Transcript 3G Altuntas, Berry

Does One Size Fit All?
Determinants of Insurer Capital Structure Around
the Globe
Muhammed Altuntas
University of Cologne
Department of Risk Management & Insurance
Thomas R. Berry-Stölzle
University of Georgia
University of Cologne
Sabine Wende
University of Cologne
Slide: 1
Agenda
1
Introduction
2
Theoretical Background
3
Data & Methodology
4
Results
5
Conclusion
Department of Risk Management & Insurance
Slide: 2
Introduction
►
Since the International Association of Insurance Supervisors (IAIS) was
established in 1994, insurance regulators and supervisors from over 140
countries have been working on promoting globally consistent supervision of
the insurance industry.
►
The ComFrame working draft released on July 2, 2012 explicitly states that the
IAIS decided, “ComFrame shall develop a partially harmonized approach to
group capital for solvency assessment purposes” (IAIS, 2012b, p. 8).
►
►
Goal of this research: examine insurers’ capital structure across a broad range
of countries including those in developing markets (impact of country specific
characteristics on insurer capital structure).
We find that:
1.…the optimal capital structure of insurers is not homogeneous across
countries
2.…country-level factors explain a substantial fraction of the cross-sectional
variation in insurers’ capitalization levels
Department of Risk Management & Insurance
University of Cologne
Slide: 3
Introduction (II)
►
If insurer capital structure is not homogeneous across countries, imposing the
same regulatory capital requirements on all insurance companies around the
globe will not make sense.
►
Cheng and Weiss (2012) focus on the U.S. property-liability insurance industry
for the years 1994-2002 and examine firm-level determinants of insurer capital
structure.
►
The contribution of our study:
We examine both firm-level and country-level determinants of insurer capital
structure:
1. Research design extends the literature by providing novel evidence on the
impact of country characteristics on insurance companies’ capital structure
choice
2. Research contributes to the literature by providing additional evidence on the
effect of firm-level determinants on insurance companies’ capital structure
across a broad range of economies including those in the developing markets
Department of Risk Management & Insurance
University of Cologne
Slide: 4
Theoretical Background
►
Doherty and Tinic (1982) extend the classic Modigliani-Miller irrelevance result
and show that changing the level of capital inside an insurance company cannot
create value in a world with perfect capital markets.
►
In the presence of market imperfections, however, Froot (2008) shows that
insurance companies arrive at an optimal capital structure that trades off the
costs and benefits of holding capital.
►
Cost of raising new capital, and hence the benefit of holding capital, is
relatively high for opaque insurance companies which are hard to evaluate by
outside investors.
►
Similarly, the benefit of holding capital is relatively high for mutual insurance
companies because mutual insurers cannot raise capital from equity markets;
mutual insurers can only raise capital through the issuance of surplus notes, a
form of highly subordinated debt.
Department of Risk Management & Insurance
University of Cologne
Slide: 5
Theoretical Background (II)
►
Overall we expect larger, more diversified insurers, and insurers operating in less
volatile business lines to hold less capital, and mutual insurers and relatively
opaque insurers to hold more capital.
►
Coase (1937) and Williamson (1985) highlights that firms do not operate in a
vacuum and suggests that a firm’s institutional environment impacts the firm’s
optimal structure (transaction cost theory).
►
We argue that:
1. Optimal capital structure of insurers is not homogenous across countries
2. Country characteristics influence firms’ costs and benefits of holding capital
and, hence, their capital structure choice
3. Country-level factors that impact firms’ capital structure decisions are a
country’s degree of capital market development, the level of property
rights protection in the country, the transparency of a country’s
accounting system, and the costs associated with financial distress
Department of Risk Management & Insurance
University of Cologne
Slide: 6
Theoretical Background (III)
►
We expect that:
1. Holding capital is less valuable in countries with well-developed capital
markets where it is relatively easy to raise external capital
2. Holding capital is more valuable in countries with poorer property rights
protection, including investor protection
3. Holding capital is more valuable in countries with less transparent
accounting standards where the cost of raising capital is relatively high due
to agency costs and information asymmetries
4. Holding capital is more valuable in countries with relatively high financial
distress costs. Financial distress costs are especially high in countries where
individuals are risk averse and willing to pay a substantial premium for policies
of financially stable insurers.
