We are able to partially address this concern by using the“missing” patents assigned to the corporate parents, asdescribed above. We also address this issue by re‐runningthe cross‐tabulations and regressions above, dropping thedivisional buyouts from our sample. Since the other casesare not “carve out” parts <strong>of</strong> firms, but rather involve thepurchase <strong>of</strong> the entirety <strong>of</strong> a corporation, this problem shouldnot be present. <strong>The</strong> key results are little changed as a result<strong>of</strong> this shift.D. Analysis <strong>of</strong> level <strong>of</strong> patentingIn the last two analyses, we move from examining the quality<strong>of</strong> individual patents and instead look at the mixture <strong>of</strong> theoverall patenting activity generated in the years before andafter the private equity investments.A natural first question is how the level <strong>of</strong> patenting activitychanges around the time <strong>of</strong> a private equity investment. If theaverage number <strong>of</strong> successful patent filings falls dramatically,our interpretation <strong>of</strong> the earlier finding that the importance <strong>of</strong>the issued patents rises considerably might be quite different.It would suggest cut‐backs <strong>of</strong> unproductive innovativeactivities rather than repositioning <strong>of</strong> research from lowerto higher impact topics.An analysis <strong>of</strong> patenting prior to and after the private equityinvestment is problematic, however, for several reasons.While we can adjust for the truncation associated with thetiming <strong>of</strong> the patent applications (the fact that, in manycases, not all patents in the five years after the private equityinvestments in our sample have been applied for, much lessawarded), it is very difficult to control for the assignment <strong>of</strong>patents to corporate parents. As noted above, we will beable to see some but not all <strong>of</strong> the patents assigned totargets that were units <strong>of</strong> larger firms prior to divisionalbuyouts or else were ultimately acquired by other concerns.<strong>The</strong> reliability <strong>of</strong> the algorithm that identifies these hiddenpatents is almost impossible to assess. Consequently, weexclude divisional buyouts for the analysis below.Despite this limitation, in Table 6 we undertake an analysis<strong>of</strong> the level <strong>of</strong> patenting. An observation is a target firm in agiven year: that is, for each transaction in 2000 and before,we use nine observations for each transaction, from threeyears prior to five years after the transaction. For transactionsin subsequent years, we use smaller number <strong>of</strong> observations,reflecting our inability to see patent filings made after 2005.<strong>The</strong> dependent variable is the number <strong>of</strong> ultimatelysuccessful patent filings made in the given calendar year.<strong>The</strong> initial analysis is in the first two columns <strong>of</strong> Table 6, whichuse in turn fixed effects for each year and firm to control forthe differing propensity to patent. In this analysis, the resultssuggest that there is a marked decline in patenting.We might worry, however, that this result is an artifact <strong>of</strong> oursample construction: in particular, while we observe somesuccessful patent filings in the final years <strong>of</strong> the sample, thereare likely to be many applications that were filed in theseyears that had not issued as at May 2007. (Recall the averagepatent pendency today is about 30 months.) Becauseobservations <strong>of</strong> patent filings in 2004 and 2005, where thisselection bias will be the worst, are disproportionately likelyto be in the years after private equity transactions, this effectmay bias our counts <strong>of</strong> patent filings.We thus repeat the analyses restricting the sample in twoways. First, in columns 3 and 4, we limit the analysis tousing only private equity investments prior to 1999. In theseregressions, effects due to not‐yet‐issued patent applicationsshould be much less severe. We find that when we use firmfixed effects, the trend <strong>of</strong> patenting over time is negative;when we use year fixed effects, the trend is insignificant.A remaining worry is that these results may be affectedby some firms not being stand‐alone firms in the yearsbefore or after this transaction, even if the transaction itselfis not a divisional buyout. To ensure we have patentinginformation about the individual firms in the