The Returns to Entrepreneurial Investment: A Private Equity Premium Puzzle?faculty.haas.berkeley.edu/vissing/tmav_aer.pdf · 2003-04-08 · The Returns to Entrepreneurial Investment:

The Returns to Entrepreneurial Investment

A Private Equity Premium Puzzle

By TOBIAS J MOSKOWITZ AND ANNETTE VISSING-JOslashRGENSEN

We document the return to investing in US nonpublicly traded equity Entrepre-neurial investment is extremely concentrated yet despite its poor diversi cation we nd that the returns to private equity are no higher than the returns to public equityGiven the large public equity premium it is puzzling why households willinglyinvest substantial amounts in a single privately held rm with a seemingly far worserisk-return trade-off We brie y discuss how large nonpecuniary bene ts a prefer-ence for skewness or overestimates of the probability of survival could potentiallyexplain investment in private equity despite these ndings (JEL G11 G12 M13)

Asset pricing and investment theory rely crit-ically on our understanding of investorsrsquo port-folio choices Yet entrepreneurial investmentwhich represents a substantial fraction of manyinvestorsrsquo portfolios is relatively understudiedand not well understood Speci cally little isknown about the aggregate return to entre-preneursrsquo equity investments and the distribu-

tion of equity returns across nonpublicly traded rms We analyze investment in and documentthe return to all nonpublicly traded equity in theUnited States The total value of private equityis similar in magnitude to the public equitymarket over our sample period Despite this theprivate equity market has received relativelylittle academic attention1 We provide the rstset of estimates of the returns and risks for theentire market of nonpublic equity

We nd investment in private equity to beextremely concentrated About 75 percent of allprivate equity is owned by households forwhom it constitutes at least half of their total networth Furthermore households with entrepre-neurial equity invest on average more than 70percent of their private holdings in a singleprivate company in which they have an activemanagement interest Despite this dramatic lackof diversi cation the average annual return toall equity in privately held companies is ratherunimpressive Private equity returns are on av-erage no higher than the market return on allpublicly traded equity

Moskowitz Graduate School of Business Universityof Chicago 1101 East 58th Street Chicago IL 60637and National Bureau of Economic Research (e-mailtobiasmoskowitzgsbuchicagoedu) Vissing-JoslashrgensenDepartment of Economics University of Chicago 1126East 59th Street Chicago IL 60637 Center for EconomicPolicy Research and NBER (e-mail vissinguchicagoedu) We are grateful to several anonymous refereesRochelle Antoniewicz Merle Erickson Eugene Fama KenFrench Lars Hansen John Heaton David Ikenberry BarryJohnson Steve Kaplan Deborah Lucas Andrew MetrickSendhil Mullainathan Ann Semer Per Stromberg and sem-inar participants at the University of Minnesota the Uni-versity of Chicago (Finance and Macro lunches and Moneyand Banking seminar) MIT UCLA Arizona State Univer-sity the New York Federal Reserve Purdue UniversityStanford University University of CaliforniandashBerkeleyMcGill University University of British Columbia OhioState University the NBER Macroeconomic Fluctuationsand Corporate Finance meetings the Berkeley Program inFinance the Wharton Conference on Household PortfolioChoice and the WFA meetings for helpful comments andsuggestions Moskowitz thanks the Center for Research inSecurity Prices and the James S Kemper Foundation for nancial support Vissing-Joslashrgensen thanks the NationalScience Foundation for nancial support Special thanks toJay Ritter for data on initial public offerings and to TaorongJiang for help with the SDC database

1 While there are papers which examine venture capital nancing of private rms venture capital accounts for atrivial fraction (less than 1 percent) of the entire privateequity market [according to the numbers in George W Fennet al (1995) as well as our numbers for total private equity]In addition venture capital pertains to a very speci c type ofinvestment in private equity that may not provide muchinsight into the typical entrepreneurrsquos investment decisionand returns

745

Using data for all private equity from both theSurvey of Consumer Finances (SCF) and theFlow of Funds Accounts and the National In-come and Product Accounts (FFANIPA) overthe period 1989 to 1998 as well as proprietorand partnership data from the FFANIPA overthe longer period 1952 to 1999 we nd that theaverage return to all private equity is similar tothat of the public market equity index This issurprising since investing in the equity of asingle private company is likely to be muchriskier than investing in the public equity indexFirst survival rates of private rms are onlyaround 34 percent over the rst ten years of the rmrsquos life Second even conditional on sur-vival the distribution of equity returns acrossentrepreneurs is wide Third the average entre-preneur holds most of his investment in thesame private rm in which he works makinghis equity return highly correlated with his hu-man capital return Fourth while it is dif cult toprecisely estimate the overall risk of privateequity our estimates suggest that the index ofprivate equity is likely as volatile as the publicequity index and that aggregate private equityreturns are highly correlated with the publicequity market Finally the amount of idiosyn-cratic risk of a single private rm implies thatthe aggregate (index) return is likely an overes-timate of the average of the returns to eachindividual entrepreneur further strengtheningthe conclusion that private equity returns arelow Our results are robust to a variety of ad-justments for the labor component of entrepre-neurial income retained earnings in the rm rm births and deaths initial public offeringsand acquisitions and potential income underre-porting due to tax evasion Overall the diversi- ed portfolio of public equity seems to offer afar more attractive risk-return trade-off than thatobtained by the typical entrepreneur

To put our results into perspective considerwhat theory suggests the expected private eq-uity return to be The higher risk from lack ofdiversi cation of private equity should lead to ahigher private equity premium than that on pub-lic equity How much higher than the averagepublic equity return would we expect the aver-age private equity return to be John Heaton andDeborah Lucas (2001) model and calibrate thehurdle rate which would make a household in-

different between investing in a portfolio of asingle private rm a public equity index andT-bills or a portfolio of just the public equityindex and T-bills For an investor with a relativerisk-aversion coef cient of 2 (as well as reason-able assumptions about the debt-to-asset ratio ofthe private rm and the fraction of entrepreneur-ial wealth invested in private equity) purelyidiosyncratic private equity risk generates a hur-dle rate of about 10 percent above the publicequity return2 Michael J Brennan and WalterN Torous (1999) estimate a certainty equiva-lent wealth loss of investing in a single (public) rm of about 64 percent over a ten-year horizonfor an investor with a relative risk-aversion co-ef cient of 2 This loss increases to 95 percentfor an investor with a relative risk-aversion co-ef cient of 3 Shlomo Benartzi (2000) nds thata return premium of 20 percent is needed for anindividual (with relative risk-aversion coef -cient of 4) to invest 45 percent of his portfolioin a single publicly traded stock when the totalportfolio is restricted to have 40 percent bondsand 60 percent stock When allowing the totalportfolio to contain 100 percent equity andreducing risk aversion to 2 the required excessreturn of the individual stock over the publicequity index return declines to 5 percent There-fore the premium required to induce investorsto hold equity in a single private rm wouldalso have to be large For simplicity we willcite 10 percent as the required premium foruse in some of our ldquoback-of-the-enveloperdquocalculations

Of course obtaining a precise measure of themean return to private equity is extremely dif- cult The notoriously dif cult exercise of esti-mating the mean on a highly volatile returnseries over a relatively short time period is wellknown This dif culty is exacerbated when us-ing fairly imprecise data on estimates of private

2 This number is based on the average of line 2 and 3 inTable 4 of Heaton and Lucas (2001) This case correspondsmost closely to the ratio of private equity to net worthdocumented below based on the SCF and the ratio of debt toassets for proprietors and partnerships in the FFA Theauthorsrsquo calculation assumes a zero correlation betweenprivate and public equity (the private investment project intheir model has no aggregate risk) A positive correlationwould increase the private equity hurdle rate

746 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 2002

rm values and pro ts Nevertheless the esti-mated realized returns to private equity are sim-ilar to those in the public market and are highlycorrelated with public equity returns Hence itis unlikely that private equity outperformedpublic equity by 10 percent per year over oursample period (including the longer period from1952 to 1999)3 The implication is that privateequity returns appear low given their risk

In addition to its sheer size it is interesting toanalyze the private equity market to help under-stand existing asset-pricing issues Consider theldquoequity premium puzzlerdquo of Lars P Hansen andKenneth J Singleton (1983) and Rajnish Mehraand Edward C Prescott (1985) Resolutions ofthe high average return on public equity whichrely on homogeneous agents with very largevalues of risk aversion [eg John Campbell andJohn Cochrane (1999)] seem at odds with thefact that many households take on much largerrisks in the private equity market without onaverage earning a higher return than the publicequity return In other words unlike the equitypremium puzzle documented in public marketsthe returns to private equity investment appearfar too low given their risk If households re-quire such a high expected return to take on therisk of publicly traded equity why are theywilling to invest substantial amounts of wealthin a single private company with a much worserisk-return trade-off Should this be considereda ldquoprivate equity premium puzzlerdquo More the-oretical and empirical work is needed to deter-mine if this is the case What we hope toconvince the reader is that a complete theory ofhousehold portfolio choice should emphasizeboth public and private equity For example

Heaton and Lucas (2000) argue that the addi-tional risk of private investment and its corre-lation with public equity market returns mayhelp explain why the (public) equity premiumis so high However while it is standard inthis literature to treat non nancial income asexogenous our ndings emphasize that acomplete understanding of investor portfoliochoice requires private equity holdings to beendogenized

An alternative interpretation of our results isthat they raise the question ldquowhy do peoplebecome entrepreneursrdquo This decision is basedon both the equity return as well as the returnon human capital Since our equity returnestimates account for the labor component ofentrepreneurial activity nding a low equityreturn makes the decision to become an en-trepreneur somewhat puzzling4 In the nalpart of the paper we brie y discuss possibletheories for what motivates entrepreneurs toenter into entrepreneurship and hold such un-diversi ed portfolios of private equity de-spite the unattractive risk-return trade-off Weconsider ve possible explanations for entre-preneurial investment high entrepreneur risktolerance large additional pecuniary bene tslarge nonpecuniary bene ts a preference forskewness and overoptimism and misper-ceived risk

The most related work to our paper is BartonH Hamilton (2000) who documents that indi-viduals in the 1984 Survey of Income ProgramParticipation (SIPP) choose self-employmentdespite facing a median (but not mean) streamof future earnings signi cantly less than thatavailable as a paid employee In addition thecross-sectional standard deviation of self-employed earnings is substantially larger thanthat of wages from paid employment Hamilton(2000) interprets these results as evidence thatlarge nonpecuniary bene ts to self-employmentexist Our data allow for a more comprehensive

3 For example suppose private and public equity eachhave annual returns with (known) standard deviation of 017and a correlation of 05 and that the sample mean returndifference is zero Then one can reject that the mean returnon private equity exceeds the mean return on public equityby 41 percent or more per year at the 5-percent signi cancelevel with 47 annual observations With zero correlationbetween private and public equity one can reject that thedifference exceeds 58 percent at the 5-percent level Thisdoes not account for measurement error in our privateequity returns or uncertainty about the variance or covari-ance Nonetheless it suggests there is hope to establish ina statistical sense that the mean private and public equityreturns are closer than predicted by existing theory

4 Even if the conditional return distribution for someentrepreneurs is attractive given their information thiswould only mean that the conditional distribution of returnsfor other entrepreneurial activities would be even less at-tractive Hence the unattractiveness of the unconditionalprivate equity return distribution indicates that the motiva-tion for at least some group of entrepreneurs is puzzling

747VOL 92 NO 4 MOSKOWITZ AND VISSING-JOslashRGENSEN ENTREPRENEURIAL INVESTMENT

treatment of the equity return component of theentrepreneurrsquos payoff over a longer time periodincluding adjustments for rm entry and exit5

The rest of the paper is organized as followsSection I brie y describes the combination ofdata sources used to analyze the diversi cationof and returns to private equity Section II doc-uments the poor diversi cation of entrepreneur-ialprivate equity investment and compares it toownership of publicly traded stock in rms forwhich a household member works Section IIIconducts a detailed analysis of the returns toprivate equity highlighting a series of issues incalculating these returns and Section IV exam-ines the idiosyncratic risks of private equityinvestment Based on this risk-return trade-offthe observed concentration of wealth in private rms appears puzzling Section V considersvarious explanations for why investors may be-come entrepreneurs and willingly hold so muchundiversi ed private equity Finally Section VIconcludes with a discussion of the results

I Data Sources

In order to analyze private equity holdingsand returns we use data from several sources

A The Survey of Consumer Finances

The rst is the 1989 1992 1995 and 1998Survey of Consumer Finances (SCF) Thesesurveys are nationally representative samples ofabout 4000 households per survey yearWeights are provided to allow aggregation toUS totals A high wealth sample is includedwhich improves the accuracy of estimates ofaggregate wealth and its components The re-spondents provide information on individualhousehold portfolio composition including in-vestment in both private and publicly traded

rms Furthermore characteristics of the house-hold are provided on employment status hoursworked per week demographics and educa-tional attainment as well as on the attributes ofprivate rms in which the household has own-ership Weighting households using the SCFweights about 11 percent of respondents reportto have some ownership in a nonpublicly traded rm (28 percent when not weighting)