Department of Risk Management & Insurance
University of Cologne
Slide: 7
Data & Methodology
Sample and Data:
►
There are substantial differences in the products offered by property-liability
insurance companies and life insurance companies and, hence, their capital structure.
►
Two firm-level samples of property-liability and life insurance companies,
respectively, from A.M. Best’s Statement File Global for the period 2001 through 2008.
►
Initial data: all listed insurers for the period 1999-2008, but database includes a large
number of data fragments without even basic information on the companies.
→ Exclude data entries for which the company description is missing
→ Exclude companies classified as reinsurers or pure holding companies
→ Exclude companies that report negative direct premiums written, premiums earned, total
assets, and policyholder surplus or investment positions
Department of Risk Management & Insurance
University of Cologne
Slide: 8
Data & Methodology (II)
►
Split the sample in two parts, separating the property-liability insurers from the life
insurers.
►
Exclude companies with missing data on the basic accounting variables used to
calculate the firm-level variables used in the regression analysis.
►
We use lagged values for some of our independent variables:
→ Exclude firm-year observations for which the preceding two years of data are not available
►
Finally, we exclude extreme outliers from the two samples. Our first outlier screen is
to eliminate firm-year observations with reported life (non-life) insurance premiums in
excess of the overall premium volume of the corresponding country’s life (nonlife) insurance market.
►
Next, we eliminate observations if the return on equity (ROE) has a value above one
or below minus one (Berger and Ofek, 1995).
Department of Risk Management & Insurance
University of Cologne
Slide: 9
Data & Methodology (III)
►
A.M. Best’s Statement File Global has a home country bias and overrepresents U.S.
insurers in the database.
→ We limit the number of unique U.S. insurance companies in our property-liability insurer sample
to 40% (life to 29%), which corresponds to the average world market share of U.S. insurers
across the 2001-2008 period
→ We randomly select insurance companies from the universe of all U.S. insurers until the total
number of U.S. insurers accounts for 40% (29%) of insurance companies in our sample, and we
remove all other U.S. insurers
►
►
Data for countries’ life and non-life insurance market premium volume are
obtained from Swiss Re’s Sigma publications.
Final Sample over the period 2001-2008:
→ P&L: 6,545 insurer-year observations from 28 different countries
→ Life: 2,001 insurer-year observations from 14 countries
Department of Risk Management & Insurance
University of Cologne
Slide: 10
Data & Methodology (IV)
Variance Decomposition Analysis of Firm-Level Determinants and Country Fixed
Effects
►
We conduct a variance decomposition analysis to assess the importance of firmlevel determinants of insurer capital structure relative to time-invariant country-level
factors (country fixed effects), see Lemmon et al. (2008):
Leveragei,c,t = α + β1Xi,c,t-1 + β2 Dc + β3 Dt + ε
►
(1)
Firm specific factors are:
Reinsurance
Stdev. of lossratio
Longtail business
Product mix
Premium growth
Size
Mutual
Group
Ratio of reinsurance ceded to reinsurance premiums assumed plus direct premiums.
Standard deviation of the net claims incurred divided by premiums earned (2000-2008).
Ratio of total gross provisions to sum of gross premiums.
Risk versus saving products.
Growth in net earned premiums.
Natural logarithm of the insurer’s total assets.
Dummy variable equal to one if the insurer is a mutual, and zero otherwise.
Dummy variable equal to one if the insurer is a member of a group, and zero otherwise.
Department of Risk Management & Insurance
University of Cologne
Slide: 11
Data & Methodology (V)
Quantifying the Impact of Specific Country-Level Factors on Capital Structure
►
►
►
►
Here, we focus on specific country characteristics rather than country fixed effects.
To examine the explanatory power of these country-level factors for firms’ leverage
levels, we include these country-level factors in a reduced form model of insurer
leverage and perform a variance decomposition analysis.
The business environment in a country may not just have a level effect on the
capital structure of all insurers operating in that country, but may also moderate the
relationship between firm-level factors and capital structure.
