Changes
Jump to navigation
Jump to search
Masulis Nahata (2011) - Venture Capital Conflicts Of Interest (view source)
Revision as of 17:41, 20 July 2012
, 17:41, 20 July 2012no edit summary
We analyze the effects of venture capital (VC) backing on profitability of private firm acquisitions. We find that VC backing leads to significantly higher acquirer announcement returns, averaging 3%, even after controlling for deal characteristics and endogeneity of venture funding. This leads us to investigate whether some VCs have interests that conflict with those of other investors. We show that such conflicts arise from VCs having financial relationships with both acquirers and targets, corporate VCs having a dominant strategic focus, and VC funds nearing maturity experiencing pressure to liquidate. Our conclusions follow from examinations of target takeover premia and acquirer announcement returns.
==Data==
The paper uses SDC M&A and VentureXpert, with observations from 1991-2006 (inclusive).
To be included in the sample, the following conditions must be satisfied:
i) Acquirers are U.S. headquartered, and their stock is publicly listed on the AMEX, NASDAQ, or NYSE.
ii) The target is a privately held U.S. incorporated company.
iii) Neither acquirer nor target is a regulated utility or a financial institution.
iv) An acquisition must be completed, the buyer has no publicly known toehold position prior to the deal announcement, and the buyer acquires 100% of target firm shares.
v) The target purchase price is at least $1,000,000, and the relative deal size (target purchase price divided by acquirer equity market value 1 month prior to the deal announcement) is at least 10%.
vi) Acquirer stock returns are available in the Center for Research in Security Prices (CRSP) database, and its daily returns are available for the 5 trading days surrounding the acquisition announcement date (event days –2 to 2).
vii) Acquirer stock prices must be at least $2 as of the acquisition announcement date (event day 0).
viii) VC-backed targets must have information available on the investment positions of 1 or more of their VC investors.
ix) Clustered acquisitions (of 2 or more) by a single acquirer within 5 days are excluded.
The main dataset is a matched set of 245:245 VC:Non-VC acqs, with hand-collected data.
==Measures==
Dependent vars are:
*CARs: 5 day, subtraction model (<math>\sum AR_i-AR_m</math>)
*Deal Premia, measured as <math>\log\frac{Price}{Target\;Book\;Val.\;Assets}</math>
Primary explanatory vars are:
*VC liquidity (Near fund end=1, otherwise=0... How near not defined).
*VC Self-dealing (VC has stake in acquirer)
*CVC strategic focus (same 2dg SIC as parent,
==Methodology==
Results are presented using matching. The primary method is propensity score matching, with a logistic score:
:<math>Prob. (Acquisition of a VC-Backed Target) = -3.2379(<0.01) + 1.2057 (<0.01) (High-Tech) + 0.9202(<0.01)(Stock) + 0.0005(<0.01)(Deal Size) - 0.0345(0.64)(Relative Deal Size)</math>
They used 6 blocks, stratified by quantiles, and interatively balanced their sample. Then for each treatment they took a nearest neighbour based on: No confounding event; industry-based matching (3dg SIC -> 1dg SIC); and minimal propensity score.
Results were checked using manual matching (3 dimensions: Deal size, relative deal size, ann. date) and a Heckman model. For the Heckman the selection the first stage was a probit using 5 indicator variables (High-Tech, MA, CA, NY, and TX) and agg. IPO proceeds last 3 months, agg. VC inv. last 3 months. These were significant in the first stage but not the second. The Heckman model gave a 2nd stage estimate of the VC-target effect of more than 2% for the CAR5 (c.f. 3% below).
Their definition of High-tech is SIC codes:
*283 (biological products, genetics, and pharmaceuticals)
*481 (high-technology communications)
*365–369 (electronic equipment)
*482–489 (communication services)
*357 (computers)
*737 (software services)
==Results==
Univariate puts VC CAR at 6.31% and non-VC at 3.38% (N=245*2), with the difference significant at 0.05.
Multivariate CAR analysis looks like:
Stock acquisition 0.002 [0.916]
VC-backed target 0.027 [0.030]**
log(acquirer size) –0.007 [0.164]
Relative deal size 0.028 [0.033]**
Intraindustry deal –0.015 [0.258]
High-technology target –0.013 [0.553]
Target industry market-to-book 0.048 [0.003]***
Acquirer stock return volatility 1.034 [0.005]***
Intercept 0.011 [0.936]
Industry and year fixed effects Present
Adjusted R2 4.26%
No. of obs. 490
VC-Backed Target CAR analysis looks like:
VC liquidity 0.022 [0.56]
VC self-dealing 0.066 [0.02]**
CVC strategic focus 0.055 [0.07]*
Early/seed stage target 0.021 [0.32]
VC inexperience 0.006 [0.79]
Stock acquisition –0.013 [0.54]
log(acquirer size) –0.006 [0.52]
Relative deal size 0.012 [0.58]
Intraindustry deal –0.014 [0.56]
High-technology target –0.019 [0.52]
Target industry market-to-book 0.049 [0.09]*
Acquirer stock return volatility 1.875 [0.03]**
Intercept 0.026 [0.72]
Industry and year fixed effects Present
Adjusted R2 7.20%
No. of obs. 239
The analysis of takeover premia is not included here, but it is noted that there seems to be no feed-back between takeover premia and CARs.
==Problems==
They don't cite any of the literature!
*[[Brander Egan (2007) - The Role of VCs in Acquisitions]]
*[[Gompers Xuan (2006) - The Role Of Venture Capitalists In The Acquisition Of Private Companies]]
*[[Gompers Xuan (2008) - Bridge Building In Venture Capital Backed Acquisitions]]
*[[Benson Ziedonis (2010) - Corporate Venture Capital And The Returns To Acquiring Portfolio Companies]]
*[[Ivanov Xie (2010) - Do Corporate Venture Capitalists Add Value To Start Up Firms]]
Their propensity score matching is questionable. The matching score is calculated using regressors that clearly fail the exclusion criteria.
Their sample is also questionable - they are using a deliberately selected 20% of their full sample and report no full sample CAR results (that I can see).
By including ''Target industry market-to-book'' and ''Acquirer stock return volatility'' they are including two common measures of information asymmetry (between target and acquirer, and acquirer and investors, respectively).
It is not clear that they matched VE to SDC-MA. That is, they appear to have taken VE's indicators for acquisitions and used SDC-MA acquisitions as controls. If this is the case, then their sample is bias towards the high-value/high-profile acquisitions reported in VE.
