A Usa Grenee Sfa 2004
A Usa Grenee Sfa 2004
A Usa Grenee Sfa 2004
By
William H. Greene
New York University
wgreene@stern.nyu.edu
Dan Segal
University of Toronto
dsegal@rotman.utoronto.ca
We appreciate the helpful comments from Joshua Livnat, Ajay Maindiratta, Stephen Ryan, James Ohlson,
and workshop participants at the Hebrew University of Jerusalem, New York University, Yale University,
London Business School, and the University of Toronto. LOMA kindly provided some of the data.
I. Introduction
The alleged linkage between operating performance and financial success is actually
quite tenuous and uncertain Kaplan and Norton (1992)
The purpose of this study is to show the linkage of operating efficiency and crosssectional variation in firm profitability and growth (henceforth value drivers) in the US
life insurance industry. That industry can be characterized as mature and highly
competitive, with fairly homogeneous products and services and comparable providers of
insurance. Few financial inventions can be patented, and most innovations are copied
shortly after their introduction. Consequently, success in this industry depends on the
insurers ability to control costs and on various intangibles, such as clientele and
business-risk preference, marketing skills, reputation, and perceived quality of service.
Hence, we hypothesize that operational efficiency explains a significant portion of the
variation in profitability and growth across life insurance companies.
We also conjecture that the relations among profitability, growth, and operational
inefficiency are conditioned on organizational form. The two main organizational forms
of life insurance companies are mutual and stock companies. The owners of a mutual
company are its policyholders, while the owners of a stock company are its shareholders.
Jensen and Meckling (1976), Fama and Jensen (1985), and Mayers and Smith (1984,
1986) argue that firms with alternative ownership structures differ in their operations and
particularly in their cost of productions. Since the mutual form of ownership gives
insurance companies mechanisms for controlling and disciplining managers that are less
effective than those available under stock ownership (primarily because the potential risk
of takeovers does not exist for mutual companies) we hypothesize that stock companies
are more efficient than mutual companies and, therefore, more profitable and able to
grow faster.
The economics literature contains numerous studies of efficiency for many
industries, including the life insurance industry. However, to the best of our knowledge,
none of these studies examines the effect of inefficiency on profitability and growth. In
addition, the studies that investigate the effect of ownership on operational inefficiency in
the life insurance industry, such as Gardner and Grace (1993) and Cummins and Zi
(1998), find that mutual and stock companies are equally efficient. Yet these studies fail
to control for demutualization (the conversion from policyholder ownership to stock
ownership) or for policy mix. In addition, these studies may incorrectly specify the
production technology, and the tests they employ may be statistically inefficient.
Our results indicate that profitability and growth are negatively and significantly
correlated with inefficiency. In particular, we find that all of the measures of inefficiency
used in this study indicate unambiguously that efficient firms enjoy returns on assets and
equity, growth rates, and ratios of operating cash flows to assets that are higher than those
of inefficient firms. We also find that after adjusting net income for inefficiency in
operating expenses the differences in the returns on assets (ROA) and the returns on
equity (ROE) between efficient and inefficient firms become insignificant. In addition,
we show that firms with consistently high ROE and high ratios of operating cash flows to
assets (CA) are more efficient than firms with low ROE and CA. The analysis of the
relationship between organizational form and efficiency indicates that stock companies
are significantly more efficient than mutual companies, have higher growth rates, and are
more profitable.
inefficiency are suboptimal usage of the firms resources through overpaying for inputs
and through employing a technologically inferior operating process. Inefficiency causes
realizable levels of earnings and cash flows that are lower than those potentially feasible
with optimal operations. The adverse effects on earnings and cash flows translate into
lower firm value either through lower dividends or through lower investments that slow
the firms growth.
Although growth is an important value driver for all firms, it is of particular
significance for life insurance firms. The efficient operation of such firm requires
considerable economies of scale generated by business volume. Without growth, an
insurer may not garner the business volume necessary to ensure the collective pooling of
insurance risks under the law of large numbers upon which the insurance operation relies.
In the domestic market, growth is achieved primarily through expansion of distribution
systems and technology improvements. Another way for insurers to grow is through
global expansion. However, to provide for future growth, an insurance company must
generate and maintain sufficient capital to satisfy regulators as well as to finance its
expansion.