To capture any indirect effects of country characteristics on firms’ capital
structure, we include interaction terms between all firm-level variables and the
country-level factors into the model. The specification of the model is as follows:
Leveragei,c,t = α + β1Xi,c,t-1 + β2 Cc,t-1 + β3 X*C + ε
Department of Risk Management & Insurance
University of Cologne
(2)
Slide: 12
Data & Methodology (VI)
►
Ease of access to capital markets: We use the Credit to Private Sector variable, the
Country Credit Rating variable and a number of indices measuring Shareholder and
Creditor Rights to proxy for the accessibility of capital markets in a country. In
addition, we use a number of indices measuring Equity Disclosure Requirements to
proxy for the degree of asymmetric information in a country.
►
Cost of financial distress: We use measures of Uncertainty Avoidance and LongTerm Orientation in a country’s population and a Savings variable as proxies for
financial distress costs in a country.
►
Property rights protection: We use the Political Risk Index variable, the Strength of
Legal System Index variable and a number of indices measuring Government Quality
to proxy for the level of property rights protection in a country.
►
Competition: We use the Market Concentration and the Insurance Penetration
variables to proxy for the level of competition in a country.
►
Macroeconomic determinants: We use the Inflation rate and the GDP growth to
proxy for the macroeconomic conditions in a country.
Department of Risk Management & Insurance
University of Cologne
Slide: 13
Data & Methodology (VII)
A Partial Adjustment Model of Leverage
►
Here, we estimate a partial adjustment model of insurer leverage with country-firm
interaction terms:
Leveragei,c,t = α + Leverage i,c,t-1 + β1Xi,c,t-1 + β2 Cc,t-1 + β3 X*C + ε
►
(3)
Following Gungoraydinoglu and Öztekin’s (2011) analysis of leverage of nonfinancial
firms, we estimate Equation (3) with Blundell and Bond’s (1998) generalized method
of moments (GMM) estimator. Among the recently proposed alternative dynamic
panel data estimators, Blundell and Bond’s (1998) system GMM estimator is expected
to have the least bias for the model in Equation (3) (Flannery and Hankins, 2012).
Department of Risk Management & Insurance
University of Cologne
Slide: 14
Univariate Comparison between Insurers with High and Low
Leverage
Panel A: Property-Liability Insurance Companies
Firm
characteristics
Reinsurance
Std. dev. of loss ratio
Longtail business
Premium growth
Size
Mutual
Group
Full sample
N
Mean Median Std. Dev.
6,545 0.340
0.263
0.749
6,947 1.287
0.091
23.635
6,928 5.227
1.099
141.324
6,947 0.260
0.114
3.257
6,947 12.299 12.272
1.876
6,947 0.181
0.000
0.385
6,947 0.701
1.000
0.458
Firms with leverage
below sample median
N
Mean Median
3,270 0.370
0.292
3,474 1.699
0.105
3,460 8.099
0.823
3,474 0.220
0.096
3,474 11.610
11.564
3,474 0.248
0.000
3,474 0.654
1.000
N
3,275
3,473
3,468
3,473
3,473
3,473
3,473
Firms with leverage
above sample median
Mean
Median
0.311 ***
0.242 ***
0.876
0.079 ***
2.362 *
1.368 ***
0.300
0.137 ***
12.990 ***
12.985 ***
0.114 ***
0.000 ***
0.747 ***
1.000 ***
N
905
1,000
1,000
1,000
1,000
1,000
Firms with leverage
above sample median
Mean
Median
0.099 ***
0.027 ***
0.117 ***
0.092 ***
0.619
0.087
15.568 ***
15.626 ***
0.174 ***
0.000 ***
0.802 ***
1.000 ***
Panel B: Life-Health Insurance Companies
Firm
characteristics
Reinsurance
Product mix
Premium growth
Size
Mutual
Group
Full sample
N
Mean Median Std. Dev.
