The Value of Catastrophe Securitization Bobby Bierley, Jim Hilliard and Rob Hoyt

Transcription

1 The Value of Catastrophe Securitization Bobby Bierley, Jim Hilliard and Rob Hoyt Institutstag IVW an der Uni Köln 6. Juni 2011

2 The Georgia RMI Program #2 RMI Program nationally in the U.S. News Rankings Largest RMI Program in the U.S. Risk Management Magazine 129 graduates last year Risk management and insurance emphasis in the MBA program Significant Ph.D. program

3 Introduction Agenda Catastrophe Bond Design & Structure The Study Purpose & Motivation Data, Methodology, & Variables Results Conclusions

4 Introduction

5 Annual Catastrophe Bond Transactions Volume Q $1 billion v. $650 million in Q Risk Capital Issued Number of Issuances Risk Capital Issues ($ Mill) $8,000 $7,000 $6,000 $5,000 $4,000 $3,000 $2,000 $1,000 Catastrophe bond issuance soared in the wake of Hurricanes Katrina and the hurricane seasons of 2004/2005, but retrenched during the financial crisis of 2008 $846.1 $984.8$1,139.0 $1,219.5 $966.9 $633.0 $1,729.8 $1,142.8 $1,991.1 $4,693.4 $7,329.6 $2,700.0 $3,400.0 $4, Number of Issuances $ Source: MMC Securities Guy Carpenter, A.M. Best; Insurance Information Institute.

6 Literature Review

7 Literature Review Corporate Demand for Insurance Mayers and Smith (1982, 1990) Demand for Insurance is a function of: tax structure expected bankruptcy and financial distress costs ownership structure investment incentives information asymmetry comparative advantage in real services Securitization is an alternative (or complement) to standard insurance, useful if it provides the advantages of insurance at a lower cost or with capacity unavailable in the insurance markets

8 Categories of Literature The Market & History Cummins (2007, 2008), McGhee et al. (2007, 2008), Cummins (1999), and Froot (2001) Design & Structure Tynes (2000), Ali (2000), Borden and Sarkar (1996) Technical Discussion Chichilnisky and Heal (1998), Cummins (2004) Our Contribution: Add to limited empirical research on catastrophe securitization

9 Catastrophe Bond Design & Structure

10 What is Catastrophe Securitization? Def: Catastrophe Bonds are fully collateralized debt securities that pay off on the occurrence of a defined catastrophic event. Catastrophe Bonds typically cover specified perils (earthquakes, hurricanes, typhoons) within specific geographic areas (California, Gulf of Mexico, East Coast, Japan, Europe, etc.)

11 Why are Catastrophe Bonds Used? CAT Bonds have been utilized primarily to provide risk transfer capacity for the Sponsor/Cedent entity s layer of loss that attaches excess of a one-in-100 year event (prob = 1%) and aggregates at the one-in-250 year event (prob <.4%) Why is this done? USAA May 2009 issue for 1-in-500 year event 1. Counterparty credit issues of reinsurers at this level 2. Reinsurance is often uneconomical at this level

12 Monthly evaluations now weekly or daily Basic Structure Quality and duration had been a problem in some deals higher quality and matched durations now common Source: MMC Securities and Guy Carpenter.

13 Basis risk Risk Trade-offs mismatch between the risk outcomes and the payoffs on the hedge Moral hazard intentional actions by the insured that make the payoff more likely or larger

14 Typical Catastrophe Bond Components Bond Term: Usually 2 to 4 years Multi-Peril vs. Single Peril Payout Triggers (3 Categories): Indemnity Triggers Index Triggers (parametric, industry loss, and modeled loss) Hybrid Triggers

15 Advantages of Catastrophe Bonds Compared to Reinsurance Potential losses are fully collateralized (reinsurance cycling not a factor) Able to lock in a price and capacity for multiple years (usually 3 years) Open up an area of new risk capital Minimize the impact of insurance market shocks from market cycles Availability in higher risk layers

16 The Study

17 Purpose & Motivation To determine whether catastrophe security issuances follow the theoretical explanations for the corporate demand for risk management (do they add value?) To examine which firm- and issue-specific factors explain the firm value impacts observed upon issuance