Table 1 reports summary statistics on theprivate equity investments in the SCF Panel Adocuments the percent of total private equity invarious lines of business The set of privateequity investments span a variety of industriesOur computation of the returns to private equityencompasses all of these entrepreneurial activ-ities However note that the data is not domi-nated by any particular industry A signi cantfraction of entrepreneurs are in manufacturing(214 percent) and service industries (30 per-cent) as well as retailwholesale (218 percent)Likewise activities that may be more consistentwith consumption or hobbies rather than invest-ment (eg restaurants bars weekend ranchesetc) represent a small fraction of our data

Panel B reports the distribution of entrepre-neurs across various household and rm char-acteristics using data for the rm in which thehousehold has its largest actively managed eq-uity share Most of the entrepreneurs are maleand 403 percent have a college degree Theaverage age of our entrepreneurs is 465 with90 percent of the sample below 65 years of ageThus the majority of private equity investorswith active management interests in our sampleare below retirement age and therefore are notindividuals looking for a ldquohobbyrdquo in retirementFinally there is a wide range of rm sizes in thesample (measured by equity sales pro ts andnumber of employees) with signi cant rightskewness

B Flow of Funds and National Income andProduct Accounts

As an additional supplement to our privateequity data we also employ equity data fromthe Federal Reserve Boardrsquos Flow of FundsAccounts (FFA) and income data from the Na-tional Income and Product Accounts (NIPA)over the 1952 to 1999 time period This data

5 Hamilton (2000) employs various income measures tocapture both the labor component of earnings and the pri-vate equity return His results on the mean payoff aresensitive to the measure used while the median payoff issubstantially below the outside option irrespective of theincome measure used Given the limited amount of equityinformation in his sample (only one year of equity returndata for a fraction of the sample) he focuses on the medianentrepreneur rather than the mean


TABLE 1mdashSUMMARY STATISTICS ON ENTREPRENEURS FROM THE SURVEY OF CONSUMER FINANCES

A Percentage of Private Equity in Each Industry (Average 1989 and 1992)

Industry Percentage of private equity

Agriculture 1302Farm nursery forest management agricultural services landscaping 1302

Retail wholesale 2184Restaurant bar 276Direct sales Amway Avon Mary Kay Tupperware Stanley Home products 003Gas station 008Foodliquor store 170Other retail andor wholesale business 1727

Professionals 1172Professional practice law medicine architecture accounting 996Business management and consulting services 176

Manufacturing 2140Manufacturing printingpublishing oil eld services 1396Contracting construction services plastering painting plumbing 632Trucking moving and storage warehousing 112

Services 2998Beauty shop barber shop 014Personal services hotel dry cleaners funeral home 504Entertainment services dance studio theater 100Communications (cable) TV or radio stations 045Auto repair car wash 149Repair services appliances TV upholstery furniture shoes 024Real estate insurance 1538Various business services advertising equipment rental computer programming 462Banks and brokerage rms mortgage nance company 163

Other 204

B Distribution Across Individual and Firm Characteristics (Average 1989ndash1998)

Characteristic MeanStandarddeviation

Percentile10th 25th Median 75th 90th

Entrepreneur age 465 129 31 37 45 55 65Firm age 107 107 1 3 7 15 25Market equity 186888 1647228 0 4000 25000 100000 300000Sales 4027681 130509000 700 6500 40000 186000 900000Pro ts 344127 8790767 0 1000 10000 50000 160000Employees (including entrepreneur) 187 3379 1 1 2 5 12

Percentage male 811Education (percentages)

Less than high school 95High school 501College graduate 403

Notes Summary statistics for households who own private equity are reported from the 1989 1992 1995 and 1998 SCFPanel A contains summary statistics on the percent of equity each industry category accounted for in the 1989 and 1992surveys Industry statistics pertain to the largest three actively held private equity positions of each household Private equityvalue is net equity if business were sold today plus loans from household to business minus loans from business tohousehold Panel B reports the distribution of entrepreneurs across demographic categories as well as the distribution of rmage (since foundedacquired) and size Panel B uses information for the rm in which the household has the largest activelymanaged position The calculations include all rms with nonzero pro ts or nonzero market equity For the entrepreneur-leveldata the entrepreneur is de ned as the respondent (the male in couples) if heshe is self-employed and the self-employedspouse otherwise All statistics reported use averages across all ve SCF imputations


source provides aggregate statistics on the valueand income of corporate and noncorporate rmson an annual basis We employ this data togenerate additional evidence on private equityreturns

C Other Data Sources

We also supplement our return calculationswith adjustments for IPOs (provided by JayRitter) merger and acquisition activity in pri-vate and public markets [from the SecuritiesData Corporation (SDC)] as well as publicstock return information from the Center forResearch in Security Prices (CRSP) and ac-counting information on public rms fromCompustat Data from the 1993 National Surveyof Small Business Finances (NSSBF) are alsoused to supplement our calculations6

II Entrepreneurial Equity Concentration

We start by comparing the level of diversi -cation of private equity investors to that ofpublic equity investors focusing on ownershipin publicly traded corporations for which ahousehold member is or has been employed asthe most severe candidate for poor diversi ca-tion We nd that private equity investors aredramatically less diversi ed than public equityinvestors

A Ownership in Privately Held Firms

Using data from the SCF Panel A of Table2 documents the poor diversi cation of house-hold portfolios in private equity The value ofprivate equity for a given household is the self-reported value of the householdrsquos share of netequity in the business if it were sold today(Possible reporting bias issues are addressed

later in the paper) We account for entrepreneur-ial leverage in the rm by adding loans from thehousehold to the business and subtracting loansfrom the business to the household We excludethe value of personal assets used as collateralfor business loans This is done to be conserva-tive but does not materially affect the resultsSummary statistics are reported for each surveyyear (1989 1992 1995 and 1998) as well as theaverage across years All gures are calculatedusing SCF weights and are thus representativeof the population of US households We aver-age dollar values across the ve SCFimputations

The rst three rows of Panel A report thepercent of total private equity owned by house-holds with various degrees of net worth devotedto private equity A little more than 75 percentof all private equity was held by householdswho had 50 percent or more of their net worthdevoted to private equity A more direct mea-sure of the poor diversi cation caused by in-vestment in private equity is captured by thenext two rows of Panel A The rows report theaverage percent of net worth invested in privateequity across all households with some privateequity holdings and positive net worth Theaverage household in this group invests 41 per-cent of its wealth (45 percent when weightingby net worth) in private equity consistent withthe ndings of William M Gentry and R GlennHubbard (2001a) This gure does not accountfor human capital and the fraction of this de-rived from labor income in the rm Moreoverthis investment is typically devoted to a singleprivate rm in which the household has anactive management interest The next two rowsof Panel A report the mean percent of privateequity held in the rm representing the house-holdrsquos largest actively managed equity positionThe average household who owns private eq-uity has 82 percent (73 percent when weightedby amount of private equity invested) of itsprivate equity investment in such a rm More-over more than 86 percent of total private eq-uity is held by investors with an activemanagement role in the company in each yearof the SCF Overall these results indicate thatnot only is private equity investment substantialrelative to net worth it is also poorly diversi edand concentrated in the hands of managers

6 The 1993 NSSBF is a rm-based survey of smallbusinesses sponsored by the Federal Reserve Board to pro-vide detailed information on a representative sample ofprivate non nancial nonfarm businesses with less than 500employees The sample represents the population of about 5million small businesses in the United States in operation asof December 1992 The sample covers 4637 smallcompanies


B Own-Company Stock Ownership inPublicly Traded Firms

For comparison to the concentration ofwealth in private equity we document the prev-alence of holdings in public rms in which ahousehold member is or has been employed

Panel B of Table 2 reports that for householdswith own-company stock holdings these con-

stitute the majority of the householdsrsquo directequity investment averaging 738 percent (502percent when weighted by amount of directlyheld public equity) As a fraction of all publicequity held both directly and indirectly throughmutual funds IRAs pension plans and annu-ities and trusts own-company stock accountsfor about 524 percent (341 percent whenweighted by amount of total public equity

TABLE 2mdashPRIVATE EQUITY AND OWN-COMPANY STOCK OWNERSHIP

A Private Equity Ownership

Measure 1989 1992b 1995 1998 Average

Percentage of total private equity owned by households witha

$ 25 percent net worth in private equity 922 924 932 917 924$ 50 percent net worth in private equity 762 733 772 747 754$ 75 percent net worth in private equity 408 469 503 479 465

Mean percentage of net worth invested in private equity for households with positive private equity and net worthSCF weights only 423 450 372 399 411Weighted by net worth 454 456 457 440 452

Mean percentage of private equity held in one actively managed rm for households with positive private equitySCF weights only 779 829 825 848 820Weighted by amount of private equity 728 707 740 735 728

B Own-Company Stock Ownership in Public Firms

Measure 1989 1992 1995 1998 Average

Percentage of total public equity owned by households with$ 25 percent of their public equity in own company 134 125 109 125 123$ 50 percent of their public equity in own company 104 90 67 62 81$ 75 percent of their public equity in own company 56 43 37 36 43

Mean percentage of net worth invested in own-company stock for households with positive own-company stock and networth

SCF weights only 87 69 108 104 92Weighted by net worth 77 89 102 127 99

Mean percentage of directly held public equity in own-company stock for households with positive own-company stockcSCF weights only 777 775 691 710 738Weighted by amount of directly held public equity 547 491 477 492 502

Mean percentage of directly and indirectly held public equity in own-company stock for households with positive own-company stockd

SCF weights only 670 556 469 402 524Weighted by total public equity held 436 318 308 303 341

Notes Private and own-company stock ownership for households are reported from the 1989 1992 1995 and 1998 SCF aswell as the average across all four survey years Panel A contains information on private equity ownership and Panel Bcontains information on own-company stock holdings in public corporations de ned as ownership in a public rm for whicha household member is or has been employed All statistics reported are averages across all ve SCF imputations

a Ownership by households with negative net worth includedb For 1992 data for two households with very small values of net worth for one of the imputations were deletedc In each year a few households report holding more directly held own-company stock than their total direct stock holdings

For these we set the percent of own-company stock in directly held equity to 100d In each year a few households report holding more directly held own-company stock than their total direct and indirect

stock holdings For these we set the percent of own-company stock in directly and indirectly held equity to 100


TABLE 3mdashTHE SIZE OF PRIVATE AND PUBLIC EQUITY MARKETS

1989 1992 1995 1998

A Private Equity ($ billion) SCF

Proprietors and partnerships (market value) 2026 1977 1991 2511S and C corporationsa (market value) 1661 1780 2302 3226

Total private equity (market value) 3687 3757 4293 5737

Public equity (market value) 1587 2102 3439 7256Ratio privatepublic equity 232 179 125 079

Pro ts ($ billion)Pretax proprietors and partnerships 335 430 458j 534After-tax S and C corporationsb 267 288 341 496Pro ts 2 retained earnings PampP (20 percent retained) 268 344 367 427Pro ts 2 retained earnings SampC (2040 percent retained) 175 194 244 355

Labor income ($ billion)Total salary paid to self-employed managers 141 191 159 300(Hours worked) 3 (estimated wage rate)c for entrepreneurs

with no self-employment salary 175 193 229 232Proprietors and partnerships 152 155 200 172S and C corporations 23 38 30 60

Price-to-earnings ratio 61 52 54 56Price-to-dividends ratiod 138 109 112 104

B Private Equity ($ billion) FFANIPA

Equity in noncorporate businesse 3102 3127 3599 43942 Value of 1ndash4 family rental properties 942 1003 1135 1272

5 Proprietors and partnerships (market value) 2160 2124 2463 3122

S and C corporations (market value) (estate multiplier 5 2) 1412 1220 1585 2067S and C corporations (market value) (estate multiplier 5 3) 2117 1830 2377 3101

Total private equity (market value) (estate multiplier 2) 3571 3344 4048 5190Total private equity (market value) (estate multiplier 3) 4277 3954 4841 6223

Ratio privatepublic equity (estate multiplier 2) 108 076 060 039Ratio private(070 public) equity 155 109 086 056

Income and dividends ($ billion)Proprietorsrsquo income 362 434 498 624Adjusted proprietorrsquos income 2 retained earningsf 209 247 336 519Dividends S and C corporationsg 147 176 236 376

C Public Equity ($ billion) Center for Research in Security Prices

Market value 3292 4376 6734 13217

New issues and takeovers three-year total ($ billion)hNew issues 42 76 110SDC MampA adjustment to private equityi 55 129 421SDC private acquisitions of public rms 34 31 58

