Accounting for acquisitions and firm value.

Schnusenberg, Oliver ; Sherman, W. Richard

ABSTRACT

The accounting of an acquisition is based on either the pooling or the purchase method. Pooling (or pooling of interests) treats the combined companies (acquirer and acquired) as though they had always operated as one company. The purpose of this study is to investigate the extent to which the accounting method used affects the value of the acquiring firm. One argument is that the accounting method used should not affect this value inasmuch as the accounting method does not directly affect cash flow. Our sample consists of 146 pooling and 46 purchase announcements from 1981 to 1995. Results indicate that valuation effects are more favorable for acquisitions using the purchase method in the eleven-day period surrounding the announcement and for at least six months following the announcements. These results stand even after conducting a cross-sectional analysis that controls for the firms' price/earnings ratio, size, market parameter estimates, earnings surprises, and leverage. These results suggest that market participants value the added flexibility and indirect tax benefits that are provided by the purchase method of accounting as opposed to the higher reported future earnings associated with the pooling method.

INTRODUCTION

The accounting of an acquisition is based on either the pooling or the purchase method. Pooling (or pooling of interests) treats the combined companies (acquirer and acquired) as though they had always operated as one company. Consequently, the financial statements of the new company merely reflect the consolidation of statements of the two previously separate entities. In contrast, purchase accounting revalues the assets and liabilities of the acquired company at their current fair market values with the possible difference between the acquisition price and the market value of the acquired company's net identifiable asset (i.e. goodwill) being amortized over a period not to exceed 40 years. This amortization creates an expense that reduces reported earnings of the acquiring firm.

The purpose of this study is to investigate the extent to which the accounting method used affects the value of the acquiring firm. One argument is that the accounting method used should not affect this value inasmuch as the accounting method does not directly affect cash flow. However, there are other arguments relating to valuation effects from future cash flows and/or indirect cash flows that provide a rationale for how and why the choice of accounting method can impact value.

ADVANTAGES OF POOLING

Research by Ball & Brown (1968), Gonedes (1975), Hoskins, Hughes & Ricks (1986), and others has shown that reported earnings can partially drive stock prices. To the extent that the accounting method affects future earnings, the valuation effects may be more favorable for acquirers using the pooling method.

Earnings for pooling firms are generally higher for a number of reasons. The first is due to the way in which the acquired firm's earnings are folded into the new entity's reported earnings. Under pooling, the net earnings for the entire year of acquisition are carried into the merged firm's income statement; under purchase accounting, only the income earned by the acquired firm after the acquisition date are reported by the acquiring firm. Depending on when during the year the acquisition takes place, this difference may be more (late in the year) or less (early in the year) significant in the reported earnings for the first year.

Pooling would also result in higher earnings reporting for reason related to the treatment of the acquired firm's assets and liabilities after acquisition. The tax aspects of mergers and acquisitions are extraordinarily complex but can be roughly divided between tax-free reorganizations and taxable acquisitions. In general, tax-free reorganizations under IRC Section 368 will be accounted for under the "pooling of interest" method. Taxable acquisitions, with re-valuation of assets to their fair market value (an election under IRC Section 338 is available for stock acquisitions), are usually reported using the "purchase" method of accounting. The discussion of the reporting aspects of pooling vs. purchase accounting assumes a parallel to the tax consequences of tax-free reorganizations vs. taxable acquisitions. It further assumes that market participants implicitly understand the relevant tax consequences related to the method of accounting disclosed in the acquisition announcement.

Under pooling, the valuation of these assets and liabilities remains unchanged from how they appeared on the pre-acquisition balance sheet of the acquired firm. Under purchase accounting, the assets and liabilities of the acquired firm are restated at their current market values. Because there is no re-valuation (i.e., "write-up") of the acquired firm's assets under pooling, depreciation expenses after the acquisition are generally lower (with the resultant reported earnings being generally higher) than the depreciation that would taken for the same acquisition under purchase accounting.

Related to this non-revaluation of assets is the fact that there is no possibility for the recognition of "goodwill" (i.e. the difference between the purchase price and the market value of the acquired firm's net identifiable assets) from an acquisition under pooling. Had a firm recognized goodwill upon acquisition (as would have been the case under purchase accounting), it would be required to amortize (i.e., expense) this intangible asset, which, in turn, would negatively impact reported earnings. In fact, under the Exposure Draft on a proposed Statement, Business Combinations and Intangible Assets, issued by the FASB in September, 1999, not only would the purchase method be required for all business combinations but any goodwill that is recognized as a result of the acquisition would not be subject to amortization. Instead, goodwill would be reviewed for impairment (i.e., its fair value is less than its carrying amount) on a regular basis (FASB, 1999).

As a final consequence of non-revaluation of assets, pooling firms will generally report higher gains (or lower losses) upon the disposal of the assets of the acquired firm due to the lower basis of these assets. In contrast, because of the revaluation to current market values at the time of acquisition, the acquired firm's assets lose whatever pre-acquisition gains which are inherent in them.

As a result of these accounting differences, a company using pooling would be expected to report higher earnings after the acquisition than a company using the purchase method. (Also see Herz & Abahoonie (1990) for a more detailed discussion of earnings differentials between the purchase and pooling methods.)