By the end of 1998, more than 90% of US life insurance companies were stock
companies, although mutual companies were, on average, larger than stock companies
and owned approximately 33% of the total industry assets and 40% of the total amount of
insurance. During the 1990s a growing number of mutual companies converted to stock
companies. The primary objective of demutualization is the potential for growth through
investments in capital and distribution channels and through mergers and acquisitions.2
Since it is likely that the problems associated with the mutual form of ownership are not
function of outputs and input prices, where deviation from the frontier are assumed to be
related to cost inefficiency and statistical noise.
To control for random error in the estimation and specification, the cost function
is typically specified with two error components:
ln Ci = ln C*(yi,pi) + ui + vi = ln C*(yi,pi) + i,
(1)
where i indexes the firms, Ci is the observed total costs for firm i, ln C*(yi,pi) is the log
cost function, yi is a vector of outputs, pi is a vector of input prices, ui is a one-sided error
term that captures cost inefficiency (ui0) and vi is a random error term that is assumed to
be normally distributed with zero mean and variance 2v. In addition, u and v are
assumed to be independent. From equation (1) it follows that exp (ui) = C/C*, so the cost
inefficiency the proportion by which the firm could have reduced its costs and still
attain the same level of outputs is computed as 1-exp(-ui).
The estimation of the stochastic frontier along with the inefficiency term involves
specifying the distribution of u as well as of the cost function. For the one-sided
inefficiency disturbance term ui, several distributions have been suggested, such as the
absolute value of a normal distribution with zero-mean (half-normal), the absolute value
of a normal distribution with nonzero mean (truncated normal), the exponential
distribution, and the gamma distribution. We use the zero-mean half-normal distribution.3
With the assumed independence of the distributions of vi and ui, the computation of the
distribution of and the maximum likelihood estimation are usually straightforward.4
We compute the firm-specific inefficiency, ui, which is not observed directly as the
conditional expectation E(ui|i) as in Jondrow et al. (1982).
We denote the conditional means of the inefficiency term under the half-normal
distributional assumption as UNOR. By construction, the conditional mean is greater or
equal to zero; the closer it is to zero, the more efficient is the firm. To estimate the
stochastic frontier, we use a translog with the homothetic technology cost function5 (see
Christensen and Greene (1976)):
ln Ct = 0 + 0LNAVPL + JJln(Pj) + JIJIln(Pj)ln(Pi) + mmYm +
mmYm2 + t,
(2)
where LNAVPL is the natural log of the average amount of insurance of life policies, Ym
is the natural-log of output m, Pi is the price of input i, and t indexes the sample firms. To
assure the linear homogeneity of the cost function in the factor prices, we divide each of
the prices and total costs by one of the prices.
10
insurance policies, only a portion of the premium covers the risk-bearing that life
insurance companies provide to the insured. The remaining portion covers the savings
element of the policy; it therefore actually belongs to the insured and cannot be
considered as revenue of the insurer.
Yungert (1993) measures outputs by additions to reserves. The major problem
with this measure is that reserves change when policies age, regardless of whether new
policies are sold. In addition, the change in reserves measures the change in liabilities,
rather than the output of the selling effort. In a more recent study, Cummins and Zi
(1998) distinguish between the two principal services provided by life insurance
companies: risk bearing/pooling, and intermediation services. As a measure of the
former, they use incurred benefits by line of business, whereas for the latter they use
additions to reserves. Here again the output measure is disputable. Benefits represent
obligations that were incurred in the past; hence they measure past cumulative output, not
current output.
Following Cummins and Zi (1998), we characterize the outputs by the service
provided. Life policies give either pure risk protection (through term life policies) or a
mix of risk protection and intermediation services (through whole life policies). Annuities
can be viewed as saving vehicles and, therefore, the service they provide can be
characterized as intermediation. A&H policies, on the other hand, provide risk protection
service alone.
The risk bearing/pooling services that companies provide on new life insurance
policies can be approximated by the total amount of insurance sold during the year.6 That
amount measures the outcome of the selling effort and the additional risk that the
11
company bears and, therefore, can represent the output of the life insurance line of
business.7 Furthermore, this output measure may be appropriate for all types of life
policies, both term life and whole life.
Profits and losses from annuities arise from the difference between the actual
return on investments and the return credited to the contracts. Assuming a positive
spread, the larger the annuity considerations (premiums) the larger is the expected profit.
Hence, a plausible proxy for this output is annuity considerations, which represent the
increase in the earning base of this line of business.
