, which are available since July 2002, and eliminate cancelled and corrected trades as in Bessembinder, Kahle, Maxwell, and Xu (2009). We then introduce Bond Liquidity as an additional control variable in Regressions (1) and (5) in Table 4 that investigate the influence of insurance company and mutual fund ownership, respectively, on bond maturity. Regressions (10) and (11) in Table 10 report the results from these regressions, and show that the coefficient estimate on bond liquidity is negative and significant, indicating that longer term bonds are less liquid. Nevertheless, the coefficient estimates on Insurer Ownership and Mutual Fund Ownership maintain their sign and significance controlling for bond liquidity. Overall, the results from the regressions in this section show that our findings are robust. 4. Conclusion The average maturity of new corporate bond issues in the U.S. has declined from 20 years in 1970s to its current level of 10 years. The known determinants of debt maturity fail to explain this declining trend in bond maturities. In this paper, we find that accounting for the changes in the composition and maturity preferences of investors in the bond market explains the maturity puzzle. The major investors in the corporate bond market are insurance companies and mutual funds, which have heterogeneous preferences for asset maturity. Due to the differences in their liability durations, insurance companies are more inclined to invest in longer term bonds than mutual funds. Since 1970s, the share of insurance companies in the bond market has declined by

25

about half with the rise of mutual fund industry. Therefore, the maturity preference of the average investor has shifted towards shorter term bonds with the declining presence of insurance companies and the increasing participation of mutual funds in the corporate bond market. We find that accounting for the share of insurance companies in the corporate bond market resolves the maturity puzzle. Although the growth in mutual funds exacerbates the maturity contraction, it does not explain the declining trend. Consistent with maturity preferences affecting the maturity shortening, the declining trend is more pronounced for bonds that insurance companies prefer (e.g., investment grade, non-convertible, fixed-coupon bonds). In addition, within the insurance sector, the market share of life insurance companies, compared to that of property & casualty insurance companies, is more closely associated with the declining maturity trend. Because life insurance companies have relatively longer liabilities than do property & casualty insurance companies, this finding further sheds light on the role maturity preferences of investors play in the bond market. The results from an instrumental regression approach and a natural experiment establish a causal relation between insurance company ownership and maturity of new bond issues. Overall, our findings contribute to the literature by resolving the maturity puzzle that the previous literature documents (e.g., Greenwood, Hanson, and Stein, 2010; Custódio, Ferreira, and Laureano, 2013; Harford, Klasa, and Maxwell, 2014). They also provide evidence that the maturity preferences of investors can have real effects on the maturity of securities offered to the market. Hence, they illustrate that the development and characteristics of financial institutions in a country may influence its real economy.

26

References Altınkılıç, Oya, and Robert S. Hansen. 2000. Are there economies of scale in underwriting fees? Evidence of rising external financing costs. Review of Financial Studies 13: 191-218. Barclay, Michael J., and Clifford W. Smith. 1995. The maturity structure of corporate debt. Journal of Finance 50: 609-631. Bates, Thomas W., Kathleen M. Kahle, and René M. Stulz. 2009. Why do US firms hold so much more cash than they used to? Journal of Finance 64: 1985-2021. Becker, Bo, and Victoria Ivashina. 2015. Reaching for yield in the bond market. Journal of Finance 70: 1863-1902. Berger, Philip G., Eli Ofek, and Itzhak Swary. 1996. Investor valuation of the abandonment option. Journal of Financial Economics 42: 257-287. Bessembinder, Hendrik, Kathleen M. Kahle, William F. Maxwell, and Danielle Xu. 2008. Measuring abnormal bond performance. Review of Financial Studies 22: 4219-4258. Coval, Joshua D., and Tobias J. Moskowitz. 1999. Home bias at home: Local equity preference in domestic portfolios. Journal of Finance 54: 2045-2073. Custódio, Cláudia, Miguel A. Ferreira, and Luís Laureano. 2013. Why are US firms using more short-term debt? Journal of Financial Economics 108:182-212. Dass, Nishant, and Massimo Massa. 2014. The variety of maturities offered by firms and institutional investment in corporate bonds. Review of Financial Studies 27: 2219-2266. Demirgüç-Kunt, Asli, and Vojislav Maksimovic. 1998. Law, finance, and firm growth. Journal of Finance 53: 2107-2137. Demirgüç-Kunt, Asli, and Vojislav Maksimovic. 1999. Institutions, financial markets, and firm debt maturity. Journal of Financial Economics 54: 295-336. Diamond, Douglas W. 1991. Debt maturity structure and liquidity risk. Quarterly Journal of Economics 106: 709-737.

27

Ellul, Andrew, Chotibhak Jotikasthira, and Christian T. Lundblad. 2011. Regulatory pressure and fire sales in the corporate bond market. Journal of Financial Economics 101:596-620. Fan, Joseph PH, Sheridan Titman, and Garry Twite. 2012. An international comparison of capital structure and debt maturity choices. Journal of Financial and Quantitative Analysis 47: 23-56. Foley, C. Fritz, Jay C. Hartzell, Sheridan Titman, and Garry Twite. 2007. Why do firms hold so much cash? A tax-based explanation. Journal of Financial Economics 86: 579-607. Giannetti, Mariassunta. 2003. Do better institutions mitigate agency problems? Evidence from corporate finance choices. Journal of Financial and Quantitative Analysis 38: 185–212. Greenwood, Robin, Samuel Hanson, and Jeremy C. Stein. 2010. A gap‐filling theory of corporate debt maturity choice. Journal of Finance 65:993-1028. Guedes, Jose, and Tim Opler. 1996. The determinants of the maturity of corporate debt issues. Journal of Finance 51: 1809-1833. Harford, Jarrad, Sandy Klasa, and William F. Maxwell. 2014. Refinancing risk and cash holdings. Journal of Finance 69: 975-1012. Ivković, Zoran, and Scott Weisbenner. 2005. Local does as local is: Information content of the geography of individual investors’ common stock investments. Journal of Finance 60: 267-306. Massa, Massimo, Ayako Yasuda, and Lei Zhang. 2013. Supply uncertainty of the bond investor base and the leverage of the firm. Journal of Financial Economics 110:185-214. Massa, Massimo, and Lei Zhang. 2011. The spillover effects of Hurricane Katrina on corporate bonds and the choice between bank and bond financing. Working Paper, INSEAD. Myers, Stewart C. 1977. Determinants of corporate borrowing. Journal of Financial Economics 5: 147-175. Rajan, Raghuram G., and Luigi Zingales. 1998. Financial dependence and growth. American Economic Review 88: 559–586. Rydqvist, Kristian, Joshua Spizman, and Ilya Strebulaev. 2014. Government policy and ownership of equity securities. Journal of Financial Economics 111: 70-85.

