Are publicly held firms less efficient? Evidence from the US property-liability insurance
industry
Xiaoying Xie*
Department of Finance, California State University, Fullerton
This version: January 12, 2010
Abstract
This paper studies the performance of publicly held firms in the US property-liability
insurance industry by analyzing companies that issued initial public offerings (IPOs) from 1994
to 2005, using private firms as the benchmark. I investigate ex ante determinants and ex post
effects of IPOs on firm efficiency, operating performance, and other financials. I also analyze
stock returns and follow-on SEO and acquisition activities to provide further information on IPO
motivation. The paper finds that the likelihood of an IPO significantly increases with firm size
and premium growth. IPO firms experience no post-issue underperformance in efficiency,
operations, or stock returns; register improvement in allocative and cost efficiency; and reduce
financial leverage and reinsurance usage. Moreover, IPO firms are active in follow-on SEO
issues and acquisition activities. The findings are mostly consistent with the theory that firms go
public for easier access to capital and to ease capital constraints.
JEL classification: G22; G32
Keywords: Initial public offerings (IPOs); Property-liability insurance industry; Efficiency; Data
envelopment analysis (DEA)
* Corresponding author. Tel.: +1 657 278 5389; fax: +1 657 278 2161.
E-mail address: [email protected] (X. Xie).
1. Introduction
A substantial body of literature on initial public offerings (IPOs) has emerged in the past
two decades (See Brau and Fawcett, 2006; Ritter and Welch, 2002); however, only a few studies
attempt to empirically investigate the rationale behind a firm’s decision to go public and address
performance differences between public firms and their private peers. Existing empirical studies
cover IPOs of US firms (Chemmanur et al., 2006; Helwege and Packer, 2003; Lerner, 1994),1
Indian firms (Gopalan and Gormley, 2008), UK and Irish firms (Aslan and Kumar, 2007),
German firms (Boehmer and Ljungqvist, 2004), and Italian firms (Pagano et al., 1998). For the
US financial services industries, particularly the US insurance industry, such an empirical study
has yet to be conducted.
The lack of empirical study on the decision to go public is largely due to the paucity of
data from private firms. The US insurance industry provides an excellent opportunity to
overcome this difficulty, because special regulatory filing requirements for US insurance firms
have made comprehensive financial information available for both publicly and privately held
entities. The National Association of Insurance Commissioners (NAIC) and A. M. Best
Company (an insurance rating firm) collect and distribute this data every year.
This paper analyzes performance differences between public and private firms in the US
property-liability insurance industry by looking at initial public offerings from 1994 to 2005. I
empirically investigate the ex ante determinants of the decision to go public and ex post effects
of IPOs on firm efficiency, operating performance, and other financials. I also analyze stock
returns, follow-on seasoned equity offerings (SEOs), and acquisition activities in order to
provide further information on IPO motivation. The research answers two main questions: first,
1
Lerner (1994) studies the biotechnology industry. Helwege and Packer (2003) examine a sample of nonfinancial
firms that issue bonds prior to an IPO. Chemmanur et al. (2006) study US manufacturing firms with data available
from the US Census Bureau.
1
what are the determinants of IPOs in the insurance industry? And second, do IPO firms
experience post-issue underperformance?
Several papers analyze post-IPO operating performance. Degeorge and Zeckhauser (1993)
measure operating performance change in firms that experience reverse leveraged buyouts
(LBOs), a special class of IPOs, and find that reverse-LBO firms perform substantially worse
than comparison firms (the industry-matched public firms) in the post-IPO period. Jain and Kini
(1994) study the post-IPO operating performance of public entities (other than reverse LBOs)
and find that these firms exhibit a substantial decline in operating performance subsequent to the
IPO. Mikkelson et al. (1997) examine the ownership structure and operating performance of IPO
firms and also observe initial underperformance, but they find that performance declines no
further during the subsequent ten years.
One weakness with these studies is that they usually use industry-matched public firms
rather than private companies as the control samples. Consequently, survival bias may affect
their results: the matched public firms have gone through the IPO process; therefore, they may be
more established and perform better. In fact, the better performance of matched public firms
versus IPO firms could also indicate that the inferior performance of IPO firms would eventually
end, as documented in Mikkelson et al. (1997). Without a direct comparison between IPO and
private firms, the rationale and effect of public offerings cannot be fully explained. This paper
utilizes data for both public and private insurance firms and directly examines the performance
of IPOs by using private entities as the benchmark.2
2
IPOs in the insurance industry have been investigated with a focus on underpricing and the post-offering stock
performance of demutualization IPOs (Lai et al., 2008; Viswanathan, 2006). These studies do not look at the
performance difference between public firms and private firms, the determinants of IPOs, and the effects of IPOs on
firm efficiency and operating performance, which this paper examines. Studies of IPOs in the banking industry can
refer to Friesen and Swift (2009) and Adams et al. (2009).
2
This study contributes to the literature in two ways: (1) it evaluates company decisions to
go public by directly comparing public and private firms. The literature generally acknowledges
that IPO firms perform worse than industry-matched public firms in terms of either long-run
stock performance (Loughran and Ritter, 1995; Ritter, 1991) or operating performance
(Mikkelson et al., 1997; Jain and Kini, 1994; Degeorge and Zeckhauser, 1993). The decision to
go public can still be justified, however, if IPO firms do not underperform in comparison to their
private peers. (2) This is the first paper that empirically incorporates frontier efficiency analysis
into the study of IPOs to provide a more comprehensive measurement of firm performance.
Based on classic microeconomic theory (Shephard, 1970), frontier efficiency analysis captures
the multidimensionality of a firm’s production process. It is more sophisticated than any single
financial ratio and has become the state-of-the-art method of measuring the performance of
businesses. Frontier efficiency methodology also has an advantage over analysis of stock prices
in that it can be estimated for both public and private entities, while stock prices are only
available for publicly traded firms. In this study, I estimate various types of efficiencies (scale,
allocative, technical, cost, and revenue efficiency) by using the data envelopment analysis (DEA)
method and test hypotheses based on these measures. I also analyze operating performance and
stock returns to provide a comparison.
The remainder of the paper is structured as follows. Section 2 develops the tested
hypotheses. Section 3 describes the data and sample. Section 4 presents frontier efficiency
methodology. Section 5 presents analyses on the determinants of the decision to go public and
the post-issue effects of IPOs (this section also covers firms’ SEO and acquisition activities after
the IPO and the analysis for delisted firms), and section 6 concludes the paper.
2. Hypotheses
3
Companies often base their rationale for going public on the assumption that the benefits
from public trading and more diversified ownership outweigh the costs of public listing. It is
essentially an ex ante cost-benefit consideration (Brau and Fawcett, 2006; Ritter and Welch,
2002; and Pagano et al., 1998). If the benefits exceed the costs, the performance of a firm will
improve; otherwise, it may suffer a lower efficiency.
This section discusses the motivations, costs, and benefits of going public. Hypotheses
about the effect of the IPO and about firm characteristics likely to be associated with listing
decisions are formulated based on the existing literature. These theories and their predictions are
not necessarily mutually exclusive because different motivations often work interactively to
produce an IPO issue (Pagano et al., 1998). Moreover, some theories have similar implications
for company decisions and performance, so it is an intricate process to disentangle them. The
testing hypotheses have been developed with regard to these observations.
2.1. Theories favoring going public
The literature on corporate finance suggests that firms go public for both financial
reasons and for management’s “private” reasons. As an examination of financial data, this study
is not well equipped to examine “private” considerations, so this discussion will not cover them.3
The following sections categorize the benefits of listing.
2.1.1. Ability to raise equity capital and overcome capital constraints
The most cited reason for going public is to raise capital. Going public enhances a firm’s
ability to raise equity capital more effectively than private equity, debt, or venture capital
(Pagano et al., 1998; Chemmanur and Fulghieri, 1999). In an initial offering, a firm can raise
new capital without the associated risks, restrictions, costs of debt, or the constraints set by
3
For example, the founders and shareholders of a private firm may seek to go public to cash out and diversify their
personal portfolios (see summary of this issue in Brau and Fawcett, 2006; Ritter and Welch, 2002; Alavi et al.,
2008).
4
venture capitalists. After the initial offering, a firm may raise additional capital through seasoned
equity offerings as needed. The infusion of equity capital can help the firm to rebalance accounts
after high growth or investment, sustain or pursue higher growth rates, finance new investments,
or reduce the level of debt in order to survive and thrive in competitive markets.
In the insurance industry, equity capital provides a cushion for risks taken by the
insurance companies and is the backbone of their underwriting capacities. Higher levels of risk
create stronger demand for equity capital. The adequacy of a firm’s capital is closely monitored
by regulators, policyholders, and insurance rating firms. Private insurance companies possess a
limited ability to raise new capital in a timely fashion and face substantial transaction costs when
raising new capital due to information asymmetries. Outside investors generally have less
information about the quality of a private insurer’s assets and the value of its reserve estimates
for unpaid losses, especially for long-tail lines such as commercial liability insurance, making it
more difficult and costly for private insurers to raise new capital (Chamberlain and Tennyson,
1998).
Based on this theory, it is predicted that insurance firms in need of more equity capital
are more likely to go public. These firms tend to engage in riskier lines of business and, therefore,
exhibit higher volatility in their underwriting performance. They are more likely to have higher
leverage, higher growth rates, more reinsurance usage, and/or higher agent balances than other
private firms (Viswanathan and Cummins, 2003). Firms with such characteristics are more
sensitive to capital inadequacy shocks and place more value on the ability to raise additional
capital.
In the property-liability insurance industry, long-tail lines are usually associated with
higher risk, especially lines involving commercial liability, because they deal with dynamic risks
5
in society. Their loss payments are dramatically affected by alterations of law and regulation,
scientific discoveries, and other developments. Firms with more business in such lines need
easier access to the market for additional capital, in case unexpected losses deplete their
underwriting capacities. Firms underwriting more business in long-tail lines also need to
maintain a higher level of loss reserves (liability of the company) and may operate with higher
leverage given the capital constraints they face as private firms. Additional capital raised from
the IPO and any follow-on equity issues will help bring down the leverage and ease capital
constraints.
Insurance firms may go public to support and sustain high growth rates. Firms may
experience high growth rates as a result of comparative advantage, competitive pressure, or
strategic development policy. Such firms will have a higher demand for equity capital and tend
to transfer risks beyond their underwriting capacity to reinsurers.