Notes The aggregate market values of all private and public equity as well as various pro t measures are reported Estimates are obtained fromtwo sources Panel A contains data from the 1989 1992 1995 and 1998 SCF averaging over all ve imputations Panel B contains data fromthe FFANIPA over the same years Panel C contains data on publicly traded equity (NYSE AMEX and NASDAQ) from the Center forResearch in Security Prices (CRSP) over the same period

a Included in this category are rms of unknown type and other types of corporationsb After-tax pro ts assume a 30-percent corporate tax rate which only applies to C and other corporations and type unknown rms Pro ts

from S corporations are included pretaxc Hours worked by head andor spouse for self-employed persons with positive equity in a business in which they have an active

management role and who did not report receiving a salary Estimated wage rates are determined by rst regressing hourly wage rates ofhousehold members who are not self-employed on educational and demographic attributes and then using the regression equation to predictwage rates of self-employed household members with no salary reported

d ldquoDividendsrdquo refer to pro ts minus retained earnings minus the labor adjustment for self-employed individuals who do not report a salarye Equity in noncorporate business is de ned as (tangible assets 1 nancial assets) 2 liabilities Tangible assets consist of real estate (at

estimated market value) and equipment software and inventories (at estimated replacement cost)f We adjust PampP income in three ways First we change the adjustment for misreporting of pro ts on income tax returns to be 75 percent

in each year from 1959 onward implying that for every $1 of pro ts reported to the IRS adjusted pro ts are $175 Second we subtract thecapital consumption adjustment included in NIPA pro ts from earnings to get a measure of the actual pro t ows to proprietors Third as ameasure of actual retained earnings in the rm we add capital expenditures plus net acquisition of nancial assets minus net increase inliabilities (excluding ldquoproprietorsrsquo net investmentrdquo) This measures the amount owners must have invested to cover rm investment whetherfrom pro ts or additional paid-in funds


invested) of a householdrsquos total public equityholdings Relative to net worth however invest-ment in own-company stock for public rms is farless important As a fraction of household networth investment in own-company stock isonly 10 percent compared to 45 percent forprivate rms Furthermore households withover 25 percent or more of their equity holdingsin own-company stock own only about 12 per-cent of total equity investment in public rmsHouseholds with at least 50 percent and 75percent of their equity holdings in own-com-pany stock comprise only 8 and 4 percent re-spectively of total public equity investmentHence owners of own-company stock in publiccompanies are not as poorly diversi ed as own-ers of private equity and own only a smallfraction of public equity7 It should be notedthat households may hold undiversi ed portfo-lios of public equity without owning any own-company stock However Vissing-Joslashrgensen(1999) shows that 913 percent of public equityheld in the 1995 SCF is owned by householdswith at least ve directly held stocks or half or

more of their equity holdings in indirect form(eg mutual funds retirement plans etc)This underscores the importance of analyzingand understanding investment in private equity

III The Returns to Private Equity Investment

Due to the lack of a comprehensive paneldata set on entrepreneur investments we exam-ine the returns to an index of all private equityby aggregating all the private rm values andpro ts to US totals Only by aggregation canwe account for rm entry and exit over time andassign the proper returns In the next section weargue that the private ldquoindexrdquo return is likely tobe an upward-biased estimate of the averageindividual rm return (when focusing on geo-metric buy-and-hold returns)

A The Size of the Private Equity Market

We begin by rst comparing the size of theprivate and public equity markets We employ twodata sources for our estimates of the size andreturns of this market The rst is the 1989 19921995 and 1998 SCF and the second is the FFAfrom 1952 to 19998 Panel A of Table 3 reportsthe size of the private equity market estimatedfrom the SCF using the household weights pro-vided Total market value of private equity held inbillions of dollars are reported for two types of rms proprietorships and partnerships and S andother corporations (with unknown rm types in-cluded in the latter category) In computing thetotal amount of private equity investment (andtheir returns) we again deduct collateral posted bythe entrepreneur for loans to the rm This is done

7 The numbers in Table 2 do not include own-companystock held indirectly through pension plans or employeestock-ownership plans (ESOPs) However the Departmentof Labor estimates (based on Form 5500 led with theInternal Revenue Service) that of the total $1024 billion inassets of de ned contribution plans with 100 or more par-ticipants in 1995 $165 billion was invested in employerstock ESOPs with 100 or more participants account foranother $100 billion of investments in employer equityBased on the 1995 SCF the total dollar amount of directlyheld own-company stock is $272 billion about the same asholdings through pension plans and ESOPs combined Thetotal amount of direct and indirect holdings of publiclytraded stock by households in the 1995 SCF is $3439billion implying that (165 1 100 1 272)3439 5 156percent of total public equity held directly or indirectly byhouseholds is owned by employees This is still consider-ably less concentrated than private equity

8 For a comparison of the SCF and FFA equity numbersas well as the numbers for many other asset categories seeRochelle L Antoniewicz (2000)

TABLE 3mdashContinued

g We estimate dividends paid out by private S and C corporations as total dividends paid by all corporations (from NIPA) minus dividends paidby public corporations (from CRSP) In addition we add 20 percent of the NIPA income underreporting adjustment made to total corporate pro ts

h Results in the three columns reported are for 1990ndash1992 1993ndash1995 and 1996ndash1998i The total change to private equity totals from merger and acquisition activity obtained from SDC and Table 5 Table 5 describes the various

adjustments to the private equity totalsj The SCF pro t total for PampP in 1995 is very sensitive to one outlier (household number 1921) The ownership share of this respondent

is imputed and generates a very implausible value for the dollar amount of rm pro ts which are attributable to the respondent We use insteadas our SCF PampP pro t total for 1995 a weighted average of the 1992 and 1998 SCF PampP pro t totals The weights re ect the percentage ofSCF SampC pro t growth from 1992 to 1998 that occured between 1992 and 1995


to be conservative so that private equity valueswill not be in ated by the inclusion of personalassets posted as collateral

As Table 3 shows the market value of privateequity has risen steadily from 1989 to 1998 inlarge part due to an increase for S and othercorporations The total dollar amount of privateequity is substantial ranging from $37 trillionin 1989 to $57 trillion in 1998 The SCF esti-mate of the total holdings of public equity byhouseholds has similarly risen sharply over thedecade covered by the four surveys (from $16trillion to $73 trillion)9 The growth in publicequity value has outpaced that of private equityThe private market was 23 times larger than thepublic market in 1989 but was only 79 percentas large as the public market by the end of 1998This suggests that the returns to public equitywere larger than those of private equity over thistime period Also reported is the average price-to-earnings ratio (PE) and price-to-dividendsratio (where dividends are pro ts minus re-tained earnings minus a labor adjustment de-scribed below) which average 56 and 116over the sample period respectively in the pri-vate market These are signi cantly smallerthan those in the public market

We also estimate the size of the private equitymarket from data obtained from the FFA Forcomparison to the SCF estimates we show theFFA data for 1989 1992 1995 and 1998 FFAnoncorporate equity is de ned as tangible and nancial assets minus liabilitiesTangible assetsconsist of real estate (at estimated market value)plus equipment software and inventories (atreplacement cost) As described in Antoniewicz(2000) the FFA noncorporate equity includesthe market value of 1ndash4 family rental proper-ties To obtain a number more comparable tothe SCF we subtract from the FFA number anestimate (based on aggregate data from the Bu-

reau of Economic Analysis) of the market valueof such properties

The resulting estimates of (noncorporate)proprietorship and partnership equity are fairlysimilar to those from the SCF in Panel A TheFFA numbers for equity in corporations aremore problematic Equity in S and C corpora-tions refer to both equity in publicly tradedcorporations and equity in privately held rmsThe FFA estimates the value of closely held(nonpublic) corporations from estate tax re-turns but do not publish separate series forpublicly traded corporate equity and nonpubliccorporate equity The speci cs of the approachare proprietary and they would not release theirseries To obtain an estimate of nonpublic cor-porate equity we considered subtracting fromthe FFA number the estimate of the marketvalue of public equity from CRSP which isreported at the bottom of Table 3 in Panel CHowever this produces an extremely volatile Sand C private equity series since it is the resid-ual which thus also captures any de nitionaldifferences between the FFA and CRSP As analternative measure (that is still independent ofthe SCF equity totals) we adopt a method usedby the IRS for estimates of wealth that is alsobased on estate tax returns see Barry W Johnson(2000) This method is useful since the vastmajority (over 90 percent) of equity in privatecorporations is owned by the population repre-sented on estate tax returns (ie those withassets over $600000) The estimation relies onan estate multiplier which re ects the probabil-ity that a given dollar of wealth shows up onestate tax returns for a given year The multi-plier used by the IRS is around 100 from 1989to 1995 We report numbers for multipliers of200 and 300 which we argue is a better multi-plier for private equity-holders who are un-likely to have the same mortality rates as thegeneral population in the same age and wealthcohort While obtaining precise multipliers isdif cult Appendix A provides some support forour multipliers based on health and expectedlife-span questions from the SCF This methodcan only be applied to the FFA gures from1989 to 1999 but not for the longer period 1952to 1999 due to data limitations Consequentlywe will focus on proprietorships and partner-ships from the FFA when examining the longer

9 These numbers include estimates of householdsrsquo own-ership of public equity through mutual funds de ned con-tribution retirement plans and trusts Since part of publicequity is owned by de ned bene t retirement plans includ-ing state and local government retirement plans or bynonpro t organizations insurance companies and foreign-ers the SCF public equity totals will be lower than theCRSP total market value for public equity


time period The FFA estimates of corporateprivate equity obtained by this method areslightly smaller than the estimates based on theSCF when using a multiplier of 200 and slightlylarger using a multiplier of 300

Using these numbers the total size of theprivate equity market based on the FFAestatetax return data is substantial and is larger thanthe public equity market in the 1989 data Ac-counting for the fact that individuals own about70 percent of corporate equity (direct and indi-rect holdings) the ratio of private-to-public eq-uity held by households is again large

B Returns to an Index of All Private Equity

We begin by calculating the returns to a val-ue-weighted index of all private equity based onthe 1989 to 1998 SCF data In order to estimatethe returns to private equity holdings we usethe household estimates of the market value andpro ts of the private rms being held as re-ported in Table 3 The pro ts reported byhouseholds are pretax earnings for the year priorto the survey Although these numbers are self-reported by households they are anonymousand not subject to tax scrutiny However wewill address later whether reporting biases arelikely to have in uenced our return calculationsand how we can account for these possibledistortions

We rst convert pretax earnings of C corpo-rations into after-tax pro ts by subtracting anestimate of the taxes due assuming a 30-percentcorporate tax rate Table 3 reports both thepretax pro ts of proprietorships and partner-ships and after-tax pro ts of corporations (withno adjustment for S corporations who are ex-empt from corporate taxation) Since earningsare reported for the year prior to each survey(and surveys occur only every three years) wereport the average of the returns obtained usingthe current and the previous surveyrsquos earningsestimates Thus the returns over the rst surveyperiod 1990 to 1992 are the average of thegeometric annualized returns using 1988 and1991 earnings respectively

To avoid double-counting earnings as both apotential dividend to investors as well as a cap-ital gain we make an assumption about thefraction of (after-tax) earnings that are retained

in the rm Since the SCF does not record howmuch of earnings are paid out to shareholderswe assume that 40 percent are retained in Ccorporations This corresponds roughly to theratio of retained earnings to after-tax pro ts forC corporations in the NIPA data over the period1989 to 1998 External nancing is likely to bemore costly for private rms than for largerpublic rms Therefore it is likely that private Ccorporations retain more in the rm than largerpublic rms Increasing the retention rate wouldlower our subsequent return estimates hencethe 40 percent retention assumption will if any-thing bias our returns upward Since S corpo-rations proprietorships and partnerships areoften smaller than C corporations one may ex-pect them to face even higher costs of external nancing and thus have higher retained earn-ings On the other hand they may have fewergrowth opportunities so we conservatively as-sume their retention is half that of C corpora-tions (ie 20 percent) Pro ts after retainedearnings are reported in Table 3

Using the market value of private equity atthe beginning and end of each survey periodplus the after-tax pro ts adjusted for retainedearnings we compute the return on private eq-uity over the years between each survey Table4 Panel A reports the geometric average annualreturn from investing in private equity over thethree survey periods From 1990 to 1992 theaverage return is 123 percent per year from1993 to 1995 the average return is 170 percentwhile it is 222 percent from 1996 to 1998

Panel B of Table 4 reports the returns to theCRSP value-weighted index of NYSE AMEXand NASDAQ public equity over the same timeperiod for comparison The geometric averageannual return to public equity is 110 146 and247 percent for the 1990 to 1992 1993 to 1995and 1996 to 1998 periods respectively Thesereturns are similar to those from private equityin the SCF (a bit lower from 1990 to 1995)Since private rms are much smaller and riskierthan large public companies represented by theCRSP value-weighted index perhaps a bettercomparison is to the returns on the smallestdecile of publicly traded rms Over the threesurvey periods the geometric average annualreturns on the smallest decile of CRSP rms is305 203 and 220 respectively These are


TABLE 4mdashTHE RETURNS TO PRIVATE EQUITY (1990ndash1998)

A Private Equity Returns

Data from the SCF

Retained earnings Adjustments Annual returns (percent per year)