If market participants do indeed use reported earnings in assessing a firm's value, one would expect abnormal returns associated with the pooling method to be larger than those associated with the purchase method of accounting for acquisitions. This leads to the following null hypothesis:

[H.sub.O]: Abnormal returns surrounding the announcement of a pooling
acquisition are larger than those associated with a purchase
acquisition.

ADVANTAGES OF THE PURCHASE METHOD

While its use will generally result in lower reported earnings, the purchase method of accounting does offer some advantages over pooling. As shown in Appendix A, firms must meet twelve criteria in order to be allowed to report an acquisition under the pooling method. (APB Opinion 16) A firm may, therefore, be restricted from restructuring in order to meet these criteria. For example, because a company utilizing the pooling method must agree not to enter into other financial arrangements for the benefit of the former stockholders of a combining company, an exchange of equity securities would be prohibited. Furthermore, the company also must agree not to dispose of a significant part of the assets of the combining companies. To the extent that the twelve criteria (none of which apply to the purchase method of accounting) can be perceived as restricting a firm's future financing or operating flexibility, market participants may penalize firms using the pooling method. Therefore, acquisitions reported under the purchase method may result in more favorable valuation effects than those reported under pooling.

The choice of accounting for an acquisition may also affect future indirect cash flows of a firm. As was discussed in the previous section, purchase accounting requires the revaluation of the acquired firm's assets to their market values as of the date of acquisition. This results in higher expenses being reported for the depreciation of the assets and for the amortization of the goodwill that may result from the acquisition. However, as Bittker & Eustice (1994) note, "as is often the case, an acquirer's desires to increase book income generally are at odds with its desires to reduce tax income". Inasmuch as both depreciation (Internal Revenue Code Sections 167 & 168) and amortization of goodwill (Internal Revenue Code Section 197) are tax deductible, the purchase method may be preferable over pooling because it reduces future cash tax outflows resulting from relatively high future earnings.

Another tax-related indirect cash flow relates to the method of payment. Due to the fact that pooling requires stock-for-stock acquisitions, no recognition of gain or loss is possible for tax purposes. No such restriction applies to purchase accounting.

These potential advantages of the purchase method form the basis for the alternative hypothesis (relative to HO) that acquisitions using the purchase method of accounting should result in more favorable valuation effects.

[H.sub.A]: Abnormal returns surrounding the announcement of a purchase
acquisition are larger than those associated with a purchase
acquisition.

LITERATURE REVIEW

Studies investigating the stock returns associated with purchase and pooling acquisition announcements are relatively few. Two studies measured the abnormal stock price movements over a long window surrounding acquisition announcements and compared the movements of the sub-sample that was identified as using purchase accounting to the sub-sample using pooling. Hong, Kaplan & Mandelker (1978) examined 138 pooling and 62 purchase method mergers from 1954-1964. The accounting method associated with these mergers was identified using proxy statements issued in connection with the mergers. The authors found no reaction associated with pooling announcements but did find significant cumulative abnormal returns during the twelve months preceding the announcement of a purchase and that these abnormal returns were maintained for eight subsequent months. The authors conclude that the pooling of interests method does not lead to abnormal stock price behavior for acquiring firms but offer no explanation for the large abnormal returns associated with the purchase method of accounting for acquisitions.

A related study by Davis (1990) examined 108 pooling and 69 purchase acquisitions over the period of 1971-1982. The accounting method associated with these acquisitions was identified in Mergers & Acquisitions. Davis' findings were similar to those reported by Hong et al. Purchase acquisitions exhibited positive and statistically significant abnormal returns while pooling acquisitions exhibited largely positive but statistically insignificant returns over a period from 26 weeks before the announcement to 26 weeks after the effective date of the acquisition.

Other empirical studies have addressed the reasons for choosing one method over the other. Nathan & Dunne (1991) studied 30 purchase and 291 pooling acquisitions that occurred between 1963 and 1985. Using the firms' proxy statements to identify the accounting method used, they found that the purchase/pooling choice was influenced by goodwill, acquirer leverage, and the issuance of APB Opinion No. 16.

Using the Wall Street Journal to identify the type of accounting used, Robinson & Shane (1990) investigated 59 pooling and 36 purchase acquisitions taking place 1972-1982. They found that bid premiums are generally larger for firms using the pooling method. The authors conclude that since the costs associated with structuring an acquisition to qualify for pooling are greater than those for the purchase method, acquirers will only structure an acquisition as a pooling of interests if the perceived benefits are greater than those that would be achieved under purchase accounting.

Finally, Haw, Jung, & Ruland (1994) found that analysts' forecast accuracy decreased greatly after mergers in general but even more so for firms who used the purchase method due to the fact that purchase accounting interrupts the past time series of earnings.

