The set of property and casualty insurance and reinsurance companies that are considered affected by hurricane Katrina is identified using data from the Holborn Corporation (2005) hurricane Katrina report, publicly available at the URL: http://partners.holborn.com/holborn/reports/katrina09-21-2005/Katrina2005.pdf (website accessed on 25 September 2014). The Holborn Corporation is an independent reinsurance brokerage firm, providing analytical tools, market access, and account services to U.S. clients. The hurricane Katrina report provides estimates of the impact of Katrina on the property and casualty insurance and reinsurance market, as well as information on individual insurers regarding the AM Best and Standard & Poor’s actions on the insurers in the aftermath of the hurricane, and net and gross loss estimates. The Holborn Corporation (2005) report lists the names of property and casualty (re)insurance companies along with 2004 insurance market shares in Louisiana, Mississippi, and Alabama, and whether they have rating or outlook changes immediately after the hurricane. The set of Katrina-affected (re)insurance companies is identified as follows. First, we consider the set of insurance companies with the largest 2004 market share in the Gulf states hit by hurricane Katrina. Altogether, these insurance companies account for nearly 70% of the property and casualty insurance market in these states. Second, we consider the entire set of reinsurance companies that experience a downgrade, or are put on negative watch, in the aftermath of hurricane Katrina (no information about reinsurance market shares is provided in the Holborn corporation report, and to the best of our knowledge, this information is not available in general). (Re)Insurance company names are then screened, and manually matched to the Lipper EMAXX database. The Katrina-affected (re)insurance companies are then defined as:
The top 10 property and casualty insurance companies, by their 2004 market share in the Gulf states (including both personal and commercial lines);
A set of 8 reinsurance companies (no other reinsurance companies with a match in the Lipper EMAXX database were downgraded or put on negative watch over the period).
We now report the list of the property and casualty (re)insurance companies thus identified, along with their market shares from the Holborn Corporation (2005) report.
B.1 Investor characteristics and investor-related variables
Bondholder concentration (H) For a given issuer-seniority class group of bonds, bondholder concentration is defined as the Herfindahl index of concentration of the holdings of institutional investors:
where denotes the par amount holdings of investor i (total par amount of the holdings reported in the Lipper EMAXX database), Outstanding Amt is the outstanding amount of the issuer’s bonds in the seniority class under consideration (equal to the total of the offering amount for the bonds, from the Mergent FISD database),and is the fraction of the outstanding amount controlled by investor i. At the issuer level of aggregation, two versions of H are considered. The first is an equal-weighted average of the seniority class level values (Equal-weighted H), and the second is a weighted average, where the weights are proportional to the total holdings in each seniority class of institutional investors in Lipper EMAXX (Holdings-weighted H).
Concentration of the Katrina-affected bondholders () Bondholder concentration H is decomposed into one component related to the holdings of property and casualty (re)insurance companies affected by Katrina (identified based on the Holborn Corporation (2005) report as described in the text and in Appendix A), , and a residual component , according to the formula:
where h denotes the fraction of the bonds in a given seniority class cell held by a given investor. Just as for the overall bondholder concentration H, two versions of are considered at the issuer level. The first is an equal-weighted average of the seniority class level values (Equal-weighted), and the second is a weighted average, where the weights are proportional to the total holdings in each seniority class of institutional investors in the Lipper EMAXX database (Holdings-weighted ).
Institutional investors holdings At the issuer-seniority class aggregation level, it is the fraction of the total outstanding bonds held by institutional investors in the Lipper EMAXX database. It is obtained by dividing the total par amount held by the investors by the offering amount indicated in the Mergent FISD database. If the total institutional holdings in the Lipper EMAXX database exceed the total outstanding amount indicated by the Mergent FISD database, the EMAXX value is set equal to the Mergent FISD value. Institutional investors holdings is set to zero for bonds for which no institutional holdings can be found in the Lipper EMAXX database.
Size Natural logarithm of total assets (Compustat quarterly data item ATQ, expressed in millions of dollars).
Leverage Book leverage of the firm, defined as the ratio of total debt to total assets. Total debt is equal to total long-term debt (Compustat quarterly data item DLTTQ) plus debt in current liabilities (Compustat quarterly data item DLCQ) divided by total assets (Compustat quarterly data item ATQ).
