SEARCH

SEARCH BY CITATION

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. I. M&A Accounting and the Diversification Discount
  4. II. Data and Measures
  5. III. Empirical Results
  6. IV. Discussion
  7. V. Conclusion
  8. REFERENCES
  9. Supporting Information

q-based measures of the diversification discount are biased upward by mergers and acquisitions and its accounting implications. Under purchase accounting, acquired assets are reported at their transaction value, which typically exceeds the target's pre-merger book value. Thus, measured q tends to be lower for the merged firm than for the portfolio of pre-merger entities. Because conglomerates are more acquisitive than focused firms, their q tends to be lower. To mitigate this bias, I subtract goodwill from the book value of assets and a substantial part of the diversification discount is eliminated. Market-to-sales-based measures do not have this bias.

A large proportion of financial economics studies, from corporate finance to asset pricing, use accounting data to compute various measures and proxies. When firms are exposed to different accounting treatments, these measures might be compromised by the lack of comparability across firms. Measures of the diversification discount based on Tobin's q are an example of this problem.

Most studies of corporate diversification find that, on average, conglomerates have a lower value than industry-matched portfolios of focused firms.1 Several explanations for this diversification discount have been examined. First, conglomerates might be less efficient than focused firms because of agency costs or inefficient internal capital markets. Alternatively, the discount might stem from self-selection and the endogeneity of diversification decisions, or even from data and measurement problems.2

This paper shows that q-based measures of the diversification discount suffer from a significant upward bias due to mergers and acquisitions (M&A) activity and its accounting implications. The leading procedure for estimating the discount consists of comparing a conglomerate's Tobin's q with that of a benchmark portfolio of focused firms. Most studies employ such q-based excess value as their main, and often sole, specification; market-to-sales and market-to-earnings ratios are much less frequently used alternatives. Under purchase accounting, the assets acquired in a merger are reported at their transaction-implied value in the acquirer's balance sheet. Since the transaction value typically exceeds the target's pre-merger book value, the market-to-book ratio of assets tends to be lower for the post-merger entity than for the portfolio combining both pre-merger entities. Because conglomerates are more acquisitive than focused firms (Maksimovic and Phillips (2008)), their market-to-book ratios, the usual empirical proxy for Tobin's q, tend to be lower.3 To mitigate this measurement bias, I subtract goodwill from the book value of assets. This correction eliminates a substantial part (but not all) of the diversification discount estimated with q-based methods.4 The results cast serious doubt on those widely used measures of the discount.

The paper proceeds as follows. I start by showing that, in principle, q-based measures overestimate the diversification discount if conglomerates are more acquisitive than focused firms, provided they use purchase accounting. Purchase accounting became mandatory in 2001 and was used in 80% of deals before the regulatory change.5 I model a hypothetical acquisition to study the effects of purchase accounting on the deal's excess value, defined as the difference between the merged firm's q and that of a portfolio that includes the acquirer and the target as standalones. This is similar to the excess value measure developed by Berger and Ofek (1995) but defined at the deal rather than firm level. The model predicts excess value to be negative for the representative deal in a sample of M&A transactions from SDC Platinum over the period 1984 to 2007. For instance, for the median acquirer buying a target with q above the sample median (2.14), the deal's excess value is predicted to be negative as long as the transaction premium represents at least 7% of the synergies. The latter condition is very likely to be satisfied in practice because the M&A literature documents negative acquirer returns, which suggests that acquirers tend to overpay for synergies.6 I find that deal excess value is negative for 70% of the deals.

I next turn to the main tests on the diversification discount. First, I replicate the usual conglomerate study regressions using the Compustat Segments sample and firm excess value. Firm excess value is the difference between the q of an actual conglomerate and that of an industry-matched portfolio of standalone firms. I find a diversification discount between 9% and 10% in regressions without firm fixed effects, and between 2% and 3% with firm fixed effects.7 I then try to undo the effect of purchase accounting. To do so, I adjust the excess value measure by subtracting goodwill from the book value of assets. On its own, this correction can reduce the diversification discount by approximately 30% (from 9% to 6%). With firm fixed effects, the diversification discount drops by as much as 76% from the unadjusted excess value regression, to levels between 0.5% and 1.7%, and is no longer statistically different from zero. This result shows that a significant part of the diversification discount can be explained by M&A activity and accounting when q is used to calculate excess value. These results are robust to different diversification measures, accounting for the degree of relatedness between business segments.

The market-to-sales ratio should not be affected by the M&A accounting effects discussed in this paper. Therefore, I also estimate the diversification discount using market-to-sales and expect the diversification discount estimated using this metric to be in line with that in the literature. The diversification discount estimates range between 18% and 21% using standard OLS regressions. With firm fixed effects, the estimates range between 10% and 13%.8 The market-to-sales estimate of the diversification discount is substantially larger than the estimate using the goodwill-adjusted q excess values. This significant difference remains to be explained.

To summarize, conglomerates' greater M&A activity and the associated accounting implications play a first-order role in the usual q-based estimates of the diversification discount, which cast doubt on these widely used methods.

This paper is related to the literature that identifies data and measurement issues relevant for the conglomerate discount. Lang and Stulz (1994) suggest that R&D's accounting treatment can cause a bias in q because it is not recognized as an asset. However, they find that this issue has little impact on estimates of the discount. Whited (2001) examines measurement error in q in investment regressions for conglomerates, in which q is an independent variable. In my study, it is the dependent variable—excess value—that is measured with error. Measurement error in the dependent variable is not a problem unless it is systematically related to an independent variable. The measurement error in excess value identified in this paper is mechanically related to the diversification dummy, which is the main independent variable of interest. Villalonga (2004b) links the diversification discount to data and reporting issues, and finds a diversification premium using an alternative data set to Compustat. Instead, mine is a measurement bias argument.

Graham, Lemmon, and Wolf (GLW) (2002) also link the conglomerate discount to conglomerates' M&A activity. They show that acquirers, which tend to have positive excess value, buy already-discounted targets, that is, targets with negative excess value. This causes a drop in acquirers' excess value and explains the diversification discount. Their selection argument is different from my measurement bias explanation. Moreover, I show that M&A accounting affects measures of the diversification discount even when acquirers buy nondiscounted targets, and their excess value is expected to increase.

The paper proceeds as follows. Section 'M&A Accounting and the Diversification Discount' discusses M&A accounting's effect on excess value and the diversification discount. Section 'Data and Measures' presents the data and methodology. Section 'Empirical Results' reports the empirical results. Section 'Discussion' discusses the results and Section 'Conclusion' concludes.