SAMPLE SELECTION

Both Hong et al. (1978) and Davis (1990) investigated the extent to which abnormal returns were associated with the "announcement" of the method by which the acquisitions would be accounted - pooling or purchase. Yet both studies identified this accounting method by sources not directly associated with the announcement date (i.e. by proxy statement or by information published in Mergers & Acquisitions). While these sources identify the accounting method accurately, information of the method used is not necessarily available to market participants at the time that the merger is announced. Thus, these two studies implicitly assume that the market participants could, at the time of the merger announcement, properly guess whether pooling or purchase accounting would be used. However, Hong, et al., and Davis were investigating mergers that occurred in 1954-1964 and 1971-1982, respectively. The method of accounting was rarely disclosed at the time of that the merger was announced until the 1990s.

To avoid this flaw in the previous studies and ensure that the accounting method was known by market participants, the sample of this study was created by identifying all acquisitions in which the method of accounting was stated at the time of the acquisition announcement. The Wall Street Journal Index and the Lexis/Nexis database were used to search for acquisitions during 1981-1985 for which there were corresponding announcements of the method being used to account for the merger. (Huang & Walking (1987) investigated acquisition attempts that were resisted and not resisted, using a sample based on announcements in the Wall Street Journal. As we do in our study, they argue that sample selection bias is avoided because only information known at the time of the acquisition announcement is used.) This resulted in an initial sample of 229 acquisitions (173 pooling and 56 purchase). This sample was screened by removing any companies for which the stock price data were not consistently reported over a period of 240 days before the acquisition announcement and 5 days after the announcement. Thirty-seven companies were removed for this reason, leaving a total of 192 acquisitions (146 pooling; 46 purchase) that could be assessed.

The sample is segmented by year in Table 1. While 76% of the acquisition reflected the use of pooling, purchase accounting was announced more frequently in the early 1980s and was more evenly distributed throughout the period studies than is the case for pooling. Furthermore, as was noted previously, the choice of accounting method in the acquisition announcement was not generally disclosed until the 1990s. 92% of all the relevant announcements (i.e., those which disclosed the method that was being used to account for the merger) were in the 1990-1995 period, with most of these announcements occurring in 1994 and 1995.

As shown by the descriptive statistics provided in Table 2, firms undertaking pooling acquisitions are, on average, larger than those firms using the purchase method of accounting. Pooling firms also experienced a higher growth rate, a higher return on assets (ROA), and also paid higher taxes as a percentage of before-tax income than those using purchase accounting. However, those firms using the purchase method showed higher returns on equity (ROE) than those using pooling.

METHODOLOGY

Valuations effects are measured for all firms contained in the sample. The abnormal return of each acquirer is estimated with prediction errors using the procedure of Mikkelson & Patch (1988). The alpha and beta of each acquirer are derived over an estimation period of from t = -240 to t = -20 by applying the market model to returns of the University of Chicago's Center for Research in Security Prices (CRSP). CRSP calculates the raw return as:

[R.sub.it] = ([P.sub.t] + [D.sub.t])/[P.sub.t-1] - 1,

where [R.sub.it] = the rate of return on security i for event day t,

[P.sub.t] = last sale price or closing bid/ask average on day t

[D.sub.t] = cash adjustment for day t, and

[P.sub.t-1] = last sale price or closing bid/ask average at time of last available price less than t, and

The intercept and beta resulting from application of the market model are then used long with the actual market movement over an examination period to derive an expected return. The prediction error (PE) for each acquirer is measured as:

[PE.sub.it] = [R.sub.it] - ([a.sub.i] - [b.sub.i][R.sub.mt]), (1)

where [R.sub.mt] = the rate of return on the S&P 500 index on event day t, and [a.sub.i], [b.sub.i] = ordinary-least-squares estimates of the intercept and slope of the market model regression from the estimation period.

The mean of prediction errors for all acquirers that announced the use of pooling is derived for each day within the examination period. The same process is used to derive the mean prediction errors for acquirers that announced the use of the purchase method. The primary focus of the examination window is on the two-day window (t-1, t), in which day t serves as the announcement date. Because wire services may have reported the news of the acquisition before stock trading closed on the day prior to the announcement, day t-1 is also assessed in order to determine if the market response (if any) could have occurred on this day.

Test for significance are based on Mikkelson and Partch's (1988) Z-statistic, which for each day is calculated as:

[Z.sub.t] = [summation][SPE.sub.it]/[N.sup.1/2] (2)

where [SPE.sub.it] = [AR.sub.it]/[S.sub.it], and [S.sub.it] is calculated according to Mikkelson and Partch (1988).

Two day-interval prediction errors are tested for significance using the following Z-statistic:

[Z.sub.t] = 1/[N.sup.1/2] [summation] [ASCAR.sub.i], (3)

where [ASCAR.sub.i] is the standardized cumulative abnormal return defined by Mikkelson and Partch (1988). The denominator required for this calculation is the variance of the cumulative prediction error of firm i.

RESULTS

Stock Price Reaction in the Announcement Period

Results for various intervals are presented in Table 3. As mentioned above, the two-day (t = -1, 0) prediction errors are of primary interest. The mean two-day PE for acquisitions in which pooling was announced is -0.49% (Z = -7.54). The mean PE for purchase acquisitions is 0.31% (Z = 1.27). A Z-test was applied to test for a significant difference in the two sub-samples for the cumulative abnormal returns over the [-1, 0] interval. The difference of 0.80% is significant with a Z-statistic of 3.39 (p = 0.00069). The results support HA and suggest that the market responds more favorably to acquisition announcements involving the purchase method of accounting. The results are attributed to the greater restructuring flexibility and the indirect tax benefits afforded by purchase accounting.

Long-Term Effects

Valuation effects are also measured over an extended period in order to estimate long-term performance following acquisitions. Unfortunately, in order to retain a relatively large sample, the initial sample was screened to retain only those firms with returns data available for up to five days following the acquisition announcement. Consequently, investigation of long-term valuation effects results in additional sample attrition due to missing returns data. As shown in Table 3, 108 pooling firms and 37 purchase firms had available data for intervals [0, 180]. This number drops to 37 pooling and 27 purchase firms with available data for intervals [0, 360].

The results summarized in Table 3 show that regardless of whether pooling or purchase accounting is used, acquirers suffer large negative returns in the year following the acquisitions. Pooling firms experience significant negative abnormal returns of 35.81% over the 360-day period after the merger (t = -1.82; p < 0.0005); firms announcing the use of purchase accounting experience significant abnormal returns of 17.26% over the same time period (t = -1.82; p = 0.0808). The difference (18.55%) in these negative returns is marginally significant (t = 1.62; p = 0.1114). Thus, even when assessing extended periods, acquisitions using the purchase method are associated with relatively more favorable valuation effects than those acquirers using pooling.

Hong et al. (1978) found that purchase firms maintained their positive cumulative abnormal returns for a period of eight months following the announcement. While the negative abnormal returns of -2.42% experienced by purchase firms in our study are not significant (t = -0.40; p = 0.6894), the returns of pooling firms (-15.97%) are highly significant (t m= -5.09; p < 0.00005), with the 13.55% difference in returns between the two sub-samples being significant at the 5% level (t = 2.12; p = 0.0361).

Cross-Sectional Analysis

Since the two sub-samples may have distinct characteristics, a complimentary cross-sectional analysis is conducted to control for these while testing whether valuation effects are related to the type of accounting method announced. Previous research has identified the following variables for which the accounting method may be proxying: Price: Earnings (P/E) ratio, size, market model parameter estimates, earnings surprise, and leverage. (See, among others, Basu (1983); Banz (1981); Beaver, Clark & Wright (1979); Bernard (1979); Elgers & Clark (1979); Hagerman & Shah (1984); and Reinganum (1981).) Each of these variables is discussed below.

Price/Earnings Ratio

In an efficient capital market, all available information is rapidly reflected in security prices. However, previous studies have found that the efficient market hypothesis may be violated in certain circumstances. For example, Basu (1977; 1983) found that low P/E ratios may be an indicator of future stock price performance. Low P/E ratios may lead to exaggerated investor optimism regarding future growth in earnings and dividends, thereby leading to higher future returns for low P/E stocks. Because the two sub-samples in our study may exhibit different P/E ratios, this ratio was included as a control variable.

Size

In addition to proxying for a firm's P/E ratio, the abnormal negative returns found in our study may be attributable to firm size rather than to the accounting method used. Prior research indicates that there may be a 'size effect'--i.e., small firms' stocks experience, on average, higher risk-adjusted returns than large firms' stocks. (For a more detailed discussion, see Banz (1981) and Reinganum (1981).) This finding may be due to the higher risk associated with small firms, which, in turn, may be a result of the limited information available about small firms. A size variable was included in our cross-sectional analysis to control for a potential 'size effect'.

Size may also be proxying for an income-reducing policy. For example, Watts & Zimmerman (1978) believe that larger firms tend to lobby for accounting standards that reduce reported income. Consequently, the size variable included here may be proxying for an income deflating policy rather than for the accounting method itself. It follows that since the purchase firms in our sub-sample are, on average, smaller than those in the pooling sub-sample, the pooling firms would be expected to pursue an income reducing policy to a greater extent than the purchase firms. Therefore, choosing the pooling method of acquisition accounting may have negative valuation consequences for these firms inasmuch as pooling will result in higher future earnings relative to purchase accounting.

Market Model Parameter Estimates

Elgers & Clark (1980) found important differences in risk effects across merger types. More specifically, conglomerate (as opposed to smaller) mergers are often undertaken due to a risk motive. This difference is somewhat puzzling in light of the capital asset pricing model (CAPM). Unless investors are constrained from achieving the same risk diversification effects by revising their own portfolios, there should be no economic reward from mergers undertaken to create risk shifts. In order to control for risk differences as well as for a disproportionate ratio of conglomerates in one of our samples, the beta of the estimation period is included in the cross-sectional analysis.

Furthermore, Blume (1971; 1975; 1979) suggests that market model parameter estimates may be non-stationary over time, viz. that the market model parameter are mean-reverting over time. In other words, any abnormal returns calculated using the market model parameter estimates of an estimation period could be attributable to the parameter estimates themselves if the parameters are non-constant. (For example, non-constant parameters could occur if the target's risk level is different from the acquirer's risk level inasmuch s the acquirer's operations include the target's operations following the acquisition.) To control for this possibility, our estimated alpha parameter is also included in the cross-sectional analysis.

Earnings Surprise

As noted previously, accounting research indicates that positive earnings forecast errors are associated with positive stock returns and, conversely, for negative earnings forecast errors. To the extent that the announced method of acquisition accounting (or the acquisition itself) was anticipated, future earnings may substantially differ from forecasted earnings. Therefore, in order to test whether the abnormal returns associated with the announcement of purchase and pooling methods as opposed to other variables for which the accounting method may be proxying, it is also necessary to control for earnings surprises in a cross-sectional analysis. It is important to note that this variable directly accomplishes the purpose of testing the null hypothesis ([H.sub.0]) as the relatively low future earnings of firms using the purchase method are the primary reason for expecting higher abnormal returns for firms using the pooling method. Thus, if this variable is insignificant in explaining the abnormal returns, H0 is rejected.

Leverage

Davis (1990) found that purchase firms have significantly higher leverage than pooling firms. This finding may be explained by Crawford (1986), who finds that firms using the purchase method are more likely than pooling firms to have debt covenants based on assets or intangible assets rather than on earnings. In this context, the higher abnormal returns experienced by purchase firms may be attributable to the higher leverage of these firms rather than to the accounting method itself. To control for this, leverage is included as a control variable in this cross-sectional analysis.

Accounting Method

To directly test whether the abnormal returns are significantly related to the accounting method employed as opposed to any of the control variables described above, the cross-sectional analysis employs an indicator variable to identify the accounting method used. If the coefficient of this variable is significant after controlling for the other variables, then it is likely that the higher abnormal returns of purchase method firms are due to the higher organizational flexibility and indirect tax cash flows which the purchase method offers.

Cross-Sectional Model

The resulting cross-sectional model was applied to both the purchase and pooling sub-samples and is stated below:

[ASCAR.sub.i] = [a.sub.0] + [a.sub.1] [ALPHA.sub.i] + [a.sub.2] [BETA.sub.i] + [a.sub.3] [EARSUR.sub.i] + [a.sub.4] [LEV.sub.i] + [a.sub.5] [PE.sub.i] + [a.sub.6] [SIZE.sub.i] + [a.sub.7] [POOL.sub.i] + [e.sub.i], (4)

where [ASCAR.sub.i] = the cumulative standardized [-1,0] prediction error for firm i,

[ALPHA.sub.i], [BETA.sub.i] = the market model parameter estimates for firm i,

[EARSUR.sub.i] = earnings surprise for firm i, defined as the change in EPS in the announcement year divided by share price at the beginning of the announcement year; thus, the surprise represents the abnormal earnings based on a naive random walk earnings forecast model,

[LEV.sub.i] = leverage of firm i, defined as the ratio of total liabilities to total assets,

[PE.sub.i] = the price/earnings ratio of firm i, defined at the end of the year prior to the announcement year,

[SIZE.sub.i] = firm i's total assets,

[POOL.sub.i] = a dummy variable equal to unity if firm i uses the pooling method and zero otherwise, and

[e.sub.i] = error term.

Correlations of Variables Used in the Cross-Sectional Model

The correlation coefficients for all pairs of independent variables are disclosed in Table 4. The ALPHA, BETA, LEV, and SIZE exhibit some degree of correlation. Consequently, the significance tests for these variables may be biased. However, the maximum variance inflation factor (VIF) of the full model is only 1.4, which, according to Neter, Kutner, Nachtsheim & Wasserman (1996) does not present a serious multi-collinearity problem. Variance inflation factors for the individual variables measure how much the variances of the estimated regression coefficients are inflated as compared to when the predictor variables are not linearly related. However, in order to account for potential multi-collinearity effects, our original model is supplemented with three reduced models.

Cross-Sectional Results

Results from the cross-sectional analysis are summarized in Table 5. The first row presents the results of the full model. Model 2 excluded the PE variable due to its extreme insignificance (p = 0.9693). Models 3 and 4 additionally exclude variables BETA and SIZE to counteract any potential multi-collinearity problems. Once a variable was omitted, it did not re-enter the cross-sectional model.

The POOL variable is of primary importance in this analysis. If this variable exhibits a significant coefficient, the accounting method itself explains a portion of the distribution of abnormal returns associated with the acquisition announcement. As shown in Table 5, the POOL variable is significant in all models examined, with a maximum p-value of 0.0314. Note that the size of the coefficient of this variable is very close to the difference over the [-1, 0] interval found in stock price reaction summarized in Table 3. (While the Z-statistic used in our original analysis was computed using standardized values (in accordance with Mikkelson & Partch (1988)), the 0.08% difference shown in Table 3 was computed using the raw cumulative abnormal returns. Consequently, it is possible for the coefficient in the cross-sectional model to be larger than this difference because the standardized cumulative abnormal returns are used as the dependent variable in this cross-sectional analysis.) These findings lend support to the results in the previous section. Apparently, in addition to any indirect tax benefits that may accrue under purchase accounting, market participants highly value that purchase method firms are not constrained by the same requirements set out in APB Opinion No. 16 for pools.