28

Stohs, Mark Hoven, and David C. Mauer. 1996. The determinants of corporate debt maturity structure. Journal of Business 69: 279-312.

29

Appendix A: Variable Definitions and Data Sources Variable Insurer Market Share

Source Flow of Funds

Mutual Fund Market Share

Flow of Funds

Insurer Ownership

Bloomberg

Mutual Fund Ownership

Bloomberg

Bond Maturity

FISD/SDC

Trend

FISD/SDC

Share of Long-Term Government Debt

CRSP

Real Short-Term Rate

FRED and BLS

Term Spread

FRED

Number of years between the bond offering year and the initial year of our sample period (1975). Ratio of the Treasury bond payments due in more than a year to the total Treasury bond payments due in all future periods, as defined by Greenwood, Hanson, and Stein (2010). Difference between the 3-month Treasury bill rate and the quarterly percentage change in the Consumer Price Index (CPI). Difference between the 10-year and 1-year Treasury rates.

Default Spread

FRED

Difference between the BAA- and AAA-rated corporate bond yields.

Market Value of Equity

COMPUSTAT

M/B Value of Assets

COMPUSTAT

Net Income/Total Assets

COMPUSTAT

Stock price at the fiscal quarter end (PRCCQ) multiplied by common shares outstanding at the same period (CSHOQ). Ratio of market value of assets (ATQ – CEQQ + PRCCQ x CSHOQ) to book value of assets (ATQ). Ratio of net income (NIQ) to total assets (ATQ).

Total Debt/Total Assets

COMPUSTAT

Ratio of total debt (DLCQ + DLTTQ) to total assets (ATQ).

Tangibility

COMPUSTAT

Stock Return

CRSP

Offering Amount

FISD/SDC

(CHEQ + 0.715 x RECTQ + 0.547 x INVTQ + 0.535 × PPENTQ)/ATQ, as defined by Berger, Ofek, and Swary (1996). Average monthly stock return during the 3-month period before the bond offering date. Initial offered principle value of a bond.

Callable Dummy

FISD/SDC

Dummy variable indicating whether a bond is callable.

Floating Dummy

FISD/SDC

Dummy variable indicating whether a bond has a floating interest rate.

Convertible Dummy

FISD/SDC

Dummy variable indicating whether a bond is convertible.

Puttable Dummy

FISD/SDC

Dummy variable indicating whether a bond is puttable.

Sinking Fund Dummy

FISD/SDC

Dummy variable indicating whether a bond has a sinking fund provision.

Global Dummy

FISD/SDC

Dummy variable indicating whether a bond is issued globally.

Credit Rating Dummies

FISD/SDC

Dummy variables that classify bonds into credit rating groups (AAA or AA, A, BBB, BB, B or Below, and Unrated) based on the median of rating grades from S&P, Moody’s, and Fitch. Dummy variables that indicate the decade in which a firm first appears in the CRSP database, as defined by Custódio, Ferreira, and Laureano (2013).

Firm IPO Decade Dummies CRSP Industry Dummies

COMPUSTAT

Definition Amount of insurance company ownership (LM513063003 + LM543063005) in U.S. corporate bonds and foreign entity bonds (issued through U.S. dealers and purchased by U.S. residents) divided by their total outstanding amount (FL893163005). Amount of mutual fund ownership (LM653063005) in U.S. corporate bonds and foreign entity bonds (issued through U.S. dealers and purchased by U.S. residents) divided by their total outstanding amount (FL893163005). Amount of insurance company ownership in a bond measured at the end of the issuance quarter divided by the bond’s issue amount. This variable is available since 1998, but used in the study since 1999 to eliminate a potential coverage bias in the dataset inception year. Amount of mutual fund ownership in a bond measured at the end of the issuance quarter divided by the bond’s issue amount. This variable is available since 1998, but used in the study since 1999 to eliminate a potential coverage bias in the dataset inception year. Number of years between the maturity and offering dates.

Dummy variables that indicate the industry of a firm based on the FamaFrench 12-industry classification.