I. M&A Accounting and the Diversification Discount

  1. Top of page
  2. ABSTRACT
  3. I. M&A Accounting and the Diversification Discount
  4. II. Data and Measures
  5. III. Empirical Results
  6. IV. Discussion
  7. V. Conclusion
  8. REFERENCES
  9. Supporting Information

A. M&A Accounting

Using the purchase method of accounting for M&A, the acquirer adds the target's net assets to its balance sheet at their “fair value.” Fair value is determined by reference to market transaction prices for similar assets or liabilities at or near the measurement date when that information is available.9 Otherwise, fair value is estimated using other valuation techniques. Any premium paid in excess of fair market value is reported as goodwill on the acquirer's balance sheet. The main implication of interest for my analysis is that, since the target's transaction-implied value (i.e., fair value of assets plus goodwill) typically exceeds its pre-merger book value, the acquired assets' book value tends to increase.

Purchase accounting is the only M&A accounting method that has been used since 2001, and it was used in 80% of the deals in my sample before then. The remaining 20% used pooling accounting, in which the book values of the target's assets and liabilities are simply added to the acquirer's. The pooling method was used only in “mergers of equals.” To qualify for this accounting treatment, the transaction had to satisfy 12 requirements mostly related to deal structure and firm characteristics. For instance, at least 90% of the transaction currency had to be stock, and the entities involved had to be autonomous and independent. An additional requirement was the absence of planned transactions after the deal that involved either common stocks issued as part of the combination or any assets of the target company.10 If these requirements were not met, the purchase method had to be used.

B. Tobin's q and Excess Value

In this section, I model a hypothetical acquisition to study the effect of purchase accounting on estimates of Tobin's q and excess value.

Consider acquirer A buying target T to form firm inline image. For inline image, firm i's pre-merger market value is inline image and its book value is inline image. Combining A and T creates (positive or negative) synergies S. To acquire T, A pays a (positive or negative) premium P relative to its market value inline image. A fraction c of the transaction price inline image is paid with internal funds, with the remainder being externally financed with equity or debt. To simplify, assume that T is an all-equity firm.11 Firm i's empirical measure of Tobin's q is inline image.

Following the merger, inline image's market value is the sum of A and T's market values (inline image plus the synergies S, net of the internal funds paid in the transaction inline image12 Under purchase accounting, inline image's book value is inline image, that is, the book value of A, plus the market value of T, plus the premium paid above the market value of the target (P), minus the internal funds spent. inline image's measured q is therefore

  • display math

I define a deal's excess value as the difference between the merged firm's q and that of a portfolio combining the acquirer and the target pre-merger:

  • display math

This is similar to the excess value measure in Berger and Ofek (1995) but defined at the deal rather than the firm level.

I first assume that the deal is fully externally financed (inline image). Eighty percent of the deals in my sample of M&A transactions from SDC Platinum over the period 1984 to 2007 are externally financed, which means that they are all-stock deals (30%) or that the acquirer has issued enough debt or equity during the year of the transaction to meet the cash payment (50%). For externally financed deals, excess value is negative as long as inline image , inline image, and inline image. The first two conditions (inline image and inline image hold for 93% of these deals. The condition inline image is likely to be satisfied in practice as the negative acquirer returns documented in the M&A literature suggest that acquirers tend to overpay for synergies.13 Moreover, for the median acquirer (inline image, buying a target with q above the sample median (inline image), the deal's excess value is negative as long as the transaction premium represents at least 7% of the synergies. Note that a measure of excess value aiming to capture the value created by a merger should depend on the synergies, but not on the pre-merger q's nor the premium.14 For instance, assuming that inline image's book value is the sum of pre-merger book values (inline image, excess value (inline image is simply inline image. Using inline image as a benchmark, deal excess value overestimates the value created in a merger whenever inline image, and underestimates it when inline image. For 96% of deals the transaction price exceeds the target's pre-merger book value, and hence the deal's excess value tends to underestimate the value created by the deal. For instance, a deal with no synergies and no transaction premium (inline image) creates no value but has negative excess value as long as inline image.

When the deal is internally financed (inline image), excess value is increasing in c for inline image. Using internal financing partially offsets the negative bias in excess value. For a deal with no synergies and no transaction premium (inline image) the intuition is as follows. The acquirer exchanges cash with inline image against assets with post-merger inline image (book value of acquired assets equals market value under purchase accounting). Therefore, the acquirer's q does not change if the deal is fully internally financed. For the typical M&A deal in my sample, the effect of internal financing is positive because the median acquirer has a q greater than one. However, few deals consume internal funds (20% of the deals use some internal funds, that is, have inline image), and therefore excess value is expected to have a negative bias for most of the deals.

II. Data and Measures

  1. Top of page
  2. ABSTRACT
  3. I. M&A Accounting and the Diversification Discount
  4. II. Data and Measures
  5. III. Empirical Results
  6. IV. Discussion
  7. V. Conclusion
  8. REFERENCES
  9. Supporting Information

A. Data

I use deals data from the Thomson Financial SDC Platinum database. The initial sample contains all M&A transactions in the U.S. stock market over the period 1984 to 2007. The final sample includes 3,363 transactions that meet the following criteria: the deal must be completed, the acquirer must own more than 50% of the shares after the transaction, the transaction price must be available, and accounting data must be available. I supplement these data with financial items from the Compustat fundamentals quarterly database to compute the acquirer's pre- and post-merger q.15 Pre-merger q is computed at the end of the fiscal quarter immediately preceding the merger announcement date. Post-merger q is computed at the end of the fiscal quarter immediately following deal completion. I exclude deals for which the target or the acquirer is in the financial sector. I also exclude deals for which the target's q or the acquirer's pre- or post-merger q is in the top or bottom 1% of the distribution.

To study how M&A accounting affects q-based estimates of the diversification discount, I use the sample of firms included in the Compustat Segments data set over the 1988 to 2007 period. These firms must also meet the following criteria: firm sales greater than $20 million; no business segments in the financial sector (SIC codes 6000 to 6999), agriculture (SIC code lower than 1000), government (SIC 9000), or other noneconomic activities (SIC 8600 and 8800); unclassified services (SIC 8900) are excluded; firms for which the sum of business segment sales or assets deviates from the firm's total sales or assets by more than 5% are also excluded, as are firms with missing segment SIC codes.16 The final sample includes 59,106 firm-year observations.

B. Deal Excess Value and Firm Excess Value

The measures of Tobin's q, deal excess value, and firm excess value are defined as follows.