Tobin’s q Tobin’s q is proxied by the firm’s market-to-book value. The firm’s market value is defined as the ratio of total assets (Compustat quarterly data item ATQ) minus common/ordinary equity, book value (Compustat quarterly data item CEQQ) plus the market value of equity, defined as the product of common shares outstanding (Compustat quarterly data item CSHOQ) times the quarter-end stock price (Compustat quarterly data item PRCCQ). The market-to-book value ratio is the ratio of the firm’s market value to total assets (Compustat quarterly data item ATQ). In order to limit the influence of outliers, on each quarter all observations are sorted by Tobin’s q, and values above the 99th percentile are discarded.
Industry The firm’s industry is defined based on the first digit of the firm’s historical SIC code (Compustat annual data item SICCH).
Hit state Indicator variable equal to 1 if the firm’s headquarters are located in a state hit by hurricane Katrina: Alabama, Florida, Georgia, Louisiana, and Mississippi. The set of states affected by the hurricane is identified from the Holborn Corporation (2005) report, and confirmed by Knabb, Rhome, and Brown (2005). The firm’s headquarters’ location is retrieved from the contact information in the Compustat fundamentals annual database.
CDS spread The credit default swap (CDS) spread associated with a given bond issuer. The data on CDS spreads are obtained from the Markit CDS database. The Markit CDS database provides CDS spreads for contract with maturities ranging between 6 months and 30 years. To focus on the most liquid CDS contracts, 5-year maturity contracts with modified restructuring (MR) clause are selected, and it is required that for each issuer the 10-year contract is also available.
Debt maturity A proxy for the time to maturity of the firm’s overall debt. It is defined as the ratio of long-term debt (Compustat quarterly data item DLTTQ) to long plus short-term debt (short-term debt is Compustat data item DLCQ).
Excess cash Cash in excess of short-term debt, defined as cash (Compustat quarterly data item CHEQ) minus short-term debt (DLCQ) divided by total assets (ATQ).
Excess cash* Cash in excess of short-term debt and dividends, defined as Excess cash minus cash dividends (quarterly change in Compustat item DVY divided by total assets).
Dividends Dividend payout ratio, defined as total dividends (quarterly change in Compustat item DVY plus preferred dividends, DVPQ) divided by lagged total assets (ATQ).
Payout inflexibility A proxy for the inflexibility of the payout ratio. Following Bonaime, Hankins, and Harford (2014), the payout ratio is less flexible when it involves more dividends and less share repurchases. The proxy is thus defined as the ratio of total dividends (quarterly change in Compustat item DVY plus preferred dividends, DVPQ) to total dividends plus repurchases (repurchases are defined as the quarterly change in Compustat item PRSTKCY).
B.3 Bond characteristics
The following variables represent corporate bond characteristics. Throughout the analysis, they are aggregated at the issuer-seniority class level or at the issuer level, as described in the text.
Option-Adjusted Yield Spread (Yield spread) Bond yield spreads are obtained as the option-adjusted yield spread reported in the Bank of America Merrill Lynch US Corporate and High Yield Master Bond Index Database. The option-adjusted yield spread measures the amount by which a risk-free spot curve must be raised or lowered so that the resulting discounted cash flows equal the market price of the bond. It simultaneously considers credit risk and contingent cash flow risk, such that bonds with different cash flow characteristics can be compared on a more equal basis. At the issuer-seniority class (issuer) aggregation level, the yield spread is the average of the option-adjusted yield spreads of all the bonds of a given issuer in a given seniority class (of the seniority class level values). In order to limit the influence of outliers, on each date all the bonds are sorted by their Option-adjusted Yield Spread, and values below the 1st or above the 99th percentile are set to missing.
log(Years to Maturity) Natural logarithm of 1 plus the number of years to the maturity of the bond. At the issuer-seniority class aggregation level, it is the outstanding amount-weighted average of the corresponding values for the individual bonds. At the issuer level, it is the outstanding amount-weighted average of the issuer-seniority class level values.
Speculative grade (Y/N), Rating indicators Bond ratings are the Moody’s rating reported in the Bank of America Merrill Lynch US Corporate and High Yield Master Bond Index Database. Two versions of bond rating indicators are considered. The “coarse” rating indicator Speculative grade (Y/N) is an indicator variable equal to 1 if the bond is of speculative grade (rating equal to BBB3 or lower), and 0 otherwise. The “fine” set of rating indicators contains indicators for each rating class.