1,837 0.143
0.048
0.212
1,997 0.337
0.108
1.668
2,001 0.441
0.080
8.997
2,001 14.351 14.542
2.361
2,001 0.135
0.000
0.342
2,001 0.730
1.000
0.444
Department of Risk Management & Insurance
Firms with leverage
below sample median
N
Mean Median
932 0.187
0.078
997 0.558
0.140
1,001 0.264
0.074
1,001 13.135 12.908
1,001 0.097
0.000
1,001 0.658
1.000
University of Cologne
Slide: 15
Explanatory Power of Firm-Level Determinants, Country Fixed
Effects, and Year Fixed Effects (P&L)
Model (1)
Variance
OLS
Decomp. Coefficients
Firm-specific factors
Reinsurance
Std. dev. of loss ratio
Longtail business
Premium growth
Size
Mutual
Group
Year and country
fixed effects
Country FE
Year FE
Summary
Firm effect
Country effect
Year effect
Adj. R-squared
Number of
observations
Number of countries
1.87
0.01
0.07
0.02
73.40
16.00
8.62
0.42456
-0.00048
-0.00021
-0.00011
0.50078
-1.10768
-0.70926
***
***
***
***
Model (5)
Variance
OLS
Decomp. Coefficients
1.72
0.05
0.04
0.02
67.48
13.07
7.84
0.44503
-0.00103
-0.00017
-0.00011
0.53472
-1.09534
-0.74048
***
***
***
***
Model (7)
Variance
OLS
Decomp. Coefficients
3.51
0.04
0.01
0.03
18.95
3.57
0.45
-
9.79
69.19
4.24
100.00
-
90.21
9.79
26.57
69.19
4.24
Department of Risk Management & Insurance
0.82109
-0.00126
-0.00009
-0.00018
0.38630
-0.74994
-0.25300
0.12
0.14
0.31
6,947
6,947
6,947
28
28
28
University of Cologne
***
***
***
***
Slide: 16
Explanatory Power of Firm-Level Determinants, Country Fixed
Effects, and Year Fixed Effects (Life)
Model (1)
Variance
OLS
Decomp. Coefficients
Firm-specific factors
Reinsurance
Product mix
Premium growth
Size
Mutual
Group
Year and country
fixed effects
Country FE
Year FE
Summary
Firm effect
Country effect
Year effect
Adj. R-squared
Number of
observations
Number of countries
1.35
1.36
0.21
80.91
1.55
14.61
-5.79453
-0.81972
-0.02770
4.15336
3.79076
8.87953
**
**
***
**
***
Model (5)
OLS
Variance
Decomp. Coefficients
1.24
1.27
0.25
78.96
1.33
14.55
-5.68775
-0.80828
-0.03106
4.20148
3.58195
9.05716
Model (7)
OLS
Variance
Decomp. Coefficients
**
**
***
**
***
0.45
0.00
0.07
20.96
0.28
0.35
-
2.40
76.39
1.50
100.00
-
97.60
2.40
22.11
76.39
1.50
6.37128
-0.01919
-0.02806
4.17014
-3.00032
2.54677
0.21
0.21
0.57
2,001
2,001
2,001
14
14
14
Department of Risk Management & Insurance
University of Cologne
***
***
**
***
Slide: 17
Explanatory Power of Firm and Country-Level Determinants
Property-Liability Industry
Firm-specific factors
Reinsurance
Std. dev. of loss ratio
Longtail business
Premium growth
Size
Mutual
Group
Institutional factors
Direct institutional effect (I)
Reinsurance*(I)
Std. dev. of loss ratio*(I)
Longtail business*(I)
Premium change*(I)
Size*(I)
Mutual*(I)
Group*(I)
Summary
Firm effect
Direct institutional effect
Indirect institutional effect
Adj. R-squared
Number of observations
Number of countries
Average
Department of Risk Management & Insurance
5.73
4.45
0.30
0.07
29.56
6.82
11.48
3.92
6.94
4.56
0.30
0.08
7.18
4.59
14.02
58.42
3.92
37.66
0.17
6,920
27
Life Industry
Firm-specific factors
Reinsurance
Product mix
Premium growth
Size
Mutual
Group
Institutional factors
Direct institutional effect (I)
Reinsurance*(I)
Product mix*(I)
Premium change*(I)
Size*(I)
Mutual*(I)
Group*(I)
Summary
Firm effect
Direct institutional effect
Indirect institutional effect
Adj. R-squared
Number of observations
Number of countries
University of Cologne
Average
1.59
2.56
0.12
27.47
9.65
12.95
8.44
1.75
2.70
0.11
11.50
10.36
10.82
54.33
8.44
37.23
0.32
1,989
14
Slide: 18
Impact of Firm- and Country-Level Determinants on the CS
Property-Liability
Industry
Leverage
Reinsurance
Std. dev. of loss ratio
Longtail business
Premium growth
Size
Mutual
Group
Inflation
GDP growth
Institution (I)