Tobin's q. The standard empirical measure of Tobin's q in the diversification discount literature is the ratio of the market value of assets to the book value of assets. The market value of assets is defined as the book value of assets minus the book value of equity plus the market value of equity. I define goodwill-adjusted q as the market value of assets divided by the book value of assets net of goodwill.

Deal excess value is defined as the logarithm of the ratio of the merged firm's q to the hypothetical q of a portfolio that includes the target and the acquirer as standalones.

Firm excess value is defined as the logarithm of the ratio of the firm's observed q to its imputed q, defined as the sales-weighted (or assets-weighted) average of the hypothetical q of the firm's business segments. The hypothetical q is the median (or average) q of standalones in the same industry-year. The industry match is done at the four-digit SIC code level when there are five or more standalones. Otherwise, it is done at the highest level where at least five standalones are available (e.g., Berger and Ofek (1995), Villalonga (2004b)). I define a firm's goodwill-adjusted excess value as its excess value where both observed and imputed q are adjusted for goodwill.17

III. Empirical Results

  1. Top of page
  2. ABSTRACT
  3. I. M&A Accounting and the Diversification Discount
  4. II. Data and Measures
  5. III. Empirical Results
  6. IV. Discussion
  7. V. Conclusion
  8. REFERENCES
  9. Supporting Information

A. Tobin's q, Deal Excess Value, and Firm Excess Value

In this subsection, I test the model's main predictions using the sample of M&A deals. Table I summarizes the data for completed deals, acquirers, and targets for both diversifying and nondiversifying acquisitions.

Table I. Summary Statistics: Deals
 MeanMedianStd. Dev.MinMaxN
Panel A: Acquisitions
Pre-deal q − Acquirer2.421.831.77 0.7915.983,363
Post-deal q − Acquirer2.131.651.41 0.8110.453,363
Deal excess value−0.27−0.111.08−12.936.593,363
q −Target3.462.144.280.6543.473,363
Total assets − Target ($MM)513.0074.902,079.080.0656,553.003,363
Net assets − Target ($MM)193.1032.15838.49−994.1923,534.003,363
Total equity market value − Target ($MM)773.70119.60 3,439.970.0589,165.593,363
Transaction value ($MM)781.23112.383,512.570.0589,167.723,363
Price-to-book difference ($MM) (Mv − Bv)608.7672.97 2,948.59−496.4884,069.423,363
Transaction premium ($MM)66.760.00384.86−3,188.509,999.973,363
Stock payment dummy0.300.000.460.001.003,363
Cash payment dummy0.310.000.460.001.003,363
External financing dummy0.801.000.400.001.002,525
Panel B: Purchase Deals
Pre-deal q −Acquirer2.251.741.57 0.7915.982,879
Post-deal q −Acquirer1.951.561.18 0.8110.312,879
Deal excess value−0.29−0.121.02−12.935.372,879
Panel C: Pooling Deals
Pre-deal q −Acquirer3.372.552.38 0.9214.22484
Post-deal q −Acquirer3.192.522.050.8910.45484
Deal excess value−0.15−0.051.33−9.656.59484
 MeanMedianStd. Dev.MinMaxN
Panel D: Diversifying Acquisitions
Pre-deal q −Acquirer2.441.881.80 0.8014.221,713
Post-deal q −Acquirer2.161.691.39 0.8110.451,713
Deal excess value−0.27−0.101.15−12.936.591,713
q −Target3.572.184.390.6643.471,713
Total assets − Target ($MM)380.0164.001,668.000.0741,078.001,713
Net assets − Target ($MM)140.4429.22574.73−994.1917,909.001,713
Total equity market value − Target ($MM)628.26107.88 2,855.380.0555,987.791,713
Transaction value ($MM)642.64101.313,008.230.0556,307.031,713
Price-to-book difference ($MM) (Mv − Bv)515.6964.17 2,644.97−305.5454,992.031,713
Transaction premium ($MM)53.540.00378.96−2,395.309,999.971,713

The average deal value is $781 million. On average, deal value exceeds the target's pre-merger book value of equity by $609 million and its pre-merger market value by $67 million.

The average target's q (3.46) is greater than the average acquirer's pre-merger q (2.42). In addition, the target's q exceeds the acquirer's pre-merger q for 61% of the deals: the median target's q is 2.14 and the median acquirer's pre-merger q is 1.83. However, even though most acquirers merge with higher q targets, the merged firm's q is lower than both the acquirer's and the target's pre-merger q. The average acquirer's q drops 12% after the deal is complete, and deal excess value is negative for about 63% of the deals. This is particularly true for deals using purchase accounting: the average (median) excess value for these deals is −0.29 (−0.12), against −0.15 (−0.05) for deals using pooling accounting. Moreover, about 70% of the deals using purchase accounting have a negative excess value. This is consistent with the model's prediction that the typical deal will have a negative excess value.

For externally financed deals, a target's q greater than one, an acquirer's q greater than one and a transaction premium that exceeds the value of synergies are sufficient conditions for a negative deal excess value. Indeed, more than 90% of the target firms in the sample have a q greater than one, and the same is true for the acquiring firms. As argued before, the relation between the value of synergies and the transaction premium is also likely to hold. In addition, a large fraction of sample deals—80%—are externally financed. Thirty percent are all-stock deals and therefore fully financed with equity, and the rest are deals for which the acquirer has issued enough debt or equity during the transaction year to meet the cash payment.18

These patterns are similar for both diversifying and nondiversifying deals. I classify a deal as diversifying if the acquirer's industries all differ from the target's industries.19 For diversifying deals, the average (median) target's q is 3.57 (2.18). This exceeds the average (median) acquirer's pre-merger q of 2.44 (1.88). Nevertheless, the post-merger q is lower than both the pre-merger acquirer's and target's q with an average (median) of 2.16 (1.69). The average (median) deal excess value for diversifying mergers is consistently negative (−0.27).

B. Conglomerates Are More Acquisitive

Table II summarizes the data for conglomerates and focused firms. Conglomerates in my sample make more and larger acquisitions than focused firms. In a given year, the average number of past deals is 1.43 per conglomerate and 1.19 per focused firm. The average deal value is $173 million for conglomerates and $112 million for focused firms. Conglomerates' deals are also larger in relative terms: the average deal represents 51% of the book value of assets for conglomerates against 14% for focused firms.