A&H policies primarily provide risk protection. Since one cannot quantify the
amount of risk associated with each new policy, we use A&H premiums as a proxy for
these policies output. In equilibrium, where the risk associated with A&H policies is
priced correctly, premiums serve as a good proxy for risk.
To sum up, we use three outputs: amount of life insurance, total annuity
considerations and total A&H premiums8.
Inputs and Inputs Prices
For this study we employ three inputs: labor, capital, and other. Labor is defined
as the number of employee-days. The price of labor is computed as the total cost of
employees and agents divided by their total number. Capital comprises two components:
financial capital, defined as book value of equity plus the asset valuation reserve (AVR);9
and physical capital, defined as the sum of capital expenses - rent, rental of equipment,
and depreciation.10 We define the price of capital as the opportunity cost of holding the
financial capital and measure it as the difference between the ratio of five years total net
12
income to total financial capital (return on equity) and the ratio of total investment
income to total assets (return on investments) over the same period.11,12
Our third input (other) consists of all operating expenses other than labor and
capital expenses. Most of these expenses are related directly to selling and servicing
policies. We use the number of policies sold and terminated during the year as a proxy
for the number of policies sold and serviced during the year. And we quantify the price of
this third input as the related expenses divided by the total number of policies sold or
terminated.13
Data
Life insurance companies are required to file two sets of financial statements.
One, intended primarily for shareholders, is prepared according to generally accepted
accounting principles (GAAP). The other, highly detailed and intended for regulators, is
prepared according to statutory accounting principles (SAP).
The primary interest of SAP is measuring the solvency of the firm--i.e, the
amount of capital needed to cover all obligations under extreme economic conditions,
emphasizing financial results under very conservative assumptions. The measurement of
operational inefficiency requires detailed financial information on the outputs and inputs
used in the production process. Given the importance of earnings according to SAP and
the level of detail prescribed by those principles, we use the regulatory reports in the
analysis.
We obtained the insurance financial data from the regulatory annual statements
filed by insurers and reported to the National Association of Insurance Commissioners
(NAIC) life insurance data tapes for 1995 through1998. Because the NAIC tapes do not
13
14
any choice of outputs. Firms that primarily issue long-term policies would appear to be
inefficient while firms that concentrate on short-term policies would appear as efficient
because the former incur higher acquisition costs for any given level of output.
To control for the type of the policies in the estimation of inefficiency we
consider and control for the insurers product mix--in particular, the relative weights of
long-term policies and short-term policies. To account for the mix of life policies we
construct a variable (mix ratio) to represent it: the ratio of total new whole life policies
amount of insurance to total term and whole new life policies amount of insurance.
We then classify the firms into two groups--those with a mix ratio greater than
half, and those with a mix ratio less than half--and estimate inefficiency separately for the
two groups.
To mitigate the influence of extreme variables on the results, we further exclude
firms with ROEs less than 50% or greater then 200% (7 observations), firms for which
we could not estimate the growth rate and firms with growth rate greater than 100% (10
observations). Thus, the final number of firm-year observations available for the analysis
of the association between inefficiency and profitability and growth is 472.
15
score of 0. We then compute each firms total score by summing its scores over all four
years. We define the group of firms with total scores of 0 or 1 as consistently inefficient,
and the group of firms with scores of 7 or 8 as consistently efficient.
For these two groups, we test whether their profitability and growth measures
differed significantly and in the expected direction. To do so, we examine the association
between inefficiency and the following measures: ROE, defined as the ratio of net
income in year (t) to the average book value of equity in years t and t-1; ROA, defined as
the ratio of net income in year t to the average of total assets in years t and t-1; CA, the
ratio of operating cash flows in year t to the average of total assets in years t and t-1; and
the two-year average growth (GR) in direct premium revenues. Formally, we test the
following hypotheses (stated in null form):
H1a: ROEEFFROEIEF
(ROE)
H1b: ROAEFFROAIEF
(ROA)
H1c: CAEFFCAIEF
(CA)
H1d: GREFFGRIEF
(GR),
where the suffix EFF (IEF) indicates consistently efficient (inefficient) firms. We use a
one-tail test for all hypotheses.
To test the hypothesis that firms that perform better in terms of profitability and
growth rate are more efficient than poorly performing firms, we repeat the methodology
with which we create portfolios of consistently efficient and inefficient firms and use it to
construct portfolios of firms with consistently high [low] ROA, ROE, GR, and CA,
denoted HROA [LROA], HROE [LROE], HGR [LGR], and HCA [LCA], respectively.