2.1.2. Market-timing theory (windows of opportunity)
Market-timing theory suggests that firms go public when markets are “good” and equity
valuations are high and remain private when markets are “bad” and equity valuations are low
(Pagano et al., 1998; Lerner, 1994). In this way, firms can take advantage of the bull market and
issue IPOs when the industry market-to-book ratio is high4. Following Pagano et al. (1998), I
proxy the “windows of opportunity” by the median market-to-book ratio of public companies in
the insurance industry and/or include year dummies in the regression to control for stock market
conditions.
2.1.3. Managerial discipline (monitoring)
4
This theory does not necessarily predict post-issue performance improvement because firms that intend to time the
market may not always issue IPOs when equity capital is most valuable to them. Colaco et al. (2009) also show that
a firm may be willing to issue an IPO without following clusters of similar firms if the firm needs to raise capital or
if the insiders need portfolio diversification.
6
Going public enables firms to utilize an external monitoring mechanism that may not
otherwise exist in private firms. Managers of publicly traded firms face the threat of hostile
takeover and high pressure to perform in a competitive labor market. Managerial decisions in
such firms are more exposed to market assessment than those of private firms. Shareholders can
also design more efficient compensation packages and use stock options or stock-price indexed
payments to align the incentive of managers with that of the owners (Holmström and Tirole,
1993). This paper does not test this hypothesis directly, but, for it to hold true, improved
performance of IPO firms is anticipated.
2.1.4. Facilitating acquisition activities
Firms may go public to facilitate acquisition activities (Zingales, 1995). A public listing
can generate public shares and trading prices for a firm and facilitate its participation in takeover
activities either as an acquirer or as a target. Public firms can make less expensive acquisitions by
paying stocks while preserving the company cash position for operation and investment. In
addition, public targets are normally valued higher than private targets and have higher takeover
premiums (Faccio et al., 2006; Chang, 1998). In this paper, I test this theory by examining the
acquisition activities of IPO firms after their public offerings.
2.1.5. Strategic movement
An IPO may serve as a firm’s strategic movement. First, going public helps broaden the
ownership base of the company, which could reduce the bargaining power of large investors
considerably (Chemmanur and Fulghieri, 1999). Second, going public helps enhance the
reputation and visibility of a firm and capture a first-mover advantage (Ritter and Welch, 2002;
Maksimovic and Pichler, 2001). The performance of IPO firms may therefore compare more
7
favorably against private firms because their added prestige and visibility is valued by customers,
suppliers, employees, and the financial community.
2.2. Theories against going public
2.2.1. Agency costs and signaling costs
Going public may help create a better external monitoring system, but the dilution of
management ownership may increase agency costs and create a moral hazard problem (Jensen
and Meckling, 1976). After going public, more dispersed share ownership and more serious
“free-rider” problems may impair internal monitoring by shareholders. The performance of
public firms could also suffer due to increased agency conflicts between managers and
shareholders. Kaplan (1989) and Smith (1990) find improvements in the operating performance
of firms that transfer from public to private through management buyouts (MBOs) and credit the
reduced conflicts of interests between managers and owners in private firms for the improvement.
As management’s ownership interests dilute and agency costs increase, the performance of IPO
firms is expected to decline.
The high signaling costs of going public may also discourage a firm from listing.
Investors are generally less informed than insiders about the true value of companies seeking to
go public. Consequently, the market may expect an adverse selection of IPOs. To send a positive
signal to the market about the high quality of the firm, the original owners may retain a
significant ownership stake in the firm (Leland and Pyle, 1977) or underprice their IPO shares
(Allen and Faulhaber, 1989; Welch, 1989). Both signals, among others, such as using a longer
lockup period and certification,5 are expensive. As Chemmanur and Fulghieri (1999) suggest,
small and young firms are more subject to the costs of information asymmetry, so successful
5
See Brau and Fawcett (2006) for a summary of literature on the use of certification.
8
listing firms tend to be old and large firms. Based on this theory, the probability of going public
relates positively to firm size.6
2.2.2. High pre-offering expenses and fees
High expenses and fees may discourage firms from going public. Such fees include, but
are not limited to, underwriter commissions, initial and pre-offering expenses (such as fees for
attorneys and accountants), and significant housekeeping and cleanup costs for the Securities and
Exchange Commission (SEC) registration. Overall, the direct costs of going public are estimated
to be about $250,000, plus an additional 7% of the gross proceeds of the IPO (Ritter, 1987), and
the costs are not sensitive to the size of the issue (Chen and Ritter, 2000). The high costs of IPOs
and their insensitivity to firm size suggest that large firms are more likely to go public.
2.2.3. Burden of post-offering duties
Firms face additional costs after they become public. First, they must comply with federal
securities laws and various filing requirements. Administrative and compliance expenses and
efforts toward developing public and investor relations are by no means negligible. For instance,
in the US, the insurance industry is heavily regulated, with the Statutory Accounting Principles
(SAP) as the dominant accounting rule. After going public, firms must also file based on
Generally Accepted Accounting Principles (GAAP), which is not required of private companies.
Second, firms must disclose information about their operational and financial situations where
secrecy may be crucial for competitive advantage, such as investment projects and merger and
acquisition strategies. Finally, firms lose flexibility in making prompt strategic decisions,
because they face scrutiny from, and have fiduciary duties to, a more diversified group of public
6
It is ideal to include the age of firms in the model; however, in the insurance industry, it is common for many
individual firms to run under one group ownership, and the members of a group change over time. It is, therefore,
difficult to define the precise year of incorporation of a group holding company.
9
shareholders. Brau and Fawcett (2006) demonstrate that the desire to maintain decision-making
control is the most important concern for firms to stay private.
Because the costs of operating as a public firm are decidedly high, especially for smaller
firms, it is predicted that size is positively related to the probability of going public. In addition, a
public firm’s performance may suffer if the costs outweigh the benefits.
In summary, it is hypothesized that IPO firms will not underperform in comparison to
private firms if the benefits of going public outweigh the related costs, and the characteristics
associated with a firm’s likelihood of going public will be consistent with the theories under
consideration.
3. Data and sample selection
3.1. Data source
I extract the list of publicly traded property-liability insurance firms from SNL
DataSource, Center for Research in Security Prices (CRSP), and A.M. Best. 7 From these
resources, I obtain a complete list of publicly traded firms (both current and historical), including
their primary insurance sector information,8 as well as the starting and ending date of listings.
The decision-making units in this analysis are groups and unaffiliated firms. 9 Each insurer’s
group code (NAIC group code) and its public holding company are matched using information
collected from the NAIC, A.M. Best, Factiva reports, Hoover’s online company profiles, and
7
Several Bermuda-based insurance holding firms that list in US stock exchanges are deleted from the sample
because their US insurance subsidiaries cannot be identified.
8
In this paper, financial guaranty, mortgage guaranty, and surety holding firms are included in the analysis because
NAIC considers all these types of firms as “property-liability” firms. “Multiline” firms (such as AIG) are also
analyzed. Property-liability operations of other public insurance firms, such as public life-health and managed care
firms are also included in the sample. Excluding these firms does not change the results.
9
See Cummins and Xie (2008) for a discussion of company affiliations in the insurance industry.
10
various other resources.10 Finally, I collect the financial data of firms (covering the years 19932006) from the regulatory annual statements maintained by the NAIC.
Table 1 shows the sample size. During the period of 1993 to 2005, 218 publicly traded
insurance companies engage in the property-liability insurance business. Among them, 89 are
continuously traded “old listing” firms (firms that started listing before 1994 and were publicly
traded throughout 1993-2005); 52 are continuously traded “new listing” firms (firms that started
listing between 1994 and 2005, and were publicly traded at least until the end of 2005); and 77
are “historical firms” (firms that were delisted between 1993 and 2005, no matter when they
started the listing).11 The continuously traded “new listing” firms include nine demutualization
IPO firms and 43 non-demutualization IPO firms. The number of new listing firms and delisted
firms by year is also reported in the table.
3.2. Study sample
The study sample of the paper is the IPO firm sample.12 This sample consists of stock
firms that issued IPOs between 1994 and 2005, which include continuously traded “new listing”
firms (43 non-demutualization IPOs) and delisted IPO firms (12 “historical firms” that issued
10
In the insurance industry, it is common for several subsidiaries to be operated under common ownership and
management. In a few cases, both the holding company and some of its controlled subsidiaries are listed (e.g. The
American International Group and its controlled firm Transatlantic Holdings, Inc. are both listed). Research on IPO
decisions, such as Pagano et al. (1998), treats the listed independent firms and the listed subsidiaries of public
parents separately. Since this analysis is at the group and unaffiliated firm level and only a very few cases involve
the public subsidiaries of publicly traded companies, the separation is less necessary. The listed parent and the
subsidiary are considered as one observation that is represented by the parent company here.
11
Among them, 17 firms issued IPOs between 1994 and 2005; 12 were non-demutualization firms and five were
demutualization firms.
12
To investigate whether public firms underperform in the long run and shed some light on the future development
of the IPO firms, I also look at public firms that started listing before 1994 (Non-IPO public firm sample), using
private firms as the benchmark. The Non-IPO public firm sample includes 88 continuously traded “old listing” firms
and 59 non-IPO “historical firms.” The overall finding is that non-IPO public firms in general do not underperform
in comparison with private firms (in terms of cost efficiency, revenue efficiency, technical efficiency, and return on
assets), but public firms differ significantly when compared to private firms in certain characteristics such as size,
leverage, usage of reinsurance, agents’ balances ratio, business mix, and diversification level.
11
non-demutualization IPOs between 1994 and 2005).13 Demutualization IPO firms are excluded
from the sample because these firms operate under the mutual organizational form before going
public.
3.3. Control sample
The control sample in this study includes private firms in the US property-liability
insurance industry. I select the private firm sample by using the following procedures:
(1) Identify all the groups and unaffiliated firms that report property-liability data to the
NAIC from 1993 to 2006.
(2) Exclude the 218 public firms from the sample in all years.
(3) Exclude the insurance groups and unaffiliated firms whose ultimate parent
companies’ primary businesses are not insurance, but are publicly traded.14
(4) Exclude mutual firms and keep stock firms. Stock firms and mutual firms differ in
many areas such as capital structure, business focus, corporate governance, and
production techniques (Harrington and Niehaus, 2002; Cummins et al., 1999a), so
mutual firms do not serve as a proper control sample for this study and are therefore
excluded.
(5) Keep private firms that continuously operated for at least 12 years out of the 14-year
period (1993-2006). In other words, the private firms in general do not exit the market
during the study period. This requirement for relatively continuous operation implies
that these firms have the potential to go public (Pagano et al., 1998).
The above selection criteria return 270 private stock firms (with 3,598 firm years).15
13
Delisted firms are included in the sample to avoid any survivorship bias. Excluding these firms from the analyses
does not change the results.
14
For example, GE Global insurance group is excluded because its ultimate parent (General Electric Company) is a
publicly traded conglomerate.