Table II. Summary Statistics: Firms
 MeanMedianStd. Dev.MinMaxN
Panel A: Full Sample
Tobin's q1.891.431.340.627.8359,106
Tobin's q − gwill adjusted 2.061.561.470.648.47 59,106
Goodwill-to-assets0.060.000.120.000.9159,106
Total assets (MM$)1,238.37170.164,556.260.86244,192.5059,106
EBIT-to-sales0.050.070.19−0.930.4159,106
CAPEX-to-sales0.090.040.160.000.9359,106
Number of deals1.230.003.360.0083.0056,542
Diversification dummy0.190.000.390.001.0059,106
Unrelated diversification dummy0.120.000.33 0.001.0059,106
Number of unrelated segments1.171.000.52 1.009.0059,106
1-Herfindahl index0.080.000.200.001.0059,106
Total entropy0.050.000.110.000.3759,106
Unrelated entropy0.020.000.080.000.3759,106
Panel B: Diversified Firms
Tobin's q1.591.320.940.627.8310,949
Tobin's q − gwill adjusted 1.821.471.120.648.47 10,949
Goodwill-to-assets0.090.030.140.000.8110,949
Total assets (MM$)2,246.05438.905,439.062.6290,806.2710,949
EBIT-to-sales0.060.070.14−0.930.4110,949
CAPEX-to-sales0.080.040.140.000.9310,949
Number of deals1.430.004.860.0083.008,417
Excess value measures      
Assets weight − industry median−0.03−0.050.44−2.096.25 10,890
Assets weight − ind. median − gw adjusted−0.01−0.03 0.47−2.156.2510,890
Sales weight − industry median−0.19−0.190.45−1.855.72 10,890
Sales weight − ind. median − gw adjusted−0.17−0.17 0.47−1.935.6610,890
Assets weight − industry average−0.04−0.060.44−2.095.18 10,937
Assets weight − ind. average − gw adjusted−0.01−0.03 0.46−2.155.1810,937
Sales weight − industry average−0.20−0.200.44−1.854.65 10,937
Sales weight − ind average − gw adjusted−0.17−0.18 0.47−1.924.5910,937

Goodwill is specific to M&A events and should therefore reflect a firm's M&A activity. Conglomerates have higher goodwill-to-assets ratios (9.45%) than standalones (5.6%), which is consistent with their making more deals and paying a higher premium over the target's fair value.

Previous cross-sectional evidence suggests that most of the discount occurs for low diversification levels, that is, the damage is done when going from one to two segments (e.g., Servaes (1996)). Interestingly, I find that most of the increase in the goodwill-to-assets ratio occurs for low degrees of diversification: goodwill represents 5.6% of assets in one-segment firms, 7.3% for firms with two segments, and 10.1% for firms with three segments. The ratio remains stable around 10.9% for firms with four segments or more. M&A accounting has the potential to explain this evidence.

The previous findings are consistent with q-based measures of conglomerates' excess value being downward biased (and the discount being upward biased) because they do not account for conglomerates being more acquisitive than focused firms or for the marking-to-market of acquired assets under purchase accounting.

C. Goodwill-Adjusted Excess Value

In this subsection, I proceed in two steps. First, I replicate the standard conglomerate studies and estimate the q-based diversification discount using the Compustat Segments sample. Then, to partially undo the marking-to-market of acquired assets, I subtract goodwill from the book value of assets and estimate the diversification discount using goodwill-adjusted excess values.

Table II reports diversification discount estimates in a univariate setting. Diversified firms have an average negative excess value between −0.03 and −0.20, depending on the specifications. Table III reports the diversification discount estimates in a multivariate setting. The dependent variable in all the regressions is firm excess value where imputed q is estimated using the industry's median q (Panel A) or average q (Panel B).

Table III. Firm Excess Value Regressions: Adjusting for Goodwill
 (1)(2)(3)(4)(5)(6)(7)(8)
   
 Excess Value (Assets Weight)Excess Value (Sales Weight)
Goodwill CorrectionNoYesNoYesNoYesNoYes
Panel A: Industry Median
Div. dummy−0.100***−0.072***−0.022**−0.006−0.103***−0.075***−0.021**−0.005
 [−10.942][−7.504][−2.116][−0.563][−11.380][−7.914][−2.012][−0.498]
Log assets0.007***0.011***−0.115***−0.101***0.006**0.011***−0.115***−0.101***
 [2.655][4.374][−20.105][−16.845][2.477][4.161][−20.321][−17.052]
Ebit-to-sales0.414***0.434***0.665***0.671***0.412***0.431***0.663***0.669***
 [18.786][19.685][25.343][25.434][18.690][19.578][25.269][25.358]
Capex-to-sales0.073***0.0190.305***0.289***0.075***0.0200.307***0.291***
 [3.632][0.927][11.842][11.160][3.697][0.999][11.915][11.205]
Observations59,05659,04759,05659,05659,09059,09459,09059,090
R20.0370.0370.0370.0900.0370.0370.0370.091
Panel B: Industry Average
Div. dummy−0.089***−0.064***−0.030***−0.017−0.092***−0.068***−0.030***−0.017
 [−9.754][−6.726][−2.928][−1.608][−10.199][−7.163][−2.840][−1.577]
Log assets0.012***0.017***−0.113***−0.097***0.012***0.016***−0.113***−0.098***
 [4.922][6.641][−20.150][−16.574][4.739][6.434][−20.289][−16.737]
Ebit-to-sales0.457***0.468***0.677***0.683***0.455***0.465***0.675***0.681***
 [21.070][21.497][26.050][26.106][20.980][21.400][25.991][26.041]
Capex-to-sales0.128***0.063***0.303***0.289***0.129***0.065***0.304***0.291***
 [6.305] [3.050][11.902][11.177][6.374][3.126][11.912][11.168]
Observations59,05659,04759,05659,05659,09059,09459,09059,090
R20.050 0.0490.0490.0990.0500.049 0.0490.099
Firm fixed effectsNoNoYesYesNoNoYesYes

First, I run the standard regressions using (unadjusted) firm excess value as the dependent variable (columns (1) and (5)). I find a diversification discount between 0.09 and 0.10, which is slightly lower than Berger and Ofek's (1995) estimate of 0.12. Including firm fixed effects causes the diversification discount to drop to around 0.03 (columns (3) and (7)). Hence, unobserved heterogeneity at the firm level seems to explain a significant part of the diversification discount estimated with q-based excess value, but not all of it (Campa and Kedia (2002)).20

Next, I run the same regressions using goodwill-adjusted firm excess value as a dependent variable (columns (2), (4), (6), and (8)). Goodwill is only part of the change in book value of the acquired assets under purchase accounting. To fully undo the effect of purchase accounting, one should also subtract from book value the (usually positive) difference between the acquired assets' fair value and their pre-merger book value.21 However, while goodwill is an independent item on the balance sheet and can be easily identified in financial statements, the fair value adjustment is not because it is imputed to specific assets (e.g., fixed assets). Furthermore, because only goodwill is specific to M&A, further adjustments may arguably be excessive. I discuss the arguments in favor of and against adjusting for the difference between fair value and pre-merger book value in Section 'Goodwill Correction'.