years surroundingthe transaction, specifications (5) and (6) are conditionalon the firm having filed an ultimately successful patentapplication both three years before and five years afterthe transaction (Event years ‐3 and +5). This reducedthe concern that we do not observe patents for the firmin the entire nine‐year window. It also introduces a concernthat companies that are stand‐alone entities before andremain stand‐alone entities after the transaction are specialin other ways, which may affect our results as well.However, as expected, the main effect <strong>of</strong> conditioning onthis sub‐sample is to reduce the below‐one coefficientsin the years before the transaction, consistent with theconcern that specifications (3) and (4) may underestimatethese coefficients.In the final set <strong>of</strong> analyses, we use a dummy variable forpatents filed after the private equity investment, rather thanseparate ones for each year. When we do so, we can nowestimate regressions with both year and firm fixed effects(in earlier estimations, when we tried such specifications, theregressions failed to converge). Here, we find that as before,with firm effects, the time trend in patenting is negative; withyear effects, it is greater than one; and with both sets <strong>of</strong>dummy variables, it is not significantly different from zero.Taken together, the results suggest that there is no clearchange in patenting. While our conclusions must besomewhat tentative due to the discussed difficulties inmeasurement and the remaining uncertainties, the lack <strong>of</strong> aconsistent pattern once we control for the biases is evident.E. Analysis <strong>of</strong> patent portfoliosIn the final section, we turn to considering the structure<strong>of</strong> the patent portfolios constructed by these firms in theyears before and after the private equity investments. Sincethe previous section shows that the increase in patentimportance is not driven by private equity‐backed firms34 Large-sample studies: Long-run investment<strong>The</strong> <strong>Global</strong> <strong>Economic</strong> <strong>Impact</strong> <strong>of</strong> <strong>Private</strong> <strong>Equity</strong> <strong>Report</strong> <strong>2008</strong>
cutting back on the number <strong>of</strong> filings, it is natural to wonderabout the dynamics behind the change in quality.<strong>The</strong> initial analysis is presented in the final line in Panel A <strong>of</strong> Table2. We compare the Herfindahl index, or concentration measure,<strong>of</strong> the patent classes in which firms’ awards are assigned. Inthis comparison, we restrict the sample to the 59 firms with atleast four patent applications filed prior to the private equityinvestment and at least four patents applied for afterwards, inorder to ensure the computed measures <strong>of</strong> concentration aremeaningful. When we undertake this comparison, we find thatafter the private equity investments, firms are likely to have moreconcentrated patent portfolios than beforehand, but the p‐valueis just above the 10% threshold.We can gain some additional insights as to how these moreconcentrated portfolios emerge from the cross‐tabulations inTable 7. We use as observations each patent, and examinecitations in the years prior to and after the private equityinvestment, just as we did in Table 2. We now divide thepatents, though, in two ways. In Panel A, we divide theobservations into those whose primary patent classassignment was more or less well‐populated prior to theinvestments: more precisely, whether the firm, in applicationsfiled in the three years prior to the private equity transaction,had above or below the median share <strong>of</strong> patenting in thatprimary patent class. In Panel B, we divide the patents bywhether the share <strong>of</strong> patenting in the primary class increasedor decreased after the private equity transaction.