The Journal of Risk and Insurance, 2001, Vol. 68, No.

3, 489506

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS

Aigbe Akhigbe Jeff Madura

ABSTRACT
Given the increasing consolidation in the insurance industry, the authors objective is to determine how the market revalues the acquirer, target, and rival insurance companies in response to merger announcements. The authors find that target and acquirer insurance companies experience favorable valuation effects at merger announcements. The authors also find positive and significant intra-industry effects in response to the announcements of insurance company mergers, which supports the signaling hypothesis. Furthermore, the authors find that the magnitude of the intra-industry effects is conditioned on the type, size, and location of the insurance companies.

INTRODUCTION
In recent years, the insurance industry has experienced substantial consolidation as mergers have been consummated to increase shareholder value.1 The merger between two insurance companies may be value-enhancing for the following reasons. First, the merger may achieve economies of scale, as the combined company provides services at a lower average cost. The underlying factors that allow economies of scale include the application of technology to a larger customer base, a more efficient distribution system, and more efficient funding that reduces financing transaction costs. Second, the merging of two insurance companies may reduce redundant operations or branches, which would allow the remaining resources to work more efficiently. Third, the combination of two insurance companies that emphasize different services may allow for each entity to benefit from the expertise of the other entity. Fourth, to the extent that unsystematic risk is valued (see Harrington, 1983; Williams, 1983; and Venezian, 1984), the combination of two insurance companies may reduce the risk (see Cox and Griepentrog, 1988). Fifth, a merger may serve as a better alternative than potential insolvency for insurance companies that are experiencing financial problems (see BarNiv and Hathorn, 1997).
Aigbe Akhigbe is Moyer Chair in Finance, College of Business Administration, University of Akron, Ohio. Jeff Madura is SunTrust Bank Professor of Finance, Florida Atlantic University, Fort Lauderdale. The potential benefits of mergers are thoroughly summarized by Halpern (1983) for firms in general. 489

490

THE JOURNAL

OF

RISK

AND INSURANCE

Although there are valid value-enhancing motives for insurance company mergers, the ultimate test of the validity of mergers from a shareholder perspective is based on how shareholder wealth is affected. Many studies have measured the valuation effects of mergers. These studies have generally focused on industrial firms or commercial banks. Although virtually all related studies have found that the target firms experience positive significant valuation effects, the results for the acquiring firms are mixed. The authors objective is to assess the valuation effects that result from large mergers between insurance companies. More specifically, the authors (1) measure the valuation effects of acquiring and target insurance companies whose stock was publicly traded, (2) measure the valuation effects of rival insurance companies (intra-industry effects) in response to announcements of mergers between insurance companies, and (3) attempt to explain the cross-sectional variation in intra-industry effects among rival insurance companies. The authors find positive and significant valuation effects for publicly traded acquirer and target insurance companies. Second, the authors find positive and significant intra-industry effects in response to the announcements of insurance company mergers, which supports the signaling hypothesis. Third, the authors find that the magnitude of the intra-industry effects is conditioned on the type of insurance company target. The authors also find that the intra-industry effects of insurance company mergers are more pronounced for insurance companies that have a similar size and are located in the same region as the target insurance company. The remainder of this article is organized as follows: the second section develops the hypotheses for intra-industry effects, the third section outlines the data and methodology, the fourth section presents the empirical results, and the final section provides conclusions and implications.

HYPOTHESES

FOR INTRA-INDUSTRY

EFFECTS

Theory of Signaling Through Merger Announcements According to Grossman and Hart (1981), an acquirer may possess special information about the target company that indicates the potential value of the target exceeds the existing value. The acquirer can extract gains by acquiring the target and revising managerial decisions. Although small investors may not be able to obtain such special information, investors that are in a position to be large shareholders (such as potential acquirers) have an incentive to develop and capitalize on the information, because they can extract gains on all the shares that they acquire. Shareholders who monitor a bid by a potential acquirer recognize that the acquirer possesses special information not known to the general public and tend to revise upward their valuation of the target. Grossman and Hart state that the job of monitoring the firm will therefore be left to a potentially large shareholder, such as a prospective bidder... One role of a potential bidder is to discover whether a firm is being run inefficiently and, if it is, to take it over and replace the current management by more efficient management. They refer to this role as an allocational bid but state that a bidder can also extract gains by engaging in an acquisitional bid, whereby the bidder has special information that may enable the target firm to improve its perfor-