When using goodwill-adjusted excess value, the diversification discount estimate drops by some 30% to values between 0.06 and 0.075, depending on the specification. With firm fixed effects, the diversification discount drops by as much as 76% from the unadjusted excess value regression, to levels between 0.005 and 0.017, and is no longer statistically different from zero. This is consistent with M&A activity and its accounting implications, explaining the part of the diversification discount not explained by firm fixed effects. These results also suggest that M&A activity and firm fixed effects together explain a substantial part of the diversification discount estimated with standard q-based excess value.

Table IV shows that the previous results are robust to alternative diversification measures: unrelated diversification dummy, number of segments, number of unrelated segments, Herfindahl index, total entropy, and unrelated entropy. For all measures, I find no diversification discount in regressions that include firm fixed effects using goodwill-adjusted excess values. In some specifications and for the measures that capture unrelated diversification, I find a diversification premium. This is the case with the unrelated diversification dummy, which corresponds to the diversification dummy defined using two-digit SIC codes, with the number of unrelated segments defined at the same SIC code level, and with unrelated entropy.22

Table IV. Firm Excess Value Regressions: Other Diversification Measures
 (1)(2)(3)(4)(5)(6)(7)(8)
   
 Excess Value (Industry Median)Excess Value (Industry Average)
Goodwill CorrectionNoYesNoYesNoYesNoYes
Unrelated div. dummy−0.069***−0.047***0.0150.025**−0.059***−0.047***0.0030.013
 [−6.674][−4.254][1.261][2.048][−5.706][−4.254][0.277][1.064]
Number of segments−0.037***−0.026***−0.007*−0.000−0.033***−0.024***−0.010***−0.005
 [−10.409][−7.033][−1.710][−0.050][−9.414][−6.370][−2.583][−1.152]
N. of unrelated segments−0.038***−0.026***0.0100.017**−0.032***−0.022***0.0020.008
 [−5.615][−3.714][1.345][2.151][−4.810][−3.132][0.227][1.034]
Herfindahl index−0.192***−0.135***−0.052**−0.016−0.174***−0.122***−0.067***−0.036
 [−10.680][−7.057][−2.447][−0.696][−9.692][−6.377][−3.087][−1.584]
Total entropy−0.350***−0.247***−0.097**−0.032−0.314***−0.221***−0.120***−0.065
 [−10.793][−7.136][−2.515][−0.797][−9.705][−6.400][−3.077][−1.621]
Unrelated entropy−0.252***−0.154***0.0550.113**−0.220***−0.125***0.0180.075
 [−5.901][−3.250][1.038][2.108][−5.199][−2.667][0.343][1.414]
Firm fixed effectsNoNoYesYesNoNoYesYes

The main implication of these findings is that the average conglomerate discount is significantly reduced after accounting for the mechanical effect of M&A accounting on q-based estimates of excess value. The results therefore cast serious doubt on these widely used methods to estimate the discount.

D. Market-to-Sales Excess Value

Most studies employ q-based excess values as their leading, and often sole, specification, with the main and much less frequently used alternatives being based on market-to-sales and market-to-earnings ratios. The market-to-earnings ratio is affected by purchase accounting through the amortization of goodwill, which affects earnings negatively.23 Instead, the market-to-sales ratio should not be affected by the M&A accounting effects discussed in this paper. Therefore, I run the regressions of Section 'Goodwill-Adjusted Excess Value'. using market-to-sales as a dependent variable; I expect the diversification discount estimated using this metric to be in line with that in the literature. Results are presented in Table V. The diversification discount estimates range between −0.182 and −0.208 with standard OLS regressions. These estimates exceed Berger and Ofek's (1995) estimate of 0.144. With firm fixed effects, the estimate ranges between −0.103 and −0.131, which is similar to Campa and Kedia's (2002) estimate of −0.14.

Table V. Firm Excess Value Regressions: Market-to-Sales
 (1)(2)(3)(4)
   
 Excess Value (Assets Weight)Excess Value (Sales Weight)
Panel A: Industry Median
Div. dummy−0.208***−0.121***−0.195***−0.103***
 [−14.799][−6.961][−14.074][−5.960]
Log assets0.083***0.097***0.092***0.099***
 [22.953][11.448][24.697][11.399]
Ebit-to-sales0.108***0.349***−0.0290.264***
 [3.129][9.572][−0.765][6.505]
Capex-to-sales0.648***0.909***0.740***0.988***
 [19.430][22.180][20.529][21.810]
Observations59,04759,04759,09459,094
R20.0920.6970.0970.694
Panel B: Industry Average
Div. dummy−0.204***−0.131***−0.182***−0.112***
 [−14.774][−7.775][−13.533][−6.843]
Log assets0.098***0.104***0.098***0.104***
 [27.042][12.616][27.380][12.840]
Ebit-to-sales0.172***0.382***0.168***0.375***
 [5.284][10.941][5.187][10.838]
Capex-to-sales0.772***0.883***0.779***0.867***
 [24.911][23.467][25.158][23.647]
Observations59,10559,10559,09459,094
R20.1300.7030.1310.702
Firm fixed effectsNoYesNoYes

The market-to-sales estimate of the diversification discount is substantially larger than the estimate using the goodwill-adjusted q excess values, which varies between zero and 0.075. This significant difference remains to be explained.

IV. Discussion

  1. Top of page
  2. ABSTRACT
  3. I. M&A Accounting and the Diversification Discount
  4. II. Data and Measures
  5. III. Empirical Results
  6. IV. Discussion
  7. V. Conclusion
  8. REFERENCES
  9. Supporting Information

A. Goodwill Correction

In this subsection, I further discuss the goodwill adjustment and study a potential bias it might create in excess value when M&As are financed internally.