30

Appendix B: Example of Bond Ownership Data Bloomberg compiles the bond holding information of institutional investors from the 13F, Schedule D, 10K, Form 990, and Form 5500 filings since 1998 at a quarterly frequency. To eliminate a potential coverage bias in the database inception year, we construct the insurance company and mutual fund ownership in each bond at the end of its issuance quarter since 1999. We classify each investor as an insurance company or a mutual fund based on the information from Bloomberg and the investor’s web-site. Following is an example of the ownership information for a $900 million face value bond (CUSIP: 459200HD6) issued by IBM on May 11th, 2012. There are 900,000 bonds outstanding with a par value of $1,000, of which 51,465 are held by insurance companies (5.7%) and 100,700 are held by mutual funds (11.2%). Refer to Appendix A for more details on variable definitions. Investor Type Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Insurance Company Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund Mutual Fund

Investor Name Voya Retirement Ins & Ann Co Reliastar Life Insurance Company Old Republic General Ins Corp New Jersey Manufacture Insurance Euler Hermes N Amer Credit Ins Old Republic Insurance Company Bitco Gen Ins Corp Bitco Natl Ins Co Great West Casualty Company American Health & Life Ins Co Usable Life Hospitals Ins Co Inc Copic Insurance Company Maidstone Ins Co Olympus Ins Co United Automobile Insurance Co Kaiser Found Hlth Plan Of WA Mamsi Life and Health Insurance Florida Peninsula Ins Co National Catholic RRG Community Health Plan Of WA Amerigroup Ins Co Alliant Health Plans Inc Amerigroup LA Inc Protective Specialty Ins Co Sagamore Insurance Company Protective Insurance Company Capital Research and Management Principal Management Corp Teachers Advisor Inc Columbia Management Investment Candriam Belgium SA Vanguard Group Incoporated Payden & Rygel Baird Financial Group, Inc Northern Trust Company Thrivent Financial Lutherans Legg Mason Partners Fund Advisor UBS Strategy Fund Mgmnt Company Hsbc Trinkaus Investment Mgmt Amundi Luxembourg SA Vanguard Group(Ireland) Limited

31

Number of Bonds 23,000 7,000 5,000 3,000 2,500 2,000 1,000 1,000 1,000 700 700 665 600 500 500 500 400 300 250 200 150 125 100 100 95 60 20 38,620 14,750 12,700 7,500 6,500 4,700 4,300 3,000 2,430 1600 1,550 1,250 1,000 500 300

Appendix C: Preferences of Insurance Companies This table presents the results from the regression that studies the preferences of insurance companies on macro-, firm-, and bond-level variables other than Bond Maturity. The dependent variable in the regression is Insurer Ownership and the regression model is identical to Regression (1) in Table 4. This regression, however, excludes Firm IPO Decade Fixed Effects from control variables because using their coefficients estimated during the post-1999 period to predict ownership during the pre-1999 period would introduce a survivorship bias problem. The sample period for this table starts in 1999 due to the availability of bond ownership data. Refer to Appendix A for the remaining variable definitions and data sources, and Table 1 for sample selection criteria. Standard errors used to compute t-statistics in parentheses are clustered at the firm level. Variables Intercept

(1) 1.05*** (12.91) 0.01 (1.17) -0.02*** (-5.40) 0.00*** (2.89) 0.11*** (2.88) -0.04*** (-3.42) -0.02 (-1.01) -0.01 (-1.45) -0.54*** (-2.73) -0.84*** (-9.76) 0.41 (1.41) -3.28*** (-3.79) 0.03*** (3.85) -0.09*** (-11.89) -0.06*** (-8.13) -0.05*** (-6.98) 0.12 (0.86) 0.03*** (3.37)

Log(Offering Amount) Log(Market Value of Equity) Market-to-Book Ratio Net Income/ Total Asset Total Debt/ Total Asset Tangibility Stock Return Real Short-Term Rate Share of Long-Term Government Debt Term Spread Default Spread Callable Dummy Floating Dummy Convertible Dummy Puttable Dummy Sinking Fund Dummy Global Dummy

Number of Observations Adjusted R2

19101 0.222

Industry Fixed Effects Firm Fixed Effects Year Fixed Effects Bond Credit Rating Fixed Effects Firm Listing Period Dummy ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

32

Yes No No Yes No

Table 1: Time-Series Trends in Bond Maturity and Market Shares of Investors This table reports the descriptive statistics on the maturity (in years) of our sample newly issued bonds, and the shares of insurance companies and mutual funds in the corporate bond market by years. To construct our sample, we extract all U.S. dollar denominated corporate bonds issued by public U.S. industrial firms between 1975 and 2015 from Mergent FISD and SDC New Issues databases. Our final sample includes 19,101 bonds issued by 2,988 firms with complete information. Insurer and mutual fund market shares in each year are the averages of their quarterly values in that year computed using the U.S. flow of funds data. Refer to Appendix A for more details on variable definitions and data sources. Year 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Number of Bonds 64 28 31 28 49 109 142 214 208 169 285 407 326 319 315 332 526 600 769 525 667 564 846 1037 587 477 748 661 816 662 454 490 559 351 535 604 561 682 736 730 888

Number of Issuers 56 27 30 28 42 95 100 147 147 118 204 302 238 195 194 175 230 242 312 192 273 302 392 467 310 229 340 301 385 361 262 277 291 187 295 326 282 343 341 344 322

Mean Maturity 21.04 23.68 22.76 19.82 19.06 19.09 15.65 13.16 13.90 13.16 13.33 15.10 14.22 10.65 11.87 9.20 11.45 12.00 11.29 8.94 10.34 12.47 13.77 11.49 10.21 8.71 10.04 9.51 11.45 12.32 11.97 11.34 11.42 10.86 9.22 9.64 9.68 10.24 9.78 9.35 9.98

33

Median Maturity 25.00 25.01 24.98 19.99 20.01 20.02 15.01 10.03 11.99 10.03 10.03 12.02 10.03 10.00 10.01 8.01 10.01 10.02 10.01 8.01 9.98 10.00 10.00 9.96 9.72 7.01 7.98 8.01 9.84 10.01 9.88 9.62 9.53 7.41 7.05 8.03 7.70 8.71 7.77 7.04 7.17