Three other variables--the market model parameter (ALPHA), the firm's leverage (LEV), and firm size (SIZE)--are significantly related to the [-1, 0] standardized cumulative abnormal returns. The significance of ALPHA suggests that the market model parameters may be non-constant over time as discussed by Blume (1979). The significance of leverage is similar to the findings by Davis (1990). In other words, purchase firms tend to exhibit higher leverage than pooling firms, lending further support to Crawford's (1986) notion that firms using the purchase method may have debt covenants based on net assets or intangibles rather than on income. (The coefficient of this variable is highly positive. When this variable is examined separately for the two sub-samples, it is significant for the purchase group but not for the pooling firms. The difference (see Model 4) is highly significant (t = 2.39; p = 0.0183).) Consequently, even though their future income is expected to be relatively low relative to that reported by pooling firms, purchase firms are more likely to employ debt in their capital structure.

The last significant variable in the cross-sectional analysis is firm size as measured by total assets. There are two possible explanations for the negative coefficient of this variable. First, larger firms seem to experience lower abnormal returns than smaller firms, lending support to the size effect (i.e., smaller firms exhibit higher returns than larger firms regardless of the time period examined) documented in the previous literature. Consequently, the large returns for relatively small acquirers may be driven by the size anomaly.

On the other hand, since pooling firms are larger than purchase firms, previous literature suggests that they would pursue an income reducing policy. However, as we have seen, the pooling method will generally result in the reporting of higher earnings than would be the case under purchase accounting causing market participants to interpret the use of pooling as suboptimal. Given the contradictory results, it is reassuring that the difference in size between the two sub-samples is not significant (t = 0.61; p = 0.5446). (Model 3 was used to test for this difference because the SIZE variable is omitted in Model 4.) Thus, it appears that the significance of the SIZE variable is attributable to a general size effect rather than to pooling firms not pursuing an income reducing policy.

SUMMARY AND CONCLUSION

The purchase method of accounting can reduce future reported earnings as a consequence of increased expenses for amortization of goodwill and depreciation of tangible assets. However, as a trade-off, this method of accounting for acquisitions allows more restructuring flexibility than does the pooling method and can also provide greater indirect tax benefits. Given these conflicting advantages of the two methods, the objective of this study is to determine whether and how the valuation effects of acquisition announcements are conditioned on the type of accounting method employed. The sample used is pure inasmuch as it focuses solely on acquisitions in which the method of accounting was disclosed at the same time as the acquisition announcement, eliminating the need to assume that market participants have other information except what was contained in the acquisition announcement itself.

Valuation effects are found to be more favorable for acquisitions using the purchase method in the eleven-day period surrounding the announcement and for at least six months following the announcements. These results stand even after conducting a cross-sectional analysis that controls for the firms' price/earnings ratio, size, market parameter estimates, earnings surprises, and leverage. These results suggest that market participants value the added flexibility and indirect tax benefits that are provided by the purchase method of accounting as opposed to the higher reported future earnings associated with the pooling method.

APPENDIX A: TWELVE CONDITIONS FOR USE OF POOLING (AS PER APB OPINION NO. 16)

1) Each of the combining companies is autonomous and has not been a subsidiary or division of another corporation within two years before the combination is initiated.

2) Each of the combining companies is independent of other combining companies. While joint ventures are permissible, independence is generally interpreted as 10% or less ownership of voting stock at any time between the initiation and consummation date of the combination.

3) The combination is effected in a single transaction or is completed in accordance with a specific plan within one year after the plan is initiated.

4) The acquiring company offers and issues only common stock with rights identical to those of the majority of its outstanding common stock in exchange for substantially all of the voting stock of the acquired company. "Substantially all" means at least 90% of the shares outstanding and is measured as of the date of consummation.

5) Within the period beginning two years before the plan's initiation and ending at the date of consummation, none of the combining companies change the equity interest of voting common stock in contemplation of effecting the combination.

6) Each of the combining companies reacquires voting common stock only for purposes other than business combinations and no company reacquires more than a normal number of shares between the plan's initiation and combination.

7) The proportionate share of ownership of each shareholder remains the same after the combination as it was before the combination.

8) The voting rights to which the common stock ownership interests in the resulting corporation are not restricted.

9) The combination is resolved at the date of the plan's consummation and no provisions of the plan relating to the issuance of securities or other compensation are pending.

10) The combined company does not agree to retire or reacquire, directly or indirectly, all or part of the common stock issued to effect the combination.

11) The combined company does not enter into other financial arrangements for the benefits of former stockholders of the combining company (e.g. guaranty of loan secured by the stock of the combined company) which in effect negates the exchange of equity securities.

12) The combined company does not intend or plan to dispose of a significant part of its assets within two years after the combination. Disposals in the ordinary course of business and to eliminate duplicate facilities or excess capacity are permissible.

REFERENCES

APB Opinion No. 16, Business Combinations.

Ball, R. & P. Brown (1968). An Empirical Evaluation of Accounting Income Numbers. Journal of Accounting Research, Autumn, 159-178.

Banz, R. W. (1981). The Relationship Between Return and Market Value of Common Stocks. Journal of Financial Economics, March, 3-18.

Basu, S. (1977). Investment Performance of Common Stocks in Relation to their Price-Earnings Ratios: A Test for Market Efficiency. Journal of Finance, June, 663-682.