We then compute the average efficiency for each efficiency measure and test whether the
16
average efficiency of HROA, HROE, HGR, and HCA is greater respectively than that of
LROA, LROE, LGR, and LCA.
This leads to the next set of hypotheses (stated in null form):
H1e: Eff(HROA) Eff(LROA)
H1f: Eff(HROE) Eff(LROE)
H1h: Eff(HGR) Eff(LGR)
H1i: Eff(HCA) Eff(LCA),
where Eff is the average efficiency of UNOR and UEXP. Testing Hypotheses H1e
through H1i together with H1a through H1d as already stated would indicate whether a
significant relationship exists between inefficiency and ROA, ROE, GR, and CA. That is,
rejection of all of the null hypotheses would indicate that inefficiency has negative
impact on ROA, ROE, GR, and CA and conversely that firms with low ROA, ROE, GR,
and CA are also less efficient.
17
di = + 1STOCKi + 2MIXi,
(3)
where STOCK is a dummy variable valued at 1 for stock companies and 0 for mutual
companies, MIX is the policy-mix ratio, and i indexes the firms. (We add MIX to the
equation since we estimate the frontier over the entire data.15)
We then test the following hypotheses (stated in null form):
H2a: 10
H2b: 20
VI. Results
Table 1 provides descriptive statistics about the sample. Panel A of Table 1 shows
that the average size (total assets) of the sample firms ranges from $4,435 million in 1995
to $5,430 million in 1998. In 1998, the aggregate total assets of these firms were about
$657 billion, approximately a third of all assets in the industry. Thus, our sample covers a
material portion of all firms in the industry. Panel B of Table 1 presents the percentage of
direct premium revenues by line of business.
[Insert Table 1]
Table 2 demonstrates the effect of inefficiency on earnings. The table presents the
median and mean cost of inefficiency as a percentage of earnings before tax and as a
percentage of revenues, denoted EFFIN and EFFREV, respectively. We compute the cost
of inefficiency as one minus the exponent of -U times the inputs. We calculate the cost of
inefficiency in operating expenses, which comprise labor-related expenses (not including
commissions), physical capital, and all other expenses (Thus, our cost of inefficiency
does not include any inefficiency in the amount of financial capital held nor in the
commissions paid to agents16).
18
[Insert Table 2]
The median of EFFIN in 1995 through 1997 is around 60%; in 1998, EFFIN is
much higher- 75%. The median of EFFREV is stable across the period, indicating that the
cost of inefficiency as percentage of revenues is 4.5%. Hence, inefficiency is substantial
relative to earnings and revenues.
Table 3 sets out the average inefficiency of the sample firms. The mean
inefficiency over the entire period is approximately 38%. This finding is consistent with
those of Cummins and Zi (1998) and Yungert (1993), which also document inefficiency
in the range of 30% to about 40%.
[Insert Table 3]
Panel A of Table 4 shows the distribution of firms across the three efficiency
groups (consistently efficient, partially efficient, and consistently inefficient), as well as
the mean inefficiency of each group. About 25% of the firms are considered to be
consistently efficient, 27% consistently inefficient and the reminder partially efficient.
The average inefficiency of the consistently inefficient (efficient) firms is 54% (25%).
Panel B of Table 4 provides descriptive statistics about the profitability and
growth measures. The means of ROA, ROE, GR, and CA over the entire period are 1.8%,
10%, 7%, and 6%, respectively. Panel C of Table 4 shows the Spearman correlations of
the efficiency measure of the entire sample with the value drivers. All correlations are
positive, i.e., efficiency is positively associated with ROA, ROE, CA, and GR and, in
general, significant at 5%.
Panel D of Table 4 presents the mean and average Wilcoxon rank scores of the
value drivers of the consistently efficient and inefficient firms. For example, the mean
19
ROA of the consistently efficient (inefficient) firms is 2.3% (1.3%), and the Wilcoxon
rank is 105 (88). The profitability and growth measures are significantly, generally at the
5% level, higher for consistently efficient firms. Efficient firms have higher return on
assets, higher return on book value of equity, higher growth rate and higher ratio of
operating cash flows to total assets.