Insurance Mutual Fund Market Share Market Share 0.35 0.02 0.36 0.02 0.38 0.02 0.40 0.02 0.41 0.02 0.40 0.02 0.39 0.02 0.38 0.02 0.37 0.02 0.38 0.02 0.36 0.02 0.34 0.03 0.34 0.04 0.36 0.03 0.36 0.03 0.37 0.03 0.37 0.04 0.36 0.06 0.34 0.07 0.33 0.07 0.33 0.06 0.32 0.07 0.33 0.08 0.32 0.09 0.30 0.09 0.29 0.08 0.28 0.07 0.28 0.07 0.28 0.07 0.28 0.07 0.27 0.07 0.24 0.07 0.21 0.07 0.19 0.07 0.19 0.08 0.23 0.11 0.25 0.12 0.25 0.14 0.25 0.15 0.25 0.16 0.24 0.15

Table 2: Summary Statistics on Firm, Bond, and Macro Level Variables This table presents the summary statistics on firm and bond characteristics for our sample. The sample includes all bonds issued by public U.S. industrial firms during the period 1975–2015. We collect Insurer Ownership and Mutual Fund Ownership from Bloomberg, which are updated quarterly and available since 1998. To eliminate a potential coverage bias in the dataset inception year, we construct the bond ownership percentages since 1999. See Appendix A for the remaining variable definitions and data sources, and Table 1 for sample selection criteria. Variables Panel A: Firm Level Variables Market Value of Equity (in Billion Dollars) M/B Value of Assets Net Income/Total Assets Total Debt/Total Assets Tangibility Stock Return Pre-1970 IPO Dummy 1970-1979 IPO Dummy 1980-1989 IPO Dummy 1990-1999 IPO Dummy 2000-2010 IPO Dummy Post-2010 IPO Dummy Consumer Non-Durables Industry Dummy Consumer Durables Industry Dummy Manufacturing Industry Dummy Energy Industry Dummy Chemicals Industry Dummy Business Equipment Industry Dummy Telecommunication Industry Dummy Shop Industry Dummy Health Industry Dummy Other Industries Dummy Panel B: Bond Level Variables Offering Amount (in Billion Dollars) Callable Dummy Floating Dummy Convertible Dummy Puttable Dummy Sinking Fund Dummy Global Dummy AAA or AA Dummy A Dummy BBB Dummy BB Dummy B or Blow Dummy Unrated Dummy Insurer Ownership Mutual Fund Ownership Panel C: Macro Level Variables Share of Long-Term Government Debt Real Short-Term Rate Term Spread Default Spread

34

N

Mean

Median

Std. Dev.

19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101

16.714 1.874 0.003 0.348 0.446 0.064 0.421 0.132 0.151 0.222 0.056 0.019 0.086 0.032 0.154 0.089 0.092 0.118 0.074 0.140 0.083 0.131

2.806 1.499 0.011 0.317 0.463 0.043 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

41.340 1.775 0.402 0.272 0.133 0.244 0.494 0.338 0.358 0.415 0.231 0.135 0.280 0.175 0.361 0.285 0.289 0.323 0.262 0.347 0.276 0.338

19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 19101 7337 7337

0.292 0.619 0.062 0.165 0.067 0.052 0.113 0.060 0.162 0.207 0.085 0.170 0.316 0.151 0.125

0.165 1.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.086 0.087

0.430 0.486 0.240 0.371 0.250 0.222 0.317 0.237 0.369 0.406 0.279 0.376 0.465 0.174 0.129

19101 19101 19101 19101

0.738 0.029 0.015 0.010

0.743 0.035 0.017 0.009

0.035 0.027 0.010 0.004

Table 3: Explaining the Declining Trend in Bond Maturities

This table reports the coefficient estimates of Log(Bond Maturityij )=α+W'jt β+Z't δ+X't γ+λTrendt +εij, where Bond Maturityij is the maturity of bond i issued by firm j, α is the intercept, Wjt, Zi, and Xt represent firm, bond, and macro level control variables, respectively, Trend is the difference between the year of bond issuance and the year when our sample period starts (1975), and εij is the error term. Bond Characteristics Fixed Effects control for Callable Dummy, Floating Dummy, Convertible Dummy, Puttable Dummy, Sinking Fund Dummy, and Global Dummy variables. Refer to Appendix A for the remaining variable definitions and data sources, and Table 1 for sample selection criteria. Standard errors used to compute tstatistics in parentheses are clustered at the firm level. Trends in Insurer Mutual Fund Maturity Market Share Market Share Variables (1) (2) (3) (4) (5) (6) Intercept 2.35*** 2.24*** 2.04*** 1.97*** 2.47*** 2.35*** (11.14) (9.19) (8.85) (7.43) (9.46) (7.90) f -1.12*** -0.91*** -0.06 0.19 -1.26*** -1.03*** (-5.37) (-3.34) (-0.21) (0.56) (-5.54) (-3.07) Insurer Market Share . . 1.86*** 1.92*** . . . . (4.80) (4.23) . . Mutual Fund Market Share . . . . 0.57 0.46 . . . . (1.05) (0.67) Log(Market Value of Equity) 0.02** 0.03** 0.02** 0.03** 0.02** 0.03** (2.40) (2.09) (2.37) (2.12) (2.43) (2.07) Total Debt/Total Assets -0.05 -0.10** -0.05 -0.11** -0.05 -0.10** (-1.21) (-2.05) (-1.25) (-2.26) (-1.19) (-2.05) Net Income/Total Assets -0.01 0.09 -0.01 0.11 -0.01 0.09 (-0.30) (0.59) (-0.29) (0.74) (-0.29) (0.58) Tangibility 0.20*** 0.27*** 0.19*** 0.26*** 0.20*** 0.27*** (3.41) (2.77) (3.29) (2.61) (3.39) (2.76) M/B Value of Assets -0.01*** -0.01* -0.01*** -0.01* -0.01*** -0.01** (-2.94) (-1.93) (-2.85) (-1.81) (-2.99) (-1.97) Stock Return 0.06*** 0.06*** 0.06*** 0.06*** 0.06*** 0.06*** (3.40) (3.27) (3.23) (3.12) (3.38) (3.27) Log(Offering Amount) 0.02** 0.00 0.02** 0.00 0.02** 0.00 (2.14) (0.01) (2.19) (0.11) (2.13) (0.02) Share of LT Government Debt -0.40* -0.22 -1.14*** -1.05*** -0.59* -0.38 (-1.73) (-0.83) (-4.41) (-3.43) (-1.84) (-1.01) Real Short-Term Rate -0.01 -0.09 -0.05 -0.08 0.14 0.03 (-0.01) (-0.14) (-0.08) (-0.12) (0.23) (0.04) Term Spread 0.81 1.21 0.48 0.88 0.95 1.31 (0.88) (1.19) (0.53) (0.89) (1.05) (1.30) Default Spread -10.90*** -10.30*** -8.69*** -8.24*** -10.69*** -10.10*** (-6.31) (-5.29) (-4.80) (-3.99) (-6.11) (-5.00) Number of Observations 19101 19101 19101 19101 19101 19101 Adjusted R2 (%) 22.15 17.21 22.33 17.40 22.16 17.21 Firm Fixed Effects No Yes No Yes No Yes Bond Characteristics Fixed Effects Yes Yes Yes Yes Yes Yes Bond Credit Rating Fixed Effects Yes Yes Yes Yes Yes Yes Industry Fixed Effects Yes No Yes No Yes No Firm IPO Decade Fixed Effects Yes No Yes No Yes No ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively. Model:

35

Insurer and Mutual Fund Market Share (7) (8) 1.90*** 1.78*** (5.81) (4.82) 0.17 0.48 (0.42) (1.02) 2.03*** 2.13*** (4.37) (4.08) -0.53 -0.69 (-0.85) (-0.91) 0.02** 0.03** (2.37) (2.13) -0.05 -0.11** (-1.27) (-2.28) -0.01 0.11 (-0.30) (0.76) 0.19*** 0.26*** (3.30) (2.62) -0.01*** -0.01* (-2.80) (1.71) 0.06*** 0.06*** (3.23) (3.12) 0.02** 0.00 (2.19) (0.10) -1.03*** -0.90*** (-3.31) (-2.41) -0.19 -0.25 (-0.31) (-0.37) 0.32 0.70 (0.36) (0.73) -8.68*** -8.32*** (-4.80) (-4.02) 19101 19101 22.34 17.41 No Yes Yes Yes Yes Yes Yes No Yes No

Table 4: Influence of Insurance Company and Mutual Fund Ownership on Bond Maturity This table presents the results from regressions that study the relation between bond-level ownership of the different types of investors and bond maturity. The regression specification is Log(Bond Maturityij )=α+W'jt β+Z't δ+X't γ+IOi +IO2i + εij, where Bond Maturityij is the maturity of bond i issued by firm j, α is the intercept, Wjt, Zi, and Xt represent firm, bond, and macro level control variables, respectively, IOi represents insurer or mutual fund ownership of bond i, and εij is the error term. The sample period for this table starts in 1999 due to the availability of bond ownership data. Bond Characteristics Fixed Effects control for Callable Dummy, Floating Dummy, Convertible Dummy, Puttable Dummy, Sinking Fund Dummy, and Global Dummy variables. Refer to Appendix A for the remaining variable definitions and data sources, and Table 1 for sample selection criteria. Standard errors used to compute t-statistics in parentheses are clustered at the firm level. Model: Variables Insurer Ownership Insurer Ownership Square Mutual Fund Ownership Mutual Fund Ownership Square

Number of Observations Adjusted R2 (%)

Insurer Ownership (1) (2) (3) (4) 0.62*** 0.69*** 1.27*** 1.22*** (9.71) (8.98) (7.99) (6.92) . . -1.04*** -0.85*** . . (-5.05) (-3.68) . . . . . . . . . . . . . . . . 7337 31.39

7337 25.76

7337 31.72

Mutual Fund Ownership (5) (6) (7) (8) . . . . . . . . . . . . . . . . -0.33*** -0.41*** -0.70*** -0.95*** (-4.70) (-4.98) (-4.51) (-5.41) . . 0.78*** 1.16*** . . (3.08) (4.09)

7337 25.99

Firm Fixed Effects No Yes No Yes Bond Characteristics Fixed Effects Yes Yes Yes Yes Firm, Bond, and Macro Controls Yes Yes Yes Yes Bond Credit Rating Fixed Effects Yes Yes Yes Yes Industry Fixed Effects Yes No Yes No Firm IPO Decade Fixed Effects Yes No Yes No Year Fixed Effects Yes Yes Yes Yes ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