<strong>The</strong> cross‐tabulations provide additional insights into thesources <strong>of</strong> the increase in patent importance. First, we seefrom Panel A that awards in the firms’ focal technologies –the areas where they had done a disproportionate amount<strong>of</strong> patenting prior to the transaction – are more likely toincrease in quality, regardless <strong>of</strong> whether raw or adjustedpatent counts are used. Panel B reveals that patent classesthat experience an increase in patenting share are alsodisproportionately where the increase in patent qualityoccurred. <strong>The</strong>se patterns are consistent with the privateequity‐backed firms focusing their innovative investments intheir core areas <strong>of</strong> strength and generating higher‐impactpatent portfolios as a result.Consistent results emerge from Table 8, which presentsNegative Binomial regression analyses akin to that in thesixth column <strong>of</strong> Table 3. We now add controls for the share<strong>of</strong> patenting in the primary patent class prior to the privateequity investment (in the first and second regressions) andfor the change in the share <strong>of</strong> patenting in that class frombefore to after the investment (in the third and fourthregressions), as well as interactions between the patentmeasure and the dummy denoting an award filed in the firstto fifth years after the private equity investment. Because themeasures <strong>of</strong> patent shares may be misleading if there are justa handful <strong>of</strong> patents assigned to a given firm, we undertakethe analysis both using the entire sample (the first and thirdcolumns) and only patents <strong>of</strong> firms which had at least fourpatents prior to the private equity investments and four after(the second and fourth columns).<strong>The</strong> significantly greater than one coefficient for the variable“Share <strong>of</strong> firm’s pre‐investment patents in class” suggest thatpatents in the firms’ “core” areas – the areas where therewas more patenting prior to the private equity investment –are disproportionately likely to be important ones. Moreover,the interaction term is greater than one. Not only are thesepatents likely to be important, but their impact is likely toincrease after the private equity investment.<strong>The</strong> variable “Change in firm’s patent in class pre‐ andpost‐investment” initially presents a more confusing picture.<strong>The</strong> coefficient is again greater than one – areas where thereis growth are more important ones – but the significance isonly marginal. In column 3, this interaction term is less thanone, but when we restrict the sample to those firms withat least four patents before and after the transaction (orsimilar cut‐<strong>of</strong>fs), the interaction turns greater than one andsignificant. Once we exclude firms with only modestpatenting activity, an increase in patenting is associatedwith a sharp (and highly significant) boost in patent quality.Thus, these analyses suggest that private equity‐backedfirms tend to focus their patent filings. This focusing processis not indiscriminate, however, but tends to concentrate oncore technologies. Moreover, the very process <strong>of</strong> focusingseems to lead to the patents in these selected classeshaving greater impact after the private equity investment.5. ConclusionsThis paper examines the nature <strong>of</strong> long‐run investments infirms backed by private equity groups, focusing on innovativeactivities. It examines patents filed by 495 firms that receivedprivate equity backing between 1983 and 2005.We find that:• Patents <strong>of</strong> private equity‐backed firms applied for in theyears after the investment are more frequently cited• <strong>Private</strong> equity‐backed firms have no deterioration afterthe investments in patent originality and generality, whichare proxies for the fundamental nature <strong>of</strong> the research• <strong>The</strong> quantity <strong>of</strong> patenting appears not to consistentlychange in the years after the private equity investment• <strong>The</strong> patent portfolios <strong>of</strong> firms become more focused inthe years after private equity investments. Breakdowns <strong>of</strong>the patenting patterns suggest that the areas where thefirms concentrate their patenting after the private equityinvestment, and the historical core strengths <strong>of</strong> the firm,tend to be the areas where the increase in patent impactis particularly greatWe see three avenues for future research into the relationship<strong>of</strong> private equity and innovation. While each will require<strong>The</strong> <strong>Global</strong> <strong>Economic</strong> <strong>Impact</strong> <strong>of</strong> <strong>Private</strong> <strong>Equity</strong> <strong>Report</strong> <strong>2008</strong> Large-sample studies: Long-run investment 35
- Page 2 and 3: The Globalization of Alternative In
- Page 5: ContributorsCo-editorsAnuradha Guru
- Page 9 and 10: PrefaceKevin SteinbergChief Operati
- Page 11 and 12: Letter on behalf of the Advisory Bo
- Page 13 and 14: Executive summaryJosh lernerHarvard
- Page 15 and 16: • Private equity-backed companies
- Page 17 and 18: C. Indian casesThe two India cases,