12
4. Frontier efficiency methodology
This paper employs frontier efficiency methodology to measure firm performance.
Frontier efficiency is rooted in classic microeconomic theory (Shephard, 1970). It accounts for
the multidimensionality of a firm’s production process and therefore provides a more
sophisticated measure of firm performance than any single financial ratio, such as return on
assets or return on equity. The frontier efficiency methodology also has an advantage over
analysis of stock prices in that it allows the study to include both public and private firms, while
stock prices are only available for publicly traded companies (Cummins et al., 1999b; Cummins
and Xie, 2008). 16 The application of this method contributes significantly to the IPO
performance literature.
The essence of frontier efficiency is distinguishing well-performing production units
from those that perform poorly. The technique estimates the “best practice” efficient frontier that
consists of the most efficient firms in an industry, and compares all firms in the industry to the
frontier. Firms operating on the frontier are fully efficient (with efficiency scores of one), and
firms not on the frontier are inefficient (with efficiency scores between zero and one).
DEA, a non-parametric technique, is adopted to estimate firm efficiency (Cooper et al.,
2000). Non-parametric techniques are superior to econometric efficiency measurements because
they require no assumptions to be made regarding the production or cost function and the
probability distribution of the error terms, which prevents potential estimation bias caused by
model misspecification. Previous studies have shown that DEA estimation has good asymptotic
15
Some firm years are dropped in the probit and other regression analyses because of the further constraints applied
when estimating firm efficiency and missing values of other control variables.
16
This paper complements the efficiency analyses with operating performance and stock returns to see whether
results would be different if using alternative measures and to make conclusions more accessible to people not
familiar with frontier efficiency methodology. Literature has shown that frontier efficiency scores capture valuable
information, as compared to operating performance and market returns (Cummins and Zi, 1998; Cummins and Xie,
2009).
13
statistical properties (Banker, 1993) and better accuracy than econometric approaches in
estimating efficiency in the presence of heteroscedasticity (Banker et al., 2004). DEA also yields
consistent estimators for contextual variables in a two-stage regression with DEA efficiency as
the dependent variable (Banker and Natarajan, 2008).
Classic microeconomic theory presents the primary goal of companies as maximizing
revenues while minimizing costs. Accordingly, this analysis estimates both cost and revenue
efficiency for sample firms, as well as technical efficiency, scale efficiency, and allocative
efficiency in order to identify the sources of cost (in)efficiency, such as economies of scale and
resource allocation in the production process. The appendix to this paper shows the relationship
between various types of efficiency and associated mathematical programming.
The first step in estimating DEA efficiency is to specify the inputs, outputs, input prices,
and output prices for firms. Four inputs are used in this study: administrative labor, agent labor,
materials and business services, and financial equity capital. A modified version of the valueadded approach is adopted to define insurance outputs (Cummins and Weiss, 2000; Cummins
and Nini, 2002; Cummins and Xie, 2008). Four insurance outputs (personal lines short-tail
insurance, personal lines long-tail insurance, commercial lines short-tail insurance, and
commercial lines long-tail insurance) are defined for risk pooling and risk bearing, and real
financial services One intermediary output (invested assets) is defined for the financial
intermediary services provided by insurance companies.
Following Cummins and Xie (2008), this paper defines the quantities of administrative
labor, agent labor, and business services as the current dollar expenditures related to these inputs
divided by their current prices. The prices for administrative labor and agent labor inputs are
defined as the US Department of Labor (DOL) average weekly wage (AWW) for employees in
14
insurance companies (SIC 6331 and NAICS 524126) and the DOL AWW for insurance agents
(SIC 6411 and NAICS 524210), respectively. The price for materials and business services is
calculated by taking the weighted average of price indices for business services from the expense
page of Best’s Aggregates and Averages. The base year for all price indices is the year 2000. The
quantity of financial equity capital input is measured by using the average of the beginning and
end-of-year equity capital, deflated by the CPI. The ideal cost of capital measure is the expected
market return on equity capital. Unfortunately, expected market returns cannot be calculated for
most insurers because the majority of insurers are not publicly traded. After considering several
measures of cost of capital in the literature, the paper selects the size adjusted capital asset
pricing model based on data from Ibbotson Associates (2008). That is, the cost of capital for
year t is calculated as the 30-day Treasury bill rate at the end of year t-1, plus the long-term
(1926 to the end of year t-1) average market risk premium on large company stocks, plus the
long-term (1926 to the end of year t-1) average size premium from Ibbotson Associates. All
insurers in the industry are classified into four size groups based on their equity capital. The
largest size category receives no size premium, and each of the smaller size categories receives
the Ibbotson long-term average size premium.17
The quantity of insurance output is measured by the present value of losses incurred. The
price of each insurance output is defined as the difference between real premiums earned and the
17
See Cummins and Xie (2008) for more details of the method. The other two methods for cost of capital estimation
used in the literature include: (1) omitting the size premium and assigning the same cost of capital to each firm in a
given year (Cummins and Nini, 2002) and (2) the three-tier approach: estimating the cost of capital for traded
insurers in various A.M. Best financial rating categories and assigning costs of capital to non-traded insurers based
on their A.M. Best ratings (Cummins et al., 1999b). Previous literature reveals the robustness of the cost efficiency
results with alternative cost of capital assumptions (Cummins et al., 1999b; Cummins and Nini, 2002; Cummins and
Xie, 2008). The size adjusted capital asset pricing model is adopted because, theoretically, it better addresses the
concern that the cost of capital for public firms should be different from that of private firms. Since public firms on
average are larger than private firms, they are assigned lower cost of capital under the size adjusted CAPM approach,
while the three-tier approach assigns the same cost of capital to public and private firms based on the A.M. Best
ratings. An earlier version of this paper tests the three-tier approach; the overall conclusions are the same.
15
real present value of losses incurred for the output divided by the real present value of losses
incurred. A smoothing procedure is applied to the four insurance outputs and their prices for each
firm in the sample to control for the potential “errors in variables” problem in output price
measurement due to the randomness of losses (Cummins and Xie, 2008). The quantity of
intermediation output is measured by the average of the beginning and end-of-year invested
assets (deflated to 2000 by CPI). The price of the intermediation output is defined as the
weighted average of the expected return on stocks and the realized return on other interestbearing invested assets.
The decision-making units of the efficiency estimation consist of group and unaffiliated
single property-liability insurers. I estimate the year-by-year efficiency scores for every eligible
firm during the period 1993 to 2006. Originally, the sample for efficiency estimation includes all
group and unaffiliated insurers with data available from the NAIC, but I eliminate the following
firms: firms with zero or negative net worth or premiums, firms with extremely high premiumto-surplus ratios, and small firms with assets below $1 million. I also exclude risk retention
groups, US Lloyds, and state worker’s compensation fund programs from the sample.
5. Results for IPO firms
5.1. Determinants of the decision to go public
Based on the discussions of motivations of going public, I conduct a univariate analysis
and a probit regression for the determinants of issuing IPOs. The probit model is:
Pr( IPOi ,t = 1) = F (α + β1SIZEi ,t −1 + β 2 LEVERAGEi ,t −1 + β 3 REINSUREi ,t −1 + β 4 AGENTBALi ,t −1
+ β 5GROWTH i ,t + β 6 PLTi ,t −1 + β 7CLTi ,t −1 + β8 PHERFi ,t −1 + β 9GHERFi ,t −1 + β10 ROAi ,t −1
+ β11 LOSSRATIOi ,t −1 + β12 EXPRATIOi ,t −1 + β13UNAFFILIATEDi ,t −1 + β14 MTBi ,t −1
(1)
+ β15 EFFi ,t −1 + γ tYEARt )
where i represents insurer i, t represents year t and,
16
IPOi,t =
SIZE =
LEVERAGE =
REINSURE =
AGENTBAL =
GROWTH =
PLT
CLT
PHERF
GHERF
ROA
LOSSRATIO
EXPRATIO
UNAFFILIATED
MTB
EFF
=
=
=
=
=
=
=
=
=
=
YEAR =
zero if a company i stays private in year t and one if it goes public
in that year;
Logarithm value of total admitted assets;
Liabilities divided by policyholders’ surplus;
Reinsurance ratio, defined as: Reinsurance ceded / (Direct
premiums written + reinsurance assumed);
Agents' balances ratio, defined as: Agents' balances / Direct
premiums written;
Premium growth, defined as: Net premiums written (t) / Net
premiums written (t-1);
Percentage of premiums in personal lines long-tail;
Percentage of premiums in commercial lines long-tail;
Herfindahl index of product lines based on net premiums written;
Herfindahl index across states based on net premiums written;
Return on asset;
Loss ratio;
Expense ratio;
one if an unaffiliated single firm, zero if an insurance group;
Median market-to-book value of equity of insurance industry;
Efficiency (scale, allocative, technical, cost, or revenue efficiency)
generated by DEA; and
Calendar year dummies.
Similar to Pagano et al. (1998), at any time t with IPO issues, the IPO=0 sample includes
all the private companies in that year as selected in Section 3.3 and other IPO firms remaining
private in that year. A firm is dropped from the IPO=0 sample once it goes public.18 Lagged
values of the regressors are used in the regression, except for the growth variable, where the
premium growth rate in year t is used.19
Table 2 provides univariate statistics for IPO firms and private firms for the year prior to
the IPO. As predicted by the theories, IPO firms are much larger than private firms in terms of
assets, premium revenue, policyholder surplus, and net income. IPO firms have higher leverage
than private firms (1.85 vs. 1.69). Additionally, IPO firms use more reinsurance services (0.43 vs.
18
That is, the “private firms” (IPO=0 sample) include firms that stayed private during the sample period and IPO
firm-years before they went public. A robustness test applying NASDAQ SmallCap market initial listing
requirements to the IPO=0 sample returns a similar quantitative result.
19
There are a few firms whose t-1 financials are missing, so I use their year t value to retain them in the sample.
Dropping these firms returns similar results.
17
0.28) than private firms, suggesting that their underwriting capacity constrains their ability to
retain risks. IPO firms also have a higher agents’ balances ratio (0.27 vs. 0.15), indicating that
such firms may face more capital constraints. The premium growth of IPO firms is much higher
than that of private firms (1.53 vs. 1.19), indicating that IPO firms are expanding and need
stronger capital support. These findings altogether suggest that firms conduct IPOs to raise
additional capital or ease capital constraints after periods of high growth.
With respect to business mix, IPO firms write significantly more business in long-tail
commercial lines (0.54 vs. 0.42) and less business in short-tail commercial lines (0.18 vs. 0.30),
which suggests that IPO firms engage more intensively in riskier lines. IPO firms are more
diversified across business lines and geographical areas than private firms, implying that the
potential reputation benefit from going public may be significant.