36

7337 29.97

7337 24.25

7337 30.04

7337 24.42

No Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes No No Yes

No Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes No No Yes

Table 5: Insurance Company Preferences and the Declining Trend in Bond Maturities This table investigates whether the declining trend in bond maturity is associated with the preferences of insurance companies. The regression model is Log(Bond Maturityij )=α+W'jt β+Z't δ+X't γ+λTrendt +εij , where Bond Maturityij is the maturity of bond i issued by firm j, α is the intercept, Wjt, Zi, and Xt represent firm, bond, and macro level control variables, respectively, Trend is the difference between the year of bond issuance and the year when our sample period starts (1975), and εij is the error term. Columns (1)–(4) report the regression results using subsamples of bonds that face high and low demand from insurance companies. To determine insurance company preferences, we first regress bond-level insurance company ownership on the control variables in our baseline analysis (see Appendix C for details). The sample period for this regression starts in 1999 due to the availability of bond ownership data. Based on the regression estimates, we predict insurance company ownership for the entire sample of bonds issued between 1975 and 2015. Then, we classify each bond into a high or low insurance company ownership subsample based on whether its predicted insurance company ownership is above or below the sample’s median predicted insurance company ownership (11.98%), respectively. In Columns (5)–(8), the identification method in regressions rests on the relative differences in maturity preferences of life insurance companies (prefer longer maturity) and property & casualty insurance companies (prefer shorter maturity). We construct Life Insurer Market Share and Property & Casualty Insurer Market Share using the U.S. flow of funds data. Bond Characteristics Fixed Effects control for Callable Dummy, Floating Dummy, Convertible Dummy, Puttable Dummy, Sinking Fund Dummy, and Global Dummy variables. Refer to Appendix A for the remaining variable definitions and data sources, and Table 1 for sample selection criteria. Standard errors used to compute t-statistics in parentheses are clustered at the firm level. Model: Variables Trend x100 Life Insurer Market Share Property & Casualty Insurer Market Share

Number of Observations Adjusted R2 (%)

Regression (1) vs. (3) Regression (2) vs. (4) Regression (5) vs. (7) Regression (6) vs. (8)

High Expected Insurer Ownership (1) (2) -1.62*** -1.83*** (-5.12) (-5.23) . . . . . . . . 9551 23.53

9551 16.99

Low Expected Insurer Ownership (3) (4) -0.31 0.27 (-1.22) (0.71) . . . . . . . . 9550 25.38

9550 17.99

Property and Casualty Insurance (5) (6) (7) (8) -0.13 0.10 -0.45* -0.18 (-0.45) (0.29) (-1.77) (-0.63) 1.94*** 1.96*** . . (4.64) (3.94) . . 11.00*** 12.85** . . (3.99) (3.94) * . . Life Insurance

19101 22.3

Chi-square Test for the Difference in Trend Coefficients 12.86*** . . . . 20.14*** . . . . 3.49* . . . . 1.74 . . . .

Firm Fixed Effects No Yes No Yes Bond Characteristics Fixed Effects Yes Yes Yes Yes Firm, Bond, and Macro Controls Yes Yes Yes Yes Bond Credit Rating Fixed Effects Yes Yes Yes Yes Industry Fixed Effects Yes No Yes No Firm IPO Decade Fixed Effects Yes No Yes No ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

37

No Yes Yes Yes Yes Yes

19101 17.36

19101 22.37

19101 17.48

. . . .

. . . .

. . . .

Yes Yes Yes Yes No No

No Yes Yes Yes Yes Yes

Yes Yes Yes Yes No No

Table 6: Granger Causality Test of Insurer Market Share and Bond Maturity This table reports the results from a Granger causality test of Insurer Market Share and Bond Maturity. To construct the data for this test, we first calculate the average Log(Bond Maturity) in each quarter between 1975 and 2015 using the sample of 19,101 new bond issues, and then match this quarterly maturity series with Insurer Market Share at the beginning of each quarter. The dependent variables in regressions reported in Columns (1) and (2) are average Log(Bond Maturity) and Insurer Market Share, respective. This table also reports the Wald tests of Granger causality between Insurer Market Share and average Log(Bond Maturity). Refer to Appendix A for the variables definitions and data sources, and Tables 1 and 4 for sample selection criteria. Z-statistics are reported in parentheses. Dependent Variable: Variables Lagged Average Log(Bond Maturity) Lagged Insurer Market Share

Average Log (Bond Maturity) (1) 0.82*** (18.58) 0.51** (2.06)

Insurer Market Share (2) 0.00 (1.55) 0.99*** (133.98)

163

163

Number of Observation

Granger Causality Wald tests Insurer Market Share Forecasts Bond Maturity 4.26** Bond Maturity Forecasts Insurer Market Share 2.39 ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

38

. .

Table 7: Instrumenting for Insurer Ownership in Bond Maturity Regressions This table reports the two-stage least squares (2SLS) estimation results for Log(Bond Maturity). The regression model is Dependent Variableij =α+W'jt β+Z't δ+X't γ+ εij, where Bond Maturityij is the maturity of bond i issued by firm j, α is the intercept, Wjt, Zi, and Xt represent firm, bond, and macro level control variables, respectively, and εij is the error term. The sample includes only those bonds with positive insurance company ownership. Regression (1)–(2) report the first stage regression results of Insurer Ownership instrumented by Closer Dummy. To construct this variable, we first measure the geographical distance between insurance companies invested in our sample bonds and the issuers of these bonds. Next, we construct a bond-level distance variable by averaging these distances at the bond-level. Finally, we define a Closer Dummy variable that equals one for bonds with the average distance less than the median distance (805.79 miles) in our sample, and zero otherwise. Regression (3)–(4) report the regression results of Log(Bond Maturity) on Instrumented Insurer Ownership. Bond Characteristics Fixed Effects control for Callable Dummy, Floating Dummy, Convertible Dummy, Puttable Dummy, Sinking Fund Dummy, and Global Dummy variables. Refer to Appendix A for the remaining variable definitions and data sources, and Tables 1 and 4 for sample selection criteria. Standard errors used to compute t-statistics in parentheses are clustered at the firm level. Model: Dependent Variable: Variables Closer Dummy Instrumented Insurer Ownership

Number of Observations Adjusted R2(%) F-test for the Instrument Variable

First Stage Insurer Ownership (1) (2) 0.02*** 0.02*** (3.20) (3.19) . . . .