18 Data, Methodology & Variables Sample: 90 combined catastrophe securitization transactions between 1997 and 2011 combined because some transactions included multiple tranches being announced on the same announcement date Source: Market Reports, Moody s, WRDS, CRSP, Factiva, Business Source Elite, Global Insight, Company Websites Removed: 1) M&A transactions when stock was not listed, 2) Takedowns, 3) Conflicting data with regard to event date or issuers, and 4) Non-cat non-life issuances e.g. motor insurance

19

20

21 Data, Methodology & Variables (con t) Methodologies: Event Study and OLS Regression Foundation: Efficient Market Theory (CAPM) Event: Catastrophe Bond Issuances Estimation Period: t-175 to t-20 Various Event Windows: including (-10,+10), (-5,+5), (-1,0) and (-1,+1) Event: (earliest of press releases, news releases, or ratings announcements)

22 Empirical Model (Part 1) Market Model Event Study: R R ijt AR t Cumulative Abnormal Returns mjt Test Statistics: 1. Rank Test z 2. Standardized Cross-sectional z Test 3. Generalized z Test. R ijt i mjt i ijt R CAR i AR, T1, T 2 T 2 i t T i 1 it

23 Empirical Model (Part 2) OLS Regression: CAR Cumulative Abnormal Returns (Dependent Variable): CAR i, T, T BX t T i Independent Variables and Predictions: Market Cycle (-) Firm Size (-) U.S. Issuer (+/-) Non-insurer Issuer (+) Relative Issue Size (+/-) Trigger Type (-) with higher basis risk or moral hazard Single Peril (+/-) T i it i, T, T i AR

24 Market Model Results (all issuers, n=90) Days CAAR StdCsect z Generalized Sign z Rank Test Z (-10,+10) 1.69% (-5,+5) 0.70% (-1,0) -0.10% (-1,+1) 0.03% * Significant at the 10% level ** Significant at the 5% level *** Significant at the 1% level

25 Market Model Results (first issuers, n=29) Days CAAR StdCsect z Generalized Sign z Rank Test Z (-10,+10) 4.28% 2.880*** 2.230** 2.644*** (-5,+5) 1.21% (-1,0) -0.12% (-1,+1) 0.30% * Significant at the 10% level ** Significant at the 5% level *** Significant at the 1% level

26 Categorical Variables in the Model Variable N % of Issues Non-Insurer 2 2.2% Trigger Type: Parametric % Industry Index % Indemnity % Single Peril % Swiss Re % First Issue %

27 Cross-Sectional Model Results (N=90) (-10,+10) (-5,+5) (-1,+1) Intercept Market Cycle Firm Size US Firm * * ** ** Non-insurer * * * * Reliatve Issue Size Parametric Trigger Industry Index Trigger ** ** Indemnity Trigger ** ** ** ** Single Peril * First Issue ** ** Swiss Re Adjusted R-Square 13.75% 12.70% 8.03% 6.91% -2.46% -3.57%

28 Conclusions - Catastrophe bond market appears to be on a path to continued growth (recent disasters have not dampened the interest in this market) - Current securities regulation dictates that bond prospectuses for privately placed bonds can only be distributed to accredited investors as defined by the SEC (creates difficulty for researchers) - Evidence of positive returns that are statistically significant and economically meaningful for first issuers

29 Conclusions (con t) - U.S. insurer is significant and negative - May reflect higher spreads for catastrophe bond issuances in this market - Non-insurer issuer variable was significant and positive - issuance by non-insurance firms may be perceived as innovative by investors and only routine by insurance firms - Indemnity and Industry Index triggers are significant and negative relative to modeled loss trigger - Better balance of basis risk and moral hazard

30 Future Developments to Watch Solvency II and other regulatory capital frameworks are likely to generate more focus on capital requirements for tail risks Possible role for catastrophe bonds in addressing these new capital needs Issues faced by governments in addressing catastrophe risks Flood and coastal wind risks in U.S. (NFIP) Natural catastrophe exposures in developing and under-developed countries (World Bank)

31 Contact Information for the Risk Management and Insurance Program at the University of Georgia Department Head, Rob Hoyt Brooks Hall 206 Our web site