[Insert Table 4]
To test whether the differences in ROA and ROE can be explained by operational
inefficiency, we compute the yearly net income of the sample firms as if they were fully
efficient. For each firm, we add to net income and to operating cash flows the cost of
inefficiency in operating expenses after tax.17 We then test whether the adjusted ROA,
ROE, and CA differ between consistently efficient and inefficient firms. Panel E of Table
4 presents the mean and average Wilcoxon rank scores of the adjusted profitability
measures for the portfolios of consistently efficient and inefficient firms. The results
indicate that the differences in the adjusted ROA and ROE between the portfolios
become insignificant. The adjusted mean CA is still significantly (10%) higher for
efficient firms. Hence, these results appear to suggest that operational inefficiency
explains the differences in profitability between the consistently efficient and inefficient
firms.
To test hypotheses H1g through H1j we created portfolios of firms with the
highest (lowest) ROA, ROE, GR, and CA, denoted HROA (LROA), HROE (LROE),
HGR (LGR), and HCA (LCA), respectively. Panel F of Table 4 provides the average of
each efficiency measure of each portfolio. The mean and Wilcoxon rank score tests
indicate that the mean efficiency of HROE is significantly (5%) greater than the mean
20
efficiency of LROE; the mean efficiency of HCA is significantly greater than the mean
efficiency of LCA, the difference is significant at 10%. The differences in efficiency
between HGR and LGR and between HROA and LROA are not significant.
In sum, the results suggest a one-to-one relationship between ROE and CA, and
the efficiency score of the firm the higher the efficiency score, the higher are ROE and
CA, and vice versa.
Our second research question relates organizational form to efficiency. We repeat
the estimation of inefficiency, assuming a positive half-normal distribution (of the
inefficiency component) where the mean is a function of organizational form and policy
mix, using the Cobb-Douglas functional form. Panel A of Table 5 shows the regression
results. We find that the coefficient of STOCK, a dummy variable set at zero for mutual
companies and one for stock companies, is negative and significant, indicating that the
latter are significantly more efficient than the former. The coefficient of MIX is, as
expected, positive and highly significant, indicating that failure to account for the type of
policy (whole vs. term) results in higher inefficiency scores for firms that primarily issue
whole life policies; the most likely reason is that SAP ignore the matching concept.
Finally, Panel B of Table 5 provides the means and average Wilcoxon rank scores
of the profitability and growth measures of the two organizational forms. Over the entire
period, the stock companies have significantly (5%) higher ROA, ROE, CA, and GR. On
a yearly basis, the ROA and CA of stock companies are significantly, higher in every
year, at 10% or better. The ROE of stock companies is significantly higher in 1995 and in
1996, while GR is significantly higher in 1996 and in 1998.
21
Overall, we find that stock companies are significantly more efficient than mutual
companies, and that their growth rates and profitability are significantly higher. Given
our prior results, the two findings may be related: that is, since stock companies are more
efficient they are more profitable and grow faster.
22
relationship between value and efficiency; efficient firms have higher value, and higher
value firms are more efficient.
The two main organizational forms of life insurance companies are mutual
(owned by policyholders) and stock (owned by shareholders). Since the mutual form of
ownership allows less effective mechanisms for controlling and disciplining managers
than the stock ownership, we hypothesize that stock companies are more efficient than
mutual companies, and therefore, are also more profitable and grow faster. Our results
support the hypothesis: stock companies are indeed more efficient. Also, they exhibit
significantly higher ROA, ROE, CA, and GR than mutual companies.
23
References
Aigner, D., K. Lovell and P. Schmidt. 1977. Formulation and Estimation of Stochastic Frontier Production
Function Models. International Economic Review 17, 377-396
Charnes, A., W. W. Cooper, and E. Rhodes. 1978. Measuring the Efficiency of Decision-Making Units.
European Journal of Operational Research 2(6), 429-444
Christensen, R. L., and W. H. Greene. 1976. Economies of Scale in U.S. Electric Power Generation.