Surprisingly, IPO firms on average do not have higher returns on assets than private firms
before going public (0.033 vs. 0.05).20 IPO firms have a higher loss ratio (0.74 vs. 0.65) and a
slightly higher expense ratio (0.42 vs. 0.38), which may be due to the greater demand in services
from commercial customers. I find no significant difference in firms’ technical, cost, and revenue
efficiency between IPO firms and private firms, even though IPO firms have lower scale and
allocative efficiency scores.
Table 3 presents probit regression results, which are largely consistent with the findings
from univariate analyses. Large firms and firms with a higher premium growth rate and a larger
proportion of long-tail commercial lines are more likely to go public. The reinsurance usage and
agent balance ratios are also positively related to IPO issues. These findings suggest that capital
need is a motivation for IPOs. The size effect is consistent with arguments regarding the burdens
20
The median return on assets of IPO firms and private firms is 0.026 vs. 0.044. The median test is significant at the
5% level. Most of the existing literature on IPOs finds that pre-issue performance of IPO firms is better than
industry-matched (public) firms (Mikkelson et al., 1997; Jain and Kini, 1994; Degeorge and Zeckhauser, 1993, etc.).
18
and challenges of going public, such as the cost of information asymmetry, high pre-offering
expenses and fees, and the burden of post-offering duties.
Variables measuring the pre-IPO performance of firms include return on assets, loss ratio,
expense ratio, and various types of efficiency. Among them, expense ratio carries a significant
positive sign, and scale and allocative efficiency carry a significant negative sign, while other
performance variables are more or less insignificant. This suggests that IPO firms do not
necessarily perform better in all dimensions than private firms and may go public to seek
improvements. The industry market-to-book ratio is positive but not statistically significant,
providing little support for the “window of opportunity” or “market-timing” theory in the
insurance industry.21
5.2. Effect of IPO on firm performance and financials
To evaluate the IPO decision, it is essential to compare the ex post performance and
financials of IPO firms with those of firms that remain private. Following Pagano et al. (1998), I
investigate this issue by constructing an unbalanced panel sample for all IPO firms and private
firms as selected in Section 3.3 for the years 1993-2006. 22 I then estimate fixed-effects
regressions, and the effects of the decision to go public are captured by dummy variables for the
year of the IPO and the three subsequent years. The model specification is as follows:
3
yi ,t = α + ∑ β j IPOt − j + β 4 IPOt − n + μi + dt + ε i ,t
(2)
j =0
21
Year dummies are excluded in table 3 because none of them are significant and a potentially high correlation
exists between market-to-book ratio and year dummies. A robustness test with year dummies in the model returns
similar results. In order to address whether a firm’s IPO decision is affected by the Sarbanes-Oxley Act of 2002, I
conduct robustness checks for the probit analyses by splitting the overall sample period into two parts, 1994-2001
and 2002-2005. The results remain stable over time.
22
In regard to IPO firms that delisted during the sample period, the firm years after delisting are excluded from the
sample.
19
where i represents insurer i, t represents year t, μi and d t are firm-specific and calendar year
specific effects, respectively. IPOt − j are dummy variables equal to one if year t − j is the IPO
year; IPOt − n is a dummy variable equal to one if the IPO took place more than three years before
t. The design of this model allows use of a pre-IPO firm as a control for itself after the deal. It
also allows separation of the short-term and long-term effects, which could be mutually
offsetting.23
The dependent variables yi ,t include efficiency (scale, allocative, technical, cost, and
revenue efficiency), ROA, LOSSRATIO, EXPRATIO, LEVERAGE, NCEQUIP, FINANCIAL1,
FINANCIAL2, GROWTH, REINSURE, and AGENTBAL, where
NCEQUIP =
FINANCIAL1 =
FINANCIAL2 =
Investment in equipment and EDP: Expenditures over equipment
and Electronic Data Processing Equipment/Total non-commission
expenses;
Financial investments - all: Total invested assets/Total assets; and
Financial investments - long-term: (Total invested assets - Cash
and short-term invested assets)/Total assets.
Other variables are defined as in the Probit model.24
5.2.1. Efficiency, operating, and underwriting performance
The first section (Efficiency) of Table 4 shows that IPO firms experience no significant
improvement in post-IPO scale efficiency and technical efficiency. However, IPOs significantly
help improve firms’ allocative efficiency, suggesting that management with better access to
capital markets is more efficient in allocating the available resources. Driven by improvement in
23
In a robustness test, I control the status of whether a firm satisfies listing requirements in year t-j by using
NASDAQ SmallCap market initial listing requirements to construct qualification dummy variables. The results of
the IPO dummies are not affected.
24
I conduct a robustness test by including control variables in the regressions to address the concerns that changes in
performance and financials may also be affected by other variables, such as previous performance and financials.
The variables controlled include: firm size, lagged values of the dependent variable, and lagged values of other
variables that are found relevant by the literature, such as business mix and firm diversification level. In addition,
lagged values of ROA, growth, leverage, reinsurance ratio, and agents’ balances ratio are added in every regression.
The results are largely consistent with the ones reported in the paper.
20
allocative efficiency, cost efficiency of IPO firms improves as well, particularly in the long run
(as shown by “year +2” to “year > +3” dummies). IPO firms also marginally improve their
revenue efficiency in the year of IPO and the third year after the IPO, but the overall effect for all
the post-IPO years is not significant (as shown by the p-value of the F-test).
The second and third sections demonstrate that the profitability (return on assets) and
underwriting performance (loss ratio and expense ratio) of insurers do not change significantly
after the IPO. In summary, I find no ex post efficiency or operating underperformance for IPO
firms in the insurance industry.
5.2.2. Leverage, investment, growth, and other financials
IPO firms deleverage within the first three years after the IPO, but not in the long run.
The overall effect of all IPO dummies is significant at the 10% level, suggesting that IPO eases
firms’ capital constraints in the short run. IPO has no significant impact on capital expenditures,
as shown in the regression for investment in equipment and EDP (electronic data processing
equipment), investment in financial assets, and long-term financial investment. Premium growth
declines after the IPO (particularly in the long run), but the overall effect is not significant.
The reinsurance ratio also declines in the early years after the IPO, with overall effects
significant at the 10% level. This also confirms the easing of capital constraints as a motivation
for IPOs because the additional capital infusion through IPO enables firms to retain more risks
and seek less reinsurance usage. Agents’ balances ratio is not significantly impacted by the IPO.
A complete picture can be drawn when combining the effect of the IPOs with their ex
ante determinants. Prior to an IPO, firms tend to experience high growth and use more
reinsurance services. After an IPO, these firms reduce their leverage and cede less risk to
reinsurers. These findings suggest that the major motivation of IPOs is to ease capital constraints,
21
increase firms’ underwriting capacity, and rebalance their capital structure following a period of
rapid growth in business.
5.3. Long-run stock performance of IPO firms
As a robustness check, I estimate the abnormal returns for IPOs during the period 1994 to
2005 to find out whether the long-run stock performance of these firms shows a different pattern
than what efficiency and operating performance measures demonstrate. The analysis follows
Ritter (1991) to compute the long run performance of IPOs. The benchmark-adjusted stock
return for firm i in event month t is defined as:
ARit = rit − rbt
(3)
where rit is the monthly return for stock i and rbt is the return for the benchmark in month t.
The average benchmark-adjusted stock return on a portfolio of n stocks for month t is:
ARt =
1 n
∑ ARit
n i =1
(4)
The cumulative benchmark-adjusted stock return from month one to month s is the summation of
the average benchmark-adjusted stock return:25
s
CAR1,s = ∑ ARt
(5)
t =1
Table 5 reports the average monthly raw returns of IPO firms, the average matching firmadjusted returns (AR (t)), and the cumulative average matching firm-adjusted returns (CAR (1,t))
for 36 months after the offering date.26 Seven out of the 36 monthly raw returns are negative, but
25
Another way to calculate CAR1,s is to first calculate the cumulative abnormal return for month 1 through s for firm
i and then take an average across the n firms (Lai et al., 2008).
26
As in Ritter (1991), this study calculates the post-issue returns for 36 months after the IPO, excluding the first day
of trading. Months are defined as successive 21-trading-day periods relative to the IPO date, i.e., month 1 consists of
event days 2-22, month 2 consists of event days 23-43, etc. For IPOs that are delisted before the 36th month, stock
return calculation ends with the CRSP's last listing. The corresponding 21-trading-day period is also applied to the
benchmark return rbt. Two benchmarks are used: (1) the CRSP value-weighted NYSE/AMEX/NASDAQ index and
22
are statistically insignificant. Sixteen out of the 36 monthly average adjusted returns are negative,
with one of them having t-statistics higher than 2.0. The cumulative average adjusted return
gradually declines over time, reaching its minimum at -13.36% in the 34th month, but is
statistically insignificant. As such, unlike Ritter (1991), this study finds little evidence of
underperformance in stock returns for insurance IPOs. This finding squares with the observations
in section 5.2 that IPO insurers experience no significant post-issue increase in investments,
where literature suggests that firms investing more will have lower expected returns and
experience underperformance (Carlson et al., 2006).
Figure
1
plots
the
CAR
of
raw
returns,
the
CRSP
value-weighted
NYSE/AMEX/NASDAQ index adjusted returns, and the matching firm-adjusted returns for
month zero to month 36, with month zero as the initial return.
underperformance
is
even
less
significant
when
using
the
27
The evidence of
CRSP
value-weighted
NYSE/AMEX/NASDAQ index as the benchmark.
5.4. Seasoned equity offer of IPO firms
To further investigate whether capital need is the major motivation for issuing IPOs, the
study examines whether IPO firms conduct follow-on SEOs to have further access to equity
capital. The SEO information is collected from SDC Platinum on capital issues of insurers. Table
6 reports the corresponding result, which provides further evidence of the capital need
motivation of IPOs. Until the end of November 2008, about 52.7% of the IPO firms (29 out of 55
IPO firms) have issued at least one SEO, and among them, 72.4% of firms (21 out of 29 firms
that issued SEOs) have had their first SEO issued within three years of the IPO.
(2) listed firms matched by size. The matching insurance firms have the closest market capitalization to the issuers
on the day of (or before) the IPO, selected by using a similar procedure described in Ritter’s (1991) appendix.
27
The initial return is calculated as (First trading day closing price – Offer price)/Offer price. The offer price comes
from SDC Platinum. For this sample, the initial raw return is 7.37%, the initial CRSP VW index-adjusted return is
7.23%, and the initial matching-firm adjusted return is 6.43%.