Seniority class Each bond’s seniority class is retrieved as the Security_level variable from the Mergent FISD ISSUE table. There are five seniority classes: Senior secured, Senior, Senior subordinate, Subordinate, Junior.
log(1 + Trades) Natural logarithm of 1 plus the value-weighted average number of trades on a given issuer’s bonds reported in the TRACE database. Trades is defined as the weighted average number of trades on the bonds of a given issuer retrieved from the TRACE database, where the weights are proportional to each bond’s outstanding amount (as indicated in the Mergent FISD database). Whenever no trades associated with a given bond are reported in the TRACE database, Trades is set equal to 0.
CDS-bond basis The CDS-bond basis is defined as the difference between the spread on CDS contracts covering a given issuer and the average yield spread (defined below) on the issuer’s outstanding bonds. The CDS contracts have 5-year maturity and modified restructuring clause. In addition, when computing the CDS-bond basis, the sample is restricted to a set of corporate bonds with maturities bracketing the 5-year maturity associated with the CDS contract, namely with time to maturity between 2.5 and 7.5 years.
Investor portfolio duration A measure of the portfolio duration of the average investor holding a given bond. It is constructed in two steps. First, for each investor in each quarter a weighted average time to maturity (in months) is computed. Second, for a given bond the investor portfolio duration is computed, as the weighted average of the portfolio durations of all the investors holding the bond. For a given issuer-seniority class and quarter, Investor portfolio duration is the average of the corresponding bond-level portfolio durations.
Illiquidity For each bond and each quarter, it is equal to 10 times the ratio of the absolute value of the return on the bond divided by the trading volume on the bond across all institutional investors in the Lipper EMAXX database, in the spirit of Amihud (2002). The trading volume on the bond across all institutional investors, in turn, is the sum of the absolute values of the net quarterly changes in holdings of the bond across all the investors. For a given issuer-seniority class, it is a weighted average of the individual bonds’ illiquidity, where the weights are proportional to the outstanding value of each bond.
Trading volume For each bond and each quarter, it is the sum of the absolute values of the net quarterly changes in holdings of the bond across all the investors in the Lipper EMAXX database, divided by the total outstanding amount held by these investors. For a given issuer-seniority class, it is a weighted average of the individual bonds’ trading volume, where the weights are proportional to the outstanding value of each bond.
Additional results and robustness checks Table C.I Effect of hurricane Katrina on bondholder concentration – issuer-level regressions
The table reports the estimates of a model:
where H denotes bondholder concentration, the component of bondholder concentration associated with property and casualty (re)insurance companies affected by hurricane Katrina, Kat is an indicator variable equal to 1 over the period affected by hurricane Katrina, and x is a set of control variables, described below. Table III in the main text reports issuer-seniority level regressions, while this table reports the corresponding issuer-level regressions. Panel A reports estimates where both bondholder concentration and its component due to Katrina-affected investors are weighted averages of the corresponding issuer-seniority class level variables, with weights proportional to the holdings of all institutional investors in each seniority class. Panel B reports estimates where and are equal-weighted averages of the corresponding issuer-seniority class level variables. In both panels, the (re)insurance companies directly affected by hurricane Katrina are identified based on information from the Holborn Corporation (2005) report, as described in the text and in Appendix A. The indicator variable Kat is set equal to 1 over a one-, two-, three-, and four-quarter period following hurricane Katrina (August 2005), and the corresponding estimates are reported in columns (1), (2), (3), and (4) respectively. The vector x includes the following variables, whose coefficients have been omitted for convenient tabulation: (lagged) Size, Leverage, Tobin’s q, the indicator variable Hit state equal to 1 if the firm is located in a US state directly hit by hurricane Katrina (Alabama, Florida, Georgia, Louisiana, Mississippi) and an interaction term , the average log-Years to Maturity of the bonds, as well as industry (one-digit SIC code), rating, and calendar quarter indicators, and issuer-seniority level (issuer in Panels B and C) fixed effects. All variables are described in detail in Appendix B. Following Bertrand, Duflo, and Mullainathan (2004) and Petersen (2009) the t-statistics are based on standard errors clustered around issuers. The row labeled F-test reports the F-test statistic for the null hypothesis that the coefficient on the interaction term is equal to 0. The sample consists of all the issuers with corporate bonds in the Bank of America Merrill Lynch Corporate and High Yield Master Bond Index Database over the period between 1998Q1 and 2007Q4, excluding 24 bonds issued by (re)insurance companies affected by hurricane Katrina as described in Appendix A. To include a given issuer in the sample, we require the availability of bond characteristics in the Mergent Fixed Income Securities Database (FISD) and firm characteristics in the Compustat Fundamentals quarterly database. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels.