The momentum effect on estimating the cost of equity capital for property-liability insurers Jennifer L. Wang (National Chengchi University) Joseph Tien.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Slide 1
The momentum effect on estimating the cost of equity capital
for property-liability insurers Jennifer L. Wang (National Chengchi
University) Joseph Tien (Tamkang University) NTUICF,
10/DEC/2010
Slide 2
Outline (1) Introduction: Motivation and Literature (1)
Introduction: Motivation and Literature (2) Data Resource and
Empirical Model (2) Data Resource and Empirical Model (3) Empirical
Results (3) Empirical Results (4) Conclusions (4) Conclusions
NTUICF, 10/DEC/2010
Slide 3
Motivation(1/2) Cost of capital estimation is becoming
important for insurers especially after financial crisis in 2008.
Cost of capital estimation is becoming important for insurers
especially after financial crisis in 2008. Moreover, insurers need
more equity capital if International Financial Reporting Standards
(IFRS) is executed. Moreover, insurers need more equity capital if
International Financial Reporting Standards (IFRS) is executed.
More applications of financial models are used in pricing,
reserving, ALM for insurance companies. More applications of
financial models are used in pricing, reserving, ALM for insurance
companies. NTUICF, 10/DEC/2010
Slide 4
Motivation(2/2) Fama-French(1997) suggest that the cost of
capital varies across industries due to heterogeneity of the risks
facing in various sectors of the economy. There is a significant
factor for insurance. Fama-French(1997) suggest that the cost of
capital varies across industries due to heterogeneity of the risks
facing in various sectors of the economy. There is a significant
factor for insurance. The supervisor needs the reasonable cost of
capital to enact the regulations. The supervisor needs the
reasonable cost of capital to enact the regulations. Few literature
discussed how to estimate costs of capital for insurers with
different business line compositions. Few literature discussed how
to estimate costs of capital for insurers with different business
line compositions. NTUICF, 10/DEC/2010
Slide 5
Literature Process 197019801990 S(1964),L(1965), B(1972) --
CAPM Refining Beta Estimation Most Results support CAPM Adding
different possible factors to explain returns Fama-French Three
Factor Model Possible reasons to explain Size, B/M and Momentum
Global market, Different industries Jegadeesh and Titman(1993) --
Momentum NTUICF, 10/DEC/2010
Slide 6
Literature(1/2) Cummins and Harrington(1985) suggest that beta
of property-liability insurers were unstable and conformed to the
CAPM in the 1980 s but not in the 1970 s. Cummins and
Harrington(1985) suggest that beta of property-liability insurers
were unstable and conformed to the CAPM in the 1980 s but not in
the 1970 s. Cummins and Lamm-Tennant(1994) figure an additional
factor, leverage, for empirical models. Their results suggest that
the long-tail commercial lines of property-liability insurance tend
to have higher costs of capital than short-tail lines. Cummins and
Lamm-Tennant(1994) figure an additional factor, leverage, for
empirical models. Their results suggest that the long-tail
commercial lines of property-liability insurance tend to have
higher costs of capital than short-tail lines. NTUICF,
10/DEC/2010
Slide 7
Literature(2/2) Lee and Cummins(1998) propose that APT and the
Wei(1988) model perform better than the CAPM in forecasting the
cost of capital for insurers. Lee and Cummins(1998) propose that
APT and the Wei(1988) model perform better than the CAPM in
forecasting the cost of capital for insurers. Using the Fama-French
model, Cummins and Phillips(2005) suggest the cost of capital for
insurers are significantly higher than estimation based upon the
CAPM. Using the Fama-French model, Cummins and Phillips(2005)
suggest the cost of capital for insurers are significantly higher
than estimation based upon the CAPM. Wang et al. (2008) use
Rubinstein-Leland (RL) model to improve the cost of equity
estimates of insurance companies due to the highly skewed and
heavy-tailed distributions associated with the insurance claims
process. Wang et al. (2008) use Rubinstein-Leland (RL) model to
improve the cost of equity estimates of insurance companies due to
the highly skewed and heavy-tailed distributions associated with
the insurance claims process. NTUICF, 10/DEC/2010
Slide 8
Purposes (1) To test whether the momentum factor plays the
significant role in estimating the cost of equity capital for PL
insurers. (1) To test whether the momentum factor plays the
significant role in estimating the cost of equity capital for PL
insurers. (2) Moreover, we further use FIB method to calculate the
capital cost for different lines of business. (2) Moreover, we
further use FIB method to calculate the capital cost for different
lines of business. (3) Finally, the sum-beta approach is adopted to
adjust the infrequent trading. (3) Finally, the sum-beta approach
is adopted to adjust the infrequent trading. NTUICF,
10/DEC/2010
Slide 9
Data and Sample Selection(1/3) To select property/casualty
insurance (NAICS code 524126) sample by North American Industry
Classification System (NAICS). To select property/casualty
insurance (NAICS code 524126) sample by North American Industry
Classification System (NAICS). The stock return of insurance
companies were obtained from CRSP. The stock return of insurance
companies were obtained from CRSP. To follow
Fama-Frenchs(1992,1997) screening rules To follow
Fama-Frenchs(1992,1997) screening rules (a) To eliminate firms
didnt have at least 36 consecutive months of return information
during estimation period (a) To eliminate firms didnt have at least
36 consecutive months of return information during estimation
period (b) The beta coefficients greater than 5 in absolute value.