Journal of Political Economy 84, 655-675
Cobb, S. and P. Douglas. 1928. A Theory of Production. American Economic Review 18, 139-165
Coelli, T. J. (1994). A Guide to FRONTIER Version 4.1: A Computer Program for Stochastic Frontier and
Cost Function Estimation mimeo, Department of Econometrics, University of New England, Armidale
Cummins, J. D., and H. Zi. 1998. Comparison of Frontier Efficiency Models: An Application to the U.S
Life Insurance Industry. Journal of Productivity Analysis 10, 131-152
Fama, F. E., and M. C. Jensen. 1993. Separation of Ownership and Control. Journal of Law and Economics
26, 301-325
Gardner, L., and M. F. Grace. 1993. X-Efficiency in the U.S Life Insurance Industry. Journal of Banking
and Finance 17, 497-510
Geehan, R. 1986. Economies of Scale in Insurance: Implications for Regulation. The Insurance Industry in
Economic Development 137-160
Grace, F. M. and S. G. Timme. 1992. An Examination of Cost Economics in the United States Life
Insurance Industry. Journal of Risk and Insurance 59, 72-103
Greene, W. H. 1990. A Gamma Distributed Stochastic Frontier Model. Journal of Econometrics 46, 141163
Huang, C. and J. Liu. 1994. Estimation of a Non-Neutral Stochastic Frontier Production Function. Journal
of Productivity Analysis
Jensen, M. C. and Meckling W. H. 1976. Theory of the Firm: Managerial Behavior Agency Costs and
Ownership Structure. Journal of Financial Economics 3, 305-360
Jondrow, J., K. Lovell, I. Materov and P. Schmidt. 1982. On the Estimation of Technical Inefficiency in the
Stochastic Frontier Production Function Model. Journal of Econometrics 19, 233-238
Kaplan, R. S. and D. P. Norton. 1992. The balanced scorecard-Measures that drive performance. Harvard
Business Review 74(1), 71-79
Kumbhakar, S., S. Ghosh and J. McGuckin. 1991. A generalized Production Frontier Approach for
Estimating Determinants of Inefficiency in U.S. Dairy Farm. Journal of Business and Economic
Statistics 9, 279-286
Life Office Management Association (LOMA), Inc. 1998. Expense Management Program (Emap) Manual,
Expense Year 1997 February 1998
Mayers, D., and W. S. Clifford, JR. 1998. Ownership Structure Across Lines of Property-Casualty
Insurance. Journal of Law and Economics 41, 351-378
Mayers, D. and W. S. Clifford, JR. 1986. Ownership Structure and Control: The Mutualization of Stock
Life Insurance Companies. Journal of Financial Economics 16, 73-98
Meeusen, W. and J. Van Den Broeck. 1977. Efficiency Estimation from Cobb-Douglas Production
Functions with Composed Error. International Economic Review 18, 435-444
Reifschnieder, D. and R. Stevenson. 1991. Systematic Departures from the Frontier: A Framework for the
Analysis of Firm Inefficiency. International Economic Review 32, 715-723
24
Stevenson, E. R. 1980. Likelihood Functions For Generalized Stochastic Frontier Estimation. Journal of
Econometrics v13(1), 57-66.
Yungert, A. M. 1993. The Measurement of Efficiency in Life Insurance: Estimates of a Mixed NormalGamma Error Model. Journal of Banking and Finance 17, 483-496
25
Mean
4,435
5,263
5,505
5,430
4,641
2,948
3,458
3,585
3,506
12,670
14,225
12,518
Min.
1.8
1.9
2.2
2.7
1.8
1.9
2.2
2.7
11
103
103
102
Max.
125,831
120,823
128,035
125,620
125,831
85,694
92,455
100,251
38,311
120,823
128,035
100,251
Mean Premium is the average direct premium revenues of the sample firms.
Life, Annuity, and A&H stand for the life insurance, annuity, and accident and health lines of businesses, respectively.
26
EFFREV
N
116
Year
95
121
96
114
97
114
98
116
95
121
96
114
97
114
98
CoI
0.63
(1.15)
0.63
(1.05)
0.58
(1)
0.75
(1.34)
0.04
(0.05)
0.044
(0.053)
0.042
(0.05)
0.045
(0.055)
Notes:
1. CoI is the cost of inefficiency, which is computed as one minus the exponent of minus the efficiency measure, ui, times total
general expenses.
2. EFFIN is the ratio of cost of inefficiency over absolute value of income before taxes.
3. EFFREV is the ratio of cost of inefficiency over revenues.
4. We omitted from the analysis observation for which the ratio total general expense to revenues was greater than one, or
observations for which the ratio of total general expense to absolute value of income before tax is greater than 25.
N
121
126
121
121
UNOR
0.35
0.39
0.38
0.38
0.38
27
Inefficient
Partially
Efficient
92
176
100
25
48
27
Mean
Inefficiency
0.54
0.36
0.25
Notes:
1. The inefficient, partially, and efficient portfolios consist of firms that are, respectively, consistently inefficient, partially
inefficient, and consistently efficient.