Second Stage Log(Bond Maturity) (3) (4) . . . . 4.19*** 3.82*** (2.67) (2.68)

6235 41.02

6235 27.61

6235 20.85

6235 23.46

22.75***

11.26***

.

.

No Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes No No Yes

Firm Fixed Effects No Yes Bond Characteristics Fixed Effects Yes Yes Firm, Bond, and Macro Level Controls Yes Yes Bond Credit Rating Fixed Effects Yes Yes Industry Fixed Effects Yes No Firm IPO Decade Fixed Effects Yes No Year Fixed Effects Yes Yes ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

39

Table 8: Using Natural Disasters as an Exogenous Shock to Insurance Demand This table presents the results from regressions that investigate the influence of large natural disasters—an exogenous shock to the demand for bonds from insurance companies—on Insurer Ownership, Log(Bond Maturity), Offering Yield, and Mutual Fund Ownership. We identify ten largest natural disasters between 1999 and 2015 based on their insured losses: Charley (2004Q3), Frances (2004Q3), Ivan (2004Q3), Katrina (2005Q3), Rita (2005Q3), Wilma (2005Q4), Ike (2008Q3), Super Outbreak (2011Q2), Irene (2011Q3), and Sandy (2012Q4). Disaster Dummy equals one for 627 bonds issued during the disaster quarters, and zero for the remaining. The regression model is Dependent Variableij =α+W'jt β+Z't δ+X't γ+Disaster Dummyi +ε , ij

where Bond Maturityij is the maturity of bond i issued by firm j, α is the intercept, Wjt, Zi, and Xt represent firm, bond, and macro level control variables, respectively, and εij is the error term. Bond Characteristics Fixed Effects control for Callable Dummy, Floating Dummy, Convertible Dummy, Puttable Dummy, Sinking Fund Dummy, and Global Dummy variables. The dependent variable in Regression (1)–(2), (3)– (4), and (5)–(6) is Insurer Ownership, Mutual Fund Ownership, and Log(Bond Maturity), respectively, and that in Regression (7)–(8) is Offering Yield. Regression (7)–(8) include the interaction of Log(Bond Maturity) and Disaster Dummy as an additional regressor. Refer to Appendix A for the remaining variable definitions and data sources, and Tables 1 and 4 for sample selection criteria. Standard errors used to compute t-statistics in parentheses are clustered at the firm level. Dependent Variable: Variables Disaster Dummy

Insurer Ownership (1)

Mutual Fund Ownership (3)

(4)

-0.02*** -0.02** (-2.72) (-2.34) Log(Bond Maturity) . . . . Log(Bond Maturity) x Disaster x 10 . . . .

0.00 (0.16) . . . .

0.00 (0.68) . . . .

Number of Observations Adjusted R2 (%)

7337 39.71

7337 22.85

7337 42.05

(2)

7337 23.79

Log(Bond Maturity) (5)

Firm Fixed Effects No Yes No Yes Bond Characteristics Fixed Effects Yes Yes Yes Yes Firm, Bond, and Macro Controls Yes Yes Yes Yes Bond Credit Rating Fixed Effects Yes Yes Yes Yes Industry Fixed Effects Yes No Yes No Firm IPO Decade Fixed Effects Yes No Yes No Year Fixed Effects Yes Yes Yes Yes ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

40

(6)

-0.07*** -0.05** (-2.67) (-2.10) . . . . . . . .

Offering Yield (7)

(8)

-0.00 (-0.55) 0.01*** (23.61) 0.02* (1.75)

-0.00 (-1.53) 0.01*** (25.54) 0.02*** (3.26)

7337 29.49

7337 23.92

5718 72.55

5718 62.84

No Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes No No Yes

No Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes No No Yes

Table 9: The Influence of Other Trending Variables on the Maturity Trend This table reports the results from regressions that investigate whether other trending variables can explain the declining trend in bond maturity. The analysis model is 𝐿og(Bond Maturityij )=α+W'jt β+Z't δ+X't γ+λTrendt + εij , where Bond Maturityij is the maturity of bond i issued by firm j, α is the intercept, Wjt, Zi, and Xt represent firm, bond, and macro level control variables, respectively, Trend is the difference between the year of bond issuance and the year when our sample period starts (1975), and εij is the error term. The trending variable in Regressions (1) and (2) is the ratio of cash to assets. Regression (3) and (4) include Pension Fund Market Share as a regressor to explain the trend. Pension Fund Market Share equals the amount of pension fund ownership (FL573063005 + FL343063005 + FL223063045) in U.S. corporate bonds and foreign entity bonds (issued through U.S. dealers and purchased by U.S. residents) divided by their total outstanding amount (FL893163005). Regression (5) and (6) include Interest Volatility as a regressor to explain the maturity trend. Interest Volatility is the standard deviation of daily interest rate of the 10-year Treasury bond measured in the quarter before the bond issuance. Bond Characteristics Fixed Effects control for Callable Dummy, Floating Dummy, Convertible Dummy, Puttable Dummy, Sinking Fund Dummy, and Global Dummy variables. Refer to Appendix A for the remaining variable definitions and data sources, and Tables 1 and 4 for sample selection criteria. Standard errors used to compute t-statistics in parentheses are clustered at the firm level. Model: Variables Trend x100 Cash/Assets Pension Fund Market Share Interest Volatility

Number of Observations Adjusted R2 (%)

Cash/Assets Pension Fund Market Share (1) (2) (3) (4) -1.04*** -0.85*** -1.07*** -0.92*** (-4.93) (-3.08) (-4.40) (-3.00) -0.31*** -0.23 . . (-3.48) (-1.44) . . . . 0.10 -0.03 . . (0.38) (-0.08) . . . . . . . . 19101 22.28