(b) The beta coefficients greater than 5 in absolute value. NTUICF,
10/DEC/2010
Slide 10
Excess return data for market systematic risk, size, financial
distress (B/M ratio), and momentum factor were obtained from
Kenneth French s website. Excess return data for market systematic
risk, size, financial distress (B/M ratio), and momentum factor
were obtained from Kenneth French s website. Consequently, we
obtained data on insurance revenue by product lines from the NAIC
annual statement CD-ROMs. Consequently, we obtained data on
insurance revenue by product lines from the NAIC annual statement
CD-ROMs. Data and Sample Selection(2/3) NTUICF, 10/DEC/2010
Slide 11
Data and Sample Selection(3/3) Estimate betas Period: 1993-2001
Estimate betas Period: 1993-2001 We abstracted monthly return data
from CRSP and use excess return data for market systematic risk,
size, financial distress (B/M ratio), and momentum factor to
proceed the regression. We abstracted monthly return data from CRSP
and use excess return data for market systematic risk, size,
financial distress (B/M ratio), and momentum factor to proceed the
regression. Estimate the cost of equity: 1999-2001 Estimate the
cost of equity: 1999-2001 NTUICF, 10/DEC/2010
Slide 12
Estimation Methodology (1/4) -- CAPM Method the return on stock
I in period t the risk-free rate in period t(30-day Treasury bill
yield) the returm on the market portfolio in period t the CAPM beta
coefficient for firm i overall beta estimate for firm i net
premiums weight for firm I in business lines k full-information
beta of type j for business line k NTUICF, 10/DEC/2010
Slide 13
Estimation Methodology (2/4) -- FF3F Model the market risk
premium for firm size in period t the market risk premium for
financial distress in period t overall beta estimate of type j for
firm i j=m, s, v full-information beta of type j for business lines
k j=m, s, v net premiums weight for firm I in business lines k
NTUICF, 10/DEC/2010
Slide 14
Estimation Methodology (3/4) -- Momentum Model the momentum
factor in the period t overall beta estimate of type j for firm i
j=m, s, v, mo full-information beta of type j for business lines k
j=m, s, v, mo net premiums weight for firm I in business lines k
NTUICF, 10/DEC/2010
Slide 15
Estimation Methodology (4/4) -- Sum-Beta Approach In order to
correct for the bias created by infrequent trading, we utilize the
sum-beta approach that has become standard in this type of analysis
( e.g., Scoles and Williams(1977), Dimson(1979)) Adding the lagged
value of variable in model: The estimated sum beta coefficient is
NTUICF, 10/DEC/2010
Slide 16
= estimated beta coefficient in first regression for firm i, =
the cost of capital of CAPM for firm i, = the expected return of
the risk-free asset, the expected return on the market portfolio,
excess return on NYSE/AMEX/Nasdaq stocks from 1926 until June of
2001 To Estimate Cost of Equity (1/2) NTUICF, 10/DEC/2010
Slide 17
= the expected excess premium for size factor, = the expected
excess premium for financial factor, = the expected excess premium
for momentum factor To Estimate Cost of Equity (2/2) NTUICF,
10/DEC/2010
Slide 18
Empirical Results of Beta Estimations (1/2) Empirical Results
of Beta Estimations (1/2) (1) The quartile results do not show that
large insurers consistently have smaller beta than small insurers.
(2) The sum-beta estimates are constantly than the ordinary beta
coefficients because infrequent trading. than the ordinary beta
coefficients because infrequent trading. NTUICF, 10/DEC/2010
Slide 19
Slide 20
Slide 21
Slide 22
Empirical Results of Beta Estimations (2/2) (3) The market beta
estimated in FF3F is larger than in CAPM. (3) The market beta
estimated in FF3F is larger than in CAPM. (4) The market beta
estimated in momentum model is larger than in FF3F. (4) The market
beta estimated in momentum model is larger than in FF3F. (5)
Generally, the market beta or B/M beta is larger than the size
beta, the momentum beta is smallest. (5) Generally, the market beta
or B/M beta is larger than the size beta, the momentum beta is
smallest. NTUICF, 10/DEC/2010
Empirical Results Overall cost of equity (2/2) The cost of
equity estimated with sum-beta adjustment is larger than without
sum-beta method. The cost of equity estimated with sum-beta
adjustment is larger than without sum-beta method. The cost of
equity estimated from FF3F is larger than from CAPM. The cost of
equity estimated from FF3F is larger than from CAPM. (CAPM:10.5%,
FF3F:16.3%) (CAPM:10.5%, FF3F:16.3%) The cost of equity estimated
from momentum model is larger than from FF3F. (Momentum:22.56%) The
cost of equity estimated from momentum model is larger than from
FF3F. (Momentum:22.56%) NTUICF, 10/DEC/2010
Slide 27
Comparing with Cummins and Phillips (2005) Financial distress
betas are substantially larger than the parameters in the Fama-
French all industry average. (0.02) Financial distress betas are
substantially larger than the parameters in the Fama- French all
industry average. (0.02) Financial distress factor is significant
in estimating the cost of equity for property- liability insurers.