Basu, S. (1983). The Relationship Between Earnings Yield, Market Value and Return for NYSE Common Stocks. Journal of Financial Economics, June, 129-156.

Beaver, W. H., R. Clarke & W. F. Wright (1979). The Association Between Unsystematic Security Returns and the Magnitude of Earnings Forecast Errors. Journal of Accounting Research, Autumn, 316-321.

Bernard, V. L. (1986). Unanticipated Inflation and the Value of the Firm. Journal of Financial Economics, March, 285321.

Bittker, B. I. & J. S. Eustice (1994). Federal Taxation of Corporations and Shareholders, 6th Edition. Boston: Federal Tax Press.

Blume, M. E. (1971). On the Assessment of Risk. Journal of Finance, March, 1-11.

Blume, M. E. (1975). Betas and Their Regression Tendencies. Journal of Finance, June, 785-795.

Blume, M. E. (1979). Betas and Their Regression Tendencies: Some Further Evidence. Journal of Finance, March, 265267.

Davis, M. L. (1990). Differential Market Reaction to Pooling and Purchase Methods. The Accounting Review, July, 696709.

Davis, M. L. (1991). APB 16: Time to Reconsider. Journal of Accountancy, October, 99-107.

Elgers, P. T. & J. J. Clark (1980). Merger Types and Shareholder Returns: Additional Evidence. Financial Management, Summer, 66-72.

FASB (1999). Exposure Draft of Statement of Financial Accounting Standard: Business Combinations. Stamford, CT: FASB.

Gonedes, N. J. (1975). Risk, Information and the Effects of Special Accounting Items on Capital Market Equilibrium. Journal of Accounting Research, Autumn, 220-256.

Hagerman, R. L. & P. Shah (1984). The Association Between the Magnitude of Quarterly Earnings Forecast Errors and Risk-Adjusted Stock Returns. Journal of Accounting Research, Autumn, 526-540.

Hagerman, R. L. & M. E. Zmijewski (1979). Some Economic Determinants of Accounting Policy Choice. Journal of Accounting and Economics, August, 141-161.

Haw, I. M., K. Jung & W. Ruland (1994). The Accuracy of Financial Analysts' Forecasts After Mergers. Journal of Accounting, Auditing, and Finance, Summer, 465-486.

Herring, C. B. & F. Norris (1990), Merger Mania. The National Public Accountant, June, 38-42.

Herz, R. H. & E. J. Abahoonie (1990). Innovations to Minimize Acquisition Goodwill. Mergers & Acquisitions, March/April, 35-40.

Hong, H, R. S. Kaplan & G. Mandelker (1978). Pooling vs. Purchase: The Effects of Accounting for Mergers on Stock Prices. The Accounting Review, January, 31-47.

Hoskin, R. E., J. Hughes & W. Ricks (1986). Evidence on the Incremental Information Content of Additional Firm Disclosures Made Concurrently With Earnings. Journal of Accounting Research, Supplement, 1-32.

Huang, Y. & R. A. Walking (1987). Target Abnormal Returns Associated With Acquisition Announcements: Payment, Acquisition Form, and Managerial Resistance. Journal of Financial Economics, December, 329-349.

Mikkelson, W. H. & M. M. Partch (1986). Valuation Effects of Security Offerings and the Issuance Process. Journal of Financial Economics, January/February, 31-60.

Mikkelson, W. H. & M. M. Partch (1988). Withdrawn Security Offerings. Journal of Financial and Quantitative Analysis, June, 119-133.

Nathan, K. & K. M. Dunne (1991). The Purchase-Pooling Choice: Some Explanatory Variables. Journal of Accounting and Public Policy, Winter, 309-323.

Neter, J., M. H. Kutner, C. J. Nachtsheim & W. Wasserman (1996). Applied Linear Statistical Models, 4th Edition. Chicago: Richard D. Irwin, Inc.

Reinganum, M. R. (1981). Misspecification of Capital Asset Pricing: Empirical Anomalies Based on Earnings' Yields and Market Values. Journal of Financial Economics, March, 19-46.

Robinson, J. R. & P. B. Shane (1990). Acquisition Accounting Method and Bid Premia for Target Firms. The Accounting Review, January, 25-48.

Watts, R. L. & J. L. Zimmerman (1978). Towards a Positive Theory of the Determination of Accounting Standards. The Accounting Review, January, 112-134.

Zmijewski, M. E. & R. L. Hagerman (1981). An Income Strategy Approach to the Positive Theory of Accounting Standard Setting/Choice. Journal of Accounting and Economics, August, 129-149.