Undoing the effect of purchase accounting requires more than subtracting goodwill from the book value of assets. One should also subtract the (usually positive) difference between the fair value of acquired assets and their pre-merger book value. However, this difference is due to a net write-up of the value of existing target assets. In this respect, it can be viewed as just part of the natural variation in book value due to differences in the timing of asset valuation. Firms' book values for the same types of assets in place can differ depending on when they were purchased. The unique issue posed by purchase accounting is that it effectively requires adding the present value of future expected cash flows to the book value, and the extent to which it does is best represented by goodwill. For example, consider two otherwise identical firms: one growing internally via capital expenditures and the other growing externally via M&A. The book value of the latter would exceed that of the former by the amount of goodwill. If such write-ups occur sufficiently frequently outside M&A events, subtracting goodwill would be the correct adjustment. If, however, firms update the value of their assets mostly in M&A, subtracting goodwill is only a conservative adjustment.

A second concern is that the goodwill adjustment might generate a bias in excess value if acquirers use internal funds to finance their acquisitions. This bias is positive if the acquirer's q exceeds one, and negative otherwise.

To illustrate, consider the case of overpayment (inline image) when the acquirer uses internal funds. Because the acquirer uses internal funds, the value of the firm decreases with the overpayment (in contrast, when the acquirer uses external financing, overpaying represents a wealth transfer between old investors and new investors). Purchase accounting reflects the overpayment in the acquirer's book value (which increases) and in its q and excess value (which both decrease). However, the goodwill adjustment would unduly undo the negative effect of overpayment on q and excess value. When the firm issues new securities and uses the proceeds to pay for the target, the goodwill correction does not create any bias, even when there is overpayment.

This bias is unlikely to affect the previous results significantly since only a few acquisitions (18%) are financed with internally generated funds and have an acquirer's q that exceeds one. Moreover, cash-only deals represent only 31% of the sample, and among those firms 67% have issued enough debt or equity in the same year as the deal to finance this cash payment.24 Hence, the bias possibly affects less than 10% of deals. Furthermore, for these deals debt and equity issues in years prior to the acquisition are not excluded.

B. Goodwill Amortization and Impairment

Before 2001, goodwill was amortized over a period of up to 40 years. Goodwill amortization (negatively) impacts book values but not market values because it has no information content. Goodwill amortization alleviates the bias in q-based estimates of excess value because over time it brings the book value of acquired assets closer to its pre-merger value. In this sense, it is a gradual version of my goodwill adjustment.25 Since 2001, goodwill has not been subject to amortization but is tested for impairment. The impairment test consists of comparing the carrying amount of goodwill with its fair value and recognizing a loss whenever the former exceeds the latter.

Unlike amortization, impairment tests may contain information and therefore affect not only q's denominator, but also possibly its numerator. The empirical evidence about the market value impact of impairment tests is mixed. Francis, Hanna, and Vincent (1996) find no statistically significant impact while Bens, Heltzer, and Segal (2007) find an average negative announcement return of about 3.4%.

Goodwill impairment is reported in only 1,432 firm-years in my sample (about 2%). In addition, goodwill impairment represents less than 1% of the book value of the assets for the firms involved.26 Therefore, goodwill impairment is unlikely to affect my main results significantly.

Another concern is that, since 2001, goodwill impairments are reported at the segment level, not the firm level, which allows for some managerial discretion (Ramanna and Watts (2011)). This could potentially cause a bias in excess value measures because conglomerates have more discretion than standalones, and the market reaction to impairments could be different. However, Bens, Heltzer, and Segal (2007) find no cross-sectional difference in market reactions based on the number of segments.

C. Discounted Targets versus M&A Accounting

GLW (2002) also link the diversification discount to conglomerates' M&A activity. They show that acquirers tend to buy already-discounted targets (i.e., those with negative excess value), which lowers their own excess value. Because conglomerates are more acquisitive, they are more affected. This selection argument is different from my measurement bias explanation. Moreover, I show that the M&A accounting effect is significant even when GLW's mechanism is not (i.e., when acquirers buy targets with positive excess value exceeding the acquirer's and their excess value is expected to increase). In a second step, I try to estimate how much of the change in excess value can be explained by purchase accounting and by GLW.

Following GLW, the projected firm excess value is defined as inline image where inline image is the market value of firm i and inline image is its imputed value. The projected change in firm excess value is the difference between the projected firm excess value and the acquirer's pre-merger excess value. The actual change is the difference between the actual pre- and post-merger excess values.

Consistent with GLW, in Table VI, Panel A, I find that acquirers tend to buy already-discounted targets with negative excess value: the average pre-merger excess value for the target is −0.039, while for the acquirer it is 0.194. However, in my sample, the projected change in excess value (−0.045) is too small to explain the actual change (−0.138): the difference (−0.093) is statistically different from zero. In GLW, the projected change is not statistically different from the actual change. However, in my sample, their mechanism cannot fully explain the change in excess value.

Table VI. Actual and Projected Changes in Excess Value of Acquirers
 MeanMedian
Panel A: Full Sample (505 firms)
Pre-merger target firm excess value−0.039**−0.038**
Pre-merger acquirer firm excess value0.194***0.127***
Actual change inline image−0.138***−0.090***
Projected change inline image−0.045***−0.017***
Difference inline image−0.093***−0.062***
Panel B: Nondiscounted Targets (145 firms)
Pre-merger target firm excess value0.407***0.255***
Pre-merger acquirer firm excess value0.182***0.141***
Actual change inline image−0.297***−0.200***
Projected change inline image0.039***0.016***
Difference inline image−0.336***−0.254***
Panel C: Purchase Method Deals (382 firms)
Pre-merger target firm excess value−0.101***−0.068***
Pre-merger acquirer firm excess value0.140***0.087***
Actual change inline image−0.146***−0.094***
Projected change inline image−0.046***−0.018***
Difference inline image−0.100***−0.072***
Projected change − adjusted inline image−0.143***−0.075***
Difference − adjusted inline image−0.0030.021

In Table VI, Panel B, I restrict the sample to acquirers buying nondiscounted targets with excess values that exceed the acquirer's, that is, deals that are expected to generate a positive change in the acquirer's excess value. For these deals the projected change is positive (0.039), but the actual change is negative (−0.297). The difference between the projected change and the actual change is even larger in this subsample (−0.336). By construction, GLW's argument cannot explain the negative change for these deals because the projected change is positive. Purchase accounting, instead, is a possible explanation. The model in Section 'M&A Accounting and the Diversification Discount' shows how merging with a high q target can result in a negative excess value.

I next study the sample of deals that use purchase accounting. I define the projected firm excess value adjusted for purchase accounting as inline image, where inline image is calculated using the transaction value instead of the target's book value. Therefore, inline image is the projected excess value, taking into account the fact that the acquirer will write up the target's assets to their transaction value.