23
I also perform a regression analysis by using the same model specification as described in
equation (2). The dependent variable is the number of SEOs issued by a firm. Since SEO is not
feasible for private firms, I identify their private placement of equity instead of SEOs. The data
sources used include SDC Platinum on private issues, SNL DataSource on capital issues, LexisNexis database, Factiva (Dow Jones), and Best’s Insurance Reports-“capitalization” section of
individual firms. Panel C of Table 6 shows that IPO firms incur significantly more capital issues
than private firms after the IPO. The result is particularly significant for two years after the IPO
(“year +2”) and in the long run (“Year > +3”).
5.5. Acquisition motivation of IPO firms
To investigate the acquisition motives for going public, I analyze subsequent merger and
acquisition (M&A) activities and their financing methods for the sample of IPOs after going
public28 and report the results in Table 7.
Panel A of Table 7 indicates that IPO firms are acquirers 44 times (with 27 IPO firms
involved) from 1994 to February 2007, and targets only eight times, suggesting that IPO firms
are acquirers more often than they are targets. In those deals for which methods of payment can
be identified (28 transactions for acquirers and eight transactions for targets), the majority of the
transactions are paid in cash, and only a few transactions are financed by stock or a mixture of
stock and cash. Cash financing generally dominates in insurance industry M&As (Cummins and
Xie, 2009), partially due to the size of acquisitions and the regulation rules that some states
impose on M&As. Panel B of Table 7 shows the first acquisition deal of firms (as an acquirer or
as a target) and finds that most IPO acquirers pay cash to their targets, but for IPO targets, 50%
28
M&A information comes from Conning & Company and SNL DataSource, which are more specialized in
insurance M&As than other generic databases, such as Thomson Financial’s SDC Platinum.
24
are financed by common stock, showing some evidence that IPOs help form a currency of stock
for takeovers. In brief, trading as a public firm facilitates a company’s acquisition activities.
Panel C of Table 7 shows a regression analysis with the dependent variable as the number
of times an acquiring firm is involved in M&As. 29 As expected, IPO firms are involved
significantly more in M&A activities than private firms, particularly within the first four years of
going public.
5.6. Analysis of delisted firms
During the study period, some publicly traded insurers delisted from their exchange.
Reasons for firms’ delisting usually include, but are not limited to, high expenses of running as a
public company, poor future performance prospects, and mergers or acquisitions, etc. (Leuz et al.,
2008; Engel et al., 2007; Benninga et al., 2005). As the reverse process, delisting could shed light
on IPO decision making. This study, therefore, conducts a similar set of probit analyses to look
into delisting decisions. Searching the CRSP yields the delist code for the 71 delisting firms in
the sample (12 IPO firms and 59 non-IPO public firms). The Factiva (Dow Jones) reports
confirm the reasons for delisting. Panel A of Table 8 summarizes the delisting reasons. Fortynine firms delisted after they were acquired or merged with other firms; 17 firms were delisted
by their current exchange mainly because they could no longer satisfy the listing requirements.
One firm delisted and traded over-the-counter, and two firms delisted because of share exchange
and liquidation. Two firms voluntarily delisted without disclosing the reason. Thus, the main
reason for delisting seems to be M&A and poor performance. In a study of M&A in the US
property-liability insurance industry, Cummins and Xie (2008) show that financial vulnerability
is a major reason for becoming a takeover target. This implies that financial vulnerability could
be the ultimate cause of most delistings.
29
Including the “target” transactions in the regression returns the same results.
25
Panel B of Table 8 presents the factors affecting a firm’s delisting move. Regression
analysis shows that delisted firms tend to be small, with low premium growth rates, low returns
on assets, and more geographic diversification. These firms also tend to be more financially
vulnerable, as reflected in the significantly positive sign of leverage and agents’ balances ratio
variables.
6. Conclusion
This paper examines the performance of publicly held firms in the US property-liability
insurance industry with the focus on firms that issued IPOs from 1994 to 2005, using private
firms as the benchmark. The objective is to understand and evaluate the decision to go public,
with particular attention given to the question of whether IPO firms underperform, as suggested
by the IPO literature. The study first examines the ex ante determinants of a firm’s decision to go
public, and then examines the ex post effect of the IPO on firm efficiency, operating
performance, and other financials. The paper also analyzes stock returns, post-issue SEOs, and
acquisition activities of IPO firms to provide further information on IPO performance and
motivation. During the sample period, some public firms delisted. This study analyzes and
compares the reasons for their delisting to the reasons for going public.
The results show that IPO firms in the US property-liability insurance industry tend to be
large firms and are more likely to face capital constraints. They have a higher growth rate, use
more reinsurance services, and underwrite more business in riskier insurance lines than firms that
remain private. It is surprising to note that IPO firms are not more profitable than private firms
before going public, as reflected in return on assets, cost efficiency, and revenue efficiency
measures.
IPO firms experience no post-issue underperformance in efficiency, return on assets, or
26
stock returns.30 There is some evidence showing that allocative efficiency and cost efficiency
improve after the IPO, suggesting that IPO firms benefit from the release of capital constraints.
IPO firms deleverage and reduce reinsurance usage immediately after going public, but
investments do not increase significantly. The high premium growth before IPO does not last
into the post-issue period. These findings suggest that insurance companies go public to ease
capital constraints following a period of rapid growth. IPO firms are also active in follow-on
SEOs and are actively involved in M&As, taking advantage of the benefits of operating as public
firms. This paper finds that delisted firms are smaller, with high leverage and low premium
growth, opposite to the determinants of IPOs.
Unlike the general findings of the IPO literature, the IPO firms of this study do not
register poor post-issue stock or operating performance. 31 One explanation is the regulatory
scrutiny that insurers face. Insurance regulations set strict rules regarding minimum capital
requirements and emphasize the long term viability of the insurance operation. Relative stability
in operation may then contribute to steady performance in efficiency, profitability, and stock
returns. Another explanation is the investment strategy of insurers. The model in Carlson et al.
(2006) suggests that underperformance following increase in investment is expected as a
consequence of growth option exercise. In this study, rapid growth occurs before the IPO, and
the post-IPO investment level of firms does not increase significantly. A third possibility is that
the insurance industry has a less serious problem of “window dressing”,32 as IPO firms exhibit
30
The consistency of the results acquired using the three different measures featured in this study provide further
evidence that frontier efficiency scores based on accounting data provide value-relevant information for
management (Cummins and Xie, 2009).
31
Lai et al. (2008) also find no underperformance in long run stock returns of demutualization IPOs in the US
insurance industry. Song (2006) finds no long term stock underperformance of SEOs in the US property-liability
insurance industry, which is also different from the general findings of the SEO literature.
32
Teoh et al. (1998) find that the post-issue underperformance of IPOs is related to firms’ earnings management
behavior before the issue (window dressing). Kao et al. (2009) also find that earnings management is related to
worse long-run stock performance of IPOs in China.
27
no better operating performance or efficiency before their initial issues.
In conclusion, the results show that insurers are more likely to conduct IPOs to ease
capital constraints after rapid growth in business. IPO firms perform no worse than their private
peers after public offering. This suggests that insurance firms are able to adopt appropriate
strategies in public trading based on their business characteristics.
28
References
Adams, B., Carow, K.A., Perry, T., 2009. Earnings management and initial public offerings: The
case of the depository industry. Journal of Banking & Finance 33, 2363–2372.
Alavi, A., Pham, P.K., Pham, T.M., 2008. Pre-IPO ownership structure and its impact on the IPO
process. Journal of Banking & Finance 32, 2361–2375.
Allen, F., Faulhaber, G.R., 1989. Signaling by underpricing in the IPO market. Journal of
Financial Economics 23, 303-323.
Aslan, H., Kumar, P., 2007. Going public and going private: What determines the choice of
ownership structure? Working paper, University of Houston.
Banker, R.D., 1993. Maximum likelihood, consistency and data envelopment analysis: A
statistical foundation. Management Science 39, 1265-1273.
Banker, R.D., Chang H., Cooper, W.W., 2004. A simulation study of DEA and parametric
frontier models in the presence of heteroscedasticity. European Journal of Operational
Research 153, 624-640.
Banker, R.D., Natarajan, R., 2008. Evaluating contextual variables affecting productivity using
DEA. Operations Research 56, 48-58.
Benninga, S., Helmantel, M., Sarig, O., 2005. The timing of initial public offerings. Journal of
Financial Economics 75, 115–132.
Boehmer, E., Ljungqvist, A., 2004. On the decision to go public: Evidence from privately-held
firms. Working paper, New York University.
Brau, J.C., Fawcett, S.E., 2006. Initial public offerings: An analysis of theory and practice.
Journal of Finance 61, 399-435.
29
Carlson, M., Fisher, A., Giammarino, R., 2006. Corporate investment and asset price dynamics:
Implications for SEO event studies and long-run performance. Journal of Finance 61,
1009-1034.
Chamberlain, S.L., Tennyson, S., 1998. Capital shocks and merger activity in the propertyliability insurance industry. Journal of Risk and Insurance 65, 563-595.
Chang, S., 1998. Takeovers of privately held targets, methods of payment, and bidder returns.
Journal of Finance 53, 773-784.
Chemmanur, T.J., Fulghieri, P., 1999. A theory of the going-public decision. Review of
Financial Studies 12, 249-279.
Chemmanur, T.J., He, S., Nandy, D., 2006. The going public decision and the product market.
Working paper, Boston College.
Chen, H., Ritter, J.R., 2000. The seven percent solution. Journal of Finance 55, 1105-1131.
Colaco, H.M.J., Ghosh, C., Knopf, J.D., Teall, J.L., 2009. IPOs, clustering, indirect learning and
filing independently. Journal of Banking & Finance 33, 2070–2079.
Cooper, W.W., Seiford, L.M., Tone, K., 2000. Data Envelopment Analysis: A Comprehensive
Text with Models, Applications, References and DEA-Solver Software. Kluwer
Academic Publishers, Boston, MA.
Cummins, J.D., Nini, G., 2002. Optimal capital utilization by financial firms: Evidence from the
property-liability insurance industry. Journal of Financial Services Research 21, 15-53.
Cummins, J.D., Weiss, M.A., 2000. Analyzing firm performance in the insurance industry using
frontier efficiency and productivity methods. In: Dionne, G. (Ed.), Handbook of
Insurance. Kluwer Academic Publishers, Boston, MA.
30
Cummins, J.D., Weiss, M.A., Zi, H., 1999a. Organizational form and efficiency: The coexistence
of stock and mutual property-liability insurers. Management Science 45, 1254-1269.
Cummins, J.D., Tennyson, S., Weiss, M.A., 1999b. Consolidation and efficiency in the US life
insurance industry. Journal of Banking & Finance 23, 325-357.