- Page 19 and 20: Part 1Large-sample studiesThe Globa
- Page 21 and 22: The new demography of private equit
- Page 23 and 24: among US publicly traded firms, it
- Page 25 and 26: should be fairly complete. While th
- Page 27 and 28: according to Moody’s (Hamilton et
- Page 29 and 30: draining public markets of firms. I
- Page 31 and 32: FIguresFigure 1A: LBO transactions
- Page 33 and 34: TablesTable 1: Capital IQ 1980s cov
- Page 35 and 36: Table 2: Magnitude and growth of LB
- Page 37 and 38: Table 4: Exits of individual LBO tr
- Page 39 and 40: Table 6: Determinants of exit succe
- Page 41 and 42: Table 7: Ultimate staying power of
- Page 43 and 44: Appendix 1: Imputed enterprise valu
- Page 45 and 46: Private equity and long-run investm
- Page 47 and 48: alternative names associated with t
- Page 49 and 50: 4. Finally, we explore whether firm
- Page 51: When we estimate these regressions,
- Page 55 and 56: Table 1: Summary statisticsPanel D:
- Page 57 and 58: Table 4: Relative citation intensit
- Page 59 and 60: figuresFigure 1: Number of private
- Page 61 and 62: Private equity and employment*steve
- Page 63 and 64: Especially when taken together, our
- Page 65 and 66: centred on the transaction year ide
- Page 67 and 68: and Vartia 1985.) Aggregate employm
- Page 69 and 70: sectors. In Retail Trade, the cumul
- Page 71 and 72: employment-weighted acquisition rat
- Page 73 and 74: FIguresFigure 1: Matches of private
- Page 75 and 76: Figure 6:Figure 6A: Comparison of n
- Page 77 and 78: Figure 8:Figure 8A: Comparison of j
- Page 79 and 80: Figure 11: Variation in impact in e
- Page 81 and 82: Figure 12: Differences in impact on
- Page 83 and 84: Private equity and corporate govern
- Page 85 and 86: et al (2007) track the evolution of
- Page 87 and 88: groups aim to improve firm performa
- Page 89 and 90: distribution of the LBO sponsors, m
- Page 91 and 92: the most difficult cases. This stor
- Page 93 and 94: to see whether these changes of CEO
- Page 95 and 96: Figure 3:This figure represents the
- Page 97 and 98: TablesTable 1: Company size descrip
- Page 99 and 100: Table 5: Changes in the board size,
- Page 101 and 102: Table 7: Board turnoverPanel A: Siz
- Page 103 and 104:
Part 2Case studiesThe Global Econom
- Page 105 and 106:
European private equity cases: intr
- Page 107 and 108:
Exhibit 1: Private equity fund size
- Page 109 and 110:
Messer Griesheimann-kristin achleit
- Page 111 and 112:
ealized it was not possible to grow
- Page 113 and 114:
The deal with Allianz Capital partn
- Page 115 and 116:
the deal, the private equity invest
- Page 117 and 118:
Exhibit 1: The Messer Griesheim dea
- Page 119 and 120:
Exhibit 5: Post buyout structureMes
- Page 121 and 122:
New Lookann-kristin achleitnerTechn
- Page 123 and 124:
feet. This restricted store space w
- Page 125 and 126:
institutional investors why this in
- Page 127 and 128:
Although a public listing did not a
- Page 129 and 130:
Exhibit 5: Employment development a
- Page 131 and 132:
Chinese private equity cases: intro
- Page 133 and 134:
Hony Capital and China Glass Holdin
- Page 135 and 136:
Hony’s Chinese name means ambitio
- Page 137 and 138:
Establishing early agreement on pos
- Page 139 and 140:
Executing the IPOEach of the initia
- Page 141 and 142:
Exhibit 1A: Summary of Hony Capital
- Page 143 and 144:
Exhibit 4: Members of the China Gla
- Page 145 and 146:
Exhibit 6A: China Glass post‐acqu
- Page 147 and 148:
Exhibit 8: China Glass stock price
- Page 149 and 150:
3i Group plc and Little Sheep*Lily
- Page 151 and 152:
y an aggressive franchise strategy,
- Page 153 and 154:
soul” of the business. But there
- Page 155 and 156:
Exhibit 1: Summary information on 3
- Page 157 and 158:
Exhibit 6: An excerpt from the 180-
- Page 159 and 160:
Indian private equity cases: introd
- Page 161 and 162:
ICICI Venture and Subhiksha *Lily F
- Page 163 and 164:
investment,” recalled Deshpande.
- Page 165 and 166:
2005 - 2007: Moderator, protector a
- Page 167 and 168:
Exhibit 3: Subhiksha’s board comp
- Page 169 and 170:
Warburg Pincus and Bharti Tele‐Ve
- Page 171 and 172:
founded two companies at this time
- Page 173 and 174:
By 2003 this restructuring task was
- Page 175 and 176:
Exhibit 1C: Private equity investme
- Page 177 and 178:
Exhibit 4B: Bharti cellular footpri
- Page 179 and 180:
Exhibit 6: Summary of Bharti’s fi
- Page 181 and 182:
Exhibit 7: Bharti’s board structu
- Page 183 and 184:
In the 1993‐94 academic year, he
- Page 185 and 186:
consumer products. She was also a R
- Page 187 and 188:
AcknowledgementsJosh LernerHarvard
- Page 189:
The World Economic Forum is an inde