19101 17.23

19101 22.15

Firm Fixed Effects No Yes No Bond Characteristics Fixed Effects Yes Yes Yes Firm, Bond, and Macro Controls Yes Yes Yes Bond Credit Rating Fixed Effects Yes Yes Yes Industry Fixed Effects Yes No Yes Firm IPO Decade Fixed Effects Yes No Yes Year Fixed Effects Yes Yes Yes ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

41

Interest Volatility (5) (6) -1.12*** -0.92*** (-5.39) (-3.35) . . . . . . . . -0.06 -0.06 (-1.10) (-1.06)

19101 17.21

19101 22.16

19101 17.21

Yes Yes Yes Yes No No Yes

No Yes Yes Yes Yes Yes Yes

Yes Yes Yes Yes No No Yes

Table 10: Robustness Tests This table presents the results from regressions that investigate the robustness of the findings reported in Tables 3 and 4. All regressions are OLS models of Log(Bond Maturity), but fixed effects regressions also produce similar results. Regressions (1)–(7) reestimate Regression (3) in Table 3, respectively, for the following specifications: (1) controlling for Trend Square as an additional regressor, (2) analyzing the subsample of bonds issued during the 1975–1994 period, (3) analyzing the subsample of bonds issued during the 1995–2015 period, (4) replacing the year-based Trend variable with a more continuous measure that equals the number of years between the first day of our sample period (January 1st,1975) and the bond offering date, (5) adjusting the values of Offering Amount and Market Value of Equity for inflation (into 2015 U.S. dollars), (6) controlling for Offering Amount/MVE instead of Log(Offering Amount), and (7) controlling for underwriter fixed effects.. Regressions (8) and (9) reestimate Regressions (3) and (5) in Table 3 by dropping Trend variable. Regressions (10) and (11) reestimate Regressions (1) and (5) in Table 4, respectively, by controlling Bond Liquidity, which is defined as the proportion of non-zero trading days observed in TRACE database (available since July 2002) during the post 3-month period following the offering date. Refer to Appendix A for the remaining variable definitions and data sources, and Tables 1 and 4 for sample selection criteria. Standard errors used to compute t-statistics in parentheses are clustered at the firm level. All regressions include firm, bond, and macro-level controls, and the fixed effects for bond characteristics, credit ratings, IPO years, and industries. Regressions (10) and (11) also include year fixed effects. Nonlinear Two Subsample Continuous Trend Trend Periods Variables (1) (2) (3) (4) -0.43 0.48 0.18 . Trend x 100 (-0.77) (0.53) (0.44) . 1.81*** 2.28* 1.85*** 1.83*** Insurer Market Share (4.54) (1.94) (4.17) (4.79) 0.01 . . . Trend Square x 100 (0.84) . . . . . . -0.00 Continuous Trend . . . (-0.30) Mutual Fund Market Share . . . . . . . . Insurer Ownership . . . . . . . . Mutual Fund Ownership . . . . . . . . Model:

Adjust Inflation (5) 0.03 (0.11) 1.84*** (4.77) . . . . . . . . . .

19101 5446 13655 19101 19101 Number of Observations 2 22.33 28.95 20.50 22.33 22.33 Adjusted R ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

42

Control AMT/MVE (6) -0.03 (-0.10) 1.81*** (4.71) . . . . . . . . . . 19101 22.21

Control U.W. Insurer F.E. Market Share (7) (8) 0.34 . (1.06) . 1.87*** 1.92*** (4.81) (7.20) . . . . . . . . . . . . . . . . . . . . 19101 23.52

19101 22.34

Mutual Control Bond Fund Liquidity Market (9) (10) (11) Share . . . . . . . . . . . . . . . . . . . . . . . . -1.08** . . (-2.20) . . . 0.66** . * . (8.39) . . . 0.43** . . (-4.75) * 19101 4855 4855 21.91 26.24 24.65

4

45

35

3

30 25 2 20 15 1

10

Market Share of Institutions (%)

Average Log(Bond Maturity)

40

5 0 1975

0 1980

1985

Log(Bond Maturity)

1990

1995

2000

Insurer Market Share

2005

2010

2015

Mutual Fund Market Share

Figure 1. Trends in Bond Maturity and Market Shares of Insurers and Mutual Funds This figure reports the average Log(Bond Maturity) for our sample of 19,101 new bond issues from 1975 to 2015, together with the evolution of insurer and mutual fund shares in the corporate bond market. Refer to Appendix A for details on variable definitions and data sources, and Table 1 for sample selection criteria.

43

1.00

Callable Bonds

0.50

Puttable Bonds

0.80

0.40

0.60

0.30

0.40

0.20

0.20

0.10

0.00

0.00 1975 1980 1985 1990 1995 2000 2005 2010 2015

0.20

Floating Bonds

1975 1980 1985 1990 1995 2000 2005 2010 2015

0.40

0.15

0.30

0.10

0.20

0.05

0.10

0.00

0.00 1975 1980 1985 1990 1995 2000 2005 2010 2015

1.00

Convertible Bonds

Sinking Fund Bonds

1975 1980 1985 1990 1995 2000 2005 2010 2015

0.50

Global Bonds

0.40

0.80

0.30

0.60 0.40

0.20

0.20

0.10

0.00

0.00

1975 1980 1985 1990 1995 2000 2005 2010 2015

1975 1980 1985 1990 1995 2000 2005 2010 2015

Figure 2. Trends in Bond Characteristics This figure reports the percentage of bonds with certain features in our sample of 19,101 new bond issues by years. Refer to Appendix A for details on variable definitions and data sources, and Table 1 for sample selection criteria.

44