Cummins, J.D., Xie, X., 2008. Mergers and acquisitions in the US property-liability insurance
industry: Productivity and efficiency effects. Journal of Banking & Finance 32, 30-55.
Cummins, J.D., Xie, X., 2009. Market values and efficiency in US property-liability insurer
acquisitions and divestitures. Managerial Finance 35, 128-155.
Cummins, J.D., Zi, H., 1998. Comparison of frontier efficiency methods: An application to the
U.S. life insurance industry. Journal of Productivity Analysis 10, 131-152.
Degeorge, F., Zeckhauser, R., 1993. The reverse LBO decision and firm performance: Theory
and evidence. Journal of Finance 48, 1323–1348.
Engel, E., Hayes, R.M., Wang, X., 2007. The Sarbanes–Oxley Act and firms’ going-private
decisions. Journal of Accounting and Economics 44, 116–145.
Faccio, M., McConnell, J.J., Stolin, D., 2006. Returns to acquirers of listed and unlisted targets.
Journal of Financial and Quantitative Analysis 41, 197-220.
Friesen, G.C., Swift, C., 2009. Overreaction in the thrift IPO aftermarket. Journal of Banking &
Finance 33, 1285–1298.
Gopalan, R., Gormley, T. A., 2008. Stock market liberalization and the decision to go public.
Working paper, Washington University in St. Louis.
Harrington, S.E., Niehaus, G., 2002. Capital structure decisions in the insurance industry: Stocks
versus mutuals. Journal of Financial Services Research 21, 145-163.
31
Helwege, J., Packer, F., 2003. The decision to go public: Evidence from mandatory SEC filings
of private firms. Working paper, Ohio State University.
Holmström, B., Tirole, J., 1993. Market liquidity and performance monitoring. Journal of
Political Economy 101, 678–709.
Ibbotson Associates, 2008. Stocks, Bonds, Bills, and Inflation: 2008 Valuation Yearbook.
Morningstar, Chicago, IL.
Jain, B.A., Kini, O., 1994. The post-issue operating performance of IPO firms. Journal of
Finance 49, 1699–1726.
Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: Managerial behavior, agency costs and
ownership structure. Journal of Financial Economics 3, 305-360.
Kao, J.L., Wu, D., Yang, Z., 2009. Regulations, earnings management, and post-IPO
performance: The Chinese evidence. Journal of Banking & Finance 33, 63–76.
Kaplan, S., 1989. The effect of management buyouts on operating performance and value.
Journal of Financial Economics 24, 217-254.
Lai, G.C., McNamara, M.J., Yu, T., 2008. The wealth effect of demutualization: Evidence from
the U.S. property-liability and life insurance industries. Journal of Risk and Insurance 75,
125-144.
Leland, H.E., Pyle, D.H., 1977. Information asymmetries, financial structure, and financial
intermediation. Journal of Finance 32, 371-387.
Lerner, J., 1994. Venture capitalists and the decision to go public. Journal of Financial
Economics 35, 293-316.
32
Leuz, C., Triantis, A., Wang, T.Y., 2008. Why do firms go dark? Causes and economic
consequences of voluntary SEC deregistrations. Journal of Accounting and Economics
45,181–208.
Loughran, T., Ritter, J.R., 1995. The new issues puzzle. Journal of Finance 50, 23-51.
Maksimovic, V., Pichler, P., 2001. Technological innovation and initial public offerings. Review
of Financial Studies 14, 459–494.
Mikkelson, W.H., Partch, M.M., Shah, K., 1997. Ownership and operating performance of
companies that go public. Journal of Financial Economics 44, 281–307.
Pagano, M., Panetta, F., Zingales, L., 1998. Why do companies go public? An empirical analysis.
Journal of Finance 53, 27–64.
Ritter, J.R., 1987. The costs of going public. Journal of Financial Economics 19, 269-281.
Ritter, J.R., 1991. The long-run performance of initial public offerings. Journal of Finance 46, 327.
Ritter J.R., Welch, I., 2002. A review of IPO activity, pricing, and allocations. Journal of Finance
57, 1795-1828.
Shephard, R.W., 1970. Theory of Cost and Production Functions. Princeton University Press,
Princeton, NJ.
Smith, A., 1990. Corporate ownership structure and performance: The case of management
buyouts. Journal of Financial Economics 27, 143-164.
Song, Q., 2006. Returns post catastrophe induced seasoned equity offerings of property and
casualty insurance companies. Working paper, University of Pennsylvania.
33
Teoh, S.H., Welch, I., Wong, T.J., 1998. Earnings management and the long-run
underperformance of initial public equity offerings. Journal of Finance 53, 1935–1974.
Viswanathan, K.S., 2006. The pricing of insurer demutualization initial public offerings. Journal
of Risk and Insurance 73, 439-468.
Viswanathan, K.S., Cummins, J.D., 2003. Ownership structure changes in the insurance industry:
An analysis of demutualization. Journal of Risk and Insurance 70, 401-437.
Welch, I., 1989. Seasoned offerings, imitation costs, and the underpricing of initial public
offerings. Journal of Finance 44, 421-450.
Zingales, L., 1995. Insider ownership and the decision to go public. Review of Economic Studies
60, 425-448.
34
Appendix: DEA mathematical programming for frontier efficiency
This study estimates five types of efficiencies: scale efficiency, technical efficiency,
allocative efficiency, cost efficiency, and revenue efficiency. Among them, cost efficiency and
revenue efficiency are the ultimate measures of firm performance. Cost efficiency can be
decomposed into (the multiplication of) technical efficiency and allocative efficiency. A firm is
fully technically efficient only if it is scale efficient (i.e., operating on the constant returns to
scale frontier with a scale efficiency score equal to 1).
All five types of efficiencies are estimated for each individual firm in each year of the
sample period. Assume that there are N firms in the industry in a given year and that each firm
uses k inputs x = ( x1 , x2 ,..., xk ) ∈ ℜk+ to produce m outputs y = ( y1 , y2 ,..., ym ) ∈ ℜm+ . The input
price vector is w = ( w1 , w2 ,..., wk ) ∈ ℜk+ , and the output price vector is p = ( p1 , p2 ,..., pm ) ∈ ℜm+ .
DEA is used to construct a frontier (production, cost, or revenue) such that all observed firms lie
on or below the frontier. Firms lying on the frontier are regarded as “best practice” firms (with an
efficiency score equal to1).
Based on the input-oriented distance functions introduced by Shephard (1970), the
following program estimates a firm’s technical efficiency with respect to constant returns to
scale technology ( TECRS ):
TE CRS ( x, y ) =Min θCRS
N
Subject to
∑λ y
j ij
(1)
≥ yij
∀ i =1,...,m
≤ θCRS xrj
∀ r =1,...,k
j=1
N
∑λ x
j rj
j=1
λj ≥ 0
∀ j=1,...,N
Input-oriented technical efficiency, TECRS , can be decomposed into pure technical
35
efficiency and scale efficiency. The same program estimates a firm’s input-oriented pure
technical efficiency with respect to the variable returns to scale ( TEVRS ) when including a
N
convexity constraint
∑λ
= 1 in the optimization. Scale efficiency for a given firm is then
j
j=1
measured as the total technical efficiency not explained by pure technical efficiency,
i.e., SE ( x, y ) =
TE CRS ( x, y )
TE VRS ( x, y )
, where SE = input-oriented scale efficiency. If SE = 1, the firm is
on the CRS frontier and is fully scale efficient. If SE < 1, the firm is not on the CRS frontier and
is therefore scale inefficient.
Cost efficiency measures whether a company minimizes its costs by choosing input
quantities while holding constant the input prices w and output quantities y . The linear
programming problem for firm j is:
k
C * ( x, y ) = Min λ , x * ∑ wrj x rj∗
j
r =1
N
∑λ y
j =1
j
ij
N
Subject to:
∑λ x
j =1
j rj
≥ yij
≤ x *rj
λj ≥ 0
∀ i = 1,..., m
∀ r = 1,..., k
(2)
∀ j = 1,..., N
where xj* = (x1j*, x2j*,…, xkj*) is firm j’s cost-minimizing input level. The cost efficiency (CE) of
k
the firm is the ratio of frontier cost over actual cost, i.e., CE ( x, y ) = ∑ wrj xrj*
r =1
k
∑w
r =1
x , where
rj rj
x j = { x1 j , x2 j ,..., xkj } is the observed input quantity vector of the firm. Fully cost efficient firms
have CE = 1 and cost inefficient firms have a CE estimate falling between zero and one.
Cost efficiency captures both technical efficiency and allocative efficiency, where
36
allocative efficiency measures the success of the firm in choosing cost-minimizing combinations
of inputs. Allocative efficiency is calculated residually from cost and technical efficiency as
follows: AE ( x, y ) =
CE ( x, y )
TE CRS ( x, y )
. The complement of AE measures the cost inefficiency of a
firm due to its failure to adopt the optimal combinations of inputs, given the input prices and
output quantities.
Revenue efficiency estimation is similar to cost efficiency, except that it adopts an
output-oriented approach to maximize revenues. The revenue efficiency of the firm is given by:
m
RE ( x, y ) = ∑ pij yij
i =1
m
∑p
i =1
ij
yij* , where yj = (y1j, y2j, …, ymj) and yj* = (y1j*, y2j*, …, ymj*) represent
firm j’s actual output quantities and the revenue-maximizing output level, respectively, and pij
represents output prices. RE is also bounded between zero and one with RE = 1 for revenue
efficient firms.
37
Table 1: Number of publicly traded property-liability (P-L) insurance firms in the US, 1993-2005
Current Firms
All Public
N
Year
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
Total
Study
Sample: IPO
Firms
N
Old listing
P-L
71
New listing 1994-2005
P-L
Other
44
8
Other
18
New listing 1994-2005
Demutualization IPO
Non-demutualization IPO
P-L
Other
P-L
Other
8
1
1
1
1
1
4
1
1
4
3
3
1
7
7
6
40
1
4
5
1
3
Current Firms
Non-demutualization IPO
P-L
Other
40
3
Historical Firms
Delisting 1993-2005
P-L
Other
59
18
Historical Firms (1)
Delisting 1993-2005
P-L (2)
Other (3)
11
8
14
8
8
5
3
2
4
2
3
2
3
1
1
2
59
18
Historical Firms
Non-demutualization IPO
P-L
Other
10
2
Note: The term "Current Firms" refers to firms that are publicly traded at least until the end of 2005. "Historical firms" refers to companies that were delisted
between 1993 and 2005. "Other" means "Other public insurance firms with P-L operation," which refers to public life-health and managed care firms that have PL business. The primary insurance sector classification information is from SNL DataSource. The omission of "Other public insurance firms with P-L operation"
from the study sample does not affect the results.
