the dual tracking puzzle

Upload: cnabity1

Post on 06-Apr-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 The Dual Tracking Puzzle

    1/44Electronic copy of this paper is available at: http://ssrn.com/abstract=899983

    The Dual Tracking Puzzle: When IPO Plans Turn into Mergers

    Qin Lian*University of Alabama

    Department of Economics, Finance & Legal StudiesTuscaloosa, AL 35487-0224

    Phone: 218-341-0217E-mail: [email protected]

    Qiming WangFMIS Department, SBE 150

    Labovitz School of Business & EconomicsUniversity of Minnesota Duluth

    412 Library DriveDuluth, MN 55812

    Phone: 218-726-7083Fax: 218-726-7517

    E-mail: [email protected]

    Current Version, March, 2007First Draft, January, 2005

    Keywords: Dual Tracking, IPO Withdrawal, Acquisition Discount, CARs

    JEL Classification: G14, G34, G39

    *Corresponding author. The authors thank Anup Agrawal, James Ligon, HarrisSchlesinger, and participants at the 2005 Financial Management Association International(FMA) meeting and the 2006 European Finance Association (EFA) meeting for theirhelpful comments.

  • 8/3/2019 The Dual Tracking Puzzle

    2/44Electronic copy of this paper is available at: http://ssrn.com/abstract=899983

    1

    The Dual Tracking Puzzle: When IPO Plans Turn into Mergers

    ABSTRACT

    We examine a sample of 132 dual tracking targets private firms entertaining acquisition

    offers at the same time as preparing for initial public offerings (IPOs) and eventually

    withdrawing their IPOs to be acquired after spending considerable time, money, and

    effort preparing for IPOs. We find that dual tracking private targets sell at a 58 percent

    acquisition premium relative to comparable private targets that never file IPO

    registrations, while their acquirers still earn a substantial average abnormal

    announcement return of 2.6 percent. Controlling for endogeneity effects does not change

    our results. The significant acquisition premium is due to neither dual tracking targets

    improved bargaining power in negotiations nor higher potential synergistic benefits for

    bidders. The premium is more consistent with the explanation that dual tracking private

    targets can signal their valuation to bidders and reduce valuation uncertainty by filing

    IPO registrations.

  • 8/3/2019 The Dual Tracking Puzzle

    3/44

    2

    Investors in a private firm can cash out by either selling shares to investors in the initial

    public offering (IPO) market or selling the firm to a public firm in the merger and

    acquisition (M&A) market. By investigating whether an IPO or a takeover is eventually

    used by a private firm to turn public, Brau, Francis, and Kohers (2003) and Poulsen and

    Stegemoller (2007) suggest that private firms choose the IPO market instead of the M&A

    market as the pathway to access the public equity market when they 1) have higher

    growth opportunities and more capital constraint, 2) are easier to value, and 3) are in a

    relatively hot IPO market.

    However, by focusing only on which route is chosen eventually by a private firm, this

    literature overlooks possible connections between IPO and M&A markets for private

    firms planning to sell their assets in public equity markets. While most private firms do

    decide whether an IPO or a takeover is the appropriate pathway to public ownership from

    the get-go and pursue the strategy from the beginning, a subgroup of private firms,

    namely dual tracking firms, seems to pursue transactions in both IPO and M&A

    markets simultaneously.1 A dual tracking private firm files an IPO registration while also

    exploring opportunities to be acquired. The firm either successfully completes its IPO or

    withdraws from the IPO registration to be acquired. However, the firm rarely declares its

    intention to dual track both IPO and M&A markets. Due to limited disclosure, we cannot

    identify a dual tracking firm that succeeds in its IPO instead of being acquired or a dual

    tracking firm that is acquired by a private firm. Therefore, in this paper, we focus on dual

    tracking target firms withdrawn-IPO firms that are acquired by a public firm after

    1 Zingales (1995) and Pagano, Pannetta, Zingales (1998) posit that an IPO is the initial step for

    entrepreneurs to eventually sell their assets in takeovers.

  • 8/3/2019 The Dual Tracking Puzzle

    4/44

    3

    withdrawing IPO registrations to study how IPO registrations affect their acquisition

    valuations.

    The choice to pursue dual tracking is puzzling. The IPO registration process is costly.

    Lee, Lochhead, Ritter, and Zhao (1996) document direct expenses (registration fee and

    printing; legal and auditing costs) of IPO registration at 3.69 percent of expected

    proceeds for IPOs from 1990 to 1994. Given that the median dual tracking private firm in

    our sample plans to raise $44 million in their IPO filings, the IPO registration cost is

    estimated to be $1.6 million. When a dual tracking private firm withdraws its IPO to be

    acquired, it essentially forfeits all the time, money, and effort spent on the registration

    process. So the ultimate question is, if a private firm sells itself to a public firm via a

    takeover, why does it file for an IPO and endure the additional costs of the IPO

    registration process in the first place?

    As the first attempt to shed light on the dual tracking puzzle, this paper investigates

    targets acquisition valuations and acquirers announcement returns for a sample of 132

    dual tracking private firms between 1984 and 2004.

    First, we examine the effects of filing and withdrawing IPO registrations on private

    targets valuations in the M&A market. We compare the valuations of dual tracking

    private targets to those of three control samples: 1) pure private targets, 2) newly public

    targets, and 3) established public targets.

    Consistent with previous studies, we find that private targets sell at an acquisition

    discount of 10-20 percent relative to similar public targets. However, our results suggest

    that the acquisition discount of private targets mainly reflects the discount of pure private

    targets relative to both newly public and established public targets. We find that dual

  • 8/3/2019 The Dual Tracking Puzzle

    5/44

    4

    tracking private targets receive valuation multiples comparable to those received by

    newly public targets. They also receive significantly higher valuation multiples than those

    received by the matched sample of established public targets.

    Most importantly, we find that, compared to pure private targets, dual tracking private

    targets sell at a significant acquisition valuation premium of 58 percent even after

    controlling for several valuation-related firm attributes. This result remains, after

    accounting for private firms self selection of the dual tracking path over a straight sale.

    Our results suggest that dual tracking private targets are willing to endure the additional

    costs of IPO registration because they will likely receive higher acquisition valuations

    than they otherwise would in takeovers.

    Then, the next question is how investors react in situations when buyers are willing to

    pay significant acquisition premiums relative to prices paid to acquire similar pure private

    targets. To answer this question, we compare average announcement period cumulative

    abnormal returns (CARs) for public firms that acquire dual tracking private targets to

    acquirers announcement returns for acquisitions of pure private targets. Recent studies

    show that acquirers experience positive average CARs when acquiring private targets and

    zero or negative average CARs when acquiring public targets. 2 Fuller, Netter and

    Stegemoller (2002) posit and Officer, Pouslen and Stegemoller (2006) show that the

    lower acquisition prices in acquisitions of private targets explain bidders superior CARs

    when acquiring private targets. Given that dual tracking private targets receive significant

    acquisition premiums relative to pure private targets and are sold at a price as if they were

    2 See Chang (1998), Fuller, Netter and Stegemoller (2002), Faccio, McConnell and Stolin (2006), Officer,

    Poulsen and Stegemoller (2006), and others.

  • 8/3/2019 The Dual Tracking Puzzle

    6/44

    5

    public, we would expect that acquirers CARs would be significantly lower when

    acquiring dual tracking private targets than when acquiring pure private targets.

    However, we find that acquirers still earn a significantly positive average CAR of 2.6

    percent around the announcement of acquisitions of dual tracking private targets and a

    significantly positive average CAR of 2.65 percent around announcement of acquisitions

    of pure private targets. This difference in CARs is statistically insignificant, even after

    controlling for the size of the acquirer, the acquisition valuation ratio, the method of

    payment, and the relative size of the target. This finding suggests that other market

    participants agree with acquirers assessment of the valuations of dual tracking private

    targets and deem the average 58 percent acquisition premium paid for such targets to be

    justified.

    Finally, we explore three explanations for the higher acquisition premiums received

    by dual tracking private targets relative to pure private targets. We find no evidence to

    support the bargaining power hypothesis that dual tracking targets use the IPO filing, thus

    increasing their visibility and generating an outside option, to improve their bargaining

    power to negotiate higher prices. The median (mean) numbers of bidders for dual

    tracking and pure private targets are same. Acquisition premiums relative to pure private

    targets for dual-tracking targets are not different between dual-tracking firms that

    withdraw IPOs before acquisition announcements and those withdrawing IPOs after

    announcements. We also find little evidence to support the argument that dual tracking

    targets receive higher acquisition premiums due to higher synergistic benefits for

    potential bidders. Information disseminated via the IPO process might lead to acquirers

    that have greater potential synergies from acquisitions. Therefore, acquirers are willing to

  • 8/3/2019 The Dual Tracking Puzzle

    7/44

    6

    pay premiums for withdrawn-IPO firms. The post-acquisition long-run stock

    performances for acquirers of dual tracking targets are no better than those for acquirers

    of pure private targets.

    Another potential explanation for the higher premium is the lower valuation

    uncertainty hypothesis. Fulfilling the SEC registration requirement reduces valuation

    uncertainty surrounding those private targets through more and better information

    disclosure. And the book-building process generates information on investors demand

    and on valuations for those targets. This argument implies that private firms that are

    difficult to value, such as firms with high growth potentials, are more likely to take the

    dual tracking route. Consistent with this implication, we find that private firms with

    higher growth rates, higher R&D costs, and lower profits are more likely to dual track

    both IPO and M&A markets. In other words, private targets with higher valuation

    uncertainty are more likely to use IPO filings, though costly, to reduce their valuation

    uncertainty and subsequently increase their acquisition prices relative to comparable pure

    private targets.

    Our results on acquisition valuation of various private and public targets contribute to

    the previous studies on acquisition discount received by private targets. Koeplin, Sarin,

    and Shapiro (1996)and Officer (2007) document that private targets are acquired at an

    average 15-30 percent discount relative to similar public targets. Our paper extends the

    previous studies by not only considering the impact of targets listing status, private or

    public, on their acquisition valuations, but also finding that by dual tracking both IPO and

    M&A markets, private targets that withdraw their IPOs command similar valuation ratios

  • 8/3/2019 The Dual Tracking Puzzle

    8/44

    7

    as public targets and significantly higher acquisition valuation ratios than other

    comparable pure private targets do.

    Our paper also contributes to the IPO withdrawal literature. Busaba (2006) and

    Busaba, Benveniste, and Guo (2001) argue that IPO withdrawals can be a valuable option

    for issuers to solicit the true information from investors in the book-building process.

    Dunbar and Foerster (2007) and Lian (2007) find that 9% of withdrawn IPOs are able to

    return to the IPO market successfully. This paper adds to the IPO withdrawals literature

    by examining dual tracking firms that are acquired after IPO withdrawals, an alternative

    of withdrawn IPOs, and provides evidence that withdrawals are valuable options for

    issuers by showing that 9% of withdrawn IPOs are acquired in M&A market shortly after

    withdrawals and at a significant premium to other pure private targets.

    The rest of the paper is organized as follows. Section I describes our data and presents

    sample descriptive statistics. Section II examines the acquisition valuation ratios, and

    Section III investigates acquirers returns around the announcement period. Section IV

    addresses concerns on selection bias, and Section V examines three explanations for the

    higher acquisition premium for dual tracking private targets. Section VI concludes.

    I. Sample Selection and Data Description

    A. Sample and Data Sources

    Our primary data sources are the Global New Issues Database and the Merger and

    Acquisition Database from the Securities Data Corporation (SDC). Our sample includes

    acquisitions over the period January 1, 1984 July 25, 2004.3

    3 We use 1984 as the beginning point because SDC started IPO withdrawal coverage that year.

  • 8/3/2019 The Dual Tracking Puzzle

    9/44

    8

    We first extract a list of IPO withdrawals, firms that file an IPO registration but later

    withdraw, from the Global New Issues Database. From the Merger and Acquisition

    Database, we construct a takeover sample that contains completed deals involving 100

    percent acquisitions of U.S. private firms by U.S. publicly traded acquirers. Using 6-digit

    CUSIPs to match the withdrawn-IPO list and the takeover list, we identify a sample of

    dual tracking private targets. After verifying the accuracy of matching, using SEC filings

    from the SEC Edgar database and the Lexis-Nexis Academic database, we obtain a

    sample of 150 dual tracking private targets that announce their acquisitions within three

    years of their initial filing dates. Following a similar procedure, we also construct a

    sample of 507 newly public targets that have completed their IPOs and announce their

    acquisitions within three years of their IPO dates.

    [INSERT TABLE I]

    Panel A of Table I reports IPO completion and withdrawal activities, as well as

    subsequent takeover activities for withdrawn IPOs and newly public firms by year.

    During the sample period, 1,652 firms (around 20 percent of all IPO filings) withdraw

    their IPOs. Of those IPO withdrawals, 150 (9 percent) are acquired by public bidders

    within three years of their IPO registrations. About two-thirds of our sample of

    acquisitions of dual tracking firms occurs during the late 1990s. Of the successful IPOs,

    507 (8 percent) are acquired within three years of their IPO dates. Panel B of Table I

    summarizes IPO and takeover activities by industry. We follow the industry

    classifications used in Busaba, Benveniste, and Guos (2001) study of IPO withdrawal

  • 8/3/2019 The Dual Tracking Puzzle

    10/44

    9

    options.4 Dual tracking targets and newly public targets are widely distributed across

    industries the service sector has the most dual tracking targets and newly public targets.

    [INSERT FIGURE 1]

    Figure 1 shows the distribution of the number of days between IPO filing and

    announcement of acquisition for dual tracking private targets and the number of days

    between IPO date and announcement of acquisition for newly public targets. In the dual

    tracking private target sample, the average (median) number of days from the IPO

    registration filing date to the takeover announcement date is 376 (294) days. However,

    the average (median) number of days for the newly public target sample is 611 (625)

    days after its IPO date. Sixty-four firms announce their acquisition before they formally

    file an IPO withdrawal form with the SEC.5 Approximately 25 percent of the targets that

    withdraw their IPOs are acquired within 115 days of their IPO registrations, but merely 5

    percent of newly public firms are acquired within 115 days of their IPO date. Dual

    tracking firms shorter average time from IPO registration to takeover suggests that a

    potential acquisition drives some of these firms to withdraw their scheduled offerings.

    Three control samples are constructed as following. For each dual tracking private

    target we find an acquisition of a comparable pure private target, newly public target, and

    established public target (a firm that has been public for more than three years).

    4 We add the financial industry (two-digit primary SIC codes: 60-69), which is excluded in Busaba,

    Benveniste, and Guo (2001).

    5 These 64 firms that are acquired before their formal IPO withdrawals truly dual track IPO and M&A

    markets simultaneously

  • 8/3/2019 The Dual Tracking Puzzle

    11/44

    10

    Comparable targets are in the same two-digit primary SIC code industry6, have the same

    acquisition payment method (stock or cash or combination of stock and cash), are

    announced within a 90 calendar-day window around the announcement of the acquisition

    of the dual tracking private target, and have the closest deal value among acquisitions that

    meet the first three conditions. We collect financial data for newly public targets and

    established targets from COMPUSTAT if available and from the SDC database otherwise.

    For dual tracking private targets, we try to obtain financial data from SEC filings by their

    acquirers7 if available and from the SDC database otherwise. We use the SDC database to

    obtain financials for pure private targets. In the final sample, we include only those dual

    tracking private targets with available deal values and financial information for the fiscal

    year preceding the takeover, and those acquirers that are covered by CRSP over the event

    period. Our final sample includes 132 dual tracking private targets, a matched sample of

    pure private targets, a matched sample of newly public targets, and a matched sample of

    established public targets.

    [INSERT TABLE II]

    6Like Officer (2007), we use this comparable industry transaction method to compare acquisition

    valuation differences between private targets and public targets, as suggested by Kaplan and Ruback (1995)

    7 Securities regulations require that public acquirers disclose financials of targets of takeovers that have

    material impact (acquisitions in which that deal value is more than 10% of the acquirers total assets) in

    their SEC filings, including S4, 8K, Proxy, or S1 filings. We obtain the financials of private targets

    acquired through the SEC Edgar, 10K Wizard, and LexisNexis databases. See Fuller, Netter and

    Stegemoller (2002) for details and we thank Mike Stegemoller for point out this data sources for private

    targets.

  • 8/3/2019 The Dual Tracking Puzzle

    12/44

    11

    B. Sample Characteristics

    Table II reports attributes for acquisitions of dual tracking private targets, pure private

    targets, newly public targets, and established public targets. Panel A presents the

    characteristics of takeover deals and acquirers. The first three rows show that: 1) deal

    value, acquirers size (measured as the market value of equity one month prior to the

    acquisition announcement), and target-to-acquirer relative size (measured as deal value

    divided by the acquirers size) for acquisitions of private targets are significantly lower

    than those for acquisitions of public targets; 2) and acquisitions of dual tracking private

    targets, on average, have a higher deal value and larger acquirer, but with lower relative

    target-to-acquirer size, than acquisitions of pure private targets do. The last two rows

    show that: 1) acquirers, on average, use more stock and less cash as acquisition

    currencies when purchasing public targets than when purchasing private targets; 2) and

    only in acquisitions of pure private targets, cash is the dominant payment method with

    fraction of cash payment at mean (median) of 50% (53%) and fraction of stock payment

    at mean (median) of 41% (20%); 3) and with fraction of cash payment at mean (median)

    of 33% (5%) and fraction of stock payment at mean (median) of 56% (76%), payment

    methods in acquisitions of dual tracking private targets seems more like that in

    acquisitions of newly public targets and established public targets.

    Panel B describes targets sales, sales growth over the takeover year, OPA (defined as

    EBITDA divided by total assets), OPS (defined as EBITDA divided by sales), leverage

    (defined as the total debt divided by total assets), and R&D (defined as research and

    development expenses divided by total assets). As shown in Panel B, dual tracking

    private targets are smaller, as measured by sales, than newly public and established

  • 8/3/2019 The Dual Tracking Puzzle

    13/44

    12

    public targets, but larger than pure private targets. Dual tracking private targets, on

    average, have significantly higher sales growth, use more financial leverage, and spend

    more on research and development (R&D) than all other comparable targets do. However,

    on average, dual tracking targets have poorer operating performance, as measured by

    OPA and OPS.8

    II. Acquisition Valuations of Dual Tracking Private Targets

    In this section, we examine the acquisition valuation difference between dual tracking

    private targets and other targets, including pure private targets, newly public targets, and

    established public targets. Section A deals with results of univariate analyses and Section

    B discusses results of multivariate tests where other determinants of valuations are

    controlled for.

    [INSERT TABLE III]

    A. Univariate Comparisons of Acquisition Multiples

    Following Officer (2007), we use three different acquisition multiples: deal value to

    sales, deal value to book value of equity, and deal value to EBITDA to compare

    acquisition valuations of different types of targets.9 Table III reports mean and median

    acquisition valuation multiples for dual tracking private targets and three samples of

    8 We also compare firm attributes of dual tracking private targets and newly public targets at the time of

    their initial IPO filings. The unreported results are similar to those presented in Panel B of Table 2.

    9 Officer (2007) uses the average of percentage differences in four acquisition valuation multiples of private

    targets relative to public targets as the acquisition discount/premium for private targets. We do not use price

    to earnings, the fourth acquisition multiple used by Officer (2007) to derive average acquisition discount,

    because including price to earnings reduces size of dual tracking private targets by more than 70%.

  • 8/3/2019 The Dual Tracking Puzzle

    14/44

    13

    comparable targets. The mean and median values of the three acquisition multiples show

    similar patterns. Because of extreme outliers in the right-hand tail for all the multiples,

    we focus on the median values in the following discussions.

    As shown in Table III, two of the three valuation multiples in acquisitions of dual

    tracking private targets are significantly higher than those in acquisitions of pure private

    targets. Acquirers typically pay $2.99 for each dollar in sales for dual tracking private

    targets. This amount is 63 percent higher than the $1.8410

    per dollar in sales paid for pure

    private targets. Acquirers, on average, pay $14.07 per dollar in book value for dual

    tracking private targets, and this figure is 58 percent higher than the $8.88 per dollar in

    book value paid for pure private targets. The deal value to EBITDA in acquisitions of

    dual tracking targets is 12.86, which is similar to the 12.79 multiple in acquisitions of

    pure private targets. Averaging the difference in these acquisition multiples, we find that

    dual tracking private targets sell at a significant 41 percent premium relative to

    comparable pure private targets in takeovers by public acquirers.

    As also shown in Table III, on average, dual tracking private targets generally receive

    similar valuation multiples compared to newly public targets, but they receive higher

    valuation multiples compared to established public targets. However, pure private targets

    sell at lower acquisition multiples compared to both newly public targets and established

    public targets. This pattern suggests that the private target acquisition discount

    documented in Koeplin, Sarin, and Shapiro (1996) and Officer (2007) is driven by lower

    acquisition valuations of pure private targets. Using deal value/sales, which is not

    10 This number is similar to Officers (2007) results that stand-alone unlisted target firms, including all

    private targets, have a typical 1.85 deal valuation ratio on sales.

  • 8/3/2019 The Dual Tracking Puzzle

    15/44

    14

    truncated, unlike deal value/book value and deal value/EBITDA, we find that, compared

    to matched public targets, dual tracking private targets sell at an insignificant 9 percent

    premium and pure private targets sell at a significant 32% discount. Following Officer

    (2007), after discarding all observations with the percentage difference in valuation

    multiples between the private target and the matched public target that is higher than +1

    to adjust for extreme outliers, we find, based on average of three valuation multiples,

    compared to similar public targets, dual tracking private targets sell at insignificant 3

    percent premium and pure private targets sell at significant 21 percent discount.

    B. Multivariate Analyses

    Our univariate analyses show that, by filing for IPOs and then withdrawing to be

    acquired, dual tracking private targets sell at a 41 percent premium relative to pure

    private targets in takeovers by public acquirers. We next perform multivariate analyses to

    determine whether the premium persists when we control for a panel of valuation-related

    independent variables.

    B.1 Regression Setup

    The dependent variable in our regression is the natural logarithm of acquisition

    valuation multiple.11 As for independent variables, we include a dummy for whether the

    target is a pure private /newly public/established public firm (1) or a dual tracking private

    target (0), a dummy for whether payment is made in stock (1) or not (0), and a dummy

    for whether payment is a combination of stock and cash (1) or not (0).

    11 Due to skewness of valuation rations, we use natural logarithm transformation of valuation multiples in

    all regressions. Our results still hold if we winsorize valuation multiples at top and bottom using 1% and

    2% thresholds.

  • 8/3/2019 The Dual Tracking Puzzle

    16/44

    15

    Bhojraj and Lee (2002) develop a systematic approach for pricing private firms on the

    basis of profitability, growth, and risk characteristics. They show that their valuation

    method accommodates both the general universe of firms as well as a sub-population of

    new-economy firms.12 Therefore, we also include as control variables factors Bhojraj

    and Lee (2002) find to be related to firms valuations.

    We include as control variables: INDVS (industry enterprise value to sales), the

    harmonic mean of the enterprise value-to-sales multiple for all firms within the same

    two-digit SIC code at the transaction year13; and INDPB (industry enterprise value to

    book value), the harmonic mean of the industry price-to-book (PB) ratio; and INDVE

    (industry enterprise value to EBITDA), the harmonic mean of the industry price-to-

    EBITDA ratio. We choose INDVS, INDPB, and INDVE to account for industry

    valuation factors.

    Bhojraj and Lee (2002) show that: 1) the more profitable a firm is relative to its

    industry, the higher the valuation, and 2) valuations are less responsive to negative

    earnings the negative valuation impact of the dollar loss per dollar of total assets is

    smaller than the positive valuation impact of the dollar profit per dollar of total assets.

    Therefore, using profit margin (OPA), defined as EBITDA divided by total assets, as our

    proxy for profitability, we include as independent variables the Industry-adjusted OPA,

    12 New economy firms refer to those loss firms especially among the tech, biotech, and telecommunication

    industries. These firms always report negative earnings.

    13 The harmonic mean is the inverse of the average of the inversed valuation multiples. Baker and Ruback

    (1999) have suggested that the harmonic mean leads to the minimum variance estimates for an industry

    valuation multiple. Enterprise value is defined as the sum of market capitalization of equity and book value

    of debt.

  • 8/3/2019 The Dual Tracking Puzzle

    17/44

    16

    defined as the difference between a targets OPA and the median of the OPA of the firms

    within the same two-digit SIC code, and Loss*industry-adjusted OPA, where Loss is a

    dummy for whether OPA is negative (1) or positive (0).

    Finally, we also use LEV, the financial leverage proxy defined as total debt divided

    by total assets;14 R&D, the fraction of R&D expenses in total assets; and Growth, the sale

    growth for the takeover year, as additional control variables.

    [INSERT TABLE IV]

    B.2 Regression Results

    Table IV reports OLS estimates of regression models of three acquisition valuation

    multiples: deal value/sales, deal value/book value, and deal value/EBITDA.

    Regressions in columns 1, 4, and 7 of Table IV are estimated for dual tracking private

    targets and matched pure private targets. In column 1, the natural logarithm of deal

    value/sales is the dependent variable. The coefficients of INDVS and Industry-adjusted

    OPA are significantly positive, while the coefficient of Loss*industry-adjusted OPA is

    significantly negative. Consistent with the results of Bhojraj and Lee (2002), target firms

    from industries with higher valuation receive higher prices in acquisition. Target firms

    acquisition valuations are also less responsive to negative earnings than to positive

    earnings. Consistent with the notion that high growth should be positively correlated with

    firm value; the coefficient of Growth is positive and significant.

    Most importantly, for the purpose of this study, consistent with our univariate test,

    dual tracking private targets are acquired at significantly higher valuations than matched

    14 Officer (2007) argues that leverage is a proxy for liquidity constraint and documents that leverage ratio is

    positively correlated with private target acquisition discount relative to similar public targets.

  • 8/3/2019 The Dual Tracking Puzzle

    18/44

    17

    pure private targets even after controlling for other determinants of valuations. The

    coefficient of pure private target dummy is negative and statistically significant at -0.46.

    The magnitude implies that dual tracking private targets sell at a 58 percent15 premium,

    measured by deal value to sales, relative to comparable pure private targets in

    acquisitions, after controlling for other determinants of valuations. Given that dual

    tracking private targets have sales of $34 million (median) and are valued at deal

    value/sales of 2.99 (median), they would receive $37 million ($34*2.99 - $34*2.99/1.58)

    less if they were not dual tracking and therefore were valued as pure private targets in

    acquisitions. The increases in acquisition valuations from filing IPO registrations are

    more than enough to compensate for approximate $1.6 million direct expenses related to

    IPO filing (Lee et al. (1996) document that direct expenses of IPO registration is 4% of

    expected proceeds for IPO. The median dual tracking target firm in our sample plans to

    raise $44 million in their IPO filings. This number is not reported in the table).

    The regressions in columns 4 and 7 are similar to those in column 1 except that the

    natural logarithm of deal value/book value (deal value/EBITDA) is the dependent

    variable in column 4 (7) and INDPB (INDVE) instead of INDVS is used as the

    independent variable in column 4 (7) to control for industry valuation. Results on other

    control variables are similar to those reported in column 1. Most importantly, the pure

    private target dummy is negative though insignificant in both columns 4 and 7.

    The regressions in columns 2, 5, and 8 of Table IV are estimated for dual tracking

    private targets and newly public targets, and the regressions in columns 3, 6, and 9 are

    15 From the coefficient estimates of -0.46, reversing the log transformation on deal value/sales, we find deal

    value/sales for dual tracking private targets are 58% (e0.46 - 1) higher than that for pure private targets.

  • 8/3/2019 The Dual Tracking Puzzle

    19/44

    18

    estimated for dual tracking private targets and established public targets. In column 5, we

    find dual tracking targets receive a significantly higher acquisition multiple in deal value

    to book value of equity than matching publicly-traded targets, which is consistent with

    Officers (2007) finding that price to book value of equity shows higher average

    premiums for unlisted targets than publicly-traded targets. This may simply reflect that

    public firms tend to have more equity financing than private firms. In general, our

    multivariate results are consistent with univariate tests. Dual tracking private targets sell

    at valuation multiples comparable to newly public targets and at higher valuation

    multiples than established targets; however, pure private targets sell at a discount in

    acquisitions relative to both newly public targets and established public targets.

    III. Acquirers Announcement Period Returns

    To investigate how other market participants respond to acquisitions of dual tracking

    private targets, in which acquirers clearly pay a significant premium relative to the price

    paid for similar pure private targets, we study stock price reactions of public acquirers

    when acquisitions are announced. We use the standard event study methodology by

    calculating CARs for the three-day period (-1, +1) around the announcement. We

    estimate the abnormal return on stock i over day t using the market-adjusted model;

    mtitit rrAR = , where rit and rmt are the returns for stock i and the market portfolio m

    (CRSPs equal-weighted NYSE/AMEX/NASDAQ index) on day t. The cumulative

    abnormal return for firm i over days (-1, +1) around the announcement date (day 0) is

    measured as +

    =

    =+

    1

    11,1

    t

    iti

    ARCAR .

    [INSERT TABLE V]

  • 8/3/2019 The Dual Tracking Puzzle

    20/44

    19

    Table V tabulates acquirers mean and median 3-day CARs for dual tracking private

    targets, pure private targets, newly public targets, and established public targets, grouped

    by the method of payment. Largely consistent with prior studies findings that listing

    status determines announcement returns, the CARs of acquirers purchasing public targets

    are negative and significant, while the CARs of acquirers purchasing private target are

    positive and significant. Despite paying significantly higher price for dual tracking

    private targets, acquirers of dual tracking private targets still earn a significant average

    abnormal announcement return of 2.6 percent, which is not significantly different from

    2.65 percent abnormal announcement return earned by acquirers of pure private targets.

    Further, regardless of the payment method, acquirers CARs in acquisitions of dual

    tracking private targets are positive and not significantly different from those in

    acquisitions of pure private targets.

    However, Table V shows that acquirers earn positive but insignificant announcement

    returns with stock payments when purchasing dual tracking private and pure private

    targets. The announcement returns with cash payments are significantly positive when

    purchasing pure private targets. These results are different from that reported by Chang

    (1998) and Fuller, Netter, and Stegemoller (2002), whose studies show that acquirers

    experience higher stock returns when buying private targets with stock than with cash.

    [INSERT TABLE VI]

    Table VI reports multivariate analyses for acquirers CARs to control for other factors.

    The dependent variable in our regressions is the cumulative abnormal return CAR over

    days (-1, +1) around the announcement. We include as independent variables factors

    found to explain acquirers announcement returns in prior studies. Regressions in Models

  • 8/3/2019 The Dual Tracking Puzzle

    21/44

    20

    1, 2, and 3 of Table VI are estimated with dual tracking private and pure private targets,

    dual tracking private and newly public targets, and dual tracking private and established

    public targets, respectively.

    Bradley, Desai, and Kim (1983); Jarrell and Poulsen (1989); Servaes (1991); and

    others document that the target size relative to the acquirer size is positively correlated

    with the acquirers CAR. To control for this factor, we include relative size, measured as

    deal value divided by the market value of the acquirer one month before deal

    announcement. Fuller, Netter, Stegemoller (2002) find that a private (public) targets size

    relative to acquirers size is positively (negatively) correlated with the acquirers return at

    the takeover announcement. To control for the different effects of the relative size on the

    acquirers announcement return in acquisitions of different types of targets, we also

    include three interaction terms between the relative size and the dummy variable

    indicating a targets type. Consistent with other studies, the coefficients of relative size

    are positive and significant in our three regression models. We also find the positive

    (negative) relationship between the relative size of acquisition and acquirers

    announcement return for acquisitions of private firms (public firms).

    Following Fuller, Netter, and Stegemoller (2002) and Faccio, McConnell, and Stolin

    (2006), to control for the impact of the method of payment on acquirers CARs, we also

    include a dummy for whether the payment is stock (1) or not (0), and a dummy variable

    for whether payment is a mix of stock and cash (1) or not (0). The coefficients of stock

    payment dummy and mixed payment dummy are not significant in our three regression

    models. This result differs from other studies, which document that acquirers earn

  • 8/3/2019 The Dual Tracking Puzzle

    22/44

    21

    significantly higher announcement returns when buying private targets with stock rather

    than with cash.

    Maquiera, Megginson, and Nail (1998) report that acquirers announcement returns

    are higher in within-industry acquisitions. We also include a dummy for whether the

    target and the acquirer have the same two-digit SIC code (1) or not (0). Neither of the

    coefficients of the same industry dummy is significant in our regressions; this result is

    similar to Faccio, McConnell, and Stolin (2006).

    Officer, Poulsen, and Stegemoller (2006) document that the acquisition discount is

    positively correlated to bidders CARs when acquiring private targets. We include one of

    the acquisition multiples, deal value/sales16, to control for this factor. The coefficients of

    deal value/sales are negative but not significant at 5% in our three regressions.

    Most importantly, the coefficient of the pure private dummy in Model 1 is not

    significant. This finding suggests that acquirers of dual tracking private targets earn the

    similar abnormal announcement returns as those earned by acquirers of pure private

    targets even after controlling for other factors such as higher prices paid for dual tracking

    private targets.

    IV. Selection Bias Adjustments

    It is necessary to account for the fact that the firm decision to file an IPO registration

    before its acquisition is endogenous, and is dependent on firm characteristics. Failure to

    adjust for this potential endogeniety might result in inconsistent and biased estimates of

    16 Using deal value/book value or deal value/EBITDA reduces our sample size by more than half and

    produces qualitatively similar results.

  • 8/3/2019 The Dual Tracking Puzzle

    23/44

    22

    the effect of dual tracking on acquisition valuation ratios in Section II. To address this

    selection bias concern, we use the two-step estimation procedure (see Heckman (1979)

    and Lee, Maddala, and Trost (1980)). The first-step Probit model estimates the

    probability of choosing the dual tracking strategy conditional on a set of private firm

    characteristics. In the second-step linear regression on acquisition valuation ratios, we

    add Lambda, the inverse Mills ratio, from the first-step Probit model estimation as a

    control variable, with the control variables used in Section II.

    [INSERT TABLE VII]

    The Model 1 of Table VII reports the coefficient estimates along with marginal

    effects calculated at the mean of each control variable for the first-step Probit model with

    the dependent variable, Dual Tracking (which equals one for dual tracking private targets

    and zero for pure private targets). We base this Probit model on prior work on a private

    firms choice of going public through IPO markets or M&A markets. Specifically, Brau,

    Francis, and Kohers (2003) and Poulsen and Stegemoller (2007) show that private firms

    choose the IPO market instead of the M&A market when they have higher growth

    opportunities and more capital constraint. Therefore, our Probit model includes Growth,

    LEV, R&D, and Industry-adjusted OPA as control variables to analyze determinants of a

    private firms decision of dual tracking. The Model 1 of Table VII shows that private

    firms with high sales growth and R&D costs are more likely to dual track. Private firms

    that are more profitable than their industry peers are less likely to file for IPOs before

    being acquired and leverage does not affect the dual tracking decision. This result

    suggests that private firms with high growth potential and less profitability, which are

  • 8/3/2019 The Dual Tracking Puzzle

    24/44

    23

    inherently difficult to be valued accurately, are more likely to use IPO processes to

    release information to potential bidders.

    For the second-step model on acquisition valuation ratios, the inverse Mills ratios

    from the first-step Probit estimation are insignificant except for in Model 2. The estimates

    in Models 2, 3, and 4 of Table VII are very similar to those reported in Table IV. In

    summary, our results that dual tracking private targets receive higher acquisition

    valuations relative to pure private targets still hold after controlling for private firms

    choices of filing for IPOs.

    V. Why Does IPO Filing Improve Dual Tracking Targets Acquisition Valuations

    In this section we investigate three possible explanations for dual tracking targets

    acquisition premium relative to pure private targets. The bargaining power hypothesis

    postulates that dual tracking firms use the IPO filing to improve their bargaining power in

    negotiation with acquirers to receive better valuations. The synergy benefits hypothesis

    argue that a private firm, seeking capital to expand, may initiate an IPO and then be

    approached by an acquirer offering a higher valuation owing to potential synergies. The

    lower valuation uncertainty posits that IPO filing can reduce valuation uncertainty.

    Therefore, bidders are willing to pay higher prices for dual tracking targets than for pure

    private targets.

    A. Bargaining Power

    IPO filing may improve a dual tracking targets bargaining power in two ways. First,

    filing IPO can improve a private firms visibility and therefore increase the number of

    potential bidders. Inconsistent with this argument, the median (mean) number of bidders

  • 8/3/2019 The Dual Tracking Puzzle

    25/44

    24

    for 127 dual tracking targets with information available is one and it is no different from

    that of their matched pure private targets.

    [INSERT TABLE VIII]

    Second, investors in the IPO market can be viewed as another bidder; therefore, IPO

    filing generates an outside option for a dual tracking target. This argument implies that

    dual tracking firms that announce their acquisitions after they have withdrawn IPO

    registrations would have acquisition valuations similar to pure private targets since the

    IPO market is no longer an implicit bidder for those dual tracking firms. To test this

    implication, we separate dual tracking private target firms into two groups: 1)

    acquisitions announced before IPO withdrawals, referred as strong dual tracking, and 2)

    acquisitions announced after IPO withdrawals, as weak dual tracking. Table VIII

    reports acquisition multiple comparisons for both groups. The results are similar to those

    reported in Table III for both strong dual tracking and weak dual tracking firms. That

    is, dual tracking private targets are acquired at a significant acquisition premium to other

    pure private targets, regardless of timing of acquisitions. However, the magnitudes of

    acquisition multiples for the strong dual trakcing firms are larger than those for the

    weak dual tracking firms in Table VIII. We repeat regression analyses in Tables IV and

    VII with the dummy variable Strong Dual Tracking indicating the timing of acquisitions.

    The coefficients for Strong Dual Tracking are not significant (results not reported,

    available at request). Overall, we do not find supporting evidence for the bargaining

    power hypothesis.

  • 8/3/2019 The Dual Tracking Puzzle

    26/44

    25

    B. Synergistic Benefits

    The thick registration statement gives a potential buyer the information to do initial

    due diligence on the target firm. This information disseminated via the IPO process might

    lead to acquirers that have greater synergies from acquisitions. Hence, acquirers are

    willing to pay higher prices for dual tracking targets. Then we would expect acquirers of

    dual tracking firms to have better long-run performance after acquisitions than acquirers

    of comparable pure private firms. To test this implication, we compare the average

    abnormal return (AR) for acquirers of dual tracking targets and that for acquirers of pure

    private targets up to three years after acquisitions. For each acquirer, over the month t

    relative to the month of acquisition announcement, we compute AR as the intercept of a

    time-series of regression of its monthly stock returns on Fama and French three factors

    and momentum factors (See Fama and French (1993) and Jegadeesh and Titman (1993)

    for details).

    [INSERT TABLE IX]

    Table IX presents the mean and median of ARs for the acquirers of dual tracking

    targets and the acquirers of pure private targets. None of the difference in ARs between

    acquirers of dual tracking and of pure private targets is significantly different from zero

    up to three years after deals. Therefore, there is little evidence of the synergy benefits

    hypothesis.

    C. Lower Valuation Uncertainty

    The information asymmetry problem between an acquirer and a private target is

    more severe than the problem between an acquirer and a public target because private

    firms generally have less information available and lower information disclosure

  • 8/3/2019 The Dual Tracking Puzzle

    27/44

    26

    standards. By fulfilling the SEC registration requirement, a dual tracking private target

    releases more information and sends signals to potential bidders. Furthermore, book

    building process generates information on investors demand and valuation for the target

    share. Hence, dual tracking firms receive a better acquisition price by reducing valuation

    uncertainty and information asymmetry. One implication of the lower valuation

    uncertainty hypothesis is that private targets that are more difficult to value thus with

    higher valuation uncertainty are more likely to choose filing for IPO before being

    acquired. We find that, consistent with this implication, 1) Panel B of Table II shows that,

    compared to pure private targets, dual tracking targets tend to have higher valuation

    uncertainty due to higher growth rate, higher R&D costs, and lower profitability before

    acquisitions; 2) Estimations of the first-step Probit model from Table VII show that

    private targets with higher growth rates in sales, higher R&D expenses, and lower profits

    are more likely to take dual tracking route.

    VI. Concluding Remarks

    This study examines the dual tracking puzzle by investigating a sample of 132 private

    firms that withdraw their IPOs to be acquired by public bidders between 1984 and 2004.

    We find that such dual tracking private targets sell at a 58 percent acquisition premium

    relative to comparable pure private targets that never file IPO registrations. This finding

    suggests that acquirers are willing to value these almost public targets as similar to

    public targets and to pay more for such dual tracking private targets than for similar pure

    private targets.

  • 8/3/2019 The Dual Tracking Puzzle

    28/44

    27

    The significant acquisition premium is not due to dual tracking targets improved

    bargaining power in negotiations nor higher potential synergistic benefits for bidders. The

    acquisition premium is more consistent with the explanation that IPO filing can reduce

    valuation uncertainty of those dual tracking private targets.

    We also document that, despite paying significantly higher prices for dual tracking

    private targets, acquirers of such targets still earn a significant average abnormal

    announcement return of about 2.6 percent, which is not significantly different from the

    2.7 percent abnormal announcement return earned by acquirers of pure private targets.

    This finding suggests that investors agree with acquirers valuation assessments for dual

    tracking private targets and deem the 58 percent acquisition premium paid for private

    targets that file for an IPO relative to the prices paid for pure private targets that never

    file for IPO to be justified.

  • 8/3/2019 The Dual Tracking Puzzle

    29/44

    28

    REFERENCES

    Baker, Malcolm, and Richard S. Ruback, 1999, Estimating industry multiples, Harvard

    University . Working Paper.

    Bradley, Michael, Anand Desai, and E. Han Kim, 1983, The rationale behind interfirm

    tender offers: Information or synergy?Journal of Financial Economics 11, 183-206.

    Bhojraj, Sanjeev and Charles M. C. Lee, 2002, Who is my peer? A valuation-based

    approach to the selection of comparable firms, Journal of Accounting Research 40,

    407-439.

    Brau, James C, Bill Francis, and Ninon Kohers, 2003, The choice of IPO versus

    takeover: Empirical evidence,Journal of Business 76, 583-612.

    Busaba, Walid Y., 2006, Bookbuilding, the option to withdrawn, and the timing of

    IPOs,Journal of Corporate Finance 12, 159-186.

    Busaba, Walid Y., Lawrence M. Benveniste, and Re-jin Guo, 2001, The option to

    withdraw IPOs during the premarket: Empirical analysis, Journal of Financial

    Economics 60, 73-102.

    Chang, Saeyoung, 1998, Takeovers of privately held targets, methods of payment, and

    bidder returns,Journal of Finance 53, 773-784.

    Dunbar, C., and S. Foerster, 2007, Second Time Lucky? Withdrawn IPOs that Return to

    the Market,Journal of Financial Economics (Forthcoming).

    Faccio, Mara, John J. McConnell, and David Stolin, 2006, Returns to acquirers of listed

    and unlisted targets,Journal of Financial and Quantitative Analysis 41, 197-220.

    Fama, E. F. and K. R. French., 1993, Common Risk Factors in the Returns on Stocks

    and Bonds.Journal of Financial Economics 33, 3-56.

  • 8/3/2019 The Dual Tracking Puzzle

    30/44

    29

    Fuller, Kathleen, Jeffry Netter, and Mike Stegemoller, 2002, What do returns to

    acquiring firms tell us? Evidence from firms make many acquisitions, Journal of

    Finance 57, 1763-1793.

    Heckman, James J., 1979, Sample Selection Bias as a Specification Error,

    Econometrica 47, 153-161.

    Jegadeesh, N., and S. Titman, 1993, Returns to Buying Winners and Selling Losers:

    Implications for Stock Market Efficiency,Journal of Finance 48, 65-91.

    Jarrell, Gregg A., and Annette B. Poulsen, 1989, The return to acquiring firms in tender

    offers: Evidence from three decades, Financial Management18, 12-19.

    Koeplin, John, Atulya Sarin, and Alan C. Shapiro, 1996, The private company

    discount,Journal of Applied Corporate Finance 12, 94-101.

    Lian, Qin, 2007, Does the Market Incorporate Previous IPO Withdrawals When Pricing

    Second-time IPOs? University of Alabama.Working Paper.

    Lee, Inmoo, Scott Lochhead, Jay R. Ritter, and Quanshui Zhao, 1996, The costs of

    raising capital,Journal of Financial Research 19, 59-74.

    Lee, Lung-Fei, G.S. Maddala, and R.P. Trost, 1980, Asymptotic Covariance Matrices of

    Two-stage Probit and Two-stage Tobit Methods for Simultaneous Equation Models

    with Selectivity,Econometrica 48, 491-503.

    Maquieira, Carlos P., William L. Megginson, and Lance Nail, 1998, Wealth creation

    versus wealth redistributions in pure stock-for-stock mergers, Journal of Financial

    Economics 48, 3-33.

    Officer, Micah S., 2007, The price of corporate liquidity: acquisition discounts for

    unlisted targets,Journal of Financial Economics (Forthcoming).

  • 8/3/2019 The Dual Tracking Puzzle

    31/44

    30

    Officer, Micah S., Annette B. Poulsen, and Mike Stegemoller, 2006, Information

    asymmetry and acquirer returns, University of Southern California. Working Paper.

    Pagano, M., F. Pannetta, and L. Zingales, 1998, Why do Companies Go Public? An

    Empirical Analysis,Journal of Finance 53, 27-64.

    Poulsen, Annette and Mike Stegemoller, 2007, Moving from private to public ownership:

    Selling out to public firms vs. initial public offerings, Financial Management

    (Forthcoming).

    Servaes, Henri, 1991, Tobins Q and the gains from takeovers, Journal of Finance 46,

    409-419.

    Zingales, Luigi, 1995, Insider ownership and the decision to go public, Review of

    Economics Studies 62, 425-448.

  • 8/3/2019 The Dual Tracking Puzzle

    32/44

    31

    Table I

    Frequency Distribution of IPO Filings and Subsequent Takeovers

    This table reports the annual incidence and industry representation of firms filing IPOs and theirsubsequent takeovers. The sample comes from the SDC New Issues Database and the M&A Databasefrom 01/01/1984 to 7/25/2004. The IPO sample excludes ADRs, unit offerings, closed end funds, and

    REITs; while the takeover sample excludes leverage buyouts, tender offers, spin-offs, recapitalizations,self-tenders, repurchases, privatizations, and reverse takeovers. Year refers to the year of filing IPOregistration. We count only withdrawn-IPOs that are acquired within three years of their IPOregistrations. Newly public target firms are the successful IPOs that are acquired within three years oftheir initial public offerings. The percentage of withdrawn-IPOs is the number of withdrawn-IPOs overthe sum of withdrawn-IPOs and successful IPOs in a given year or an industry. The percent ofwithdrawn-IPO firms that are acquired (newly public firms that are acquired) refers to the number ofwithdrawn-IPO firms that are acquired (newly public firms that are acquired) over the number ofwithdrawn-IPOs (successful IPOs) for a year or an industry.

    Panel A. Distribution by Year

    Number of Percent of

    YearWithdrawn-IPOs

    SuccessfulIPOs

    Withdrawn-IPO firmsthat areacquired

    Newly

    publicfirms thatareacquired

    Withdrawn-IPOs

    Withdrawn-IPO firmsthat areacquired

    Newly

    publicfirmsthat areacquired

    1984 86 204 0 12 30% 0% 6%

    1985 32 260 1 14 11% 3% 5%

    1986 87 567 4 24 13% 5% 4%

    1987 89 333 1 13 21% 1% 4%

    1988 34 150 0 8 18% 0% 5%

    1989 11 138 0 5 7% 0% 4%

    1990 32 138 4 8 19% 13% 6%

    1991 36 359 2 11 9% 6% 3%

    1992 102 396 5 21 20% 5% 5%

    1993 78 542 6 31 13% 8% 6%

    1994 104 383 5 33 21% 5% 9%

    1995 53 504 5 66 10% 9% 13%

    1996 120 694 14 85 15% 12% 12%

    1997 105 453 15 45 19% 14% 10%

    1998 129 277 10 27 32% 8% 10%

    1999 106 489 29 64 18% 27% 13%

    2000 318 283 33 29 53% 10% 10%

    2001 35 60 6 5 37% 17% 8%

    2002 46 64 7 5 42% 15% 8%

    2003 8 91 0 1 8% 0% 1%

    2004 41 60 3 0 41% 7% 0%

    Total 1652 6445 150 507 20% 9% 8%

  • 8/3/2019 The Dual Tracking Puzzle

    33/44

    32

    Table I (Cont.)

    Panel B. Distribution by industry classification

    Number of

    Industry Two-digit SIC codes Withdrawn-IPOs

    SuccessfulIPOs

    Withdrawn-IPO firms thatare acquired

    Newly publicfirms that areacquired

    WithdrIPOs

    Crops, natural resource extraction 01-05,10-14 43 120 3 15 26%

    Construction 15-17 18 63 3 1 22%

    Other manufacturing 20-27, 29-34, 37, 39 160 685 8 29 19%

    Chemicals and allied products 28 110 395 6 14 22%

    Industrial machinery 35 52 321 7 15 14%

    Electronic and electric equipment 36 96 504 9 35 16%

    Instruments and related products 38 95 372 9 35 20%

    Transportation 40-49 151 503 11 48 23%

    Wholesale 50, 51 76 252 5 12 23%

    Retail 52-59 112 523 9 30 18%

    Financial 60-69 198 841 10 62 19%

    Services 70-80 463 1616 59 196 22%

    Unclassified other 78 250 11 15 24%

    Total 1652 6445 150 507 20%

  • 8/3/2019 The Dual Tracking Puzzle

    34/44

    33

    Table II

    Characteristics of Acquisitions of Dual Tracking Private Targets and Matching Targets

    This table compares characteristics of dual tracking private targets and matching target firms. Panel A presentsthe deal and acquirer's characteristics. Panel B contains targets characteristics. For each dual tracking privatetarget, we identify an acquisition of a pure private firm, a newly-public firm, and an established public firm,

    respectively, which is closest in size (measured as deal value) within the same industry (same first 2-digitprimary SIC code) purchased during the same time period and paid for with the same payment method. Dealvalue equals the total value of consideration paid by the acquirer, excluding fees and expenses. The size of theacquirer is measured as the market value of acquirer equity one month before the acquisition announcement.Relative size of acquisition is the deal value over the market value of acquirer equity. Growth is the salesgrowth rate prior to the takeover year. OPA and OPS are the EBITDA divided by total assets and salesrespectively.LEVis the leverage defined as the total debt divided by the total assets. R&D is the research anddevelopment expense as a percent of the total assets. Comparisons are conducted by using matching-pair t-testsfor differences in means and Wilcoxon signed-rank tests for differences in medians. Medians are reported inparentheses below the mean values. N is the number of matching pairs.

    MatchingDualtrackingprivate

    targets

    Pureprivate

    targets

    Newlypublic

    targets

    Establishedpublic

    targets

    P-value for the difference N

    (1) (2) (3) (4) (1) - (2) (1) - (3) (1) - (4)

    Panel A: Deal and Acquirer Characteristics

    Deal value ($ million) 252 88 958 560 0.01 0.00 0.06

    (Median) (112) (40) (234) (148) (0.00) (0.00) (0.07)132

    Size of acquirer ($ million) 8,416 993 12,695 6,493 0.02 0.36 0.60

    (Median) (947) (222) (1,257) (1,031) (0.00) (0.07) (0.62)121

    Relative size of acquisition 0.27 0.31 0.44 0.69 0.49 0.06 0.24

    (Median) (0.11) (0.20) (0.24) (0.16) (0.04) (0.00) (0.07)121

    Cash in payment (%) 33 50 18 30 0.00 0.00 0.43

    (Median) (5) (53) (0) (0) (0.00) (0.00) (0.38)

    130

    Stock in payment (%) 56 41 76 64 0.00 0.00 0.08

    (Median) (76) (20) (100) (81) (0.00) (0.00) (0.07)130

    Panel B: Target Characteristics

    Sales 97.39 53.68 237.80 285.11 0.05 0.01 0.01

    (Median) (33.89) (23.91) (68.34) (91.69) (0.10) (0.00) (0.00)110

    Growth 4.71 0.5617 0.76 0.25 0.03 0.01 0.01

    (Median) (0.61) (0.24) (0.43) (0.11) (0.01) (0.02) (0.00)96

    OPA -0.38 0.32 0.13 0.11 0.00 0.00 0.00

    (Median) (0.02) (0.19) (0.12) (0.10) (0.00) (0.00) (0.00)105

    OPS -1.92 0.18 0.18 0.17 0.00 0.00 0.00(Median) (0.03) (0.11) (0.15) (0.12) (0.00) (0.00) (0.00)

    104

    LEV 0.35 0.37 0.23 0.19 0.88 0.00 0.00

    (Median) (0.23) (0.15) (0.09) (0.09) (0.08) (0.00) (0.00)107

    R&D 0.20 0.00 0.06 0.05 0.00 0.00 0.00

    (Median) (0.01) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)107

    17 The sample size for this group is 66.

  • 8/3/2019 The Dual Tracking Puzzle

    35/44

    34

    Table III

    Valuation Ratios of Dual Tracking Private Targets and Matching Targets

    This table presents univariate comparisons of valuation ratios of dual tracking private targets and theirmatching pure private, newly-public, and established public targets. For each dual tracking private target, weidentify an acquisition of a pure private firm, a newly-public firm, and an established public firm,

    respectively, which is closest in size (measured as deal value) within the same industry (same first 2-digitprimary SIC code) purchased during the same time period and paid for with the same payment method. InPanel B (C), only firms with positive book value of equity (EBITDA) are included. Comparisons areconducted by using matching-pair t-tests for differences in means and Wilcoxon signed-rank tests fordifferences in medians. N is the number of matching pairs.

    N Mean Median

    Panel A: Deal value to sales(1) Dual tracking private targets 108 156.31 2.99

    (2) Pure private targets 108 3.20 1.84

    (3) Newly public targets 108 7.99 3.20

    (4) Established public targets 108 2.93 2.25

    p-value of t-test / Wilcoxon test: (1) - (2) 0.26 0.00

    p-value of t-test / Wilcoxon test: (1) - (3) 0.28 0.58

    p-value of t-test / Wilcoxon test: (1) - (4) 0.26 0.00

    Panel B: Deal value to book value of equity(1) Dual tracking private targets 60 41.89 14.07

    (2) Pure private targets 60 23.48 8.88

    (3) Newly public targets 60 12.59 4.98

    (4) Established public targets 60 7.11 3.47

    p-value of t-test / Wilcoxon test: (1) - (2) 0.16 0.02

    p-value of t-test / Wilcoxon test: (1) - (3) 0.02 0.00

    p-value of t-test / Wilcoxon test: (1) - (4) 0.01 0.00

    Panel C: Deal value to EBITDA

    (1) Dual tracking private targets 55 24.65 12.86(2) Pure private targets 55 30.07 12.79

    (3) Newly public targets 55 63.65 17.15

    (4) Established public targets 55 31.13 12.51

    p-value of t-test / Wilcoxon test: (1) - (2) 0.53 0.75

    p-value of t-test / Wilcoxon test: (1) - (3) 0.08 0.11

    p-value of t-test / Wilcoxon test: (1) - (4) 0.48 0.62

  • 8/3/2019 The Dual Tracking Puzzle

    36/44

    35

    Table IV

    Regressions of Valuation Ratios for Dual Tracking Private Targets and Matching Targets

    The valuation ratio is deal value divided by sales, book value, and EBITDA, respectively. Regressions shownin Models 2 and 3 only include observations with positive book values or EBITDA, respectively. The

    dependent variables are the log (valuation ratio).INDVS is the industry harmonic mean of enterprise-value-to-sales based on the two-digit SIC code for the year of acquisition. INDPB is the industry harmonic mean ofmarket-value-to-book-value based on two-digit SIC code for the year of acquisition. INDVE is the industryharmonic mean of enterprise-value-to-EBITDA based on the two-digit SIC code for the year of acquisition.Growth is the sales growth rate prior to the takeover year. LEVis the leverage defined as the total debt dividedby the total assets.R&D is the research and development expense as a percent of total assets. Industry-adjustedOPA is the difference between a target firm's OPA and the median value of the same 2-digit SIC industryOPAs. The OPA is defined as the EBITDA divided by total assets. Loss* Industry-adjusted OPA is theinteraction term ofLoss and Industry-adjusted OPA, while the Loss equals 1 if the industry-adjusted OPA isless than or equal to zero, otherwise it takes the value of zero. Absolute values of t-statistics are reported inparentheses. T-statistics are calculated using robust standard errors. * significant at 5%; ** significant at 1%.

    Model 1

    Deal value/sales

    Model 2

    Deal value/book value

    Model 3

    Deal value/EBITDA

    (1) (2) (3) (4) (5) (6) (7) (8) (9)-0.46* -0.45 -0.04

    Pure private(2.28) (1.55) (0.16)

    0.09 -0.86** 0.29Newly public

    (0.42) (3.37) (1.24)

    -0.59** -1.27** -0.06Establishedpublic (2.74) (4.86) (0.26)

    0.31* 0.31* 0.44**INDVS

    (2.53) (2.34) (4.04)

    20.01 -33.62 77.00INDPB

    (0.73) (0.32) (0.66)

    0.21** 0.10* 0.06

    INDVE (3.40) (2.14) (1.28)

    0.04** 0.04** 0.04** 0.00 0.01 0.01 0.04 0.06 0.03Growth

    (4.32) (4.42) (4.48) (0.25) (0.58) (0.89) (1.87) (1.40) (1.12)

    0.07 -0.68 -0.66 0.13** 0.64 0.57 0.10* -0.33 -0.59LEV

    (0.88) (1.95) (1.79) (4.44) (1.34) (1.27) (2.30) (1.19) (1.66)

    0.28 0.44 0.36 1.21 1.56* 1.26 -0.32 -0.01 1.00R&D

    (0.46) (0.79) (0.65) (1.42) (2.17) (1.74) (0.32) (0.01) (1.04)

    0.61** 0.30 0.73 0.29 1.50 1.54Industry-adjusted OPA (3.91) (0.38) (1.11) (1.31) (1.61) (1.83)

    -1.00** -0.59 -1.05 -0.64 -1.94 -2.07*Loss*Industry-adjusted OPA (3.15) (0.69) (1.44) (1.65) (1.71) (2.01)

    -0.33 -0.36 -0.17 0.40 0.27 0.24 0.02 0.02 0.06Mixed payment(1.36) (1.39) (0.73) (1.06) (0.72) (0.67) (0.08) (0.09) (0.21)

    0.84** 0.46 0.73** 0.88** 0.70* 0.61 0.28 0.27 0.23Stock payment

    (3.15) (1.69) (3.13) (2.65) (2.10) (1.96) (1.02) (1.04) (0.90)

    0.22 0.68 0.31 1.96** 1.74** 1.75** 0.47 1.65** 2.11**Constant

    (0.78) (1.91) (0.95) (5.66) (3.57) (3.98) (0.76) (3.05) (4.26)Number ofObservations

    164 198 198 94 114 114 88 105 104

    Adjusted R2 0.33 0.21 0.32 0.10 0.24 0.37 0.15 0.06 0.04

  • 8/3/2019 The Dual Tracking Puzzle

    37/44

    36

    Table V

    Acquirers' Announcement Period CARs over Days (-1, +1)

    This sample represents acquisitions of dual tracking private targets and their matched targets.For each dual tracking privacquisition of a pure private firm, a newly-public firm, and an established public firm, respectively, which is closest in sizewithin the same industry (same first 2-digit primary SIC code) purchased during the same time period and paid for with thAbnormal returns are calculated using a modified market model, the return on an acquirer minus the equal-weighted markday announcement period abnormal returns are measured from day -1 through day 1. Day 0 is the announcement date. Comusing matched-pair t-tests for differences in means and Wilcoxon signed-rank tests for difference in medians. The alternativt-test is that the mean of IPO-withdrawn target valuation ratios is not equal to that of matching samples. P-value of t-tereported in parentheses. ** and * represent significance at 1% level and 5% level for testing whether CARs are different frof matching pairs.

    Mean CAR(p-value)

    Median CAR(p-value)

    Matching MatchingDual tracking

    privatetargets

    Pure privatetargets

    Newlypublictargets

    Establishedpublictargets

    Dualtrackingprivatetargets

    Pure privatetargets

    Newlypublictargets

    0.94% 4.18%** -2.75%** -1.09% 0.84% 2.28%** -1.63%* CashOffers (0.13) (0.05) (0.24) (0.42) (0.06)

    4.46%** 2.39% -0.57% -2.69%** 1.78%** 1.92% -0.29% Hybrid

    (0.30) (0.02) (0.00) (0.20) (0.01)

    1.76% 1.65% -3.05% -3.11% 0.28% 0.27% -1.19% StockOffers (0.98) (0.13) (0.18) (0.82) (0.23)

    2.6%* 2.65%** -1.98%* -2.38%** 0.96%* 0.89%** -0.91%** All

    (0.92) (0.00) (0.00) (0.72) (0.00)

  • 8/3/2019 The Dual Tracking Puzzle

    38/44

    37

    Table VIRegressions for Acquirers' CARs for Acquisitions of Dual Tracking and Matching

    Private Targets

    The dependant variable is the acquirers three-day (-1, +1) CARs around the acquisition announcement.Relative size of acquisition is the deal value over the market value of acquirer equity one month beforeacquisition announcement. Interaction terms are measured as firm types interact with log (relative size of

    acquisition). The Within industry dummy takes a value of one when acquirers are in the same industry astargets if they share the same 2-digit primary SIC code. Absolute values of t-statistics are reported inparentheses. T-statistics are calculated using the robust standard errors. * significant at 5%; **significant at 1%.

    Model 1 Model 2 Model 3

    -0.02Pure private

    (0.67)

    -0.09**Newly public

    (3.00)

    -0.14**Established public

    (4.89)

    0.02* 0.02* 0.02*Log(relative size)

    (2.19) (2.27) (2.16)

    -0.01Pure private * Log(relative size)

    (0.48)

    -0.03**Newly public * Log(relative size)

    (2.95)

    -0.04**Established public * Log(relative size)

    (3.98)

    0.01 -0.02 0.01Within industry(0.45) (1.23) (0.56)

    0.01 0.02 0.01Mix payment dummy

    (0.79) (1.15) (0.62)

    0.00 -0.02 0.00Stock payment dummy

    (0.20) (1.00) (0.02)

    -0.01 -0.01 -0.01Log(deal value/sales )

    (1.42) (1.30) (1.87)

    0.07* 0.08** 0.07*Intercept

    (2.31) (2.81) (2.44)

    Number of Observations 206 206 206Adjusted R-squared 0.03 0.10 0.13

  • 8/3/2019 The Dual Tracking Puzzle

    39/44

    38

    Table VII

    Selection Bias Adjustments: Valuation Ratios of Dual Track and Pure Private Targets

    This table presents estimates of the effect of filing IPO on dual tracking private targets subsequentacquisition valuation, after controlling for selection bias. We include the dual tracking target sample andmatched pure private target sample in this analysis. We use the following two-step estimation procedure.

    In the first-step Probit regression (Model 1), the dependent variable, Dual Tracking, is one for dualtracking firms and zero for pure private targets. The second-step estimation uses ordinary least squares(OLS), where the dependent variable is the natural logarithm of acquisition valuation rations (Dealvalue/sales, Deal value/book value, and Deal value/EBITDA, in Models 2, 3, and 4, respectively).Lambda is the inverse Mills ratio from the first-step. As for other control variables, INDVS is theindustry harmonic mean of enterprise-value-to-sales based on two-digit SIC code for the year ofacquisition. INDPB is the industry harmonic mean of market-value-to-book-value based on two-digitSIC code for the year of acquisition. INDVE is the industry harmonic mean of enterprise-value-to-EBITDA based on two-digit SIC code for the year of acquisition. Growth is the sales growth rate prior tothe takeover year.LEVis the leverage defined as the total debt and short-term divided by the total assets.R&D is the research and development expense as a percent of total assets. Industry-adjusted OPA is thedifference between a target firm's OPA and the median value of the same 2-digit SIC industry OPAs.The OPA is defined as the EBITDA divided by total assets. Loss*Industry-adjusted OPA is the

    interaction term ofLoss andIindustry-adjusted OPA, whileLoss equals to 1 ifIndustry-adjusted OPA isless or equals to zero, otherwise it takes the value of zero. Absolute values of Z-statistics are reported inparentheses. Z-statistics are calculated using the robust standard errors. * significant at 5%; **significant at 1%.

  • 8/3/2019 The Dual Tracking Puzzle

    40/44

    39

    Table VII (Cont.)

    First-Step Probit Estimate Second-Step OLS Estimates

    Model 1 Model 2 Model 3 Model 4

    Coefficient

    Marginal

    effect

    Deal

    value/sales

    Deal value/book

    value

    Deal

    value/EBITDA1.35** 0.86 1.73Dual Tracking

    Dummy (2.87) (1.10) (1.71)

    0.30**INDVS

    (2.83)

    17.95INDPB

    (0.51)

    0.21**INDVE

    (4.43)

    0.34** 0.04** 0.03** 0.00 -0.02Growth

    (2.65) (3.94) (0.01) (0.30)

    0.04 0.00 0.07 0.14 0.12LEV

    (0.44) (0.80) (1.45) (1.24)6.34** 0.76** -0.26 0.97 -2.47

    R&D(3.27) (0.46) (1.21) (1.42)

    -1.33** -0.16** 0.88** 0.40Industry-adjusted OPA (3.97) (2.87) (1.02)

    -1.23** -0.67Loss*Industry-adjusted OPA (3.28) (1.21)

    -0.37 0.40 -0.05Mixed payment

    (1.57) (1.08) (0.19)

    0.81** 0.87* 0.28Stock payment

    (3.33) (2.45) (1.09)

    -0.68 -0.29 -1.11

    Lambda (2.18)* (0.57) (1.78)

    -0.37* -0.65 1.29* -0.37Constant

    -2.38 (1.89) (2.44) (0.53)Number ofObservations

    164 164 94 88

    P-value forWald-test

    0.00 0.00 0.00 0.00

  • 8/3/2019 The Dual Tracking Puzzle

    41/44

    40

    Table VIII

    Valuation Ratios of Dual Tracking Private and Matching Targets by Timing of

    Acquisitions

    This table presents univariate comparisons of acquisition valuation ratios of dual tracking private targetsand their matching pure private, newly-public, and established public targets. For each dual tracking

    private target, we identify an acquisition of a pure private firm, a newly-public firm, and an establishedpublic firm, respectively, which is closest in size (measured as deal value) within the same industry(same first 2-digit primary SIC code) purchased during the same time period and paid for with the samepayment method. Sub-panel A (B) analyze withdrawn-IPOs announce to be acquired before (after)withdrawing their IPO filings. Comparisons are conducted by using matching-pair t-tests for differencesin means and Wilcoxon signed-rank tests for differences in medians. N is the number of matching pairs.

    Panel 1: Deal value to sales N Mean Median

    Panel 1.A: Acquisitions announced before IPO withdrawals

    (1) Dual tracking private targets 48 325.83 3.08

    (2) Pure private targets 48 3.13 1.70

    (3) Newly public targets 48 10.09 4.01(4) Established public targets 48 2.83 2.55

    p-value of t-test / Wilcoxon test: (1) - (2) 0.30 0.00p-value of t-test / Wilcoxon test: (1) - (3) 0.31 0.63p-value of t-test / Wilcoxon test: (1) - (4) 0.30 0.00

    Panel 1.B: Acquisitions announced after IPO withdrawals

    (1) Dual tracking private targets 60 20.69 2.41

    (2) Pure private targets 60 3.25 1.84

    (3) Newly public targets 60 6.32 2.75(4) Established public targets 60 3.01 1.54

    p-value of t-test / Wilcoxon test: (1) - (2) 0.07 0.01p-value of t-test / Wilcoxon test: (1) - (3) 0.14 0.63p-value of t-test / Wilcoxon test: (1) - (4) 0.07 0.00

    Panel 2: Deal value to book value of equity

    Panel 2.A: Acquisitions announced before IPO withdrawals

    (1) Dual tracking private targets 29 33.21 14.29

    (2) Pure private targets 29 26.27 10.20

    (3) Newly public targets 29 16.31 4.92(4) Established public targets 29 9.96 3.78

    p-value of t-test / Wilcoxon test: (1) - (2) 0.52 0.13

    p-value of t-test / Wilcoxon test: (1) - (3) 0.13 0.00

    p-value of t-test / Wilcoxon test: (1) - (4) 0.00 0.00

    Panel 2.B: Acquisitions announced after IPO withdrawals

    (1) Dual tracking private targets 31 50.02 13.12(2) Pure private targets 31 20.86 7.31

    (3) Newly public targets 31 9.11 5.04(4) Established public targets 31 4.45 3.23

    p-value of t-test / Wilcoxon test: (1) - (2) 0.21 0.12p-value of t-test / Wilcoxon test: (1) - (3) 0.07 0.00

    p-value of t-test / Wilcoxon test: (1) - (4) 0.05 0.00

  • 8/3/2019 The Dual Tracking Puzzle

    42/44

    41

    Table VIII (Cont.)

    Panel 3: Deal value to EBITDA N Mean Median

    Panel 3.A: Acquisitions announced before IPO withdrawals

    (1) Dual tracking private targets 24 23.66 14.21

    (2) Pure private targets 24 30.15 12.79

    (3) Newly public targets 24 89.37 19.54

    (4) Established public targets 24 28.60 10.87p-value of t-test / Wilcoxon test: (1) - (2) 0.55 0.72p-value of t-test / Wilcoxon test: (1) - (3) 0.09 0.16

    p-value of t-test / Wilcoxon test: (1) - (4) 0.67 0.76

    Panel 3.B: Acquisitions announced after IPO withdrawals

    (1) Dual tracking private targets 31 25.93 11.86(2) Pure private targets 31 29.96 12.90(3) Newly public targets 31 28.63 16.41(4) Established public targets 31 34.40 12.85

    p-value of t-test / Wilcoxon test: (1) - (2) 0.78 0.33p-value of t-test / Wilcoxon test: (1) - (3) 0.84 0.42

    p-value of t-test / Wilcoxon test: (1) - (4) 0.59 0.31

  • 8/3/2019 The Dual Tracking Puzzle

    43/44

    42

    Table IX

    Acquirers Long-run Post-acquisition Stock Performance (AR)

    This table presents acquirers average abnormal return (AR) up to three years after acquisitions. For eachacquirer over months +1 to +t following the deal-announcement month, we compute AR as the interceptof a time-series of regression of monthly stock returns on Fama and French three factors and momentum

    factor. The symbols **, * denote significance at the 1% and 5% level (two-sided), respectively, fortesting whether mean or median of ARs are different from zero. P-value of t-test and Wilcoxcon test isreported in parentheses for testing the difference between acquirers of dual tracking targets and acquirersof pure private targets. N is the number of matching pairs.

    Acquirers ofdual tracking

    firms

    Acquirers ofpure private

    firms p-value N

    Mean -1.10% -0.22% 0.48Month 1 - 12

    Median -0.11% 0.18% 0.16123

    Mean -0.92% -0.21% 0.46Month 1 - 24

    Median 0.22% -0.05% 0.39123

    Mean -0.79% -0.18% 0.51Month 1 - 36

    Median 0.39% 0.00% 0.71123

  • 8/3/2019 The Dual Tracking Puzzle

    44/44

    Figure 1.Length of Time from IPO to Takeover

    For withdrawn-IPO firms, the number of days is the length of time between the IPOregistration and takeover announcement. The average (median) number of days is 376(294) days for withdrawn-IPO firms. For newly public firms, the number of days is thelength of time from IPO date to takeover announcement. The average (median) numberof days is 615 (629) days for newly public firms.

    0%

    5%10%

    15%

    20%

    25%

    30%

    120 240 360 480 600 720 840 960 1080 1200

    Number of days

    Withdrawn-IPO firms Newly public firms