C corporations P PampS LaboraFirmbirths IPOs MampAb

Taxevasion

PampPndashSampC 1990ndash1992 1993ndash1995 1996ndash1998

1) All 040 020 mdash mdash mdash mdash yes 123 170 2222) PampP 020 mdash mdash mdash mdash yes mdash 126 156 2303) SampC 040 mdash mdash mdash mdash yes mdash 120 185 2144) All 040 020 yes mdash mdash mdash yes 82 127 1845) PampP 020 yes mdash mdash mdash yes mdash 64 94 1596) SampC 040 yes mdash mdash mdash yes mdash 109 169 2067) All 040 020 yes yes mdash mdash yes 75 116 1648) All 040 020 yes yes yes mdash yes 78 121 1709) All 040 020 yes yes yes yes yes 82 130 194

10) PampP 020 yes yes yes yes yes yes 74 89 15411) SampC 040 yes yes yes yes yes yes 97 176 22812) All 040 0 yes yes yes yes yes 103 154 217

Data from the FFANIPA

SampC PampP

13) Alld actual actual yes mdash mdash mdash yes 41 167 22414) Alle actual actual yes mdash mdash mdash yes 21 147 19415) PampP actual yes mdash mdash mdash yes mdash 19 123 19816) SampCd actual yes mdash mdash mdash yes mdash 65 226 25517) SampCe actual yes mdash mdash mdash yes mdash 24 177 197

B Public Equity Returns

Source

18) CRSP data value-weighted index 110 146 24719) CRSP data smallest decile 305 203 22020) SCF data 132 207 30021) SCF data with IPO and takeover adjustmentc 131 203 298

Notes Panel A reports the returns to all private equity based on estimates of the size of privately held equity and their earningsfrom Table 3 The return estimates pertain to data from the 1989 1992 1995 and 1998 SCF as well as the FFANIPA Returnsare calculated using various assumptions about retained earnings the labor component of pro ts sample composition changesdue to entry and exit of rms and underreported pro ts due to tax evasion When separating returns by proprietorships andpartnerships (PampP) versus S and C corporations (SampC) we assume 21 percent of PampPs transfer to private corporations inorder to account for the in ow and out ow of equity values to both types of rms (denoted by a ldquoyesrdquo in the PampPndashSampCcolumn) Panel B reports returns to publicly traded equity over the same time period from CRSP All returns are nominalgeometric average returns over the three subperiods from 1990 to 1998

a When salaries are not reported for self-employed households the salary adjustment is the hours worked by head or spousefor self-employed persons times the estimated hourly wage rate for the person Estimated wage rates are determined by rstregressing hourly wage rates of household members who are not self-employed on educational and demographic attributesand then using the regression equation to predict wage rates of self-employed household members who do not report a salary

b Obtained from Securities Data Corporation for each year over the survey period A summary of the adjustments aredescribed and reported in Table 5

c IPO and takeover adjustments assume households own 70 percent of all public equity This corresponds approximatelyto the share of corporate equity owned by households (directly and indirectly) over this period in the FFA

d Estate multiplier 5 2e Estate multiplier 5 3


considerably higher than the private equity re-turns for the 1990 to 1992 period and quitesimilar for the other two periods Other small- rm indices performed worse than the CRSPindex in the 1990rsquos however Given the dispar-ity in performance across various small- rmindices in the 1990rsquos we compare the privateequity returns for this period to the returns onthe entire public index

These are our basic private equity return es-timates which are likely to be biased in severalways In the rest of this section we quantifythese biases as best we can Correcting for someof the biases leads to higher private equity re-turns while correcting for others leads to lowerprivate equity returns We will argue howeverthat our most accurate private equity returns arelower than those reported above

1 Accounting for Labor IncomemdashThe mostimportant effect not accounted for above isthat the private equity returns contain the partof pro ts that re ects the labor input of theentrepreneur This component is not return toequity but rather captures the fact that manyentrepreneurs do not pay themselves a salaryFor these entrepreneurs part of their compa-niesrsquo pro ts should be viewed as payment forhours worked rather than return on equity

Speci cally our baseline return estimates ac-count for salaries withdrawn from the private rms by self-employed managers since they arealready subtracted from the earnings numbersreported (for reference the amount of such sal-aries are reported in Table 3) However theSCF private equity-holders include many re-spondents with actively managed equity posi-tions who do not report a salary to themselvesTherefore we make an adjustment to earningsfor this labor component for individuals (headandor spouse) who report being self-employedhave ownership in a private company in whichthey have an active management interest butfail to report a salary taken To do so we use thereported weeks worked per year and hoursworked per week We multiply the annual hoursworked by an estimated wage rate for similarindividuals in the survey who worked in paidemployment Speci cally for respondents whoreported to work in paid employment (ie notself-employed) we regress their hourly wage

rate on a constant their age age squared adummy variable for having a high-school di-ploma but not a college degree a dummy forgraduating college and a dummy for their gen-der We run one regression for heads of house-holds (de ned as the male in couples) and oneregression for spouses Using the regression co-ef cients we then estimate the wage rate forself-employed individuals who do not report asalary by multiplying their demographic andeducation characteristics by the estimated coef- cients and using the predicted value as theirhourly wage rate This procedure does not ac-count for any unobserved differences betweenself-employed and other individuals In fact theresults of Hamilton (2000) suggest that thisshould lead to a labor adjustment that is too smallthus biasing our private equity return estimatesupward He shows using a sample selectionmodel that the mean wages of employees are lessthan the expected wages of entrepreneurs had theybeen paid employees Furthermore entrepreneursreturning to paid employment are found to earn ahigher wage than other employees with the sameobservable characteristics These ndings suggestthat more talented individuals self-select intoentrepreneurship10

We then subtract the estimated annual wagefor those not reporting a salary from earningsand recompute returns The fourth row of Table4 Panel A shows that the labor adjustment re-duces the estimated returns by about 4 percentper year (65 percent for proprietors and part-nerships and 12 percent for S and C corpora-tions) indicating its importance in thesecalculations With this adjustment returns toprivate equity are considerably smaller thanthose for public equity

10 As a check on our procedure we also compare thesalaries taken by self-employed households who do report asalary to what our regression approach would have pre-dicted their salary to be The average reported salary acrossall entrepreneurs who report a salary is 116 times the salaryour regression approach suggests (For proprietorships part-nerships and S corporations this ratio is 110 for C corpo-rations it is 133) This likely con rms the selection issuesemphasized by Hamilton (2000) For C corporations it mayalternatively re ect excessive salaries reported by someentrepreneurs for tax reasons Using estimated rather thanactual reported salaries for C corporations only has a smalleffect on returns


2 Accounting for Firm Entry Births andNew EquitymdashThe previous computations as-sume that the composition of rms in the SCFis the same at the beginning of each three-year survey period as it is at the end Whilethe SCF employs the same sampling proce-dure and questions for each of the surveysthere will be sample composition differencesbetween survey years that may distort thereturn estimates

First a possible distortion of the compositionof rms that comprise the beginning and end-of-period private equity values occurs whennew private rms are ldquobornrdquo between the twosurvey years Since end-of-period gures con-tain rms created after the previous survey thevalues should not be attributed to initial equity-holders from the previous survey year To takethis into account we recompute returns bydropping rms at the end of the period that werefounded (but not those that were bought orinherited) less than three years ago This is donefor the earnings estimates and labor componentcomputations as well The returns drop by 07 to2 percent per year

Similarly new equity invested in existing rms should not be attributed as a capital gainto original private equity-holders To estimatethe average value of new equity injected intoprivate rms each year we employ data fromthe 1993 NSSBF In this survey respondentsare asked ldquoDuring the last three years has the rm obtained additional equity capital fromexisting owners their relatives or from newor existing partnersrdquo And if yes how muchUsing the NSSBF weights one can aggregatethe responses to US totals and divide by 3 toget annual numbers The aggregated annualtotal for 1993 was 28 billion dollars whenexcluding funds raised for ldquobusiness expan-sion acquisitionrdquo (which we address below)and excluding the few public rms in theNSSBF Since the population of rms coveredby the NSSBF have fewer than 500 employ-ees equity raised by the biggest private rmswill not be covered Thus our returns may beoverstated As we do not have annual data forthis adjustment it is not included in Table3 However this effect likely cancels with anomitted effect from rm exit which we de-scribe below

3 Accounting for Firm Exit IPOs Mergersand Acquisitions Failures and LiquidationsmdashAs will be documented in the next section exitrates for private rms are large and include saleto new owners (including acquisitions andIPOs) as well as liquidations and failures If a rm goes public between two surveys then itwill no longer be contained in the end-of-period gures for private equity Since IPOs are gen-erally the most successful private companiesignoring these would understate the returns toprivate equity To take this into account we addthe total market value of all initial public offer-ings over the three years between surveys to theend-of-period value of private equity The effectof IPOs is rather small increasing average re-turns by only about 50 basis points per year

Another possible distortion concerns mergerand acquisition activity between the surveyyears Speci cally when a private rm isbought out by a public company between sur-veys the value of that private rm will nolonger be contained in the end-of-period privateequity value Ignoring this will understate re-turns As for sale to new private owners noadjustment to private equity returns is needed ifthe new owners hold as much equity in the rmas did the previous owners If the previousowners get more equity out than the new ownersput in (ie due to increased nancing with debtor internal funds or from foreign equity inves-tors) then our private equity returns should beincreased by the amount of the differenceTherefore we need to determine the extent towhich private rms are acquired by public com-panies (whether foreign or domestic) by for-eign private companies (irrespective of howfunded) and by domestic private companiesfunded by debt or internal funds and add backthese components to private equity values

On the other hand if domestic private rmsraise new equity to acquire foreign targets thisshould be subtracted from our private equitytotals since the gains from such acquisitionswill accrue to foreign entrepreneurs Likewisepublic rms acquired by private rms fundedwith newly raised equity will also overstate ourreturns Hence we need to subtract these fromprivate equity totals

To account for these effects we examine thetotal dollar amount and number of transactions


of merger and acquisition activity in private andpublic rms using data from Securities DataCorporation (SDC) over the period 1989 to1998 We focus only on completed transactionsand whether the acquirer and target is a privateor public rm whether foreign or domestic andwhether the acquisition was funded with equityor with debt or internal funds11

Table 5 reports the total dollar amount in mil-lions and total number of transactions involvingpublic rm acquisitions of private rms private rm acquisitions of other private rms and pri-vate acquisitions of public rms over each of thethree subperiods from 1990 to 1998 One problemwith the SDC data is that a signi cant number ofdeals have missing values Consequently the totalvalue reported only pertains to those deals withavailable price information which are typicallythe largest transactions Rather than employing theaverage value for the missing observations whichwould overstate our private equity returns weestimate the value of missing deals using a pre-dictive regression approach similar to that em-ployed for entrepreneurs with missing salariesThe details are provided in Appendix B Theseestimated values are added to the value of dealswith price information to produce a total orldquoscaledrdquo value for each subcategory Table 5 re-ports the sum of these values over the threesubperiods The sum of all changes are added tothe end-of-period total value for private equity inTable 3

As indicated in the ninth row of Panel A ofTable 4 accounting for mergers and acquisi-tions adds an additional 04 percent per year toprivate equity returns over the 1990 to 1992period about 1 percent per year from 1993 to1995 and 24 percent per year from 1996 to1998 However the modi ed returns remainsubstantially below the returns to public equity

The SDC database covers the largest mergersand acquisitions Data on sales of small busi-nesses to new owners as well as equity recov-ered in liquidations is not available annually Toevaluate the impact of such transactions we usethe 1993 NSSBF According to the US SmallBusiness Administration (2000) about 500000employer rms discontinued each year duringthe 1989 to 1998 period The upper bound onthe decrease in rm equity at sale or liquidationis the amount of assets held by such rms In the1993 NSSBF the median asset holdings for all rms with less than 500 employees (usingNSSBF weights) is about $70000 Thus if thetypical discontinued rm was of median sizethe upper bound on the total adjustment neces-sary is 35 billion dollars per year In realitymost of the discontinued rms are liquidationsor failures rather than sales to new owners (seeSection IV) Thus the relevant adjustment ismuch smaller than 35 billion dollars and there-fore likely cancels with the 28 billion dollars ofnewly raised equity by existing rms discussedin the previous subsection

We believe the returns in line 9 of Table 4 arethe most accurate returns to private equity Thefollowing summarizes our computations andvarious adjustments to earnings and private eq-uity values in Table 4

(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3

1 MampA tt 1 3 2 MVt 1 3age3

(3) AE tt 1 3 5 ~E tt 1 3 2 E tt 1 3age3~1 2 tc

3 ~1 2 rRE 2 LC tt 1 3

tc 5 tax rate ~030 for C Corps

0 for S Corps and PampPs)

rRE 5 earnings retention rate

~040 for C Corps


11 SDC records a host of information about globalmerger and acquisition activity from 1983 to 2001 includ-ing public status of the target and acquirer where it islocated and the source of funds employed in the deal Thesources of funds include borrowing from outside lendersbridge loans debt issues foreign lenders junk bonds creditlines and mezzanine nancing which we code as ldquodebtrdquosources as well as funding from internal sources We ag-gregate all deals with debt or internal funds sources into onecategory The rest are deals funded by common and pre-ferred equity