Oliver Schnusenberg, St. Joseph's University

W. Richard Sherman, St. Joseph's University

Table 1: Distribution of Pooling and
Purchase Acquisitions Over Time

 Acquisitions Acquisitions
 Involving the Involving the Total Acquisitions
 Pooling Method Purchase Method

1981 1 0 1
1982 0 0 0
1983 0 1 1
1984 0 2 2
1985 2 1 3
1986 2 3 5
1987 0 3 3
1988 0 2 2
1989 6 2 8
1990 7 0 7
1991 12 3 15
1992 25 2 27
1993 17 6 23
1994 32 11 43
1995 42 10 52
Total 146 46 192

Table 2: Summary Information for Pooling and Purchase Acquisitions

 Pooling Purchase Total
 Sample Sample Sample

Mean Asset Size of Acquirer ($million) 8,758 6,053 8,053
Mean Taxes Paid as A % of EBT in Year
 Before Acquisition 34.86% 32.58% 34.26%
Three-Year Growth Rate of Acquirer
 (Sales) 20.15% 19.18% 19.90%
Mean ROA of Acquirer 3.18% 4.24% 3.45%
Mean ROE of Acquirer 10.40% 10.12% 10.33%

Table 3: Summary of CARs Over Various Intervals for Pooling vs.
Purchase Methods of Accounting for an Acquisition (Test-Statistic
in Parentheses)

Interval CAR of CAR of Test for
 Acquisitions Acquisitions Difference
 Using the Pooling Using the
 Method (a) Purchase
 Method (b)

[-5,+5] -1.57% 3.31% 4.88%
 (-3.60) *** (2.34) ** (1.15)
[-3,0] -0.62% 2.24% 2.86%
 (-4.19) *** (2.82) *** (2.26) **
[-1,0] -0.49% 0.31% 0.80%
 (-7.54) *** (1.27) (3.39) ***
[0,180] -15.97% -2.42% 13.55%
 (-5.09) *** (-0.40) (2.12) **
[0,360] -35.81% -17.26% 18.55%
 (-5.60) *** (-1.82) * (1.62) *

* Significant at the 10% level

** Significant at the 5% level

*** Significant at the 1% level

(a) The number of pooling firms utilized over the five
intervals are 146, 146, 146, 108, and 87, respectively.

(b) The number of purchase firms utilized over the five
intervals are 46, 46, 46, 37, and 27, respectively.

Table 4: Correlation Coefficients of Independent Variables (a)

 ALPHA BETA EARSUR POOL LEV PE

BETA 0.4448
EARSUR 0.1471 -0.0201
POOL 0.1527 0.1839 -0.0889
LEV -0.2034 -0.3084 0.1733 0.0544
PE 0.0539 -0.0506 -0.0573 0.1409 0.02
SIZE -0.1863 0.017 0.0178 0.0772 0.3133 0.0032

(a) [ALPHA.sub.i], [BETA.sub.i] = the market model parameter
estimates for firm i,

[EARSUR.sub.i] = earnings surprise for firm i, defined as the
change in EPS in the announcement year divided by share price at
the beginning of the announcement year.

[POOL.sub.i] = a dummy variable equal to unity if firm i uses the
pooling method and zero otherwise.

[LEV.sub.i] = leverage of firm i, defined as the ratio of total
liabilities to total assets,

[PE.sub.i] = the price/earnings ratio of firm i, defined at the
end of the year prior to the announcement year,

[SIZE.sub.i] = firm i's total assets

Table 5
Cross-Sectional Analysis for Valuation Effects of Pooling
and Purchase Acquisitions (t-Statistic Parentheses) (a)

Model Intercept ALPHA BETA

Full Model -0.836 409.94 -0.537
(N = 119) (-0.93) (2.23) ** (-1.27)
Model 2 -0.834 409.34 -0.535
(N = 119) (-0.93) (2.25) ** (-1.27)
Model 3 -1.402 314.30 --
(N = 119) (-1.81) * (1.89) *
Model 4 -1.243 365.382 --
(N = 119) (-1.58) (2.18) **

Model EARSUR LEV PE

Full Model -4.179 2.904 -0.000091
(N = 119) (-1.48) (2.06) ** (-0.04)
Model 2 -4.172 2.904 --
(N = 119) (-1.49) (2.07) **
Model 3 -4.079 3.429 --
(N = 119) (-1.45) (2.55) **
Model 4 -4.064 2.631 --
(N = 119) (-1.42) (2.00) **

 Adj.
Model SIZE INDIC [R.sup.2] F

Full Model -0.000096 -0.893 7.40% 2.35 **
(N = 119) (-1.89) * (-1.68) *
Model 2 -0.000096 -0.896 8.22% 2.76 **
(N = 119) (-1.90) * (-1.71) *
Model 3 -0.000108 -0.989 7.72% 2.97 **
(N = 119) (-2.17) ** (-1.90) *
Model 4 -- -1.087 4.71% 2.46 **
(N = 119) (-2.06) **

* Significant at the 10% level

** Significant at the 5% level

(a) [ALPHA.sub.i], [BETA.sub.i] = the market model parameter
estimates for firm i,

[EARSUR.sub.i] = earnings surprise for firm i, defined as the
change in EPS in the announcement year divided by share price at
the beginning of the announcement year.

[POOL.sub.i] = a dummy variable equal to unity if firm i uses the
pooling method and zero otherwise.

[LEV.sub.i] = leverage of firm i, defined as the ratio of total
liabilities to total assets,

[PE.sub.i] = the price/earnings ratio of firm i, defined at the end
of the year prior to the announcement year,

[SIZE.sub.i] = firm i's total assets