GR
0.06
(0.05)
0.07
(0.05)
0.08
(0.05)
0.06
(0.04)
7%
CA
0.06
(0.06)
0.07
(0.05)
0.05
(0.05)
0.05
(0.05)
6%
Notes:
1. ROA is net income (t) over average total assets at the end of year t-1 and year t.
2. ROE is net income (t) over average book value of equity (including the AVR) at the end of year t-1 and year t.
3. GR is two years average growth in direct premiums.
4. CA is operating cash flows (t) over average total assets at the end of year t-1 and year t.
Table 4, Panel C Spearman Correlations between Efficiency Measures and Value Drivers
(N=368)
Value Driver
UNOR
ROA
0.075*
ROE
0.105**
CA
0.14**
GR
0.087**
Notes:
1. * (**) indicates significance level of 10% (5%) for the test of equality between the efficient and inefficient portfolio or for
correlations between variables.
2. For definitions of ROA, ROE, CA and GR refer to the notes to Table 4, Panel B.
3. For definition of UNOR refer to the notes to Table 3.
28
Table 4, Panel D - Means and Average Wilcoxon Rank Scores (in parentheses) of Value Drivers, by
Portfolio
Value
UNOR
Portfolio
Driver
ROA
Efficient
0.023**(105)**
Inefficient
0.013 (88)
ROE
Efficient
0.13** (109)**
Inefficient
0.08
(83)
GR
Efficient
0.087* (103)**
Inefficient
0.055 (90)
CA
Efficient
0.08** (105)**
Inefficient
0.05
(87)
Notes:
1. * (**) indicates significance level of 10% (5%) for the test of equality between the efficient and inefficient portfolio or for
correlations between variables.
2. For definitions of ROA, ROE, CA and GR refer to the notes to Table 4, Panel B.
3. The Efficient (Inefficient) portfolio consists of consistently efficient (inefficient) firms.
Table 4, Panel E Means and Average Wilcoxon Rank Scores (in parentheses) of Profitability
Measures Adjusted for Inefficiency, by Portfolio
Profitability
Portfolio
UNOR
Measure
Efficient
0.035 (96)
ROA*
Inefficient 0.029 (98)
Efficient
0.21 (96)
ROE*
Inefficient 0.17 (97)
Efficient
0.10* (101)
CA*
Inefficient 0.07 (91)
Notes:
1. * (**) indicates significance level of 10% (5%) for the test of equality between the efficient and inefficient portfolio or for
correlations between variables.
2. ROE*, ROA* and CA* are computed as described in the notes to Panel B but with the cost of efficiency (after tax) with respect
to total general expenses added to net income and operating cash flows.
3. The Efficient (Inefficient) portfolio consists of firms with the highest (lowest) profitability measure.
4. For definition of UNOR refer to the notes to Table 3.
Table 4, Panel F Mean Efficiency Measures for the Portfolios of Firms with the Highest and Lowest
ROA, ROE, and CA
Portfolio
N
UNOR
HROA
LROA
HROE
LROE
HGR
LGR
HCA
LCA
92
76
84
88
88
88
92
104
0.53 (80)
0.55 (90)
0.46** (67)**
0.55 (81)
0.57 (77)
0.55 (79)
0.47* (80)*
0.52 (91)
Notes:
1. * (**) indicates significance level of 10% (5%) for the test of equality between the efficient and inefficient portfolio or for
correlations between variables.
2. HROA (LROA) is the portfolio of firms with the largest (smallest) ratio of return on assets, computed as net income in year t
over average total assets at the end of year t-1 and year t.
3. HROE (LROE) is the portfolio of firms with the largest (smallest) ratio of net income in year t over average book value of equity
at the end of year t-1 and t.
4. HGR (LGR) is the portfolio of firms with the highest (smallest) two-year average growth in direct premiums revenue.
5. HCA (LCA) is the portfolio of firm with the largest (smallest) ratio of operating cash flow in year t over average total assets at
the end of year t-1 and year t.