TABLE 5mdashMERGER AND ACQUISITION ACTIVITY IN PRIVATE AND PUBLIC FIRMS

Acquirer

1990ndash1992 1993ndash1995 1996ndash1998

Public Private Private Public Private Private Public Private PrivateTarget Private Private Public Private Private Public Private Private Public

All Acquirers All TargetsValue ($ million) $ 62236 $24059 $70989 $109702 $32358 $ 90217 $287669 $ 69727 $136736Number of deals 6290 4338 2397 10451 5716 3828 18942 8118 3723Number of deals

wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987

All Acquirers Domestic TargetsValue ($ million) $ 30579 $11116 $30310 $ 67448 $14193 $ 26764 $192238 $ 27519 $ 50155Number of deals 3141 1181 1221 5737 1535 1814 10711 2467 1787Number of deals

wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073

Domestic Acquirers Domestic Targets Debt or Internally FundedValue ($ million) $ 3483 $ 3068 $ 8794 $ 12015 $ 3568 $ 4632 $ 28592 $ 5832 $ 16806Number of deals 163 88 70 391 102 57 511 84 86Number of deals

wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198

Foreign Acquirers Domestic TargetsValue ($ million) $ 6400 $ 5919 $12574 $ 7654 $ 6110 $ 10831 $ 17836 $ 11738 $ 19858Number of deals 432 239 588 425 304 1013 737 447 970Number of deals

wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073

Domestic Acquirers Foreign Targets Equity FundedValue ($ million) $ 2081 $ 222 $ 8635 $ 6138 $ 631 $ 9306 $ 16907 $ 1893 $ 4595Number of deals 374 100 84 728 195 151 1548 299 110Number of deals

wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083

Domestic Acquirers All Targets Equity FundedValue ($ million) $ 23291 $ 4216 $20262 $ 55227 $ 6201 $ 21784 $165406 $ 15420 $ 25138Number of deals 2938 988 666 5683 1359 911 11054 2258 872Number of deals

wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

Total D Private Equity Value(1) 1 (2) 2 (3) 5 $54795 $128936 $421066

Notes The total dollar amount (in $ millions) and total number of transactions of merger and acquisition activity in privateand public rms are reported above over the three subperiods 1990 to 1992 1993 to 1995 and 1996 to 1998 Data are fromSecurities Data Corporation (SDC) and correspond only to completed transactions Statistics are reported separately for public rm acquisitions of private rms private rm acquisitions of other private rms and private rm acquisitions of public rmseach broken down further into domestic acquirers and targets foreign acquirers and targets and acquisitions funded with debtor internal cash and equity Also reported are the number of transactions with available price information and a scaled dollarvalue for all deals using an estimated value for deals with missing transaction value as detailed in Appendix B The totalchange in private equity value from this activity is reported at the bottom of the table

a Calculated as follows For column (1) (Private-to-Public) 5 scaled value of all acquisitions of domestic targets Forcolumn (2) (Private-to-Private) 5 scaled value of domestic acquisitions of domestic targets funded by debt or internal funds 1scaled value of foreign acquisitions of domestic targets 2 scaled value of domestic acquisitions of foreign targets funded byequity For column (3) (Public-to-Private) 5 scaled value of domestic acquisitions of all targets funded by equity


where R tt13 is the return over the three-yearperiod between surveys (which is reported as ageometric average annual return) AMV t13 isthe aggregate market value of all private rmsthree years or older at time t 1 3 plus the valueof private rms in existence at date t who wentpublic or were acquired by a public rm be-tween dates t and t 1 3 AE tt13 is the adjustedaggregate earnings of all private rms from datet to t 1 3 IPOtt13 MampAtt13 and LCtt13are the total value of IPOs acquisitions of pri-vate rms and the labor component of pro tsrespectively over the period t to t 1 3 Differ-ent return estimates in Table 4 include or ex-clude these various adjustments

C Returns Across Firm Type

The returns to private equity we have docu-mented pertain to all rms not held publiclyWhile we would like to compute private equityreturns across industries this cannot reliably bedone using the SCF data given the fairly smallnumber of observations in each of the industrycategories As noted in Table 1 our sample ofentrepreneurs are not dominated by any partic-ular industry

We can however compute returns separatelyfor proprietors and partnerships and S and Ccorporations using the 1993 NSSBF to estimatethe percent of proprietor and partnership equitywhich ldquomigratesrdquo to S and C corporation equityeach year The NSSBF provides both currentand 1992 scal year corporate status fromwhich we can quantify the migration of rmsfrom PampP to SampC This is important sincemany of the most successful PampP rms becomeS and C corporations as they expand We esti-mate the migration rate from PampP to SampC to be21 percent of proprietor and partnership equityper year12 Using this rate as well as attributingall IPO and merger activity to S and C corpo-rations and employing a labor adjustment of 65percent for PampP and 12 percent for SampC lines10 and 11 of Table 4 report returns across thetwo rm types With all of the return adjust-ments returns to equity in S and C corporations

are 23 percent per year higher from 1990 to1992 87 percent higher from 1993 to 1995 and74 percent higher from 1996 to 1998 than re-turns to equity in PampP rms However even thehigher SampC returns are lower than those of thepublic market in two of the three subperiodsPublic equity outperformed PampP private equityin all three subperiods by between 36 and 93percent per year We now consider further ro-bustness checks on the SCF private equityreturns

D Robustness of the Return Estimates

We consider robustness issues and possiblereporting biases in the SCF to gauge whetherthese could distort our return estimates

1 Retained Earnings SensitivitymdashFor ro-bustness and as an overestimate of the returnsto private equity the twelfth row of Panel Aassumes that proprietors partnerships and Scorporations do not retain any earnings This isan extreme assumption since it implies that ac-tual retained earnings for these rms will bedouble-counted as both a dividend and capitalgain However the private equity returns arestill below those of the public market in two ofthe three time periods

2 Understated Pro ts Due to Tax EvasionmdashSince the SCF is based on interviews and nottax returns it is not clear whether respondentsreport their true pro ts or the pro ts as stated ontheir tax forms However as long as respon-dents trust that the SCF will not release infor-mation to other government agencies (which theSCF goes to great lengths ensuring) householdshave no incentive to hide their true pro ts Thisis supported by the fact that the SCF pro ts forPampPs are quite close to the corresponding NIPApro ts (proprietorrsquos income) The latter arebased on pro ts as reported to the IRS with a75-percent adjustment for income underreport-ing on tax returns (more detail below) The SCFpro ts are almost identical to the adjusted NIPApro ts in 1992 and within 15 percent of theNIPA pro ts in the other three years Further-more evidence from evaluation studies of the1977 economic censuses also suggests thathouseholds do in fact report higher income to

12 This may even be overstated since the survey was elded between March 1994 and January 1995 Thus thetwo rm-type observations are more than one year apart


surveys than to tax authorities For these cen-suses the Census Bureau conducted additionalspecial surveys of small rms for which taxreturn information had been used in the originaleconomic censuses The income reported in thespecial surveys consistently exceeded the infor-mation based on tax returns13

3 Reporting BiasesmdashThe SCF is consid-ered quite accurate and relatively free of bi-ases14 Nevertheless to address possible report-ing biases and potential issues involving surveyweights and imputations we calculate returnsbased on data from the FFANIPA in the nextsubsection and nd returns similar to those ofthe SCF

To determine whether there is any generalreporting bias in the SCF equity numbers orproblems with using survey weights or imputa-tions we use the SCF to construct public equityreturns and then compare them to those fromCRSP As Panel B of Table 4 reports the publicequity return numbers from the SCF are 27ndash61percent higher than the CRSP returns Since theCRSP data implicitly takes into account IPOsand merger activity but the SCF data may notwe make an adjustment for this (subtracting thevalue of IPOs but adding the value of public rms taken over by private rms) This has asmall effect Thus if there is a reporting orweighting bias it seems to run in the wrongdirection to reconcile our low private equityreturn numbers15

However since price information is morereadily available in public markets it is possiblethat reporting distortions may be more prevalentin the private equity gures Respondents mayreport stale values of private equity that may lag

the public market Since public equity per-formed remarkably well from 1989 to 1998 thismay explain the low SCF private equity returnsLike private equity owner-occupied homes areilliquid assets that are likely to suffer fromsimilar reporting biases To defend the surveynumbers we therefore examine housing returnsby calculating the capital gain on detached sin-gle family homes using the SCF data and com-paring it to the capital gain on such propertiesbased on data from the Of ce of Federal Hous-ing Enterprise Oversight (OFHEO) The twosets of numbers differ in that the SCF numbersare based on householdsrsquo self-reported esti-mates of what they think they could sell theirhouse for whereas the OFHEO numbers arebased on actual repeat-sales housing transac-tions data from Freddie Mac and Fannie MaeThe comparison can be done for the periods1993 to 1995 and 1996 to 1998 since the 19921995 and 1998 SCFs provide information onthe type of property in which the respondenthouseholds reside16

The resulting capital gains based on the SCFhousehold surveys are 53 percent per year from1993 to 1995 and 59 percent per year from1996 to 1998 The actual capital gains based onOFHEO data are only 26 percent per year from1993 to 1995 and 43 percent per year from1996 to 1998 This suggests that household self-reported estimates of the market value of theirhomes if anything leads to higher capital-gainestimates If self-reported private equity valuesexhibit a similar bias it is likely our privateequity return estimates overstate the true re-turns See also Michael Collins et al (2001) fora summary of the literature on homeownersrsquo

13 See Robert P Parker (1984) and Carol S King andEdward K Ricketts (1980) for information on these issues

14 See Robert B Avery et al (1988) Kennickel andMartha Starr-McCluer (1994) Kennickel et al (1997) andKennickel et al (2000) for a discussion of the survey andweighting schemes as well as the SCF codebook

15 It should be noted that for some account types inwhich public equity is held the SCF only provides categor-ical information about holdings eg ldquomostly stocksrdquoldquomostly bondsrdquo or ldquoa combination of stocks and bondsrdquoThis by itself could lead the public equity returns calculatedusing the SCF to differ a bit from the CRSP returns butshould not cause a systematic bias

16 One adjustment to the SCF data is needed The valueof new homes sold in between survey years enters thecurrent SCF calculation in the same way as new rmscreated between survey years affected the calculation of thereturn to private equity We therefore subtract an estimate ofthe value of new single family houses sold between surveyyears from the end-of-period SCF value of single familyhouses to obtain the correct capital gain The estimate of thevalue of new single family houses is obtained from the USBureau of the Census The capital gain for the period 1993to 1995 is thus calculated as [(SCF based 1995 total valueof single family houses 2 US Bureau of Census estimateof the value of new single family houses sold in 1993 1994and 1995)(SCF based 1992 total value of single familyhouses)]13 Similarly for the 1996 to 1998 period


estimates of the value of their homes Thisliterature nds only small valuation biases ofdifferent sign in different surveys

Another possibility is that households simplyemploy a static valuation model or ldquorule ofthumbrdquo to estimate their private equity valueFor example households may simply report thebook value of their private equity holdings ifthey nd it dif cult to estimate market valuesThis would tend to understate returns in periodswhen the market-to-book ratio is increas-ing However in the 1989 survey both mar-ket and book values are reported for the three rms in which the household has its largestactively managed equity share The aggregatemarket-to-book ratio for proprietorships andpartnerships is 174 and for S and C cor-porations is 124 indicating that householdsare distinguishing between market and bookvalues Furthermore the dispersion of house-hold market-to-book ratios is substantial Thelower quartile of reported market-to-book ratiosfor proprietorships and partnerships is 095while the median and upper quartile is 125 and458 respectively The lower quartile medianand upper quartile for S and C corporations is 1147 and 641 respectively (leaving out house-holds with zero book equity values) This indi-cates that the majority of households are notsimply reporting book values

Finally the private and public equity returnsseem to move together over the three subperi-ods Moreover in the next subsection we showthat the two return series are highly correlatedover the longer time period from 1952 to 1999

E Another Data Sourcemdashthe FFANIPA

For further robustness Table 4 also computesthe return to private equity using data from theFFANIPA The national accounts do not rely onsurvey information and are therefore free of po-tential household reporting biases and provide anindependent check on our return estimates