(1) Seventeen historical firms issued IPOs between 1994 and 2005, among them, four "P-L" and one "other" were demutualization IPOs; (2) Four
demutualization IPO firms delisted; (3) two demutualization IPO firms delisted; The demutualization IPO information comes from Viswanathan (2006) and
SNL Financial -- Insurance Industry Vital Statistics.
"Study Sample: IPO Firms" includes current firms and historical firms that issued IPOs between 1994 and 2005; the demutualization IPO firms are excluded.
Table 2: Summary statistics on IPO firms and private firms
IPO firms
Variables
Mean
N
Private firms
Std. dev
Mean
45
Std. dev
Mean Difference:
IPO-Private
1499
Size and Earnings
Total assets ($ millions)
2398.6
5194.8
366.8
1488.3
2031.8**
Premiums ($ millions)
1377.7
3051.9
205.9
824.4
1171.8**
770.8
1686.9
126.7
484.5
644.1**
44.4
191.0
8.6
40.6
35.8
Leverage
1.849
1.205
1.688
1.082
0.161
Reinsurance ratio
0.428
0.253
0.283
0.227
0.145***
Agents' balances ratio
0.265
0.267
0.153
0.146
0.112***
1.529
1.291
1.186
0.701
0.344*
0.079
0.115
0.091
0.156
-0.012
0.203
0.246
0.187
0.269
0.016
0.177
0.250
0.298
0.390
-0.121***
0.541
0.358
0.424
0.413
0.117*
0.410
0.312
0.611
0.292
-0.201***
0.384
0.336
0.609
0.371
-0.225***
-0.017
Policyholder surplus ($ millions)
Net income ($ millions)
Financial Status
Premium Growth
Change in premiums, t-1 to t
Business Mix
Percent of premiums in personal
lines short-tail
Percent of premiums in personal
lines long-tail
Percent of premiums in commercial
lines short-tail
Percent of premiums in commercial
lines long-tail
Diversification Level
Product line Herfindahl, premiums
written
Geographic Herfindahl, premiums
written
Operating and Underwriting Performance
Return on assets
0.033
0.080
0.050
0.079
Loss ratio
0.739
0.346
0.646
0.264
0.094*
Expense ratio
0.422
0.392
0.383
0.272
0.039
0.244
0.435
0.634
0.482
-0.390***
Scale efficiency
0.758
0.216
0.867
0.161
-0.110***
Other Characteristics
Percent unaffiliated companies
Efficiency Scores
Allocative efficiency
0.668
0.203
0.722
0.172
-0.054*
Technical efficiency
0.554
0.258
0.534
0.232
0.020
Cost efficiency
0.373
0.183
0.380
0.184
-0.007
Revenue efficiency
0.330
0.213
0.345
0.227
-0.015
Note: The “private firms” (IPO = 0) include both firms that stayed private during the sample period and pre-IPO
firms. Company characteristics and efficiency scores take values one year prior to the IPO issue, except for
"Premium Growth," where "Change in premiums t-1 to t" is defined as net premiums written at time t divided by net
premiums written at time t-1. Premiums = Direct premiums written + reinsurance assumed; Reinsurance ratio =
Reinsurance ceded / Premiums. *Significant at the 10% level. **Significant at the 5% level. ***Significant at the
1% level. The asterisks illustrate whether the mean difference between IPO firms and private firms is significant
based on the t-test.
Table 3: Probit analysis: Determinants of the decision to go public
Variables
Size: Ln (assets)
Leverage
Reinsurance ratio
Agents' balances ratio
Premium growth
Percent of premiums in
personal lines long-tail
Percent of premiums in
commercial lines long-tail
Product line Herfindahl,
premiums written
Geographic Herfindahl,
premiums written
Return on assets
Loss ratio
Expense ratio
Unaffiliated companies
Industry market-to-book ratio
Model 1
0.203***
[0.062]
-0.055
[0.093]
0.597*
[0.340]
0.668*
[0.403]
0.160**
[0.065]
0.658
[0.465]
0.599*
[0.310]
-0.127
[0.303]
-0.177
[0.214]
0.405
[1.853]
-0.180
[0.517]
0.619***
[0.221]
-0.159
[0.191]
0.641
[0.431]
Scale efficiency
Model 2
0.175***
[0.065]
-0.081
[0.099]
0.571
[0.353]
0.588
[0.401]
0.159**
[0.069]
0.732
[0.464]
0.551*
[0.312]
-0.131
[0.304]
-0.186
[0.222]
0.252
[1.704]
-0.141
[0.539]
0.536**
[0.241]
-0.194
[0.200]
0.520
[0.473]
-0.971*
[0.518]
Technical efficiency
Model 3
0.199***
[0.063]
-0.057
[0.091]
0.589*
[0.335]
0.658
[0.408]
0.157**
[0.066]
0.624
[0.442]
0.621*
[0.321]
-0.151
[0.321]
-0.184
[0.217]
0.371
[1.927]
-0.198
[0.529]
0.637***
[0.221]
-0.165
[0.189]
0.655
[0.442]
Model 4
0.235***
[0.067]
-0.103
[0.091]
0.476
[0.341]
0.634
[0.404]
0.145**
[0.064]
0.964*
[0.525]
0.703**
[0.340]
-0.189
[0.307]
-0.130
[0.222]
0.087
[1.633]
-0.217
[0.542]
0.633***
[0.222]
-0.190
[0.194]
0.492
[0.488]
Model 5
0.240***
[0.066]
-0.059
[0.094]
0.581*
[0.346]
0.707*
[0.409]
0.161**
[0.068]
0.930**
[0.474]
0.536*
[0.320]
-0.022
[0.311]
-0.119
[0.228]
0.586
[1.760]
-0.111
[0.524]
0.551**
[0.245]
-0.128
[0.191]
0.519
[0.486]
0.138
[0.484]
Allocative efficiency
-1.568***
[0.503]
Cost efficiency
-0.954
[0.648]
Revenue efficiency
No. of observations
Pseudo R-square
Model 6
0.206***
[0.063]
-0.052
[0.091]
0.605*
[0.338]
0.677*
[0.409]
0.164**
[0.067]
0.654
[0.470]
0.590*
[0.314]
-0.099
[0.322]
-0.172
[0.218]
0.461
[1.919]
-0.163
[0.532]
0.610***
[0.224]
-0.155
[0.194]
0.638
[0.433]
1528
0.195
1528
0.206
1528
0.195
1528
0.223
1528
0.201
-0.127
[0.504]
1528
0.195
Note: This table presents probit analysis to test the factors associated with a firm's decision to go public during the
period 1994-2005. The dependent variable is zero if a firm is not listed and one on the year of IPO (a firm is dropped
from the sample after it goes public), i.e., the IPO = 0 sample includes firms that stayed private during the sample
period and pre-IPO firms. The independent variables take values one year prior to the IPO issue (i.e. year t-1 value),
except for "premium growth," for which current year value is used, as in Pagano et al. (1998). Industry market-tobook ratio is the median market-to-book value of equity of public firms in the insurance industry at the year-end of
t-1. *Significant at the 10% level. **Significant at the 5% level. ***Significant at the 1% level. Standard errors in
brackets. Constant term is not reported. Time dummies are excluded because none of them are significant. A
robustness test with time dummies returns similar results.
Table 4: Effects of the decision to go public
Variables
Year 0
Year +1
Year +2
Year +3
Year> +3
F-test
Obs
-0.002
-0.032
-0.033
0.001
-0.005
0.4425
2264
[0.021]
[0.021]
[0.023]
[0.024]
[0.021]
0.088***
0.057**
0.086***
0.059*
0.099***
0.0046
2264
[0.029]
[0.028]
[0.031]
[0.032]
[0.028]
-0.015
-0.029
-0.004
0.015
-0.011
0.8283
2264
[0.029]
[0.028]
[0.031]
[0.032]
[0.028]
0.047**
0.015
0.056**
0.052**
0.056***
0.0268
2264
[0.020]
[0.020]
[0.022]
[0.023]
[0.020]
0.050*
0.032
0.045
0.059*
0.046
0.4118
2264
[0.029]
[0.029]
[0.031]
[0.033]
[0.028]
0.001
-0.013
0.013
-0.007
-0.022
0.2183
2266
[0.014]
[0.014]
[0.015]
[0.016]
[0.014]
0.018
0.053
0.072*
0.073*
0.051
0.4609
2257
[0.038]
[0.038]
[0.041]
[0.043]
[0.037]
-0.022
-0.048
0.002
-0.026
-0.008
0.8516
2238
[0.041]
[0.041]
[0.044]
[0.047]
[0.040]
-0.312**
-0.240*
-0.265**
-0.168
-0.156
0.0749
2266
[0.125]
[0.123]
[0.134]
[0.141]
[0.121]
Investment in equipment and
EDP
0.136
0.398
-0.021
0.118
-0.123
0.8905
2266
[0.433]
[0.425]
[0.462]
[0.488]
[0.418]
Financial investments - all
0.014
0.000
0.000
-0.018
-0.015
0.5720
2266
[0.016]
[0.016]
[0.017]
[0.018]
[0.015]
Financial investments –
Long-term
0.017
0.013
-0.002
-0.029
-0.021
0.5322
2259
[0.024]
[0.024]
[0.026]
[0.027]
[0.023]
Premium growth
-0.064
-0.185
-0.010
-0.290*
-0.299**
0.2249
2265
[0.150]
[0.147]
[0.160]
[0.169]
[0.146]
-0.063**
-0.057**
-0.066**
-0.021
-0.015
0.0667
2242
[0.029]
[0.028]
[0.031]
[0.034]
[0.028]
0.026
0.020
0.006
0.011
0.014
0.8677
2192
[0.022]
[0.021]
[0.023]
[0.025]
[0.021]
Efficiency
Scale efficiency
Allocative efficiency
Technical efficiency
Cost efficiency
Revenue efficiency
Operating Performance
Return on assets
Underwriting Performance
Loss ratio
Expense ratio
Other Financials
Leverage
Reinsurance ratio
Agents' balances ratio
Note: This table presents the fixed effects model for the effects of the decision to go public. The dependent variables
are the firm’s various types of efficiencies, operating performance, underwriting performance, and other financial
variables. The key independent variables are the post-IPO dummies (which are reported in the table).
The F-test provides a test for the hypothesis that the sum of the coefficients of all the post-IPO dummies is equal to
zero (p-value is reported). *Significant at the 10% level. **Significant at the 5% level. ***Significant at the 1%
level. Standard errors in brackets.