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS 491

mance even with its existing management. Grossman and Hart suggest that allocational bids are not necessarily distinguishable from acquisitional bids. Halpern (1983) supports the theory of Grossman and Hart by suggesting that the bidder has information about the target firm that is not available to other participants in the market, and not reflected in the current share price of the target. The information may be that the target shares are undervalued based on publicly available information or there are more efficient operating strategies that could be used by the targets management... The announcement of an acquisition bid should be a signal to the market place and the asymmetry in information should be ameliorated (p. 300). Application of Signaling Theory to Insurance Company Mergers The theory of signaling through acquisitions suggested by Grossman and Hart and by Halpern can be applied to insurance company mergers. A bidding insurance company may have special information that allows for either an allocational or acquisitional bid of a target insurance company. The bidder insurance company recognizes that the target insurance company may have potential to improve on its distribution system, its pricing of insurance, its expansion into new insurance products, or other operations (especially once the bidder insurance company controls the target). Once the merger has been announced, shareholders recognize that the bidder has special information about the target insurance company and bid upward the share price of the target. Application of Signaling Theory to Intra-Industry Effects The theory of signaling through acquisitions by Grossman and Hart (1981) and by Halpern (1983) can be adapted to explain the potential for intra-industry effects as a result of the merger announcement between insurance companies. Because small shareholders do not possess the information of a potential insurance company bidder, they tend to rely on signals (announcements of intentions to purchase a target) by the potential bidder. While the signal will obviously be most pronounced for the target insurance company, it may also be transmitted to other rival insurance companies. That is, the information that one particular insurance company will be acquired may not only cause a revaluation of the target insurance company, but it can also revise the probability that other insurance companies similar to the target exhibit characteristics that may allow a potential bidder to extract gains. Thus, market participants revalue the shares of these rival insurance companies as well in anticipation that they will be improved either with their existing management or as a result of being acquired. Malatesta and Thompson (1985) also provide indirect justification for the authors hypothesis that a merger announcement can create valuation effects for rival firms. Malatesta and Thompson find that investors can partially anticipate acquisition attempts made by active acquirers and revalue the acquirer in response to the anticipation. Mandelker (1974) found that target firms experience positive abnormal returns several months before the official merger announcement, which confirms that investors bid up the share prices of those firms that are likely to be acquired. A review of the merger activity segmented by industry over time (as reported in issues of Mergers & Acquisitions) confirms that mergers sometimes occur in waves within a given industry. Mitchell and Mulherin (1996) found that the takeover and restructuring activity in a

492

THE JOURNAL

OF

RISK

AND INSURANCE

particular industry tends to cluster within a narrow time period. A merger within a particular industry may prompt the likelihood of additional mergers in that industry whether they are initiated by one or several different acquirers in that industry. Thus, as investors attempt to anticipate potential targets, they bid up the shares of those firms that are more likely to be acquired. Research by Eckbo (1985) documents the existence of intra-industry effects for manufacturing firms. He determined that merger announcements of manufacturing firms resulted in positive and significant valuation effects for rival firms.2 The majority of his sample covers mergers over the period from 1963 through 1978. The goal of the study was to determine whether the enforcement of U.S. antimerger laws was associated with relatively large industry wide monopoly rents (p. 325). Based on his analysis, he concludes that the industry wealth effect is unrelated to the change in concentration ratio resulting from the merger.3 While his study cannot be used to make inferences about signals within the insurance industry, it can be used to justify the hypothesis that the announcement of a merger can signal information about rival firms. Eckbo suggests that the information may occur in the form of opportunities for productivity increases that are available to the nonmerging rival firms (p. 348). When applying Eckbos argument to the authors sample of insurance companies, the authors hypothesis of positive intra-industry effects in response to a merger announcement could occur simply because of a signal about the potential for higher productivity of rival insurance companies, even if the perceived probability of a takeover of these rivals is not altered by the merger announcement. Signaling Throughout the Insurance Industry The characteristics of insurance company operations may cause market participants to value insurance company shares based on signals related to other insurance companies with similar operations. The financial statements of insurance companies offer limited information for predicting the future performance of the firm. Their cash outlays for property-liability and health insurance payments are subject to much uncertainty, and the valuation of their assets is highly sensitive to market conditions that affect the prices of their security holdings. Furthermore, their performance is highly sensitive to regulations of insurance operations and regulations imposed on insurance reimbursement. Given the limited financial information available, market participants may attempt to rely on events pertaining to an individual insurance company as a possible signal about other insurance companies with similar operations. Some evidence exists that information about an insurance company can emit a signal about other insurance companies. Fenn and Cole (1994) found that the announcements of financial problems of First Executive and Travelers resulted in negative valu-

Other recent studies have documented that specific announcements pertaining to a single firm can contain signals about the corresponding industry. For example, industry effects have been detected around going-private transactions (Slovin, Sushka, and Bendeck, 1991), seasoned equity issues (Slovin, Sushka, and Polonchek, 1992; Szewczyk, 1992), and bankruptcies (Lang and Stulz, 1992). Also Stillman (1983), Eckbo (1983), Eckbo and Wier (1985), and Eckbo (1992) examine the effect of horizontal mergers on industry rivals and find little evidence for collusion.

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS 493

ation effects of other life insurance companies, which reflects a contagion effect. Given that market participants used negative information about an insurance company to revalue shares of other insurance companies downward, it is possible that market participants would use favorable news about an insurance company (that the company will be acquired and used to improve its efficiency) to revalue shares of other insurance companies upward. The transfer of the signal (whether the signal reflects good news or bad news) across insurance companies is driven by the possibility that the event occurring for a specific insurance company now will recur for other similar companies in the future. Research by Shaked (1985) suggests that the probability of insurance company failure is commonly assessed by analysts, and that these analysts use various indicators to revise their perception that an insurance company will fail. But analysts assessments are not only focused on unfavorable conditions. Analysts also assess the probability that an insurance company will be acquired, because it is well documented that targets experience substantial share price appreciation in response to an acquisition. Eakins (1993) found that acquisitions can be an effective means of corporate control in the insurance industry. However, the expectations that corporate control will be enforced may increase over time in response to information about a new acquisition. Thus, share prices could increase throughout the industry when the market is prompted by a new acquisition. The rationale for responding to a prompt is that the initiative for an insurance company to enact an acquisition may be stimulated by news of another acquisition within the industry. An insurance company may be pressured to expand through acquisitions when it notices consolidation within the industry. Its board of directors may be more willing to approve an acquisition as a defensive strategy when competitors are engaging in acquisitions.

DATA

AND

METHODOLOGY

Data The analyses necessary to test the authors hypotheses require a sample of publicized insurance company mergers. To qualify for the sample, the acquisition had to (1) occur during the period from 1985 through 1995, (2) be reported in Mergers & Acquisitions, (3) have a purchase price of $50 million or more, and (4) be announced in The Wall Street Journal without any public leakage of information before the announcement. The minimum purchase price of $50 million is intended to screen out acquisitions that would not receive much media attention. The announcement criterion is necessary to pinpoint the time at which the intra-industry effects (if any) would occur. A total of 88 large insurance company mergers qualify for the sample. A description of the sample is disclosed in Table 1. More than one-half of the acquisitions occurred within the period from 1992 through 1995, which reflects the increasing popularity of insurance company mergers in recent years. The mean purchase price has varied substantially among years but is generally higher in the 1990s. This is attributed to an emphasis on acquiring larger insurance companies, as well as the higher valuations of insurance companies in the 1990s.

494

THE JOURNAL

OF

RISK

AND INSURANCE

TABLE 1 The Distribution of 88 Acquisitions of Insurance Companies, 1985 Through 1995

Year 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 TOTAL Number 3 13 5 7 7 2 5 11 8 12 15 88 Mean Purchase Price ($ Million) 110.97 129.01 138.74 291.50 244.53 460.25 103.44 302.56 78.88 370.49 212.30 221.40

Methodology The event study methodology is used to measure the average daily abnormal returns to insurance merger announcements. The authors event date, t0, is defined as the first report date of the merger in The Wall Street Journal. Abnormal return is calculated as the difference between the actual return and the expected return. Expected return is generated from the market model parameters, estimated with daily returns from the period t220 to t20 relative to the announcement date. Abnormal returns are generated for the acquirer, target, and rival firms pooled into an equally weighted portfolio. Following the methodology of Mikkelson and Partch (1988), the z-statistic is computed and used to test for statistical significance of the average standardized cumulative abnormal return.

EMPIRICAL RESULTS
Valuation Effects for the Acquirer Insurance Companies The valuation effects of insurance companies involved in mergers are disclosed in Table 2. For 61 of the 88 insurance company mergers that qualified for the sample, the acquiring firm was a publicly traded insurance company. For these cases, the acquiring insurance company experienced a mean two-day cumulative abnormal return (CAR) of 2.21 percent, on average, which is statistically significant. The positive valuation effects on acquiring insurance companies suggests that the market tends to distinguish acquirer insurance companies from other acquirers. Most studies on acquisitions that are not focused on a particular type of firm find that acquirers experience either a significant negative valuation effect or nonsignificant valuation effects. Such results are sometimes attributed to a lack of potential synergy or agency problems that may prevent the combined firm from capitalizing on potential efficiencies. The favorable valuation effects of the acquirer insurance companies may occur be-

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS 495

TABLE 2 Cumulative Abnormal Returns (CARs) to Acquirers and Target Insurance Firms in Response to Acquisition Announcements
This table presents the abnormal returns to acquirers and their target insurance firms in acquisition announcements. The sample period is 1985 through 1995. Abnormal return is calculated as the difference between the actual return and expected return. Expected return is generated from the market model parameters, estimated with daily returns from the period t220 to t20, where t0 is the first report of the acquisition in The Wall Street Journal. The z-statistic tests the null hypothesis that the average abnormal return equals zero. Panel A: CARs of Acquirers in Response to Acquisition Announcements Number of Events 61 61 61 Event Period [11, 2] [1, 0] [+1, +10] CAR (%) 0.28 2.21 0.23 z-statistic 1.62 0.06
2

% Pos 44 55 47

4.961

Panel B: CARs of Targets in Response to Acquisition Announcements Number of Events 22 22 22 Event Period [11, 2] [1, 0] [+1, +10] CAR (%) 1.49 20.78 2.28 z-statistic 1.111 21.43 0.36
1

% Pos 63 81 36

Panel C: CARs for the Matched Pairs of Acquirers and Targets. Number of Events 16 16 16
1 2

Event Period [11, 2] [1, 0] [+1, +10]

CAR (%) 0.62 13.11 2.00

z-statistic 0.42 2.401 1.06

% Pos 56 72 41

Significant at better than the 0.05 level Significant at better than the 0.10 level

cause of the nature of the insurance company operations. The services offered by insurance companies are somewhat standardized and can be more easily merged or commingled among units, which leads to a more favorable market reaction. Conversely, the merging of non-standardized manufacturing operations may not be as efficient, causing the market to be more skeptical in its response to mergers among manufacturing firms. Valuation Effects for the Target Insurance Companies Twenty-two of the 88 target insurance companies in the sample were publicly traded. The mean two-day CAR of these targets is 20.74 percent (see Panel B of Table 2). This result is somewhat consistent with findings from previous studies on mergers. The average increase in dollar value among the publicly traded target insurance companies is $36.63 million. However, the small size of the sample of publicly traded target insurance companies limits the inferences that can be made.

496

THE JOURNAL

OF

RISK

AND INSURANCE

Valuation Effects for the Acquirer and Target Insurance Companies Combined The combined valuation effects for insurance company mergers are measured as the CARs of the matched pairs of the acquirers and the targets. The combined valuation effect is 13.11 percent (significant at the 0.01 level), which suggests that the merger of the two entities has a higher value than the sum of the two individual entities, on average.4 However, only 16 mergers within the sample of large insurance company mergers met the requirement that the acquirer and the target be publicly traded to be included in this sample. Given the small sample size for this specific measurement, any inferences regarding the combined valuation effects must be made with caution. Cross-Sectional Analysis of Acquirer Insurance Companies To explain the variation of valuation effects among insurance company acquirers, a cross-sectional analysis is conducted. Previous studies (see Halpern, 1983) have suggested that the form of payment can influence the valuation effect on the acquirer, as acquirers willing to make cash payments may be valued more favorably by the market at the time of the announcement. Also, the relative size of the target (measured in proportion to the size of the industry median size) may lead to a more pronounced effect on the acquirer because the relative investment is larger. Furthermore, the benefits extracted by the acquirer may be conditioned on the operations of the insurance company. The three-digit SIC codes are used to classify the insurance company into life, non-life, or multi-line.5 The authors distinguish among the three types of insurance companies by including a dummy variable that designates life insurance acquirers and a second dummy variable that designates nonlife insurance companies. The authors also include a dummy variable to distinguish between publicly traded target and privately held insurance company targets. When a target is publicly traded, the authors expect lower valuation effects to the acquirer, because the value of the target was already marked to market because of ongoing trading of that targets shares even before the acquisition. Conversely, the privately held target is not continuously valued by the market, so that an acquirer may more easily acquire a privately held target at a relatively low price. The variables just identified are tested with the following cross-sectional model: CARj = l0 + l1CASHj + l2RELSIZEj + l3LIFEj + l4NONLIFEj + l5PUBLICj + j,

When the pairs of acquirers and targets are weighted by their respective market capitalizations, the combined valuation effect is 1.81 percent, significant at the 0.05 level. Life insurers are represented by the three-digit SIC code 631. Non-life insurers specialize in one insurance type other than life and are denoted by the three-digit SIC codes 632, 633, 635, 636, and 641. Multi-line insurers specialize in more than one insurance line of business. The (historic) SIC codes are obtained from the Center for Research on Securities Prices (CRSP) files, at the date of the merger announcement (see Hertzel, 1991; Slovin, Sushka, and Bendeck, 1991; Slovin, Sushka, and Polonchek, 1992; and Szewczyk, 1992).

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS 497

where CAR j CASHj RELSIZEj LIFEj PUBLICj = two-day (1,0) CAR of the acquirer j in response to the merger announcement, = 1 if the acquirer j paid cash for the acquisition and 0 otherwise, = the market capitalization of target j, divided by the industry median market capitalization, = 1 if the acquirer j is a life insurance firm and 0 otherwise,

NONLIFEj = 1 if the acquirer j is not a life insurance firm and 0 otherwise, and = 1 if the target j is publicly traded and listed on the CRSP tape and 0 otherwise.

The regression model is tested using Whites (1980) heteroskedasticity-consistent covariance matrix. The default for the dummy variables representing the type of insurance company is the set of multi-line insurance companies.6 Results of the crosssectional model as applied to valuation effects of acquirer insurance companies are disclosed in Table 3. The non-life variable is positive and significant, suggesting that non-life insurance company acquirers experienced more favorable valuation effects, after controlling for the other factors. In addition, the acquisitions of publicly traded insurance company targets resulted in less favorable valuation effects for acquirers, supporting the hypothesis presented earlier. The CASH and RELSIZE variables are not significantly related to the acquirer CAR. Thus, the valuation effects among insurance acquirers do not appear to be driven by the method of payment and the relative size of the target. Although the results regarding the acquirer and the target insurance companies offer insight into the distribution of wealth effects for those insurance companies involved in the merger, the authors main focus is on the measurement of intra-industry effects, which is discussed next. Intra-Industry Valuation Effects While the authors measure the valuation effects of the bidder, the target, and the combined insurance companies to determine the distribution of shareholder gains from the merger, they also attempt to determine whether the announcement of the merger signals information about the rival insurance companies. To the extent that one merger may prompt additional mergers within the industry, an insurance company merger announcement may increase the probability that an insurance company may be acquired, which could result in higher valuations of rival insurance companies. The effect of an insurance company merger on rivals is more likely to occur when the merger announcement has a positive effect on the target insurance company. Although virtually all mergers have a positive effect on the target, the news can leak before the first official announcement. In addition, speculation may have occurred well before
6

Note that, of the three possibilities representing types of insurance companies, only life and non-life are included in the regression model. If all three binary variables are used, as well as an intercept term, there is multicollinearity. To avoid this problem, the multi-line insurance type is dropped.

498

THE JOURNAL

OF

RISK

AND INSURANCE

TABLE 3 Regression Results Explaining Cumulative Abnormal Returns (CAR1,0) of Acquirers Using Whites Heteroskedasticity-Consistent Covariance Estimator.
CARj = l0 + l1CASHj + l2RELSIZEj + l3LIFEj + l4NONLIFEj + l5PUBLICj + j Variables INTERCEPT CASH RELSIZE LIFE NONLIFE PUBLIC Sample F-value R2 Adj R2 CARj CASHj Coefficient 0.0769 0.0083 0.0124 0.0181 0.0653 0.0713 61 1.6132 0.1279 0.0486 t-statistic 2.751 0.37 0.79 0.67 1.991 2.601

= two-day (1,0) CAR of the acquirer j in response to the merger announcement = 1 if the acquirer j paid cash for the acquisition and 0 otherwise

RELSIZEj = the market capitalization of target j, divided by the industry median market capitalization LIFEj PUBLICj
1 2

= 1 if the acquirer j is a life insurance firm and 0 otherwise = 1 if the target j is publicly traded and listed on the CRSP tape and 0 otherwise

NONLIFEj = 1 if the acquirer j is not a life insurance firm and 0 otherwise Significant at better than the 0.05 level Significant at better than the 0.10 level

the first announcement. Thus, if the complete set of information has already been transmitted to the market before the first official announcement of a merger, any possible signal for rivals would have already been transmitted as well (see Malatesta and Thompson, 1985). It is difficult to determine whether the news of the merger leaked out completely before the announcement for those target insurance companies that were privately held, because it is not possible to monitor their share price at the time of the announcement. However, given the abrupt upward valuation of 20.78 percent (on average) for publicly traded target insurance companies at the time of the merger announcement, the market had not fully anticipated the merger announcement. The authors measurement of intra-industry effects is used to determine whether the information effect is limited to the target insurance company or spreads to other insurance companies that conduct similar operations. Results from estimating intra-industry effects in response to insurance company merger announcements are disclosed in Table 4. The mean intra-industry effect, based on the two-day CAR, is 0.06 percent for all insurance companies with the same two-

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS 499

digit SIC code as the target company.7 This effect is statistically significant at the 0.05 level and reflects an aggregate dollar gain of $4.6 billion for the insurance industry. The positive and significant intra-industry effects support the hypothesis that an insurance company merger announcement signals favorable information for corresponding rival insurance companies. TABLE 4 Cumulative Abnormal Returns to Rival Firms in Response to Acquisition Announcements
This table presents the abnormal returns to target insurance firms and their corresponding rivals portfolios in 88 acquisition announcements. Rival portfolios contain all insurance firms that were publicly traded at the time, grouped into portfolio by the event. The sample period is 1985 through 1995. Abnormal return is calculated as the difference between the actual return and expected return. Expected return is generated from the market model parameters, estimated with daily returns from the period t220 to t20, where t0 is the first report of the acquisition in The Wall Street Journal. The z-statistic tests the null hypothesis that the average cumulative abnormal return equals zero. Panel A: CARs of Rival Firms Matched by 2-digit SIC Code Number of Events 88 88 88 Event Period [11, 2] [1, 0] [+1, +10] CAR (%) 0.03 0.06 0.04 z-statistic 0.17 2.191 0.37 % Pos 49 59 56

Panel B: CARs of Rival Firms Matched by 3-digit SIC Code Number of Events 88 88 88 Event Period [11, 2] [1, 0] [+1, +10] CAR (%) 0.34 0.40 0.10 z-statistic 0.50 4.091 0.70 % Pos 51 61 57

Panel C: CARs of Rival Firms Matched by 3-digit SIC Code and Segmented by Line of Business Type of Company Life Non-life Multi-line
1

Number of Events 29 27 32

Event Period [1, 0] [1, 0] [1, 0]

CAR (%) 0.21 1.01 0.06

z-statistic % Pos 2.021 4.041 1.14 69 56 59

Significant at better than the 0.05 level

A total of 13,322 insurance companies were assessed for the 88 merger events. The mean number of rival insurance companies per merger announcement based on this matching method was 151.38, the median was 167.5, the minimum was 15, and the maximum was 185.

500

THE JOURNAL

OF

RISK

AND INSURANCE

The signal resulting from insurance company merger is more likely to carry throughout rival insurance companies if the insurance industry is defined by segments. That is, analysts may view the trends and competition for life insurance companies differently from those of property and casualty insurance companies. For this reason, the authors partition all insurance companies by the three-digit SIC code so that they can measure the degree of signal throughout those insurance companies that emphasize the same line of insurance business. The three-digit SIC classification system categorizes insurance companies according to whether they focus on providing life insurance, on specialized insurance other than life insurance (non-life), or on various types of insurance (multi-line). As shown in Panel B of Table 4, the mean intra-industry effect when using this three-digit SIC matching system is 0.40 percent (based on the two-day CAR), which is statistically significant at the 0.01 level. The intra-industry effects from using the three-digit matching system (Panel B) are compared to those when using the two-digit matching system (in Panel A), based on the two-day CAR. The difference between the mean two-day intra-industry effects of the two subsamples is 0.34 percent, which is statistically significant at the 0.01 level, based on a t-test for difference between means. The difference is even more pronounced when including the ten days before the two-day window. This comparison confirms that the intra-industry effects are more pronounced when narrowing the criteria for identifying rivals of the target insurance company. That is, the market appears to discriminate when revaluing insurance company rivals in response to a merger announcement. They focus their revaluation on those insurance companies that have similar operations as the target insurance company. This implies that their markets expectations of the probability of future takeover or of the potential improvement in operations are revised more for insurance companies that are perceived to be more similar to the target insurance company. To determine whether the intra-industry effects are conditioned by the type of insurance company that was acquired, two-day CARs are separately estimated for each of the three groups classified by three-digit SIC code (see Panel C). For announcements regarding the acquisition of a life insurance company, the mean intra-industry effect is 0.21 percent (significantly different from zero), while announcements regarding the acquisition of non-life companies were accompanied by a mean intra-industry effect of 1.01 percent (significantly different from zero), and announcements regarding the acquisition of multiple-line insurance companies resulted in a mean intraindustry effect of 0.06 percent (not significantly different from zero). Based on a pairwise comparison of these mean intra-industry effects among subsamples, the intra-industry effects resulting from acquisitions of life insurance companies are not significantly different than the intra-industry effects resulting from the acquisitions of non-life insurance companies or multi-line insurance companies. However, there is a significant difference between the intra-industry effects resulting from acquisitions of non-life insurance companies versus multi-line insurance companies. The market relies more heavily on the acquisitions of non-life insurance companies as signals about the respective rival insurance companies. Conversely, the market does extract a signal from the acquisition of a multi-line company, perhaps because of the relatively ambiguous operations of that company and the difficulty in identifying the companies for which a signal may be appropriate. That is, it is not only difficult to

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS 501

pinpoint the specific segment of a multi-line insurance company target that will realize efficiencies from the merger, but it is also difficult to pinpoint the other multi-line insurance companies that exhibit the most potential for increasing those same efficiencies. In general, the analysis of intra-industry effects according to type of insurance company confirms that the degree of the intra-industry signal resulting from an insurance company merger can be conditioned on the type of insurance company that is being acquired. Cross-Sectional Analysis of Intra-Industry Effects The dispersion of the rival portfolio valuation effects (based on two-day CARs) among merger announcements is displayed in Table 5. A review of this distribution suggests that the rival valuation effects in response to insurance mergers vary among merger announcements. This variation in rival valuation effects may be due to the firm-specific factors of individual rivals. Thus, the authors assess firm-specific variables that may cause the valuation effects among individual rival firms to vary. It is possible that some rivals may respond more favorably to a merger announcement if they exhibit characteristics that make them a more viable takeover candidate than other rivals. The authors hypothesize that individual rival valuation effects resulting from each merger announcement are conditioned by the rival firm-specific factors. TABLE 5 Distribution of Rival Firms Two-Day Cumulative Abnormal Returns Associated With the 88 Insurance Company Mergers
Intervals for CARS 2% < CAR 1% < CAR < 2% 0% < CAR < 1% 1% < CAR < 0% 2% < CAR < 1% CAR < 2% TOTAL Number of CARs 6 8 40 28 4 2 88

To test whether firm-specific factors help to identify rival insurers that are subject to more favorable signals, the authors follow the portfolio approach outlined in the methodology section. First, for each of the 88 merger events, the authors segment all rival insurers into two equally weighted portfolios by the firm-specific factor. The factors examined are size, location, recent price performance, and market share of individual rivals.8 Second, the authors perform event tests on the returns to the rival portfolios in each segment.9 The empirical results of these tests are reported in Table 6.
8

The variables representing size, price performance, and market share were constructed from the 1995 CRSP tape. The variable representing location was obtained by searching the 1995 Compustat (Industrial and Research) files. The results are qualitatively similar when event tests are performed on the returns to individual rival firms in each segment.

502

THE JOURNAL

OF

RISK

AND INSURANCE

TABLE 6 Explaining the Cumulative Abnormal Returns Among Individual Rival Firms
This table presents the valuation effects for individual rival firms grouped into portfolios by size, location, price performance, and market share. For each of the 88 merger announcements, the authors segment all rivals into two portfolios by the firm-specific factor. Event tests are performed on the returns to the rival portfolios in each segment. The size, price performance, and market share variables are obtained from the CRSP files. The location variable is obtained from the Compustat files. Panel A: Size Classification of Rival Insurance Company Size of Insurance Company Similar size Different size Event Period [1, 0] [1, 0] CAR (%) 1.59 0.16 z-statistic 4.80
1

% Pos 58 54

2.231

Panel B: Location of Rival Insurance Company Location of Insurance Company Similar location Different location Event Period [1, 0] [1, 0] CAR (%) 0.32 0.04 z-statistic 3.46 0.17
1

% Pos 63 50

Panel C: Prior Price Performance of Rival Insurance Company Prior Performance Positive performance Negative Performance Event Period [1, 0] [1, 0] CAR (%) 0.50 0.31 z-statistic 3.64
1

% Pos 63 61

3.401

Panel D: Market Share of Rival Insurance Company Market Share Large market share Small market share
1

Event Period [1, 0] [1, 0]

CAR (%) 0.33 0.45

z-statistic 2.51
1

% Pos 63 63

2.441

Significant at better than the 0.05 level

Size Classification To the extent that the market believes that the merger signals favorable prospects for rival insurance companies, those rival insurance companies with similar size characteristics may be most likely to benefit. The mean intra-industry effects are measured for those rival insurance companies that are close in size to the respective targets, and they are also measured for those rivals that are not close to the respective targets in size. A rival firm is classified as close in size to the target if the market capitalization falls within 25 percent of the targets market capitalization. Results of this comparison are disclosed in Panel A of Table 6. The mean intra-industry effects are positive and significant for both groups. However, the mean intraindustry effects are much higher for the group of rivals that are close in size to the

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS 503

targets. Furthermore, the t-statistic based on a test of a difference in means between the two groups is 2.29, which is statistically significant. This supports the authors hypothesis that the signal emitted from the insurance company merger announcement is more pronounced for those rivals that have a similar size as the target that is being acquired. Location Classification To the extent that the market believes that the merger signals favorable prospects for rivals, those rivals with similar location characteristics may be most likely to benefit. The entire U.S. geographic area is divided into four regionsnorthwest, northeast, southwest, and southeast. Target and rival firms are assigned to the regions on the basis of the state of incorporation. The mean intra-industry effects are measured for those rivals in the same location (region) of the respective targets and are also measured for those rivals in a different location than the respective targets. Results of this comparison are disclosed in Panel B of Table 6. The mean intra-industry effects are positive and significant for the group of rivals in the same region, but they are not significant for the group of rivals in a different region. Moreover, the tstatistic from a t-test comparing the means of the two groups is 2.02, which is statistically significant. This supports the authors hypothesis that the signal emitted from the insurance company merger announcement is more pronounced for those rivals that are located in a similar region as the target that is being acquired. Recent Price Performance Rival insurance companies that have experienced weak performance recently are expected to experience more favorable valuation effects in response to a merger announcement. To the extent that a merger prompts the market to assess how rival insurance companies may be valued if acquired, the relatively weak insurance companies have more potential for improved performance. The price performance of each rival insurance company is measured as the cumulative abnormal return in the 12-month period before the announced acquisition. Intra-industry effects are measured for those rivals with positive performance just before the merger announcement and for rivals with negative performance just before the announcement. A comparison of the intra-industry effects is disclosed in Panel C of Table 6. Both groups experienced positive and favorable valuation effects, and the t-statistic based on a t-test for a difference in means is not significant. Thus, the intra-industry effects among rivals are not conditioned on the pre-merger performance of rivals. Rival Market Share If the merger is expected to result in less competition in the insurance industry, those rivals with large market shares may be more likely to benefit and experience more favorable announcement-period valuation effects. To examine this possibility, the authors segment all rivals into two subsamples on the basis of their market shares. Rival market share is calculated as the ratio of the market capitalization of the rival firm to the total market capitalization of the insurance industry at the time of the announcement. The mean intra-industry effects are measured for those rivals with

504

THE JOURNAL

OF

RISK

AND INSURANCE

relatively large market shares, as well as for those with relatively small market shares. Results of the analysis are disclosed in Panel D of Table 6. The rival insurers with relatively large market shares experienced a significant mean two-day CAR of 0.33 percent, while rival insurers with relatively small market shares experienced a significant mean two-day CAR of 0.45 percent. However, the difference between the mean two-day CARs of the two subsamples is not significant. Therefore, mergers of insurance companies are not expected to result in less competition in the insurance industry. This result is consistent with previous studies, which found no evidence of collusive, anticompetitive effects in horizontal mergers (see Stillman, 1983; Eckbo, 1983; Eckbo and Wier, 1985; and Eckbo, 1992).