The FFA market equity estimates for propri-etors and partnerships and S and C corporationsare described in Section III subsection A Forthe income component of returns we adjustNIPA PampP income in three ways First wechange the adjustment for misreporting of prof-its on income tax returns to be 75 percent in

each year from 1959 onward implying that forevery $1 of pro ts reported to the IRS adjustedpro ts are $17517 This differs from the incomeunderreporting adjustment made in NIPAwhich uctuates dramatically over time from alow of 33 percent in 1959 to a high of 200percent in 1982 see NIPA Table 823 Whilesome uctuations in income underreporting tothe IRS is possible this level of volatility seemsimplausible Appendix C discusses the mainsource of information about income underre-porting on tax returns which are studies per-formed by the IRS under the Tax ComplianceMeasurement Program (TCMP) Given the sub-stantial uncertainty about the actual amount ofincome underreporting to the IRS in any givenyear we employ a constant 75-percent adjust-ment each year Our resulting returns for PampPover the 1952 to 1999 period are very similar towhat would be obtained using the same incomeunderreporting adjustment as NIPA Second wesubtract the capital consumption adjustment in-cluded in NIPA pro ts from earnings to get ameasure of the actual pro t ows to proprietorsTo the extent that tax laws allow for differentdepreciation than the true economic depreciationthe difference will show up in the capital gaincomponent of returns Third as a measure ofactual retained earnings in the rm we use capitalexpenditures plus net acquisition of nancial as-sets minus net increase in liabilities (excludingldquoproprietorsrsquo net investmentrdquo) This measures theamount owners must have invested to cover rminvestment whether from pro ts or additionalpaid-in funds The ratio of retained earnings topro ts averages 23 percent for the 1952 to 1999sample and 25 percent for 1989 to 1998

For private S and C corporations we estimatedividend income as total dividends paid by allcorporations (from NIPA) minus dividends paidby public corporations (from CRSP)18 In addi-tion we add 20 percent of the NIPA income

17 The NIPA data do not rely on IRS data prior to 1959see Parker (1984)

18 Since neither the NIPA nor the CRSP dividend seriesadjusts for intercorporate holdings our measure of private Sand C dividends will also double-count dividends due tointercorporate holdings However since our measure ofequity also double-counts intercorporate holdings our re-turn estimates should not be biased


underreporting adjustment made to total corpo-rate pro ts19 Appendix C details the exact ta-bles and line items we use from the FFANIPA

Using these equity and dividend series PanelA of Table 4 reports an average annual return toprivate equity of 41 167 and 224 percentfrom 1990 to 1992 1993 to 1995 and 1996 to1998 respectively using an estate multiplier of200 for S and C corporations When employingan estate multiplier of 300 the returns drop to21 147 and 194 respectively These returnssubtract out the average labor adjustment fromthe SCF (65 percent per year for PampP and 12percent for SampC) and should be compared toline 4 in Panel A for the SCF The FFANIPAreturns are lower in the rst subperiod butslightly higher in the latter two periods Com-pared to the public returns the private FFANIPA returns are lower in two of the threesubperiods We do not adjust for rm entry orexit in the FFANIPA (since an entry adjust-ment is not feasible) but the SCF numberssuggest that the total effect of this is small(compare lines 4 and 9 in Table 4)

Separating out PampP returns from SampC it isagain the PampP returns that are the lowest How-ever even the SampC returns using an estatemultiplier of 200 (our highest return estimates)do not consistently outperform the public index

An advantage of the FFANIPA data is that itis available since 1952 allowing a comparisonof private and public equity returns over alonger time period Since public equity experi-enced large growth over the 1990rsquos it is usefulto examine private and public equity returnsover a longer period The drawback from the

longer analysis is that we can only examineproprietors and partnerships (as discussed ear-lier) Again we do not account for rm entryand exit in this calculation but comparing lines5 and 10 in Table 4 the SCF numbers suggestthat these effects largely cancel out for propri-etors and partnerships The SCF numbers omitthe effects of new equity to existing rms andequity recovered by discontinued rms We ar-gued that these effects are small and likelycancel out for all private equity This is likelythe case for proprietors and partnerships aswell20

Table 6 Panel A reports the arithmetic andgeometric average annual returns and standarddeviation to private equity for PampP over the1952 to 1999 time period Panel B reports theaverage public equity return and standard devi-ation over the same period The private andpublic equity returns are similar Moreoverwhen comparing the private returns to thesmallest decile of CRSP stocks the public eq-uity returns signi cantly outperform private eq-uity over the longer period

Since the PampP equity contains tangible as-sets at market value but does not capture thevalue of intangibles it is useful to compare itsreturn to book equity returns in the publicmarket Using Compustat data on public bookvalues [which is only available from 1963 onand is de ned as in Eugene F Fama andKenneth R French (1993) to be book value ofstockholderrsquos equity plus balance-sheet de-ferred taxes and investment tax credit minusthe book value of preferred stock] we com-pare public value-weighted book equity re-turns to PampP returns from the FFA from 1963to 1999 A comparison with public book eq-uity returns also abstracts from public marketrealizations which Fama and French (2001)argue has in ated estimates of the public eq-uity premium over the last half-century Thebook equity returns on public equity are about

19 Based on SCF market value of private S and C cor-porations these corporations account for between 24 and 51percent of all corporate equity Since part of the hiddenincome is likely retained in the rm (and thus shows up ascapital gains) we add only 20 percent of the NIPA corpo-rate income underreporting adjustment to private S and Cpro ts The NIPA income underreporting adjustment forcorporations is around 15 percent during the 1989 to 1998period For large C corporations (assets greater than $10million with no distinction between public and private Ccorporations) the IRS TCMP does not report recommendedchanges in income only the changes in taxes The resultsbased on audit yields imply recommended dollar tax in-creases of 214 percent using 1985 data With progressivetaxes the underlying income changes will be smaller con-sistent with the NIPA adjustment

20 In the 1993 NSSBF new equity to existing PampP rmsis 10 billion annually We estimated that salesliquidationsamount to 35 billion (likely an upper bound) If half of thisis attributed to proprietor and partnerships the net effect is175 2 10 5 75 billion per year This is about 04 percentof PampP equity in the 1992 FFA implying only a smalldownward bias in our return estimates


2 to 3 percent per year higher than the returnsto equity in private PampPs

In sum these numbers based on the FFANIPA are reassuring con rming our previousconclusion that the returns to private and publicequity are similar

F The Risk of Private Equity

Is the private market riskier in aggregate thanthe public market This is hard to evaluate withthe available data The PampP equity in the FFA isa ldquomixrdquo of book and market equity since itcaptures tangible assets at market value but doesnot capture intangibles As reported in Table6 the standard deviation of the PampP equityreturn series is about twice that of the publicequity book return series and a bit less than halfthat of the public market-value return seriesFigure 1 plots the FFANIPA return series ofprivate proprietors and partnerships and thebook equity returns series for public rms Theseries exhibit a strong correlation of 070 overthe 1963 to 1999 period suggesting that it maybe more relevant to compare the PampP return

volatility to the public equity book return vola-tility Finally to gauge the riskiness of marketequity returns note that the annual standarddeviation of the smallest decile of public rmreturns is 411 percent A portfolio of evensmaller private rms is likely to be as volatileMore importantly since entrepreneurs typicallyown equity in a single private rm the riskfaced by the average entrepreneur may behigher still

In the next section we analyze rm-levelentrepreneurial risk and returns We argue thatthe risk-return trade-off faced by the typicalentrepreneur is much worse than that of theprivate equity index and therefore also likelyto be much worse than that of the public equityindex

IV The Distribution of ReturnsAcross Private Firms

Since most entrepreneurs own equity in asingle private rm for which they have an activemanagement interest we are interested in char-acterizing the distribution of returns across


Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation

A Private Equity Returns (from the FFANIPA)

Proprietors and partnerships equity returns1953ndash1999

131 128 69


132 128 77

B Public Equity Returns (from CRSP)

Value-weighted index market equity returns1953ndash1999

140 127 170

Value-weighted index book equity returns1963ndash1999

156 156 37

Value-weighted smallest decile marketequity returns 1953ndash1999

242 182 411

Correlation between PampP and CRSP (book) equity returns 1963ndash1999 070

Notes Panel A reports the returns to private equity in proprietorships and partnerships Returnestimates pertain to data from the FFANIPA over the period 1952 to 1999 Returns arecalculated assuming labor income adjustments of 65 percent Proprietorsrsquo income is calcu-lated as stated in Appendix C Panel B reports returns to publicly traded equity over the sametime period from CRSP All returns are nominal


individual entrepreneurs In this section we rstdiscuss the conditions under which the indexreturn will be a good estimate of the averageindividual return We argue that the averagegeometric (buy-and-hold) return in the cross-section of rms is likely substantially lowerthan the geometric average return of the pri-vate equity index To document the dramaticamounts of idiosyncratic private rm risk wethen examine the returns to an individual entre-preneur by considering rm survival rates andthe distribution of individual entrepreneur re-turns conditional on rm survival

A When Are Aggregate Returns a GoodMeasure of the Returns to the Average

Single Private Firm

The documented poor diversi cation of pri-vate equity holdings suggests that the typical

investor cares about the return to investing in asingle rm rather than an index of private eq-uity Unfortunately available data do not allowus to directly compute the average geometricreturn across rms We only have estimates of rm survival rates and rm-level returns condi-tional on survival but do not have rm-levelinformation about the return to rms who werediscontinued (bankrupt sold etc) To ourknowledge no comprehensive data of this sortexists In this subsection we argue howeverthat the index return we calculate most likelyoverstates the average of the returns across in-dividual entrepreneurs

Data from the SCF indicate that the typicalinvestment horizon of an entrepreneur is longThe average surviving entrepreneur has ownedhis rm for about ten years at the time of thesurvey implying a typical horizon of at least tenyears Illiquidity of private equity is one factor

FIGURE 1 THE RETURNS TO PRIVATE AND PUBLIC EQUITY (1963ndash1999)

Notes The annual returns to the index of FFANIPA private proprietor and partnership equity and book equity returns to theindex of public corporations from the CRSPndashCompustat universe are plotted over the period 1963ndash1999


contributing to long holding periods Longholding periods suggest that entrepreneurs areprimarily concerned with the buy-and-hold re-turn of their investment For example if returnsconsisted only of capital gains and horizonswere exogenous entrepreneurs would careabout the geometric return over their holdingperiod Moreover the theoretical models ofHeaton and Lucas (2001) Brennan and Torous(1999) and Benartzi (2000) (motivated in the In-troduction) all focus on buy-and-hold returns ofindividuals Consequently we focus on whetherthe geometric return on the index is an upward-biased estimate of the average geometric returnacross individuals To the extent that returns havea stochastic dividend component the entrepreneurwill care not only about the properties of thegeometric return but also about other features ofthe return path In this case determining whetherthe private equity index returns and poor diversi- cation documented earlier constitutes a puzzlerequires further theoretical work We leave this forfuture study and focus here on whether the aver-age geometric return across rms is lower than thegeometric value-weighted return We argue thatthis is likely to be the case strengthening theconclusion that the returns to private equity aresurprisingly low

The key feature of the return distributionwhich leads to the geometric index return beingan upward-biased estimate of the average geo-metric return across rms is the presence ofidiosyncratic rm risk To illustrate this con-sider rst the case with no idiosyncratic riskSuppose the typical rm lives for N periodswhere the initial investment is $1 and the rmgrows exponentially to be worth $K at date NThe setting is one with ldquooverlapping rm gen-erationsrdquo in which one rm is born each yearand one rm is sold in each period at age NThus N is the holding period of the founder Tosimplify the calculations assume that private rms are sold to public rms after N periodsThe geometric return obtained by each founderis simply K1N which is therefore also the av-erage geometric return across entrepreneursThe geometric index return 1 1 rgeometricindexis the return to buying all N private rms inexistence at date t (the newborn rm the1-year-old rm up to the N 2 1-year-old rm) and holding these rms until date t 1

121 The denominator in the calculation of1 1 rgeometricindex is the total purchase price forthe N rms at date t The numerator is the totalvalue of these N rms at date t 1 1 includingthe K obtained from selling the oldest rm to apublic company

Under this scenario of gradual rm growththe geometric index return and the average geo-metric return across rms are identical (andboth are constant over time)

1 1 raverage geometric 5 K1N

1 1 rgeometric index

5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N

If growth is not gradual (and still with noidiosyncratic risk) the geometric index returnwill not be identical to the average geometricreturn across rms In the case of early growththe index return will understate the averagegeometric return across rms while the oppo-site will be true under late growth For exampleif rm value grows to K after only one periodand then stays constant (early growth) the re-turns are


1 1 rgeometric index 5NK

1 1 ~N 2 1K K1N

On the other hand if rm value stays constant at$1 until date N 2 1 and then jumps to $K atdate N (late growth) the returns are


1 1 rgeometric index 5~N 2 1 1 K

N K1N

21 With the adjustment to date t 1 1 value for thenewborn rm at date t 1 1 (as in the index calculationsabove) this rm will not affect our calculations