29
Table 5, Panel B Means and Wilcoxon Rank Scores (in parenthesis) of Profitability and Growth
Measures by Organizational Form
Year
Type of firm
N
ROA
ROE
CA
GR
95
96
97
98
Overall
Stock
Mutual
Stock
Mutual
Stock
Mutual
Stock
Mutual
Stock
Mutual
37
72
25
77
23
74
22
74
100
389
0.02* (48)**
0.014 (59)
0.017 (56)**
0.007 (36)
0.02** (52)**
0.01 (39)
0.017* (51)*
0.01 (40)
0.02**(216)**
0.01 (164)
0.11* (50)
0.09 (58)
0.10* (54)**
0.06 (41)
0.11 (49)
0.09 (46)
0.1
(49)
0.8
(46)
0.1** (210)**
0.08 (181)
0.067 (50)*
0.055 (58)
0.08** (57)**
0.03
(33)
0.058**(53)**
0.028 (49)
0.053* (53)**
0.023 (32)
0.065**(220)**
0.037 (153)
0.077 (53)
0.051 (56)
0.077* (53)*
0.031 (44)
0.086 (50)
0.045 (45)
0.086**(50)*
0.029 (42)
0.08** (209)**
0.04 (183)
30
However, they could sell insurance and have made major inroads into the annuity market.
2
The truncated normal, which Stevenson (1980) suggested, avoids the restriction of a
zero mean for the normal distribution. However, it is not clear whether this restriction has
any effect on the efficiency estimates. Moreover, based on our experience, when (the
mean of u) is unrestricted, the log-likelihood seems to be ill behaved, the standard errors
of the parameters are inflated, and the function cannot converge. The normal/gamma
distribution, which Greene (1990) suggested, is superior to the other distributions since it
does not restrict either the location or the shape of the distribution. However, the loglikelihood is currently highly complicated to estimate. In general, the ranking of the firms
according to the efficiency score is preserved across the different distributions of u.
4
Although OLS provides consistent estimates of the parameters with the exception of the
constant term, maximum likelihood estimation provides more efficient estimates of the
parameters.
5
The choice between this cost function and the regular translog function relies on the
statistical power of the estimated regression. The full translog function would increase
the number of variables significantly. Given our sample size (see the Data Section) that
would hamper seriously the statistical properties of the estimated regression and therefore
of the inefficiency estimates.
6
By using this measure we implicitly ignore the intermediary output associated with
whole life policies. In this type of policies, insurance companies make a profit both on
31
the insurance and on the investments of the savings portion of the policy. However, we
believe that the main output of the life insurance line of business is the insurance risk
assumed by the insurer. Second, given the data limitations, it is impossible to separate the
premiums on whole life policies into their insurance and savings components.
7
Another potential proxy is the change in the amount of insurance in force during the
year. It would measure the net additional amount of risk that the company assumes
during the year. However, this measure could take on negative values in cases of
reinsurance or when the amount of insurance paid is greater than the amount of insurance
sold in any given year.
8
Cummins and Zi (1998) and Grace and Timme (1992) control also for group and
individual policies in the cost function. Given our sample size, we do not control for
group and individual policies because of lack of degrees of freedom. Another important
aspect that might affect the results is the marketing distribution system of the firm.
Insurers use various marketing distribution systems such as branch offices, agencies and
direct marketing. The results reported here are possibly associated with the distribution
system. Most insurers, however, employ more than one distribution system and hence
one cannot determine the unique distribution system of each firm.
9
The AVR does not reflect future obligations (as do other reserves) but is set aside to
protect against an extreme decline in the value of the assets that back up liabilities.
10
We are aware that the financial capital is a stock variable while physical capital is a
We measure these ratios over five years, rather than averaging the yearly ratios, in
order to mitigate the influence of extreme fluctuations in the returns ratios on the price
32
of capital. If the price of capital in a particular case is negative--that is, if the five-year
investment return was greater than the return on equity--we compute the price of capital
as the average price of capital of the sample for that year.
12
We do not account for the price of the physical capital in the aggregate price of capital
since the related expenses are rather negligible compared to the magnitude of the
financial capital.
13
The data do not contain information as to the number of insured under A&H group
master policies. Therefore, we used the number of master policies in the computation.
14
EMaP is a detailed expense study of life insurance companies that chose to participate
in the program. LOMA agreed to provide the data as part of a study of the cost structure
of the life insurance industry.
15
We did not use this procedure in estimating the SF measures for two reasons related to
the software (Frontier (Coelli (1984)): (1) the program allows for only the half normal
distribution assumption of the inefficiency component; and (2) the program uses only a
simple Cobb-Douglas cost function, which imposes constant returns to scale.
16
We did not include commissions and amount of financial capital in the computation of
the cost of inefficiency because we believe that these variables are subject to less
discretion by management as compared with other operating expenses.
17
We computed the tax rate as the four-year mean of the ratio of tax expense to earning
before income tax. If the computed tax rate is greater than 35% or negative, we changed
it to 35% and 0%, respectively.
33