Reinsurance*(I)
Std. dev. of loss ratio*(I)
Longtail business*(I)
Premium change*(I)
Size*(I)
Mutual*(I)
Group*(I)
Inflation*(I)
GDP growth*(I)
Country FE
Year FE
AR(1)/AR(2)
Sargan
Observations
Countries
Market
capitalization
0.67453***
0.29015
-0.01111***
-0.00245**
0.05615
-0.12960***
-0.30602**
-0.27780*
-0.01974***
0.36541
-0.02093***
-0.00167
0.00003**
0.00002**
0.00036
0.00147***
0.00065
0.00122
-0.00033**
0.05099**
Yes
Yes
0.00/0.66
0.15
5,279
27
Department of Risk Management & Insurance
Financial
efficiency
0.67684***
0.16035
0.02896
-0.00081
0.06128
-0.06661
0.27666
-0.43426
-0.02588**
11.55314***
-5.70218
-0.06820
-0.01767
0.00030
-0.02505
0.10752***
-0.30117
0.25394
0.00416
-4.08814***
Yes
Yes
0.00/0.60
0.64
5,199
24
Life-Industry
Leverage
Reinsurance
Product mix
Premium growth
Size
Mutual
Group
Inflation
GDP growth
Institution (I)
Reinsurance*(I)
Product mix*(I)
Premium change*(I)
Size*(I)
Mutual*(I)
Group*(I)
Inflation*(I)
GDP growth*(I)
Country FE
Year FE
AR(1)/AR(2)
Sargan
Observations
Countries
University of Cologne
Market
capitalization
0.85871***
11.93258***
0.94777***
0.83077***
0.81073***
-0.07830
-2.16598***
-0.11085
-6.16280
0.00342
-0.09155***
-0.00845***
-0.00142
-0.00281***
-0.02822***
0.01969***
0.00654***
0.26215***
Yes
Yes
0.00/0.73
0.45
1,509
14
Financial
efficiency
0.85458***
-3.96689**
0.60888
-1.57405***
-0.61915***
-18.14966***
1.10680
0.61948***
44.96211***
0.00863
5.95024***
-0.38246
0.94596***
0.29796***
5.95761***
0.65467***
-0.23128***
-28.41800***
Yes
Yes
0.00/0.90
0.40
1,489
13
Slide: 19
Conclusion
►
►
Previous research has ignored country-level determinants of insurer capital structure.
This research examines the determinants of insurance companies’ capital structure
across a broad range of economies including both, developed and emerging market
countries.
►
We find that country characteristics matter and that country-level factors explain a
substantial fraction of the variation in insurance companies’ capital structure.
►
This research has an important public policy implication. If insurer capital structure
is not homogeneous across countries, imposing the same regulatory capital
requirements on all insurance companies around would lead to substantial market
distortion.
►
Regulators should rather work on improving cooperation and information sharing
with respect to supervision of multinational insurance companies and on procedures to
quickly resolve multinational insurance companies then on harmonizing capital
standards across heterogeneous markets.
Department of Risk Management & Insurance
University of Cologne
Slide: 20
Back-up: Description of Firm- and Country-Level Determinants
Variable name
Variable description, and source
Firm-level determinants
Leverage
Book leverage; ratio of total assets minus capital surplus to capital surplus. Source: A.M. Best’s Statement File Global.
Reinsurance
Ratio of reinsurance ceded to reinsurance premiums assumed plus direct premiums. Source: A.M. Best’s Statement File Global.
Std. dev. of loss ratio
Standard deviation of the net claims incurred divided by premiums earned for the years 2000-2008. Source: A.M. Best’s Statement File Global.
Longtail business
Ratio of total gross provisions to sum of gross premiums. Source: A.M. Best’s Statement File Global.
Product mix
Risk versus saving products. Source: A.M. Best’s Statement File Global.
Premium growth
Growth in net earned premiums. Source: A.M. Best’s Statement File Global.
Size
Natural logarithm of the insurer’s total assets. Source: A.M. Best’s Statement File Global.
Mutual
Dummy variable equal to one if the insurer is a mutual, and zero otherwise. Source: A.M. Best’s Statement File Global.