Comparing the projected excess value with the adjusted excess value shows how much of the actual change can be explained by each mechanism. The average actual change in firm excess value for this sample is −0.146. With GLW's selection mechanism, the projected change is −0.046, which is about 30% of the actual change. If I also account for purchase accounting, the projected change is −0.143, which is much closer to the actual change. In fact, the difference between the adjusted projected change and the actual change is not statistically significant. This result suggests that M&A accounting can explain as much as 70% of the change in firm excess value, and GLW as little as 30% in the sample of deals using purchase accounting.27

V. Conclusion

  1. Top of page
  2. ABSTRACT
  3. I. M&A Accounting and the Diversification Discount
  4. II. Data and Measures
  5. III. Empirical Results
  6. IV. Discussion
  7. V. Conclusion
  8. REFERENCES
  9. Supporting Information

This paper shows that q-based measures of the diversification discount are biased upward by M&A accounting implications. The most common procedure for estimating the discount is to compare a conglomerate's q with that of a benchmark portfolio of focused firms. Under purchase accounting, the acquired assets are reported at their transaction-implied value in the acquirer's balance sheet. Since the transaction value typically exceeds the target's pre-merger book value, measured q tends to be lower for the merged firm than for the portfolio that combines both pre-merger entities. Because conglomerates are more acquisitive than focused firms, their measured q tends to be lower. To mitigate this measurement bias, I subtract goodwill from the book value of assets. This correction eliminates a substantial part (but not all) of the diversification discount estimated with q-based methods. The results cast serious doubt on these widely used methods of measuring the diversification discount.

Measures based on market-to-sales ratios are not affected by M&A accounting. Further research is needed to explain the difference between the goodwill-adjusted estimate of the discount and estimates based on market-to-sales ratios.

The measurement bias related to M&A accounting could have further implications for the diversification discount. In the cross section, the negative bias should be greater in industries with high goodwill. In the time series, it should be greater in times of high goodwill and when more firms use purchase accounting (after 2001). Investment efficiency measures based on investment sensitivity to q might also be biased against conglomerates. Investigating these implications is left for future research.

Editor: Campbell Harvey

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. I. M&A Accounting and the Diversification Discount
  4. II. Data and Measures
  5. III. Empirical Results
  6. IV. Discussion
  7. V. Conclusion
  8. REFERENCES
  9. Supporting Information
  • Andrade, Gregor, Mark Mitchell, and Erik Stafford, 2001, New evidence and perspectives on mergers, Journal of Economic Perspectives 15, 103120.
  • Bens, Daniel A., Wendy Heltzer, and Benjamin Segal, 2007, The information content of goodwill impairments and the adoption of SFAS 142, Working paper, INSEAD.
  • Berger, Philip, and Eli Ofek, 1995, Diversification's effect on firm value, Journal of Financial Economics 37, 3965.
  • Bevelander, Jeffrey, 2002, Tobin's q, corporate diversification, and firm age, Working paper.
  • Byrd, John W., and Kent A. Hickman, 1992, Do outside directors monitor managers? Evidence from tender offer bids, Journal of Financial Economics 32, 195221.
  • Campa, Jose Manuel, and Simi Kedia, 2002, Explaining the diversification discount, Journal of Finance 57, 17311762.
  • Comment, Robert, and Gregg Jarrell, 1995, Corporate focus and stock returns, Journal of Financial Economics 37, 6787.
  • Denis, David J., Diane K. Denis, and Atulya Sarin, 1997, Agency problems, equity ownership, and corporate diversification, Journal of Finance 52, 135160.
  • Francis, Jennifer, J. Douglas Hanna, and Linda Vincent, 1996, Causes and effects of discretionary asset write-offs, Journal of Accounting Research 34, 117134.
  • Glaser, Markus, and Sebastian Muller, 2010, Is the diversification discount caused by the book value bias of debt? Journal of Banking and Finance 34, 23072317.
  • Graham, John, Michael Lemmon, and Jack Wolf, 2002, Does corporate diversification destroy value? Journal of Finance 57, 695720.
  • Healy, Paul M., Krishna G. Palepu, and Richard S. Ruback, 1992, Does corporate performance improve after mergers? Journal of Financial Economics 31, 133278.
  • Henning, Steven, Barry L. Lewis, and Wayne H. Shaw, 2000, Valuation of the components of purchased goodwill, Journal of Accounting Research 38, 375386.
  • Jensen, Michael, 1986, Agency costs of free cash flow, corporate finance, and takeovers, American Economic Review 76, 323329.
  • Kaplan, Steven, and Michael Weisbach, 1992, The success of acquisitions: Evidence from divestitures, Journal of Finance 47, 107138.
  • Laeven, Luc, and Ross Levine, 2007, Is there a diversification discount in financial conglomerates, Journal of Financial Economics 85, 331367.
  • Lamont, Owen, and Christopher Polk, 2002, Does diversification destroy value? Evidence from industry shocks, Journal of Financial Economics 63, 5177.
  • Lang, Larry, and René Stulz, 1994, Tobin's q, corporate diversification and firm performance, Journal of Political Economy 102, 12481280.
  • Lins, Karl, and Henri Servaes, 1999, International evidence on the value of corporate diversification, Journal of Finance 54, 22152240.
  • Maksimovic, Vojislav, and Gordon Phillips, 2002, Do conglomerate firms allocate resources inefficiently across industries? Theory and evidence, Journal of Finance 57, 725767.
  • Maksimovic, Vojislav, and Gordon Phillips, 2007, Conglomerate firms and internal capital markets, in B. Espen Eckbo, ed.: Handbook of Corporate Finance: Empirical Corporate Finance (North-Holland, Amsterdam).
  • Maksimovic, Vojislav, and Gordon Phillips, 2008, The industry life-cycle, acquisitions and investment: Does firm organization matter? Journal of Finance 63, 673707.
  • Montgomery, Cynthia A., and Birger Wernerfelt, 1988, Tobin's q and the importance of focus in firm performance, American Economic Review 78, 246250.
  • Mulherin, J. Harold, and Audra L. Boone, 2000, Comparing acquisitions and divestitures, Journal of Corporate Finance 6, 117139.
  • Rajan, Raghuram, Henri Servaes, and Luigi Zingales, 2000, The cost of diversity: The diversification discount and inefficient investment, Journal of Finance, 55, 3580.
  • Ramanna, Karthik, and Ross L. Watts, 2011, Evidence on the use of unverifiable estimates in required goodwill impairment, Working paper, Harvard Business School.
  • Scharfstein, David, and Jeremy Stein, 2000, The dark side of internal capital markets: Divisional rent-seeking and inefficient investment, Journal of Finance 55, 25372564.
  • Servaes, Henri, 1996, The value of diversification during the conglomerate merger wave, Journal of Finance 51, 12011225.
  • Villalonga, Belen, 2004a, Does diversification cause the “diversification discount”? Financial Management 33, 527.
  • Villalonga, Belen, 2004b, Diversification discount or premium? New evidence from the business information tracking series, Journal of Finance 59, 475502.
  • Whited, Toni, 2001, Is it inefficient investment that causes the diversification discount? Journal of Finance 56, 16671692.
  1. 1