Without idiosyncratic risk the bias in theindex return depends on the growth pro le of rms However when adding idiosyncratic riskthe geometric index return is likely to be lowerthan the average geometric return across rmseven in cases with substantial early growthConsider augmenting the above setting as fol-lows Suppose rms face a constant bankruptcyprobability over time and that equity investorsin bankrupt rms lose half of their investmentThe probability of bankruptcy p is calibratedto a 35-percent survival rate of rms within the rst ten years of life Furthermore in eachperiod surviving rms face a two-point distri-bution of returns The two points of this distri-bution are chosen to generate pre-chosen valuesfor the mean and standard deviation of a rmrsquosreturn To capture early growth assume themean return conditional on survival declineswith rm age according to the formula mt 51 1 [041 1 (t 2 1)b] where b 5 03 togenerate a strong decline in mean returns over rm life (eg from 40 percent per year at age 1to 18 percent per year at age 5) If volatility stis constant at 30 percent per year [likely a fairlylow number for the typical private rm giventhat the annual standard deviation of a typicalsingle public rmrsquos equity return is 50 to 60percent according to Campbell et al (2001)]and N 5 20 then the geometric index return is109 percent per year while the average geomet-ric return across rms is 47 percent per year Asan alternative scenario if volatility is allowed todecline with rm age such that the Sharpe ratio(mtst) is constant over a rmrsquos life (equal to03) then the geometric index return is 109percent per year while the average geometricreturn across rms is as low as 2117 percentper year22

These calculations illustrate how even a lowlevel of idiosyncratic risk will bias the indexreturn upward even with early rm growth Thedifference between the index return and theaverage individual rm return would be even

larger with gradual or late growth Although wedo not have adequate rm-level information todirectly determine whether early gradual orlate growth occurs the fact that risk seems todecline with age suggests that early growth andearly risk are probably most consistent with thedata

While the calculations are admittedly sim-ple they illustrate that our geometric indexreturn is likely to be a substantially upward-biased estimate of the typical geometric re-turn to a single rm Hence the true return toa poorly diversi ed individual entrepreneur islikely much lower than our previous calcula-tions suggest We now turn to documentingthe amount of idiosyncratic risk of a singleprivate rm

B Private Firm Survival Rates

Certainly a large part of the risk associatedwith starting a new business is the risk of fail-ure as opposed to a risky distribution of returnsconditional on survival In order to gauge thiswe appeal to outside evidence on rm survivalrates Timothy Dunne et al (1988) construct rm survival rates based on the 1967 19721977 and 1982 Census of Manufacturers and nd that on average 615 percent of rms exit inthe ve years following the rst census in whichthey were observed On average 796 percent of rms exit within ten years Popkin and Kirchhoff(1991) analyze survival rates by age of businessfrom 1976 to 1986 using the United StatesEstablishment Longitudinal Microdata le(USELM) which is based on Dun and Bradstreetrsquosmarketing le They estimate that the two-yearsurvival rate of rms who were less than twoyears old in 1976 is 769 percent and the ten-year survival rate is 344 percent Survival ratesincrease with initial rm age Firms who werebetween 10 and 19 years old had a two-yearsurvival rate of 739 percent and a ten-yearsurvival rate of 469 percent

It is dif cult to evaluate how much ownerslose when their business is discontinued Dataprovided by the US Small Business Adminis-tration (2000) document that the average annualnumber of rm bankruptcies over the 1990 to1997 period was 59393 (source The Adminis-trative Of ce of the US Courts) The number

22 Several empirical facts suggest the presence of ldquoearlyriskrdquo Firstly bankruptcy rates decline with rm age [JoelPopkin and Bruce A Kirchoff (1991)] Secondly the cross-sectional standard deviation of average geometric returnsacross surviving rms is declining with holding period inthe SCF


of bankruptcies is somewhat lower than theaverage number of business failures of 78711over this period (source Dun and BradstreetCorporation) A business failure is de ned as anenterprise that ceases operation with a loss toone or more creditors The average number offailures constitute 153 percent of the averagetotal number of employer rm terminationswhich was 515273 over the same time periodOwners in failed companies probably lose all oftheir initial equity investment (since they dis-continue with debt outstanding) Entrepreneurscan in fact lose more than their equity invest-ment since rm debt is often backed by personalcollateral (typically home equity) Assumingthey lose all of their equity in failed rmscombining the survival rates with the share ofdiscontinued rms who fail the founder of anew private company faces a (1 2 0344) 30153 3 100 5 100 percent risk of losing all ofhisher investment within the rst ten years

For the remainder of discontinued rms it isdif cult to evaluate how much of the initialequity investment by owners has been lost ifany Some rms may be discontinuedwith a fullor partial equity investment loss due to poorfuture prospects Others are successful and maybe sold to new owners or ldquocashed outrdquo Thenumber of rm salestakeovers is quite lowBased on the 1993 NSSBF about 70000 rmswere acquired within the last two years (twoyears to account for possible lag in introductionto the Dun and Bradstreet database on which theNSSBF sample is based) This implies that ap-proximately 350000 (or about 70 percent of)terminated rms liquidated It is likely that en-trepreneurs lose at least some if not all of theirinvestment upon liquidation Clearly failureliquidation poses a great risk

C Entrepreneur-Level ReturnsConditional on Survival

The rest of this section focuses on the condi-tional distribution of entrepreneurial returns todocument that substantial idiosyncratic risk ex-ists even conditional on survival Using data onindividual household investment in private eq-uity from the SCF we calculate the distributionacross households of returns since they found-edacquired a private rm We examine those

private companies in which the household hasits largest actively managed equity positionThe following information is available from theSCF the year in which the rm was foundedacquired rm pro ts in the year before thesurvey interview the market value of the own-ership share in the interview year (estimated bythe respondent) and the basis value for taxpurposes of the current ownership share Weuse the latter as an estimate of the initial valueof the entrepreneurrsquos equity investment

We estimate the geometric average annualcapital gain over the period since the rm wasfoundedacquired Assuming the current pro tto equity ratio is representative of those in pre-vious years we also construct an estimate of theincome stream to the household from the invest-ment These returns represent the price appre-ciation and income received from the initialinvestment date to the time of the survey Weare not able to construct estimates of the returnobtained through the full period of ownershipof course since households may keep theirownership share in the company for manyyears after the survey We are also not able toconstruct return estimates for household invest-ments that did not survive Hence we empha-size that the distribution of returns we calculateis conditional on survival and does not repre-sent the unconditional distribution of returns

We plot in Figure 2 the distribution of returnsfrom private equity investment The graphs per-tain to the distribution of household returns fromthe 1989 SCF Other survey years were similar23

The rst graph plots the histogram of averageannual capital gains accrued across householdsover the period since the rm was foundedacquired For each household we compute thegeometric average annual capital gain as

(4)

1Value at the

time of the survey

Value oforiginal investment

21~Years since foundedacquired

2 1

23 We focus on households with initial investments of atleast $1000 (1983 dollars using the CPI for all urbanconsumers) This implies dropping about 5 percent of theentrepreneur households All graphs employ SCF weights


The distribution of capital gains conditional onsurvival is wide24 Using the 1989 survey themedian of the capital gain distribution is 69percent per year while the rst quartile is 0 andthe third quartile is 186 percent per year As for

the holding periods over which these annualizedcapital gains have been obtained 43 percent ofhouseholds had invested in private equity for ve years or less at the time of the survey 473percent had invested for between ve and 25years and 96 percent had invested for morethan 25 years (averaged across all four surveyyears)

The second graph plots the histogram of earn-ings rates de ned as earnings in the year beforethe survey divided by the total market value of

24 We plot households who lost all of their initial capitalbut still say they are in business at 2100 percent in this gure These households are not included in the subsequentgraphs since it is not possible to de ne pro tequity forcompanies with zero equity

FIGURE 2 THE CONDITIONAL DISTRIBUTION OF RETURNS TO PRIVATE EQUITY ACROSS HOUSEHOLDS

Notes Household data from the 1989 SCF are used to plot the returns to private equity investment in surviving rms Thetop left plot shows the histogram of geometric average annual capital gains accrued across households The top right plotshows the histogram of earnings rates (earnings in the year prior to the survey divided by market value of equity) accruedacross households The bottom left plot shows the histogram across households of the geometric average return on investmentif households had instead invested their wealth in the CRSP value-weighted index of all publicly traded equity over the samehorizon as their private equity investment The bottom right plot shows the histogram across households of the total averagereturn (capital gain plus earnings where 30 percent of earnings are assumed to be retained in the rm) on private equity inexcess of the CRSP index return over each householdrsquos holding period


the rm There is substantial variation in earn-ings rates although most households report zeroor positive earnings rates The third graph ineach panel plots the histogram of the geometricaverage returns households would have ob-tained had they invested their wealth in theCRSP index of all publicly traded equity overthe same horizon as their private equity invest-ment For example for an investor who heldprivate equity in his company for 30 years at thetime of the 1989 survey we compute the geo-metric average annual return to investing in theCRSP index over those same 30 years (ie from1959 to 1989) As shown in the graph the distri-bution of returns on a diversi ed public equityindex over the same investment horizon is tightwith a minimum return of 56 percent per year anda maximum return of 199 percent per year

The nal graph combines the capital gain andincome components for the private rms to con-struct a total return where we assume earningsrates are constant over time and equal those inthe interview year and that (for simplicity) 30percent of pro ts are retained in the rm acrossall rm types25 We then subtract from this totalreturn the return the household could have ob-tained by investing in the CRSP index over thesame period This essentially combines the rstthree plots into one

Even though this distribution is conditional onsurvival around 30 percent of households wouldhave been better off investing in the CRSP indexrather than their own company Moreover there issubstantial variation in the excess returns to pri-vate over public equity investment even condi-tional on survival The excess return distribution ishighly skewed While the median excess returnis 182 percent per year the average excess returnis 1396 percent per year due to a fairly smallfraction of households with very large annualizedexcess returns These high meanmedian excessreturns are to a large extent due to householdswithsmall initial investments When households areweighted by the size of their initial investment themedian excess return is 220 percent per yearwhile the mean excess return is 244 percent

D Conditional versus Unconditional Meanand Variance

Finally our conclusions that entrepreneurialreturns appear unattractive are based on an es-timate of the unconditional distribution of pri-vate equity returns That is for a randomlychosen entrepreneur investment in private eq-uity seems like a bad deal However entrepre-neurs may have superior information about their rmrsquos prospects In this case the conditionalvariance of returns to each entrepreneur may bemuch lower than suggested by the poor diver-si cation and high rm-level risk Thus forsome individuals entering entrepreneurshipmay be a very good deal However if entrepre-neurship is attractive for some entrepreneursthen it must be even less attractive for otherentrepreneurs than what our index return esti-mates suggest Hence if the low returns appearpuzzling on average they must be even morepuzzling for a segment of the entrepreneurpopulation

V Why Do People Become Entrepreneurs

In this section we brie y discuss possibleexplanations for why private equity investorswillingly invest in concentrated private equityportfolios despite the seemingly poor riskndashreturn trade-off

A Optimal Contracting and the Abilityto Diversify

Concentrated private equity investmentscould be motivated by issues of moral hazard orasymmetric information Institutional and gov-ernmental monitoring is also far less prevalentin the private market making assignment ofcontrol rights of the rm even more criticalHowever this cannot explain why individualsenter into entrepreneurship initially given thepoor riskndashreturn trade-off

B Why Are Entrepreneurs Willing toParticipate in the First Place

We consider ve possible explanations forentry into entrepreneurship despite the poorriskndashreturn trade-off of existing entrepreneurs

25 Since we wish to have uniform assumptions across rm types and since our previous calculations employed40-percent retention for C corporations and 20 percent forall other rm types a 30-percent retention rate is used


high entrepreneur risk tolerance large additionalpecuniary bene ts non-pecuniary bene ts a pref-erence for skewness and overoptimism and mis-perceived risk

1 Risk TolerancemdashIf entrepreneurs havevery low risk aversion then disutility from poordiversi cation may be small and the returns toprivate equity need not be higher than those ofpublic equity Gentry and Hubbard (2001a)compare the composition of entrepreneurportfolios to those of non-entrepreneurs usingthe 1989 SCF They nd that (apart from thesizeable investment in the private equity of theirown rm) the rest of entrepreneursrsquo portfoliosare quite similar to non-entrepreneurs even forthose in the top 5 percent of the wealth distri-bution Since entrepreneurs do not invest theremainder of their wealth any more conserva-tively than non-entrepreneurs they may bemore risk tolerant However it is possible thatprivate equity-holders might be expected tohold larger shares of their remaining wealth inpublic equity This is suggested by the results ofHeaton and Lucas (2001) and is due to the factthat private equity income provides not onlyldquobackground riskrdquo but also positive income ow on average26

2 Other Pecuniary Bene ts and CostsmdashSalaries derived from private companies arealready accounted for in our return calculationsTo assess the bene ts derived from possibleperquisite taking we compute how large thesebene ts would have to be to provide a 10 per-cent per year return premium in private equityover public equity This amounts to 143 percentof total annual household income (or $460000)

for the median entrepreneur (using data fromthe 1998 SCF focusing on entrepreneurs with atleast $5000 of private equity holdings andweighting households by the size of their hold-ings) This seems high given that salaries andunreported income from tax evasion are alreadyaccounted for

In addition we should consider the fact thatinvestors compare asset returns after personaltaxes Previously we used survey data or NIPAdata with an adjustment for income underre-porting on tax returns to produce more accuratepre-personal tax returns comparable to the re-turns from CRSP It remains to considerwhether personal taxes differ between privateand public equity-holders Certainly since en-trepreneurs save taxes on income they hide fromthe IRS their effective tax rate is lower than thestatutory rate This effect is likely to be small27