Financial distress factor is significant in estimating the cost of
equity for property- liability insurers. NTUICF, 10/DEC/2010
Slide 28
Our New Findings Although the momentum beta could be the
smallest of all beta estimations, the momentum effect still plays a
significant role in estimating cost of equity. Although the
momentum beta could be the smallest of all beta estimations, the
momentum effect still plays a significant role in estimating cost
of equity. The estimations of beta and cost equity are dropped
sharply in 2001. The estimations of beta and cost equity are
dropped sharply in 2001. NTUICF, 10/DEC/2010
Slide 29
Empirical Results for different lines of business (1) Long-tail
and Short-tail (1) Long-tail and Short-tail (2) Commercial and
Personal (2) Commercial and Personal (3) Workers Compensation,
Automobile Insurance and All Other Property-Liability Insurance (3)
Workers Compensation, Automobile Insurance and All Other
Property-Liability Insurance NTUICF, 10/DEC/2010
Slide 30
Results of Long-tail and Short-tail Long-tail other liability,
products liability (occurrence, claim made), private passenger auto
liability, aircraft, commercial auto liability. Long-tail other
liability, products liability (occurrence, claim made), private
passenger auto liability, aircraft, commercial auto liability.
Short-tail fire, allied lines, homeowners, multiperil, automobile
physical damage, accident and health coverage, fidelity, surety,
mortgage guaranty. Short-tail fire, allied lines, homeowners,
multiperil, automobile physical damage, accident and health
coverage, fidelity, surety, mortgage guaranty. NTUICF,
10/DEC/2010
Slide 31
Slide 32
Empirical Results of Commercial and Personal (1/2) Personal
Line- Personal Line- Homeowner, Framowner, Earthquake, Personal
Auto Liability, Homeowner, Framowner, Earthquake, Personal Auto
Liability, Personal Auto Damage Personal Auto Damage Commercial
Line- Commercial Line- All other lines of insurance are considered
commercial lines. All other lines of insurance are considered
commercial lines. NTUICF, 10/DEC/2010
Slide 33
Slide 34
Empirical Results of Commercial and Personal (2/2) Generally
Speaking, equity cost in commercial lines is larger than in
personal lines. Shareholders ask more risk compensation for
operating the commercial line products. Generally Speaking, equity
cost in commercial lines is larger than in personal lines.
Shareholders ask more risk compensation for operating the
commercial line products. CAPM CAPM FF3F FF3F Momt Momt Personal
Line 10.6512813.7862014.82667 Commercial Line
11.0261117.4460521.68727 NTUICF, 10/DEC/2010
Slide 35
Results of Workers , Auto and All other (1/2) CAPM FF3F
Momentum Automobile insurance 11.08254 11.92098 14.96861 Workers
compensation 9.84042 10.40837 15.49364 All other P&L lines of
insurance 10.73661 19.03513 20.88048 NTUICF, 10/DEC/2010
Slide 36
Results of Workers , Auto and All other (2/2) (1) The equity
cost of worker compensation is slightly larger than automobile. But
the equity cost of other P&L is largest among these three
categories. (1) The equity cost of worker compensation is slightly
larger than automobile. But the equity cost of other P&L is
largest among these three categories. NTUICF, 10/DEC/2010
Slide 37
Conclusions (1/2) (1) The cost of equity capital estimation
based the FF3F (mometum) method is significantly higher than the
estimation based on the CAPM (FF3F). (1) The cost of equity capital
estimation based the FF3F (mometum) method is significantly higher
than the estimation based on the CAPM (FF3F). (2) It is important
to adjust infrequent trading when estimating betas for PL insurers.
(2) It is important to adjust infrequent trading when estimating
betas for PL insurers. NTUICF, 10/DEC/2010
Slide 38
Conclusions (2/2) (3) The cost of equity capital varies
significantly by line of insurance. (3) The cost of equity capital
varies significantly by line of insurance. (4) Financial distress
factor is significant in estimating the cost of equity for
property- liability insurers. (4) Financial distress factor is
significant in estimating the cost of equity for property-
liability insurers. NTUICF, 10/DEC/2010