SUMMARY
The authors objectives are to determine the valuation effects of merger announcements on insurance company acquirers, targets, and rivals and to explain the variation in these valuation effects. The authors find that insurance company acquirers experience positive and significant valuation effects. This finding distinguishes insurance company acquirers from other acquirers and is attributed to the unique characteristics of insurance company operations. The merging of standardized insurance products may be viewed more favorably by the market than the merging of manufacturing facilities between industrial firms. The authors also found that targets of insurance company mergers experienced very favorable valuation effects. Thus, the distinct favorable effects on insurance company acquirers do not appear to be offset by a reduction in favorable effects on targets. The authors primary objective is to determine whether the merger between insurance companies signals information about the prospects of rival insurance companies. The authors find positive and significant intra-industry effects in response to the announcements of insurance company mergers, which supports the signaling hypothesis. Furthermore, the authors find that the magnitude of the intra-industry effects are conditioned on the type of insurance company target. The announcement that a non-life insurance company is being acquired signals more information about its respective rivals than the announcement that a multi-line insurance company is being acquired. The authors attribute this difference to the higher degree of ambiguity in a signal about multi-line companies, as the specific nature of promising prospects in those companies is more difficult to identify. The authors also find that the intra-industry effects of insurance company mergers are more pronounced for insurance companies that have a similar size and are located in the same region as the target insurance company. This provides some evidence that the market compares the similarity in operating characteristics when revaluing rival insurance companies in response to the acquisition of a target insurance company.

REFERENCES
BarNiv, R., and J. Hathorn, 1997, The Merger or Insolvency Alternative in the Insurance Industry, Journal of Risk and Insurance, 64(1): 89113. Cox, L. A., and G. L. Griepentrog, 1988, Systematic Risk, Unsystematic Risk, and Property-Liability Rate Regulation, Journal of Risk and Insurance, 55(4): 606627.

INTRA-INDUSTRY SIGNALS RESULTING FROM INSURANCE COMPANY MERGERS 505

Eakins, S., 1993, Institutional Investor Support of Managers: An Investigation of Tender Offers, Quarterly Journal of Business and Economics, 32(3): 7584. Eckbo, B. E., 1983, Horizontal Mergers, Collusion, and Stockholder Wealth, Journal of Financial Economics, 11(1): 241273. Eckbo, B. E., 1985, Mergers and the Market for the Concentration Doctrine: Evidence From the Capital Market, Journal of Business, 58(3): 325349. Eckbo, B. E., 1992, Mergers and the Value of Antitrust Deterrence, Journal of Finance, 47(3): 10051029. Eckbo, B. E., and P. Wier, 1985, Antimerger Policy Under the Hart-Scott-Rodino Act: A Reexamination of the Market Power Hypothesis, Journal of Law and Economics, 28(1): 119149. Fenn, G. W., and R. A. Cole, 1994, Announcements of Asset Quality Problems and Contagion Effects in the Life Insurance Industry, Journal of Financial Economics, 35(2): 181198. Grossman, S. J., and O. D. Hart, 1981, The Allocational Role of Takeover Bids in Situations of Asymmetric Information, Journal of Finance, 36(2): 253270. Halpern, P., 1983, Corporate Acquisitions: A Theory of Special Cases? A Review of Event Studies Applied to Acquisitions, Journal of Finance, 38(2): 297317. Harrington, S. E., 1983, The Relationship Between Risk and Return: Evidence for Life Insurance Stocks, Journal of Risk and Insurance, 50(4): 587610. Hertzel, M. G., 1991, The Effects of Stock Repurchases on Rival Firms, Journal of Finance, 46(2): 707716. Lang, L. H. P., and R. M. Stulz, 1992, Contagion and Competitive Intra-Industry Effects of Bankruptcy Announcements, Journal of Financial Economics, 32(1): 4560. Malatesta, P. H., and R. Thompson, 1985, Partially Anticipated Events: A Model of Stock Price Reactions With an Application to Corporate Acquisitions, Journal of Financial Economics, 14(2): 237250. Mandelker, G., 1974, Risk and Return: The Case of Merging Firms, Journal of Financial Economics, 1(4): 303335. Mitchell, M. L., and J. H. Mulherin, 1996, The Impact of Industry Shocks on Takeover and Restructuring Activity, Journal of Financial Economics, 41(2): 193229. Shaked, I., 1985, Measuring Prospective Probabilities of Insolvency: An Application to the Life Insurance Industry, Journal of Risk and Insurance, 52(1): 5980. Slovin, M. B., M. E. Sushka, and Y. M. Bendeck, 1991, The Intra-Industry Effects of Going-Private Transactions, Journal of Finance, 46(4): 15371549. Slovin, M. B., M. E. Sushka, and J. A. Polonchek, 1992, Informational Externalities of Seasoned Equity Issues, Journal of Financial Economics, 32(1): 87102. Szewczyk, S. H., 1992, The Intra-Industry Transfer of Information Inferred From Announcement of Corporate Security Offerings, Journal of Finance, 47(5): 19351946. Venezian, E. C., 1984, Are Insurers Under-Earning?, Journal of Risk and Insurance, 51(1): 150156. White, H., 1980, A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity, Econometrica, 48(4): 817838. Williams, C. A., 1983, Regulating Property and Liability Insurance Rates Through Excess Profits Statutes, Journal of Risk and Insurance, 50(3): 445472.