Beruflich Dokumente
Kultur Dokumente
Jarrad Harford
Foster School of Business
University of Washington
Seattle, WA 98195
jarrad@uw.edu
and
Adam Kolasinski
Foster School of Business
University of Washington
Seattle, WA 98195
adamkola@uw.edu
We examine a comprehensive sample of 788 large US private equity buyout transactions from 1993-2001,
tracking exit status through 2009. About 90% are exited other than through IPO. We test whether, on
average, private equity sponsors create value or transfer value to themselves. All of our evidence is
consistent with value creation. While private, portfolio companies reduce overinvestment, but do not
underinvest. Special dividends to sponsors are rare and are not correlated with future portfolio company
distress. The most common exit is to a strategic buyer, where announcement and long-run return results
establish that value is created. Almost a third of portfolio companies are sold to another financial buyer.
Such companies tend to be held longer and also tend to be in concentrated industries, suggesting that such
secondary buyouts are motivated by liquidity needs as well as a desire to help monopolistic companies
avoid the disclosure costs and public scrutiny that accompany public listing.
*
We thank Rocky Higgins, Paul Malatesta, Jon Karpoff, Jennifer Koski and participants at a University of
Washington Finance Brown Bag for helpful comments. Jared Stanfield, Robert Schonlau, and Wei Ming
Lee provided excellent research assistance.
community, such as Warrant Buffet, have opined that the returns generated by private
equity sponsors are a result of wealth transfers rather than value creation (The Times of
London, 2006). Articles in the popular press frequently use words such as “locusts” and
“rape-and-pillage” when referring to private equity sponsors (eg., Forbes, 2006, Der
Spiegel, 2006). In contrast, in a recent survey article, Kaplan and Stromberg (2009)
conclude that the bulk of the evidence suggests private equity sponsors create value.
However, they note that much of this is based on either reverse leveraged buyouts (e.g.,
improvements (eg., Kaplan, 1989, Guo, Hotchkiss and Song, 2010, Litchenberg and
Siegel, 1990, Smith, 1990). Critics would respond that reverse leveraged buyouts are a
performance improvements under private equity sponsor ownership leaves open the
equity buyouts performed from 1993 to 2001. We test and reject the hypothesis that
at the expense of the long-term. We further investigate several specific avenues of value
Specifically, for the companies bought out by private equity sponsors (i.e.
“portfolio companies”) in our sample with outstanding public debt, we confirm the short-
ROA that is higher than a control group’s following the buyout. We also add a new
by making investment less responsive to positive cash flow following the buyout. At the
same time, the portfolio company’s investment is no more responsive to negative cash
We test and reject the hypothesis that on average private equity sponsors transfer
wealth or damage the portfolio company’s long-term value. While previous research has
rejected this hypothesis for the 10% of buyouts exited via IPO, we test it for buyouts
exited via the most common method, sale to a strategic acquirer, which account for 36%
of exits, as well as for portfolio companies that end up in financial distress, which
account for around 14% of exits. We find that when the sponsor sells a portfolio
company to a public strategic acquirer, the latter’s stock price reaction is positive, even
more so when we control for the size of the acquisition relative to the acquirer’s size, as
well as acquirer governance. We follow the acquirers’ long-run stock price performance
and find it is indistinguishable from that of other acquirers. We also test the hypothesis
that private equity sponsors extract capital through special dividends and leave the target
firm in distress. First, despite the impression given by the public press, we find that
special dividends are rare. Second, those that occur are uncorrelated with future portfolio
portfolio company is sold to another private equity sponsor (or “financial buyer,” to use
industry parlance), we examine whether some portfolio companies are simply suited to
perpetually remain so. We find that having been bought from another private equity
sponsor before does not make sale to a financial buyer any more likely than an IPO or
2
sale to a strategic buyer, so secondary buyouts are not indicative of a portfolio company-
type characteristic. Rather, secondary buyouts are common when the sponsor has held
the portfolio firm longer, suggesting that inability to exit in a timely manner, leads to exit
to a financial buyer through a secondary buyout. This is consistent with Kaplan and
Schoar (2005), who find that private equity sponsors face pressure to exit early. On the
other hand, we find that portfolio companies are more likely to stay private, either by exit
to a financial buyer or no exit at all, when they are in concentrated industries. This
suggests that private equity sponsors can add value by helping monopolistic companies
stay private, thereby allowing them to avoid potentially costly disclosure to competitors,
The rest of this study is organized as follows. In section I, we review the private
equity buyout literature in greater depth. In Section II, we discuss our data sources and
sample selection procedure. In Section III, we discuss our tests and results. Section IV
concludes.
I. Literature Review
necessarily limited because of lack of data. While databases such as SDC and Capital IQ
reliably track private equity buyouts, they do not track exit outcomes. Furthermore, the
As a result, research on exit outcomes of buyouts has been largely limited to portfolio
companies that were exited via IPO (eg., Cao and Lerner, 2009, Degeorge and
Zeckhuaser, 1993, Holthausen and Larcker, 1996, Mian and Rosenfeld, 1993, Muscarella
3
and Vetsuypens, 1990). Likewise, as Kaplan and Stromberg (2009) point out in their
survey article, research on the operating performance of portfolio companies has been
largely limited either to portfolio companies that were public before or after the buyout
(e.g., Kaplan, 1989, Smith, 1990, Guo, Hotchkiss, and Song, 2010), or to countries that
require private companies to disclose financial statements (e.g., Boucly, Sraer and
Thesmar, 2008, Grubb and Jonsson, 2007, Harris, Siegel and Wright, 2005).
buyouts, in excess of $50 million, and tracking each of them to exit. Kaplan and
Stromberg (2009), also provide a distribution of exit outcomes for all buyouts, large and
small, but they fail to determine and outcome for 11% of the sample. In contrast we were
We further add to the literature by providing the first evidence on buyouts exited
via sale to strategic buyer. We confirm that the strategic buyers benefit when they
private equity sponsors add value not only for them themselves, but their counterparties
as well. These results complement Masulis and Nahata (2009), who find positive
sold by venture capital funds. Their results imply that venture capital investors do not
profit at the expense of the shareholders of the strategic acquirers to whom they sell.
While our results have a similar flavor, they are distinct and important in their own right
since the business of buyouts and the types of portfolio companies held by buyout-
focused private equity sponsors are fundamentally different from those held by venture
funds.
4
Our paper presents the first evidence on the relation between value extraction by
financial buyers and portfolio company financial distress. Specifically, we find that,
among portfolio companies with public debt, special dividends are rare and do not help
company operating performance. First, this study is the first using US data that is not
limited to portfolio companies that were public at one time. Second, it is the first to
examine the effect of private equity sponsors on the efficiency of portfolio company
those that were public at some point. In contrast, we use the Capital IQ, which contains
comprehensive financial statement data on all US private companies with public debt.
Hence we are able to obtain data on operating performance, capital expenditures, special
dividends, and the like for all portfolio companies financed with public debt. Hence we
Only Litchenberg and Siegel (1990), who utilize Census Bureau plant-level data, study
operating performance in a sample larger than ours. However, they are naturally limited
of profitability (e.g., Cotter and Peck, 2001, Guo et al., 2010, Kaplan, 1989, and Smith,
1999), and productivity (Lichtenberg and Siegel, 1990). Kaplan (1989) also provides
5
descriptive evidence on how a portfolio company’s ratio of capital expenditures to total
sales changes during a buyout. We expand this literature by providing first evidence on
not only study how investment policy changes, but how buyouts affect portfolio
Specifically, we find evidence that private equity sponsors add value by preventing
portfolio companies from overinvesting when cash flows are high. At the same time, we
test and fail to find evidence that private equity sponsors cause their portfolio companies
to underinvest.
The last result is significant because it clarifies past evidence that private equity
Stromberg point out, just because profitability and productivity improve while portfolio
companies are private does not necessarily mean private equity sponsors add value. It is
possible that such short-run performance improvements come at the expense of long-run
performance. Our finding that private equity sponsors do not appear to cause their
this finding, together with our evidence that strategic acquirers benefit from buying
portfolio companies, our evidence on special dividends and portfolio company financial
distress, as well as the results or Cao and Lerner on post-IPO performance and Lerner,
Sorensen and Stromberg’s (2008) evidence on portfolio company patent productivity, the
6
II. Data sources and sample selection
From Thomson Financial’s SDC Platinum and Capital IQ, we extract information
going private transactions that have a disclosed value of greater than $50 million, and
addition, for each transaction, we examine by hand the entities listed as acquirers and
keep only those transactions for which at least one of the acquirers is a private equity
sponsor, defined as a legal entity that engages in buyouts in its ordinary course of
business. Private equity sponsors include corporations and partnerships in the business of
subsidiaries, such GE Capital or Goldman Sachs Merchant Banking, whose parent firms
are in other lines of business. We use sponsor websites, SEC filings, and Hoover’s
After applying the above filters, our sample comes to 788 buyouts. We obtain a portfolio
company’s primary NAICS code from SDC if the deal is in that database. If a deal is
only in Capital IQ, we assign a NAICS code based on the detailed business description
provided in the database. Using SDC and stories in the business press (Factiva), we
determine the pre-buyout status of the portfolio company and categorize it in one of the
following ways: public, a subsidiary of a firm that is not a private equity sponsor,
privately held firm. Statistics on the pre-buyout status of the portfolio companies in our
7
Next, for each buyout in our sample, we determine whether the private equity
sponsor or syndicate exited the investment, and if so, how. We check for exit via IPO
and merger and acquisition (M&A) using Capital IQ and SDC. We classify IPOs and
M&As as exits if a controlling stake of the portfolio company was sold to the public or
some entity or syndicate that did not participate in the initial buyout.1 For those portfolio
companies for which no exit can be found in the databases, we obtain exit information
from stories in the business press, company and private equity sponsor websites,
bankruptcy filing databases, and, in a few instances, with a phone call to the portfolio
company or private equity sponsor.2 We place each exit outcome in one of five
categories: IPO, sale to strategic buyer, sale to financial buyer, restructured, or still held.
conduct buyouts in the ordinary course of business. “Financial buyers” are private equity
sponsors. We classify as “restructured” all buyouts in which the private equity sponsor
or syndicate lost its controlling stake due to a bankruptcy, workout, or other debt
restructuring resulting from financial distress. We count as “still held” those portfolio
companies where the original sponsor or syndicate, or some subset thereof, continued to
hold a controlling stake as of September 30, 2009. In each case, we confirm that a
company for which we could find no exit outcome was still held. In the case of a
“rollup”, in which two or more portfolio companies held by the same sponsor or
syndicate are merged with one another before exit, we assign to each individual portfolio
1
IPO’s in which only a minority stake is sold to the public are not counted as IPO’s. It is not uncommon in
our sample for private equity sponsors to sell a small stake in an IPO, only to have a strategic buyer acquire
100% of the portfolio company a short time later. We label these exits as sales to strategic buyers.
2
In one instance, the private equity sponsor revealed that when its portfolio company was nearing financial
distress, it had a fortuitously timed fire. The insurance payout provided enough capital for the portfolio
company to recover, and it was eventually sold to another financial buyer.
8
company the exit outcome of the rolled-up entity. We use an indicator variable for
whether a portfolio company was part of a rollup in some of our tests. We were able to
determine the status of every portfolio company in our sample as of Sept. 30, 2009. The
public debt outstanding during the time they were held by the buyout syndicate. For each
of these companies, we obtain data on Sales, total assets, EBITDA, special dividends, and
capital expenditures for the year of the buyout, the year prior, and the two years after the
buyout took place. Three hundred eighty eight portfolio companies had public debt and
financial statement data available on Capital IQ. This sample does not perfectly overlap
with our main sample because in order to fully track exits, our main sample consists only
of buyouts with transaction dates before the end of 2001. Because we do not need exit
data for our tests of while-private behavior, and because Capital IQ has better data on
portfolio firms later in our sample period, we do not impose this restriction on the public
debt sample.
In addition, we construct for our sample of portfolio companies with public debt a
“standalone” if all of its business and operating segments in the Computstat segments file
fall within the same four-digit NAICS code, or if the firm does not report business or
operating segment results. We match each portfolio company to the standalone public
firm in the same 4-digit NACIS industry that is also closest in terms total sales as of the
year before the buyout. In addition, as a proxy for investment opportunities within the
9
industry for each year, we compute the median ratio of market equity plus total liabilities
to total assets for all standalone public firms in a given industry in a give year, and label
it as indQ.
In order to test the wealth transfer hypothesis on buyouts exited via sale to
strategic buyers, we obtain data on each publicly traded strategic buyer around the time
of the buyout exit. If the strategic buyer is listed on a US exchange at the time of the exit,
we use Compustat data to compute the strategic acquirer’s ratio of market value of equity
plus book liabilities to total assets as of the most recent quarter end. We label this
variable Q. Using the CRSP daily file, we compute the cumulative abnormal return to
the strategic buyer’s stock during the three day window around the acquisition
announcement and label it ret. We define abnormal return on a given day as the
acquirer’s return less that of its CRSP size decile portfolio. Also using the CRSP daily
file, we determine the acquirer’s market capitalization as of 2 days prior to the deal and
label it size. We define relsize as ratio of the exit transaction value (taken either from
SDC, Capital IQ, or news stories) to size. If we are unable to obtain the transaction
value, it indicates that the transaction was not material from the acquirer’s point of view,
so in these cases we set relsize equal to 1st percentile of the sample of strategic M&A exit
transactions for which the transaction value is available. We define an indicator variable,
stock, which equals 1 if 50% or more of the acquisition was paid for in stock, and zero
otherwise. We use SDC, Capital IQ, or news stories to determine the acquisition
statement and stock return data, in dollars, from Datastream. Of the 287 strategic buyers
in our sample, 191 are public, and we are able to obtain financial statement and stock
10
return data for all of them. Descriptive statistics for our sample of strategic buyers’ exit
In addition, for each strategic acquirer, we obtain governance data from Risk-
Metrics/IRRC as of the year of the exit event. We determine the board’s size and the
equals 1 if a majority of directors are independent and zero otherwise. We also obtain
information on whether the strategic buyer has any of the following six anti-takeover
provisions: poison pill, staggered board, CEO golden parachute, and supermajority
count the number of the above provisions that a strategic buyer has in effect at the time of
the acquisition, and label it as its entrenchment index, or Eindex, which can take integer
values between 0 and 6. If a strategic buyer is not covered by IRRC, we collect the same
data by hand from SEC filings, or, if the firm is not listed in the US, from foreign filings,
corporate websites, and Bloomberg. We were able to obtain these data for all public
Finally, we obtain time series data on each portfolio company’s industry. Using
the COMPUSTAT segments database, we obtain data on sales, operating income, and
total assets for each non-financial business segment (or operating segment, if business
segments are unavailable) for each firm and year, aggregating segments within firms by
four digit NAICS codes. If a firm is not in the segments file, we assume it operates in
only one segment, whose industry corresponds to its NAICS, and treat its consolidated
financial data as if it were segment data. We then use segment sales to compute a
Herfindahl index for each industry-year as a proxy for industry concentration, and label it
11
herf. For each segment-year, we also compute sales growth and take the median by
industry to obtain industry salesgrowth. Finally, we do the same for return on assets,
defined as segment operating income divided by beginning-of-year total assets, and label
it roa. Finally, for each portfolio company, we take the time series average of the
industry variables starting two years prior and ending two years after the buyout. We
were able to obtain industry data on 787 of 788 portfolio companies in our sample.
III. Results
Panel A of Table I shows the pre-transaction status of the buyouts in our sample.
The largest category, “other private,” which accounts for nearly 40% of our sample,
includes portfolio companies that were bought from owners who were neither private
equity sponsors nor any business entity. Nearly 35% of the portfolio companies were
approximately 8%, were owned by other private equity sponsors prior to the buyout. In
our sample, buyouts of public companies are generally bigger than the other categories,
but not by much. The mean size of a buyout of a public company is $477 million,
whereas it is just over $334 million in the next largest category, in which buyout target
well as the mean time to exit for each category. For portfolio companies not exited, or
“still held,” we simply compute the number of years between the announcement of the
12
buyout and Sept. 30, 2009, the last date for which we have information. One striking
characteristic of the exits is that about 90% of all exits are not through IPO. Thus, while
we have learned much by studying the sample of buyouts that go public again, “Reverse
LBOs,” we are basing most of our inferences about buyouts on about 10% of the
population. The largest exit type is a sale to a strategic buyer, followed by sale to another
restructuring actions account for around 14% of all exits. IPO’s are, in point of fact, the
least common form of exit, despite being the most-studied in the literature. The fastest
exits are through IPO, followed by strategic buyer. Portfolio companies exited via sale to
a financial buyer are held longer than other successful exists, supporting the conjecture
that private equity sponsors only sell to each other when they have not been successful
with arranging another exit. Panel C of Table I provides information on the operating
B. Do Private Equity Sponsors Pay Themselves Special Dividends and Cripple the
Portfolio Firm?
Despite their portrayal in the popular press, special dividends to private equity sponsors
are extremely rare. For sample firms with public debt, we can observe the financials of
the portfolio firms. We find only 42 instances of special dividends out of 2435 firm-
years. Table II presents the profitability and ROA for the portfolio firms with public debt,
breaking the sample out by whether the portfolio firm had a special dividend. The table
demonstrates that there is no difference between the subsamples and no evidence that the
special dividend firms are constrained or in distress. In the next section, when we
13
investigate the investment-cash flow sensitivity of portfolio firms, we test whether firms
One way for private equity sponsors to add value is through improved governance
that reduces agency problems and overinvestment. Conversely, PE sponsors could induce
offers the chance to distinguish between the two predictions. In Table IV we present the
before to after the buyout transaction. In order to make a statement about the optimality
of their sensitivities, we match each private equity portfolio company to the standalone
public firm in its 4-digit NAICS industry that is closest in pre-transaction sales.
investment activity of portfolio companies that happen have public debt, as well as their
matched control sample. Both the portfolio and control firms are profitable on average
with EBITDA/Assets of around 15%. The control firms tend to have higher capital
expenditures than the portfolio firms even before the transaction. Post-buyout, the
Next, we examine the sensitivity of their investment to cash flow pre- and post-
estimate the following panel regression using OLS with firm fixed-effects:
14
+ −
CAPX EBITDA EBITDA
= α + β1Q + β 2 + β3 + β 4 PostBuyout
Assets Assets Assets
+ −
EBITDA EBITDA
β5 PostBuyout * + β 6 PostBuyout * +ε (1)
Assets Assets
Where Assets are total book assets as of the beginning of the fiscal year, CAPX is total
capital expenditures over the fiscal year, and Postbuyout is a dummy variable that equals
1 for each private equity portfolio company while it is private and is zero during the pre-
buyout years. It is zero for all firm-years for the matched control firms. We use earnings
before interest, taxes, depreciation and amortization (EBITDA) as our proxy for cash
flow. We partition EBITDA according to whether it is positive and negative cash flow
because firms typically have an asymmetric response to the two. Specifically, EBITDA+
and EBITDA− are set to equal EBITDA when the latter is, respectively, positive and
negative. Finally, as is standard in this literature, we control for industry q to reduce the
possibility that cash flows are proxying for investment opportunities. We focus on the
The results, presented in Table IV, demonstrate that private equity sponsors
reduce overinvestment in their portfolio companies. The value of the constant term of
0.045 indicates that there is a baseline capital expenditure rate of about 4.5% of assets.
The Post-buyout dummy taking a value statistically indistinguishable from zero indicates
that this rate does not change following the buyout. In addition to the baseline rate, firms
invest 17 cents per dollar of positive operating cash flow (EBITDA), as indicated by the
The point estimate is approximately equal and opposite for negative EBITDA, indicating
that firms on average continue to increase investment even when they are generating
15
negative cash flows. Turning to the interactions of PostBuyout with the cash flow
variables, we see that the interaction for positive cash flow is negative and marginally
insignificant. The coefficient is estimated with considerable error because there are very
few cases of negative operating cash flows in the first few years following and buyout.
We estimate a second specification that includes a dummy variable for firms that
pay special dividends as well as an interaction for those firms with positive EBITDA
realizations (there are no negative EBITDA realizations for those firms, so that variable is
dropped due to collinearity). The results show that neither variable is significant.
Portfolio firms that pay special dividends do not invest differently from those that do not.
Table V presents summary statistics for the acquirer’s cumulative abnormal return
during the 3-day window around the announcement of the sale of private equity portfolio
companies to pubic strategic acquirers. Also included in this table are various control
hypothesis that private equity sponsors take advantage of strategic buyers, these
purchases of portfolio companies are generally seen as value-increasing for the buyer: the
average CAR is 2.1% and the median is 1.1%, and both are statistically significant at the
5% level. Thus, on average, there is no evidence to support the claim that private equity
sponsors strip portfolio firms and then unload them on naïve strategic acquirers willing to
16
overpay. Rather the descriptive statistics suggest these deals are good for the acquiring
firm.
Where mktcap is the acquirer’ market capitalization as of two days prior to the deal.
relsize is the ratio of the transaction size to the acquirer market capitalization, stock is a
dummy variable indicating that acquirer stock was included in the consideration paid, and
Q is the ratio of the acquirer’s market value of equity plus liabilities to book assets.
Timeheld is the number of years the portfolio company was held by the private equity
sponsor prior to the sale. We control for the acquirer’s market capitalization, Q and
acquisition currency because prior literature has found all to be important. We include
Timeheld because some have argued that private equity sponsors are motivated to unload
portfolio companies after they have held them for a long time. We are, however,
particularly interested in relsize, the relative size of the portfolio company to the acquirer.
If these acquisitions are generally good for the acquirer, we expect the value creation, as
measured by CAR, to be directly proportional to relsize. If, on the other hand, private
equity sponsors profit at the expense of strategic acquirers, then we expect the coefficient
on relsize to be negative.
Our results are in column (1) of Table VI. The coefficient on relsize is strongly
positive and significant at the 1% level, suggesting that strategic buyers benefit
substantially when they purchase portfolio companies from private equity sponsors. The
17
together with the constant term of 0.033 implies that if a strategic acquirer purchases a
portfolio company half its size, it will increase shareholder value by 5%. Consistent with
prior research, the coefficient on stock is negative and significant. However, the rest of
In addition, we find, consistent with other studies (e.g. Masulis, Wang and Xie,
above, except that we include the acquirer’s board size (ndirectors), a dummy indicating
managerial entrenchment index (Eindex) discussed above. We interact each with relsize
since the effect on CAR should be related to the relative size of the acquisition target. As
can be seen in column (2) of Table VI, the coefficient on the interaction between relsize
and ind_board is positive and significant at the 1% level. Hence an independent board is
critical to the deal being value-increasing for the acquiring shareholders. Given that the
average deal is positive, this suggests that a minority of sales may be the result of a
private equity sponsor taking advantage of poorly governed public firms. However, this
also shows that the majority of deals, involving well-governed public firms, create even
greater value than indicated by the mean CAR. The evidence is that strategic buyers do
In the two specifications above, we keep strategic acquisitions in the sample for
which the transaction value was not disclosed. While we cannot compute relsize in these
cases, we know that it must be small relative to the acquirer, since lack of disclosure
implies the transaction was deemed to be immaterial for financial reporting purposes.
Thus, in the cases of non-disclosure, in the above specifications we set relsize equal to
18
first percentile in the sample of deals for which a transaction value was disclosed (see
Table V). As a robustness check, we re-estimate the above specifications dropping deals
for which the transaction value was not disclosed, and we report the results in columns
(Loughran and Vijh, 1997; Rau and Vermaelen, 1998). Thus, it is possible that the
announcement returns do not present the full picture of the effect of purchasing the
portfolio firm. In this section we test the hypothesis that public acquirers of portfolio
and short the buyers of public companies.3 Portfolios are rebalanced monthly, so an
acquirer is added to the portfolio during the calendar month after the acquisition took
place and is kept in the portfolio for 36 months. We equally weight the portfolios, but
our results are robust to market-cap weighting. After constructing a time series of returns
for the portfolios of strategic acquirers and public acquirers, we subtract the risk-free rate,
the one-month t-bill return over that month. We obtain the French-French factors from
We present two sets of results in Table VII. The first regresses the returns of the
long portfolio of strategic buyers of private equity portfolio companies on the Fama-
French factors. The long portfolio’s alpha is negative, but insignificant. The second
3
In the present draft, for ease of computation, we limit this analysis to domestic acquirers and foreign
acquirers that are either listed on a US exchange or have ADR’s. In future drafts we will expand this
analysis to include all foreign acquirers.
19
regression uses the returns of the long-short portfolio, that is, the return of the portfolio of
acquirers of private equity buyout targets less the return on the portfolio of all acquirers
of public companies. We see that the portfolio is essentially market-neutral and has a
very small and insignificantly positive alpha. We conclude that there is no evidence that
public acquirers of portfolio firms underperform in the long run. Combined with the
positive announcement effect, the evidence establishes that these purchases are value-
Given the high returns private equity sponsors are believed to have, the fact that
about 30% of all buyouts are exited via a sale to another private equity sponsor seems
the one hand, it is plausible that some firms are just better suited to be portfolio
companies of private equity sponsors, so the most efficient exit for such an investment is
simply to sell to another sponsor. If this explanation is valid, then we expect portfolio
companies exited via secondary buyout to have been predominantly owned by private
equity sponsors pre-buyout. On the other hand, it is possible that sale to a financial buyer
is a last resort. That is, private equity sponsors only sell to financial buyers if they cannot
exit an investment quickly through other means. If this explanation is true, we expect the
time held to be the main factor influencing exit via secondary buyout. It is also possible
20
⎡α + β1 prefund + β 2 presub + β3 preotherpri + β 4 log(lbovalue) + ⎤
P(outcome) = Λ ⎢
β5 roa + β 6 sgrowth + β 7 herf + β8TimeHeld ⎥⎦
(3)
⎣
outcome, is a multinomial categorical variable that can take one of five values
corresponding to the exit outcome: sale to financial buyer, sale to strategic buyer, IPO,
restructuring, or still held. The variables prefund, presub, and preotherpri are dummy
variables indicating that, prior to the current buyout, the portfolio company was,
or high net worth individuals. The variables roa and sgrowth are the median return on
assets and sales growth in the portfolio firm’s industry during the first two full calendar
years it was held by the private equity sponsor, and herf is a measure of industry
concentration during that time. TimeHeld is the number of years the portfolio company
was held by the sponsor prior to exit. We set IPO exit as the base outcome, so our
coefficient estimates are measures of how the above variables affect the odds of a given
exit outcome, such as sale to a financial buyer, relative to the odds of an IPO. If certain
firms are just better suited to be in private equity sponsor portfolios, then we would
expect the coefficient on prefund to be significantly more positive for financial buyer
exits than other exits. On the other hand, if private equity sponsors sell to other sponsors
coefficient TimeHeld to be more negative for portfolio companies exited via sales to
financial buyers than other types of exits. If poor performance is driving secondary
buyouts, then we expect the coefficients on industry ROA and salesgrowth to be strongest
21
First note that, for firms exited via sale to a financial buyer, the coefficient on
indistinguishable from its value in the strategic buyer column. This result implies that
having been previously held by a private equity sponsor does not make it more likely that
a portfolio company investment will be exited via sale to financial buyer than IPO or sale
to strategic buyer. However, the coefficient in question is statistically larger for the
financial buyer column than it is for still held and restructured. Taken together, all the
results imply that being previously held by a financial buyer does increase the odds of the
three favorable exits, namely financial buyer, strategic buyer, or IPO, but it does not
Note further that the coefficient on TimeHeld is positive and significant in the
financial buyer column. It is also statistically larger for financial buyer column than for
the strategic buyer column. This implies that when a portfolio company has been held
longer, it is more likely to be exited via sale to financial buyer than via IPO or sale to
strategic buyer. This result supports the hypothesis that private equity sponsors sell to
other sponsors when a timely exit via IPO or strategic buyer is impossible. However, the
effect of roa and salesgrowth is not significant, so poor portfolio company performance
On the other hand, we find strong evidence that industry concentration has a large
effect on keeping portfolio companies private. We draw this inference from the
coefficient on herf, which is positive and significant at the 10% level in the financial
buyer column and significant at the 1% level in the still held column. In addition, in both
financial buyer and still held columns, it is statistically larger than in it is in the strategic
22
buyer column. Economic theory provides several motivations for why it may be superior
to keep monopolistic firms private. SEC disclosure could plausibly give competitors
information that can aid them in entering the market. Alternatively, private firms
generally attract less attention from the public at large and, by plausible extension, the
depth.
IV Conclusions
Using a comprehensive sample 788 of large private equity buyouts from 1993 to
2001 that includes all exits, not just by IPO, we conclude that private equity sponsors
create value. They take the firms private in a transaction that pays a premium to existing
shareholders. While private, the portfolio firms do not underinvest and if anything,
overinvest less, suggesting that private equity sponsors reduce agency problems, but do
not cut back investment too far. Further, special dividends are rare, and are uncorrelated
with portfolio company distress. The typical exit is by sale to either a strategic or
financial buyer. Exits to strategic buyers create value on average, particularly when the
buyer is well-governed. Coupled with the extant evidence examining the 10% of buyouts
that exit by IPO, our evidence is strongly in favor of wealth creation rather than wealth-
exit time. We find that strong industry sales growth helps predict exit via IPO. We also
find that firms exited via sale to financial buyer tend to have been held longer, suggesting
that private equity sponsors tend to sell to other sponsors when a timely exit of another
23
sort was not possible. On the other hand, we find that operating in a concentrated
industry is a strong predictor of the portfolio company staying private, that is, either
being sold to another private equity sponsor or not being exited at all. This latter result
suggests that private equity sponsors may add value by allowing monopolistic firms to
stay private, thus helping them avoid disclosures that can aid potential competitors and
avoid the scrutiny of anti-trust authorities by keeping out of the public eye.
24
References
Boulcly, Q., D. Srarer and D. Thesmar, 2008, Do leveraged buyouts appropriate worker
rents? Evidence from French data. Working paper, HEC Paris.
Bergstrom, Grub, and Jonsson, 2007, The operating impact of buyouts on Sweden: a
study in value creation, Journal of Private Equity 11, 22-39.
Cao, and J. Lerner., 2009, The performance of reverse leveraged buyouts, Journal of
Financial Economics 91, 139-157.
Degeorge, Z. and R. Zeckhauser, 1993. The reverse LBO decision and firm performance:
theory and evidence. Journal of Finance 48, 1323-1348.
Der Spiegel, 2006. The locusts. Staff writer, March 22, 2006.
Forbes, 2006. Private inequity. Weinberg, N. and N. Vardi, March 13, 2006.
Holthhausen, R., and D. Larcker, 1996. The financial performance of reverse leveraged
buyouts. Journal of Financial Economics 42, 293-332.
Mian, S. and J. Rosenfeld, 1993. Takeover activity and the long-run performance of
reverse leveraged buyouts. Financial Management 22, 46-57.
Guo, S., E. Hotchkiss and W. Song, 2010, Journal of Financial Economics, Forthcoming.
Harris, R., D. Siegel and M. Wright, 2005, Assessing the impact of management buyouts
on economic efficiency: plant-level evidence from the United Kingdom, The
Review of Economics and Statistics 87, 148-153.
Kaplan, S., 1989, The effects of management buyouts on operating performance and
value, Journal of Financial Economics 24, 217-254.
Kaplan, S. and A. Schoar, 2005, Private equity returns: persistence and capital flows,
Journal of Finance 60, 1791-1823.
Kaplan, S. and P. Stromberg, 2009, Leveraged buyouts and private equity, Journal of
Economic Perspectives 23, 121-146.
Lichtenber, F. and D. Siegel, 1990, The effects of leveraged buyouts on productivity and
related aspects of firm behavior, Journal of Financial Economics, 27 165-194.
25
Masulis, R., C. Wang and F. Xie, 2007, Corporate governance and acquirer return,
Journal of Finance 62, 1851-1889.
Masulis, R. and R. Nahata, 2010, Venture Capital Conflicts of Interest: Evidence from
Acquisitions of Venture Backed Firms, forthcoming in Journal of Financial and
Quantitative Analysis.
Smith, A., 1990, Capital ownership structure and performance: the case of management
buyouts, Journal of Financial Economics 27, 143-521.
Times of London, 2006, No silverware and Buffet’s buffet. Staff writer, May 8, 2006.
26
Table I
Panel A Panel B
Pre-LBO Status of Portfolio Companies Distribution of LBO Exit Outcomes
% Size % Mean
# Deals Frequency ($MM) # Deals Frequency Years Held
Portfolio of other sponsor 66 8.38% 254.65 IPO 77 9.77% 3.70
Operating subsidary 272 34.52% 334.02 Strategic Buyer 287 36.42% 4.26
Other private owner 300 38.07% 260.43 Financial Buyer 239 30.33% 5.44
Bankrupt 9 1.14% 307.13 Restructured 117 14.85% 4.85
Public 141 17.89% 477.73 Still Held 68 8.63% 10.26
Panel C
Descriptive Statistics on Portfolio Company Industry
Variable N Mean Std Dev Minimum 1st Pctl 25th Pctl 50th Pctl 75th Pctl 99th Pctl Max
roa 787 4.98% 12.49% -129.88% -37.55% 4.71% 7.07% 9.37% 13.83% 55%
Concentration 787 0.160 0.141 0.019 0.024 0.068 0.106 0.214 0.747 1
sales growth 787 11.62% 20.38% -10.73% -1.90% 6.03% 8.86% 14.04% 47.71% 365%
27
Table II
This table presents pre- and post-buyout industry-adjusted capital expenditures, operating
margins and ROA for portfolio firms with public debt. We subtract the 4-digit NAICS industry
median in order to make the industry adjustment. “Chang in” variables are computed as the firm-
specific difference in the variable for the post-buyout years (up to 3) compared to year -1 relative
to the buyout. None of the “Change In” variables are statistically different between the subsample
that paid a special dividend and the subsample that did not.
Mean Median Std Dev 25th pctl. 75th pctl.
All Firms
Pre-buyout Ind-Adjusted CapEx 0.018 0.004 0.081 -0.014 0.029
Pre-buyout Ind-Adjusted Op. Margin 0.285 0.186 3.216 0.044 1.000
Pre-buyout Ind-Adjusted ROA 0.117 0.076 0.304 0.022 0.145
28
Table III
Descriptive statistics for portfolio companies with public debt & the control group
Each portfolio company with public debt is matched to a public firm in the same 4-digit
NAICs industry with the closest pre-transaction sales. The summary statistics for the
variables used in the investment-to-cash flow sensitivity regression in Table IV are
presented here.
25th `75th
Variable Mean Median Std Dev Percentile Percentile
Portfolio firms
CapEx/Assets 0.061 0.038 0.080 0.021 0.071
EBITDA/Assets 0.151 0.136 0.150 0.095 0.191
Industry Median q 1.827 1.648 1.121 1.306 2.065
Control Firms
CapEx/Assets 0.090 0.053 0.140 0.026 0.106
EBITDA/Assets 0.161 0.163 0.385 0.089 0.255
Industry Median q 1.777 1.626 0.748 1.304 2.061
29
Table IV
Investment-cashflow sensitivity regressions
The dependent variable is capital expenditures scaled by assets. The sample consists of
all LBO firms that have public debt, so that we can track their investment while private,
and matching firms chosen for each buyout firm as the public firm in its 4-digit NAICs
industry closest in pre-transaction sales. Industry median q is calculated as the median
market-to-book of assets for the firm’s 4-digit NAICS industry that year. EBITDA/Assets
(Negative, Positive) are EBITDA scaled by assets if negative and if positive,
respectively. Post-buyout is a dummy equal to 1 for buyout firms while it is private and is
zero otherwise. Post-buyout is also interacted with the EBITDA variables. Paid Special
Dividend is a dummy variable set equal to 1 if the firm paid a special dividend while
private and is zero otherwise. The regression is estimated with fixed firm effects and the
standard errors are reported in parentheses. ***, **, and * indicate significance at the 1, 5
and 10% levels.
30
Table V
Descriptive statistics on strategic acquisitions & acquirers
The sample consists of all acquisitions of private equity portfolio companies by public strategic acquirers, defined as
firms not in the business of buyouts. CAR is the cumulative abnormal return on the acquirer’s stock price during the 3-
day window around the acquisition announcement date. Other variables are as follows: relsize is the ratio of the
transaction value to the acquirer’s equity market capitalization two days prior to the acquisition; stock is a dummy
indicating that acquirer stock was used as consideration; ndirectors are the number of directors on the acquirer’s board
before the acquisition; ind_board is an indicator variable for a majority of independent directors; eindex is the
managerial entrenchment index that counts the following antitakeover provisions for the acquirer: poison pill, staggered
board, CEO golden parachute, and super majority requirements for mergers, charter amendments or bylaw
amendments; Q is the ratio of the acquirer’s equity market capitalization plus liabilities to total assets; TimeHeld is the
length of time the acquisition target was held as private equity portfolio company prior to the acquisition; mktcap is the
acquirers equity market capitalization as of two days prior to the acquisition.
n mean stddev min 1st pctl 25th pctl Median 75th pctl 99th pctl max
CAR 191 0.021 0.077 -0.203 -0.159 -0.012 0.010 0.047 0.386 0.429
relsize 157 0.500 0.860 0.002 0.002 0.059 0.228 0.550 3.034 8.266
stock 191 0.188 0.392 0.000 0.000 0.000 0.000 0.000 1.000 1.000
ndirectors 191 9.874 3.208 3.000 4.000 8.000 9.000 12.000 20.000 21.000
ind_board 191 0.717 0.452 0.000 0.000 0.000 1.000 1.000 1.000 1.000
EINDEX 191 1.848 1.488 0.000 0.000 1.000 2.000 3.000 6.000 6.000
Q 191 2.470 2.657 0.000 0.487 1.357 1.852 2.878 12.715 30.848
TimeHeld 191 3.950 2.207 0.515 0.559 2.192 3.600 5.463 10.603 10.767
log(mktcap) 191 7.809 1.956 2.184 2.890 6.689 7.567 9.083 12.810 13.139
31
Table VI
Regression analysis of strategic buyer announcement returns
The dependent variable is the cumulative abnormal return to the strategic buyer
acquiring an private equity portfolio company during the 3-day window around
the acquisition announcement date. Independent variables are defined in Table IV
above. Columns (1) and (2) are run on a sample of all strategic acquisitions,
including those for which a transaction value was not disclosed. In the latter case,
the 1st percentile of relsize was imputed. Columns (3) and (4) are run on a sample
that exclude acquisitions for which the transaction value was not disclosed.
(1) (2) (3) (4)
relsize 0.037*** 0.043 0.037*** 0.043
(0.007) (0.037) (0.008) (0.040)
log(mktcap) -0.003 -0.007* -0.003 -0.008*
(0.003) (0.003) (0.004) (0.004)
stock -0.039*** -0.044*** -0.038*** -0.045***
(0.013) (0.013) (0.014) (0.014)
Q -0.003 -0.002 -0.003 -0.002
(0.002) (0.002) (0.002) (0.002)
TimeHeld 0.003 0.003 0.004 0.003
(0.002) (0.002) (0.003) (0.003)
ndirectors 0.003 0.004
(0.002) (0.003)
ind_board -0.012 -0.016
(0.013) (0.017)
eindex 0.001 0.001
(0.004) (0.005)
relsizexind_dum 0.057** 0.061**
(0.023) (0.026)
relsizexbn -0.004 -0.005
(0.003) (0.003)
relsizexeindex -0.009 -0.010
(0.007) (0.008)
Constant 0.033 0.036 0.030 0.041
(0.026) (0.030) (0.033) (0.039)
Observations 191 191 157 157
R-squared 0.24 0.28 0.26 0.30
Standard errors in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1
32
Table VII
Analysis of the strategic buyer portfolio returns
Column (1) below regresses the excess returns of the equal-weighted portfolio of
strategic acquirers of private equity buyout targets on the Fama-French factors.
Column (2) contains an analysis of a long-short portfolio, where the dependent
variable is the return on the portfolio of strategic acquirers of private equity
buyout targets less the return on the portfolio that contains all acquirers of public
companies. Acquirers are added to their respective portfolio in the calendar
month immediately following the acquisition and dropped 36 months later.
(1) (2)
alpha -0.0046 0.0018
(0.0031) (0.0037)
MktRF 0.9562*** -0.2748***
(0.0680) (0.0821)
HML 0.4639*** 0.0340
(0.0944) (0.1153)
SMB 0.2781*** -0.2813
(0.0875) (0.1057)
Obs 171 171
Rsquared 0.57 0.14
33
Table VIII
Multinomial logistic regression analysis wherein we model the probability of five
different buyout exit types: IPO, sale to a strategic buyer, sale to financial buyer,
distressed restructuring, or still held by the sponsor as of Sept. 30, 2009. IPO is the
reference exit type, so the parameter estimates in the table constitute the respective
variable’s marginal effect on the log-odds of the exit type indicated at the top of the
column relative to that of an IPO. The variables prefund, preotherpri, and presub are
dummies indicating that the portfolio company, before the buyout, was, respectively,
owned by a private equity fund, held by some other private owner, or was an operating
subsidiary of a parent company that does not conduct buyouts in the ordinary course of
business. The variable lbovalue is the enterprise value of the buyout in question. The
variables roa and salesgrowth are the cross-sectional median return on assets and sales
growth of the portfolio company’s 4-digit industry, averaged over the time period
beginning two years prior to the buyout and ending two years afterward. The variable
herf is the herfindhal index of the industry averaged over the same time period.
Compustat segments data were used to compute all industry variables.
Exit type
Financial Strategic
Restructured Still Held Buyer Buyer
prefund -0.212 0.585 0.913 1.131*
(0.681) (0.897) (0.625) (0.614)
preotherpri -0.235 0.178 0.593 0.820**
(0.394) (0.546) (0.384) (0.379)
presub -1.110*** -0.366 0.254 0.676*
(0.397) (0.560) (0.364) (0.355)
log(lbovalue) -0.229 -0.222 -0.358** -0.361***
(0.154) (0.209) (0.142) (0.137)
roa -4.107 -5.430** -3.856 -4.257
(2.681) (2.696) (2.612) (2.596)
salesgrowth -0.264 -2.255 -1.161 -1.914*
(0.754) (2.046) (0.860) (1.135)
herf 1.407 3.949*** 2.014* 0.885
(1.260) (1.455) (1.171) (1.171)
TimeHeld 0.213*** 0.901*** 0.291*** 0.115*
(0.065) (0.092) (0.060) (0.060)
Observations 787 787 787 787
Standard errors in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1%
34