    Montgomery and Wernerfelt (1988), Lang and Stulz (1994), Berger and Ofek (1995), Comment and Jarrel (1995), Servaes (1996), Lins and Servaes (1999), Rajan, Servaes, and Zingales (2000), and Lamont and Polk (2002) document a diversification discount. Laeven and Levine (2007) find a discount for financial conglomerates.

  2. 2

    Jensen (1986) and Denis, Denis, and Sarin (1997) suggest that conglomerates are less efficient due to agency costs; Scharfstein and Stein (2000) and Rajan, Servaes, and Zingales (2000) suggest that internal capital markets are inefficient. Campa and Kedia (2002), Graham, Lemmon, and Wolf (GLW; 2002), Maksimovic and Phillips (2002), and Villalonga (2004a) address self-selection and endogeneity issues. Villalonga (2004b) argues that the diversification discount is generated by segment data problems. Glaser and Muller (2010) find that using the book value of debt to measure firm value generates a downward bias in the value of conglomerates. See Maksimovic and Phillips (2006) for a detailed survey of the literature on corporate diversification.

  3. 3

    Henceforth, I use q to refer to the market-to-book ratio of assets, the empirical measure of Tobin's q commonly used in the diversification discount literature, rather than the theoretical economic concept.

  4. 4

    Firms can write-up existing assets to fair value in situations other than M&As, but accounting for goodwill is unique to purchase accounting.

  5. 5

    Until 2001, the alternative was pooling accounting, in which the book value of acquired assets is set to their pretransaction value. The gist of my arguments is that the typical q-based measures of the diversification discount compare actual conglomerates, which generally use purchase accounting, with hypothetical conglomerates formed by a merger of standalones using pooling accounting.

  6. 6

    Most event studies on M&A find that these transactions tend to add value for shareholders, but that most of the gains accrue to the target. Acquirers tend to show negative abnormal returns after the announcement of a merger. See Andrade, Mitchell, and Stafford (2001) for a survey.

  7. 7

    A discount of 10% means that conglomerates have, on average, 10% lower excess value than standalone firms. Berger and Ofek (1995) find a diversification discount of 12% over the period 1986 to 1991. Bevelander (2002) finds an 8% discount for the period 1980 to 1998.

  8. 8

    The ordinary least squares (OLS) estimate exceeds Berger and Ofek's (1995) estimate of 0.14. The firm fixed effects estimate is similar to Campa and Kedia's (2002) estimate of −0.14.

  9. 9

    See Statement of Financial Accounting Standards 141 and Financial Accounting Standards Board December 11, 2002 meeting minutes on “Business Combinations: Purchase Method Procedures” (http://www.fasb.org/jsp/FASB/Page/12-11-02.pdf) for further details.

  10. 10

    See Accounting Principles Board Opinion 16 for further details.

  11. 11

    This assumption does not affect the main results.

  12. 12

    I assume that the market value of internal funds equals their book value.

  13. 13

    Byrd and Hickman (1992), Healy, Palepu, and Ruback (1992), Kaplan and Weisbach (1992), Mulherin and Boone (2000), and Andrade, Mitchell, and Stafford (2001) all find negative cumulative abnormal returns for the acquirer between −3.8% and −0.37% and combined cumulative abnormal returns between +1.8% and +9.1%.

  14. 14

    When inline image, excess value should not be affected by the transaction premium, not even when inline image; that is, overpaying is a wealth transfer from existing shareholders to new shareholders and no value is destroyed.

  15. 15

    I use financial information from Compustat for the acquirer and from SDC for the target because the sample includes private targets that are not covered by Compustat.

  16. 16

    Results are robust to replacing the missing segments' SIC codes with the main SIC code reported by the firm in order not to reduce the sample size of diversified firms. Results are also robust to keeping the smaller firms in the sample; the unadjusted average diversification discount is smaller in this case.

  17. 17

    Ideally, I would also adjust the assets' weights for goodwill when calculating excess value. However, segment-specific goodwill is not observed in the data.

  18. 18

    I use the items dltis and sstk from Compustat to measure debt and equity issues.

  19. 19

    I use four-digit SIC codes to be consistent with the following analysis on the diversification discount, where a conglomerate is defined at the same SIC code level. The results are robust to using two-digit SIC codes.

  20. 20

    Because these regressions include firm fixed effects and a diversification dummy, the identification only arises from firms that either become diversified or refocus during the sample period. The regressions using alternative diversification variables do not face this limitation.

  21. 21

    Henning, Lewis, and Shaw (2000) estimate goodwill to be 62.5% of the difference between the transaction price and the target's book value.

  22. 22

    The results in this subsection are also robust to replacing firm excess value with q.

  23. 23

    Because goodwill amortization occurs over a period of up to 40 years, and because goodwill is no longer amortized (since 2001), the effect of purchase accounting is likely smaller than for q-based estimates of the discount.

  24. 24

    I use the items dltis and sstk from Compustat to measure debt and equity issues.

  25. 25

    I adjust for goodwill net of amortization and hence there is no risk of “overadjusting” excess value.

  26. 26

    Bens, Heltzer, and Segal (2007) find that the average goodwill impairment represents about 18% of the book value of assets. This difference is possibly due to sample selection, since they drop all the observations for which goodwill impairment represents less than 5% of the assets.

  27. 27

    When I restrict my sample to the period before 1996 to get closer to the period covered by GLW (2002), I find that their mechanism explains as much as 56% of the change in firm excess value.

Supporting Information

  1. Top of page
  2. ABSTRACT
  3. I. M&A Accounting and the Diversification Discount
  4. II. Data and Measures
  5. III. Empirical Results
  6. IV. Discussion
  7. V. Conclusion
  8. REFERENCES
  9. Supporting Information
FilenameFormatSizeDescription
jofi12108-sup-0001-SupMat.pdf242KSlides: M+A Accounting and the Diversification Discount.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.