Group
Dummy variable equal to one if the insurer is a member of a group, and zero otherwise. Source: A.M. Best’s Statement File Global.
Country-level determinants
Access to financial markets
Market capitalization
Market capitalization of listed companies (% of GDP). Source: World Development Indicators.
Financial efficiency
Financial system’s efficiency. Measured by the logarithm of the total value-traded ratio divided by overhead costs. The total value-traded ratio
captures the efficiency of stock markets and the overhead costs capture the efficiency of the banking sector. Source: Levine (2002).
Credit to private sector
Amount of credit banks provide to private sector as a percent of GDP. Source: World Development Indicators.
Country credit rating
Average of two ratings published semi-annually. The ratings are based on surveys of bankers and are on a scale from 0 to 100, with higher
values indicating a better rating. Source: Institutional Investor.
Shareholder rights
Anti-director rights index ranges from 0 (weak shareholder rights) to 5 (strong shareholder rights). Source: La Porta et al. (1998).
Creditor rights
Creditor rights index ranges from 0 (weak creditor rights) to 4 (strong creditor rights). Source: La Porta et al. (1998).
Corporate transparency The index is created by examining and rating companies’ 1990 annual reports on their inclusion or omission of 90 items. These items fall into
seven categories (general information, income statements, balance sheets, funds flow statement, accounting standards, stock data, and special
items). Source: La Porta et al. (1998).
Equity disclosure
The index equals the arithmetic mean of prospectus, compensation, shareholders, inside ownership, and transactions. Source: La Porta et al.
(2006).
Equity liability
The index equals the arithmetic mean of (1) liability standard for the issuer and its directors, (2) liability standard for distributors, and (3)
liability standard for accountants. Higher scores indicate stronger liability standards. Source: La Porta et al. (2006).
Department of Risk Management & Insurance
University of Cologne
Slide: 21
Back-up: Description of Firm- and Country-Level Determinants (II)
Variable name
Variable description, and source
Country-level determinants
Cost of financial distress
Uncertainty avoidance
Measures the extent to which people feel uncomfortable with uncertainty and ambiguity. Source: Hofstede et al. (2010) and Hofstede’s
Homepage.
Long-term orientation
Reflects long-term pragmatic attitudes versus short-term normative attitudes. Cultures scoring high on this dimension show emphasis on future
rewards, in particular saving, persistence and adapting to changing circumstances. Source: Hofstede et al. (2010) and Hofstede’s Homepage.
Savings
Gross savings in percent of GDP. Gross savings are calculated as gross national income less total consumption, plus net transfers. Source:
World Development Indicators.
Property rights protection
Government
Reflects perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political
effectiveness
pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies. This
index ranges from approximately -2.5 (weak) to 2.5 (strong) governance performance. Source: World Development Indicators.
Political risk index
Index is an assessment of government accountability and stability, quality of bureaucracy and law enforcement, investment climate, and
various sources of political and social conflicts. The index takes on values between zero and 100, with lower values representing unstable
institutions and higher risk. Source: PRS International Country Risk Guide Researchers dataset.
Strength of legal
Strength of legal rights index measures the degree to which collateral and bankruptcy laws protect the rights of borrowers and lenders and thus
System
facilitate lending. The index ranges from 0 to 10, with higher scores indicating that these laws are better designed to expand access to credit.
Source: World Development Indicators.
Time to enforce a
Procedures to enforce a contract are the number of independent actions, mandated by law or courts that demand interaction between the parties
contract
of a contract or between them and the judge or court officer. Source: World Development Indicators.
Competition
Market concentration
Market share of the 5 largest insurers. It is calculated as the sum of premiums earned for the 5 largest insurers in the sample divided by the
industry’s premiums written. Source: A.M. Best’s Statement File Global, and Swiss Re Sigma publications.
Insurance penetration
Insurance penetration is the ratio of the industry’s premiums written to GDP. Source: World Development Indicators, and Swiss Re Sigma
publications.
Macroeconomic determinants
Inflation rate
Annual inflation rate. Growth in Consumer Price Index (CPI). Source: World Development Indicators.
GDP growth
Economic growth. Growth in nominal Gross Domestic Product (GDP). Source: World Development Indicators.
Department of Risk Management & Insurance
University of Cologne
Slide: 22