Table 5: Abnormal returns for initial public offerings from 1994 to 2005
Raw Return
Raw Return
t-stat
(%)
Month
# of
Firms
1
55
3.87
2
55
2.78
Matching Firm Adjustment
AR(t) (%)
t-stat
CAR(1,t) (%)
t-stat
2.63
1.50
0.88
1.50
0.76
1.84
-3.43
-0.94
-1.93
-0.83
3
55
0.93
0.65
-0.42
-0.21
-2.36
-0.90
4
54
2.47
1.81
0.39
0.19
-1.97
-0.67
5
54
1.97
1.32
0.35
0.20
-1.61
-0.51
6
54
-0.54
-0.55
0.74
0.55
-0.88
-0.26
7
54
2.02
1.44
0.56
0.32
-0.32
-0.09
8
54
3.05
2.03
1.81
0.93
1.49
0.39
9
54
3.53
2.88
1.07
0.75
2.56
0.64
10
54
-0.78
-0.76
-2.09
-1.46
0.47
0.11
11
54
-0.81
-0.47
-1.56
-0.64
-1.09
-0.25
12
54
2.03
1.33
2.07
1.39
0.98
0.21
13
54
0.32
0.21
-0.89
-0.44
0.08
0.02
14
54
0.79
0.79
0.05
0.02
0.14
0.03
15
54
0.19
0.17
-1.33
-0.88
-1.20
-0.24
16
54
2.96
2.13
1.39
0.66
0.19
0.04
17
54
2.48
1.89
0.16
0.09
0.35
0.07
18
53
0.22
0.18
-1.44
-0.87
-1.10
-0.20
19
53
0.57
0.41
-0.54
-0.33
-1.64
-0.29
20
53
1.18
1.20
0.21
0.17
-1.43
-0.25
21
52
0.59
0.33
0.52
0.26
-0.90
-0.15
22
52
0.84
0.48
-2.27
-1.12
-3.18
-0.52
23
52
0.54
0.41
0.07
0.05
-3.11
-0.50
24
52
1.64
1.24
0.06
0.04
-3.05
-0.48
25
52
1.77
1.52
-0.37
-0.23
-3.42
-0.53
26
52
-0.35
-0.30
-2.42
-0.92
-5.84
-0.88
27
51
0.56
0.34
0.74
0.42
-5.09
-0.75
28
51
-0.95
-0.69
-4.23
-2.51
-9.33
-1.35
29
50
0.77
0.62
-2.51
-1.39
-11.83
-1.67
30
49
1.95
0.98
3.13
1.31
-8.70
-1.19
31
49
0.30
0.19
-2.42
-1.06
-11.12
-1.50
32
49
-1.20
-0.63
-1.95
-1.06
-13.07
-1.74
33
49
3.05
1.44
2.76
0.93
-10.31
-1.35
34
48
-2.54
-1.52
-3.05
-1.20
-13.36
-1.71
35
48
0.60
0.36
2.76
1.25
-10.60
-1.34
36
48
0.46
0.20
0.40
0.15
-10.20
-1.27
Note: This table presents the average monthly raw returns of IPO firms, the average matching firm-adjusted returns
(AR (t)), and the cumulative average matching firm-adjusted returns (CAR (1,t)) for 36 months after going public,
excluding the initial return.
Table 6: Number of IPO firms that issued follow-on SEOs (1994 - Nov. 2008)
Panel A: Firms with issues of SEO
Non-Delisted IPOs
Delisted IPOs
No issue
20
6
One issue
11
5
Two issues
4
1
Three issues
2
> 3 issues
6
Panel B: Number of firms whose first SEO occurred at
Year after IPO
Non-Delisted IPOs
Delisted IPOs
Year 0
2
1
Year +1
7
1
Year +2
2
4
Year +3
4
Year > +3
8
Panel C: Regression of number of SEOs
Variable
Year 0
Year +1
Year +2
Year +3
-0.02
0.018
0.176***
0.003
Number of
SEOs
[0.039]
[0.038]
[0.041]
[0.043]
Total Firms
26
16
5
2
6
Total Firms
3
8
6
4
8
Year> +3
0.098***
[0.037]
F-test
0.0000
Obs
2278
Table 7: Number of IPO firms involved in M&A activities (1994 - Feb. 2007)
Panel A: Firms with M&A activities
Total Firms
Total Activities
Stock
Payment Method
Cash
Cash and Stock
Acquirer
None
28
0
One time
17
17
7
1
Two times
8
16
9
2
> Two times
2
11
1
7
1
Acquirer Total
27
44
1
23
4
Target
None
4
0
One time
8
8
4
3
1
Panel B: Number of firms whose first acquisition occurred at
Payment Method for the First Deal
Year after IPO
Total Firm
Stock
Cash
Cash and Stock
Acquirer
Year 0
1
1
Year +1
9
2
1
Year +2
9
6
1
Year +3
1
1
Year > +3
7
3
Target
Year 0
0
Year +1
1
1
Year +2
0
Year +3
2
2
Year > +3
5
3
1
1
Panel C: Regression of number of times firms are acquirers
Variable
Year 0
Year +1
Year +2
Year +3
Year> +3
F-test
Obs
-0.019
0.169***
0.266***
0.190***
0.009
0.0000
2278
Number of
Acquirers
[0.060]
[0.059]
[0.064]
[0.067]
[0.058]
Table 8: Analysis of delisted firms
Panel A: The reason of delisting from the CRSP for 71 delisted firms
Delisting Reason
Number of Firms
Merger 1
Merger 2
Merger 3
Exchanges
Liquidations
Dropped 1
Dropped 2
Dropped 3
Dropped 4
Dropped 5
Dropped 6
Dropped 7
Dropped 8
17
25
7
1
1
1
1
5
1
1
1
1
2
Dropped 9
7
Detailed Reason
When merged, shareholders primarily receive common stock or ADRs.
When merged, shareholders receive cash payments.
When merged, shareholders primarily receive common stock and cash, issue on CRSP file.
Issue exchanged, primarily for another class of common stock.
Issue liquidated, no final distribution is verified, issue closed to further research.
Issue stopped trading current exchange - trading Over-the-Counter.
Delisted by current exchange - insufficient number of shareholders.
Delisted by current exchange - price fell below acceptable level.
Delisted by current exchange - insufficient capital, surplus, and/or equity.
Delisted by current exchange - insufficient (or non-compliance with rules of) float or assets.
Delisted by current exchange - company request (no reason given).
Delisted by current exchange - company request, deregistration (gone private).
Delisted by current exchange - delinquent in filing, non-payment of fees.
Delisted by current exchange - does not meet exchange's financial guidelines for continued
listing.
Panel B: Regression analysis of factors affecting delisting
Variables
Size: Ln (assets)
Leverage
Reinsurance ratio
Agents' balances ratio
Premium growth
Percent of premiums in personal lines
long-tail
Percent of premiums in commercial lines
long-tail
Product line Herfindahl, premiums written
Geographic Herfindahl, premiums written
Return on assets
Loss ratio
Expense ratio
Unaffiliated companies
Model 1
-0.303***
[0.073]
0.295***
[0.108]
0.603
[0.432]
1.266***
[0.457]
-0.574**
[0.292]
0.374
[0.509]
0.087
[0.317]
0.299
[0.351]
-1.455***
[0.485]
-1.939
[1.518]
-0.52
[0.426]
0.025
[0.443]
0.098
[0.342]
Technical efficiency
Model 2
-0.314***
[0.075]
0.298***
[0.112]
0.744*
[0.444]
1.289***
[0.464]
-0.598*
[0.310]
0.382
[0.536]
0.21
[0.331]
0.166
[0.388]
-1.506***
[0.492]
-1.878
[1.577]
-0.568
[0.457]
-0.029
[0.480]
0.26
[0.352]
0.332
[0.514]
Cost efficiency
Model 3
-0.317***
[0.076]
0.305***
[0.110]
0.770*
[0.453]
1.315***
[0.463]
-0.589*
[0.311]
0.355
[0.564]
0.196
[0.329]
0.208
[0.373]
-1.485***
[0.494]
-1.868
[1.581]
-0.549
[0.453]
-0.041
[0.486]
0.267
[0.354]
Model 5
-0.281***
[0.068]
0.258**
[0.104]
0.345
[0.394]
1.217***
[0.442]
-0.454*
[0.260]
Model 6
-0.244***
[0.062]
0.217**
[0.096]
0.224
[0.337]
-1.317***
[0.438]
-1.143
[1.241]
0.097
[0.324]
0.119
[0.314]
-1.295***
[0.403]
-2.686*
[1.387]
-0.375
[0.384]
0.035
[0.410]
-0.075
[0.309]
702
0.128
747
0.115
-0.528*
[0.275]
0.293
[0.572]
Revenue efficiency
No. of observations
Pseudo R-square
Model 4
-0.305***
[0.075]
0.313***
[0.111]
0.706
[0.442]
1.210**
[0.474]
-0.600*
[0.309]
0.535
[0.523]
0.201
[0.326]
0.097
[0.389]
-1.522***
[0.493]
-2.103
[1.600]
-0.661
[0.472]
0.068
[0.482]
0.262
[0.351]
687
0.142
687
0.148
687
0.148
0.612
[0.528]
687
0.151
Note: This table presents probit analysis to test factors associated with a firm's delisting status during the period 1994-2005. The dependent
variable is zero if a public firm is not delisted and one on the year of delisting (a firm is not included in the sample before it goes public and is
dropped from the sample after it is delisted). Among the 71 delisted firms, 12 conducted IPOs from 1994 to 2005. The independent variables take
values one year prior to the delisting (i.e. year t-1 value). *Significant at the 10% level. **Significant at the 5% level. ***Significant at the 1%
level. Standard errors in brackets. Constant term and time dummies are not reported.
Figure 1. Long-run Stock Performance of IPOs with
Initial Return
50
40
CAR (%)
30
20
10
0
-10
0
5
RAW Returns
10
15
20
25
Months Relative To the Date of IPO
CRSP VW index-adjusted
30
35
Matching firm-adjusted
This figure presents the cumulative average adjusted returns (CAR) for 55 IPO firms from 1994
to 2005, with monthly rebalancing. Three CAR series are plotted for the 36 months after the IPO
date: 1) no adjustment (Raw Returns), 2) CRSP value-weighted AMEX/NYSE/NASDAQ index
adjustment (CRSP VW index-adjusted), and 3) matching firm adjustment (Matching firmadjusted). Month zero is the initial return interval. The initial return is calculated as (First trading
day closing price – Offer price)/Offer price. For the sample of IPOs, the initial raw return is
7.37%, the initial CRSP VW index-adjusted return is 7.23%, and the initial matching-firm
adjusted return is 6.43%.