Furthermore a substantial fraction of publicequity is held in tax-advantaged accounts re-ducing the effective tax rates paid on publicequity

On the cost side at least 25 billion dollars inpro ts in each of the SCF years pertain tohouseholds who report a zero market value anda zero tax basis for their equity share It may bemore reasonable to exclude these householdsfrom our analysis which would lower our re-turn estimates by about 05 percent per year Alarge fraction of these pro ts are in partner-ships The zero equity value may simply re ectthe fact that equity shares are not tradable inthese rms but rather are payments for laborinput to employees who make partner

3 Nonpecuniary Bene tsmdashIn addition non-pecuniary bene ts derived from entrepreneur-ship may explain the concentrated equityholdings Over 21 percent of survey respon-dents in the 1992 Economic Census Character-istics of Business Owners stated being their ownboss as the main reason for starting the rm as

26 Furthermore even the wealthiest managers appear farfrom risk neutral A recent article in the Wall Street Journal(ldquoYour Money Matters Hedging a Single Stock Has UpsDownsrdquo by Ruth Simon 2 February 2000) cites the risingpopularity of hedging strategies offered by investment rmsto reduce exposure to own-company stock performance fortop executives (as many as a couple thousand such strate-gies are executed each year) This suggests that executivesdo care about the volatility of their own company stockholdings and take steps to reduce their exposure to the rmOne of the more notable participants in these strategies isTed Turner despite his more than $9 billion wealth (at thetime of the article)

27 For example if the statutory personal tax rate is 30percent and 30 percent of income is sheltered from taxauthorities the effective tax rate is 21 percent This in-creases the income component of after-tax returns of privatecompanies relative to public companies assuming the latterdoes not hide income by 9 percent (eg from 10 percentper year to 109 percent)


opposed to having a primary or secondarysource of income as the main reason Otherstudies have also identi ed the exibility andautonomy of self-employment as a major non-pecuniary bene t [see David G Blanch owerand Andrew J Oswald (1992)] Indeed Hamil-ton (2000) interprets his results for the medianentrepreneur as evidence of large nonpecuniarybene ts

Using the calculation from above a 10-percent (of private equity investment) nonpecu-niary bene t would have to amount to 143percent of total annual income or $460000While a substantial amount this may not beunreasonable Certainly many nancial econo-mists willingly give up substantial amounts bychoosing to remain in academia where the ac-ademic lifestyle may be considered a nonpecu-niary bene t

4 Preference for SkewnessmdashRather thantry to augment the rst moment of the returndistribution of private equity through additionalpecuniary or nonpecuniary bene ts a motiva-tion for entrepreneurship may lie in higher mo-ments of the distribution For instance Fig-ure 2 shows that the distribution of entrepre-neurial returns is highly skewed with a fat righttail If entrepreneurs have a preference forskewness then they may be willing to accepta lower mean return despite the high varianceA preference for skewness could explain theresult in Gentry and Hubbard (2001b) thatprogressive marginal tax rates discouragesentry into entrepreneurship

Alan Kraus and Robert Litzenberger (1976)and Campbell R Harvey and Akhtar Siddique(2000) argue that investors have a strong skew-ness preference However skewness in returnscan also be obtained more easily through theoptions market or various trading strategies inpublic markets Hence the skewness of privateequity returns may not be the only attributeattracting investors

5 Overoptimism and Misperceived RiskmdashFinally entrepreneurs may behave in a mannerthat is not perfectly rational For instance theymay be overly optimistic about the rmrsquos meanprospects or they may irrationally believe thathaving control of the rm lowers risk

We showed previously that the average re-turn conditional on survival from private eq-uity is about 24 percent greater than the publicmarket return Hence if entrepreneurs simplybelieve their probability of survival is suf -ciently high then the distribution of future re-turns would look very attractive Surveyevidence of entrepreneurs is consistent with thisnotion Arnold C Cooper et al (1988) nd that68 percent of entrepreneurs think that the oddsof their business succeeding is better than theodds for another business like theirs only 5percent think their odds are worse In additiona third of entrepreneurs believe their probabilityof success (eg surviving) is 1 and 72 percentof entrepreneurs think their probability of suc-cess is at least 080 J Edward Russo and PaulJ H Schoemaker (1992) nd that managers aredramatically overcon dent28

Most likely it is some combination of all veexplanations that contributes to entrepreneurialactivity Quantifying the impact each has on thepropensity to become an entrepreneur as wellas on subsequent returns is an interesting issueleft for future research

VI Concluding Remarks (Is There a Puzzle)

We nd that the majority of household in-vestment in private companies is concentratedin a single risky privately held rm in whichthe household has an active management inter-est Despite the risks these investors face intaking on large amounts of idiosyncratic riskthe returns to private equity are surprisinglylow We conduct the rst comprehensive studyof the unconditional returns to all nonpubliclytraded equity Controlling for the labor compo-nent of returns adjusting for entry and exit of rm equity over time (as best possible) andaddressing issues related to potentially distortedestimates of market values and rm pro ts (egdue to tax evasion motives) we nd that theaverage return to private equity is similar to thatof public equity Given the large equity pre-mium demanded by investors in public markets

28 Antonio Bernardo and Ivo Welch (1998) argue whyindividuals remain overcon dent in an entrepreneurialsetting


it seems surprising that entrepreneurs are will-ing to invest so heavily in a single private rmwhich offers a far worse risk-return trade-off

We recognize that a precise measure of themean return to private equity is extremely dif- cult to obtain Expected returns are notoriouslydif cult to estimate and our estimates are basedon relatively short sample periods (nine yearsfor the SCF and 47 years for the FFANIPA)This dif culty is exacerbated when using fairlyimprecise data on estimates of private rmvalues and pro ts Nevertheless the estimatedrealized returns to private equity are quitehighly correlated with public equity returns in-dicating it is less likely that the realized returnsrepresent an abnormal draw for one of the twomarkets only or simply measurement error inour data Moreover we argued earlier that it isunlikely that the private equity mean returnexceeds the public equity mean return by 10percent per year (as theory suggests it should)Our ndings for the private equity marketpresent a challenge to theories seeking to ex-plain the size of the equity premium in publicmarkets within a homogeneous agent framework

Whether or not our results constitute a puz-zle remains an open question On the empir-ical side more information about the amountof equity recovered in liquidated rms wouldenable a more precise estimate of the uncon-ditional returns to private equity and thecross-sectional distribution of those returns Itwould also be interesting to obtain a longerreturn series for S and C corporations to de-termine if the fact that S and C corporationsoutperform proprietors and partnerships is ro-bust to other sample periods outside of the1990rsquos On the theory side models that cap-ture the correlation of human and nancialcapital returns and allow for consumption bythe entrepreneur before the terminal date areneeded

Finally distinguishing among other motivesfor entrepreneurship (ie private bene ts ofcontrol preferences for skewness and misper-ceptions of the probability of failure) may haveimportant policy implications For example ifentrepreneurs are enticed by small probabilitiesof very large returns high tax rates for high-income individuals could have strong adversegrowth effects On the other hand if many

entrepreneurs enter business with overoptimis-tic expectations government educational efforts(as opposed to government-subsidized smallbusiness loans) may be warranted

APPENDIX A ESTIMATING THE VALUE OF EQUITY

IN PRIVATE S AND C CORPORATIONS BASED ON

ESTATE TAX RETURNS

To obtain an estimate of the value of equity inprivate S and C corporations which is indepen-dent of the SCF equity numbers we follow amethod used by the IRS to estimate wealthbased on estate tax returns The approach isdescribed in Section III-A This Appendix pro-vides evidence that owners of private equityhave lower mortality than others at the same ageand with similar wealth Thus a multiplierhigher than that used by the IRS should be usedfor this category of wealth

Since most private equity is owned by house-holds with active management interests it isunlikely that holders of private equity have thesame mortality rates as others at the same ageand with similar wealth (as is assumed in theIRS multiplier) Entrepreneurs are likely to selloff their private businesses when their healthdeteriorates making active management dif -cult Consequently a smaller percentage ofprivate equity (than of other wealth compo-nents) shows up on estate tax returns for a givenyear

Two measures of respondent health are avail-able in the SCF to support this Question X6030asks ldquoWould you say your health is excellentgood fair or poorrdquo and question X7381 asksldquoAbout how old do you think you will live toberdquo Responses to the rst question are avail-able for the 1989 1992 1995 and 1998 surveysand for the second for 1995 and 1998 Mergingthe data across years and restricting attention tohouseholds with assets greater than $600000we nd that the percent of household headsreporting to be in poor health (for couples therespondent is the male) is 23 percent for non-business owners and 08 percent for owners ofequity in private S and C corporations usingSCF weights and further weighting by amountof private equity owned This ratio (2308)equals 29 In addition the percent of house-holds expecting to live ve (ten) years or less is


39 (108) percent for nonbusiness owners and15 (52) percent for owners of private S and Ccorporation equity corresponding to a ratio of26 (21) Using the same weights as above theowners of private S and C corporation equityare about three years younger than nonbusinessowners Taking this into account would lowerthe differential in mortality a bit

In sum if mortality is approximately linear inthese measures of health this suggests using amultiplier for S and C private equity which isbetween two and three times higher than thatused for other wealth components This is ourmotivation for employing multipliers of 200and 300 to estimate the total value of S and Cequity based on estate tax returns

APPENDIX B ESTIMATING THE VALUE OF MISSING

MERGERS AND ACQUISITIONS IN THE

SDC DATABASE

For each deal in the SDC database with miss-ing price information we search for data on thetransaction to indicate its size We found fourdata items with broader coverage than dealvalue These are book value property plantand equipment total assets and number of em-ployees of the target We then take the dealswith price data and run a cross-sectional regres-sion of all deal values on a constant and each ofthese variables individually as well as every

combination of the variables producing 15 setsof regression coef cients This is done for eachyear and category separately These regressioncoef cients are then used to predict the value ofthose deals with missing price information buthaving at least one of the other variables Forexample if a deal is missing its value but hasinformation on book value we estimate itsvalue by multiplying its book value times thecoef cient estimated from the univariate regres-sion of deal market value on book value for alldeals with prices If a deal has more than onedata item then we employ the correspondingmultivariate regression coef cients from dealswith prices In other words we use the regres-sion coef cients from the appropriate combina-tion of data items for which the deal hasrecorded information This provides an estimateof the value of missing deals while taking intoaccount the characteristics of such deals (iethat they are typically smaller) Finally forthose deals with missing value and no addi-tional information on the other four data itemswe simply assign the average of the estimatedvalues of missing deals to these transactions Ifanything this is likely to overstate our numbersslightly These estimated values are computedfor each subcategory of merger and acquisitionactivity in the same manner and added to thevalue of deals with price information to producea total or ldquoscaledrdquo value for each subcategory

APPENDIX C DETAILS ON NUMBERS FROM THE FFA AND NIPA

A Series Used in Our Calculations Based on the FFA and NIPA

We calculate the baseline annual returns to proprietorships and partnerships (PampP) as

PampP~Equity t 1 1 1 PampP~Profits t 1 1 2 CCA t 1 1 2 RE t 1 1 1 DTax adj t 1 1

PampP~Equity t

where

1 PampP(Equity) 5 (FFA Table btab100d FL153080015) 2 (Value of 1 to 4 family rental properties not owned bycorporations from the Bureau of Economic Analysis xed assets detailed residential table)

2 PampP(Pro ts) 5 NIPA Table 114 line 93 CCA 5 Capital consumption adjustment 5 NIPA Table 114 line 12 plus line 164 RE 5 Retained earnings 5 (FFA Table utab103d FU116300005 1 FU113180005) 1 (FFA Table utab104d

FU136000105 1 FU133180005)


5 DTax adj 5 Change in tax adjustment 5 (075 2 NIPA PampP tax adjustment percent used) 3 (NIPA nonfarm PampP pro tsas reported to the IRS) where NIPA PampP tax adjustment percent used 5 (NIPA Table 823 line 2NIPA Table 823 line1) and NIPA nonfarm PampP pro ts are as reported to the IRS in NIPA Table 823 line 1

We calculate the baseline annual returns to private SampC corporations as

SampCprivate ~Equityt 1 1 1 SampCall~Div t 1 1 2 SampCpublic~Div t 1 1 1 02~SampCall~Tax adj t 1 1

SampCprivate~Equity t

where

1 SampCprivate(Equity) is estimated based on estate tax returns as described in Appendix A2 SampCall(Div) 5 NIPA dividends paid in cash or assets according to the IRS (NIPA Table 825 line 29) plus

Posttabulation amendments and revisions (NIPA Table 825 line 30)3 SampCpublic(Div) 5 dividends paid by companies listed on the NYSE AMEX or NASDAQ calculated as the income

return on the CRSP value-weighted index times the total market value of NYSE AMEX and NASDAQ equity4 SampCall(Tax adj) 5 NIPA adjustment for misreporting on income tax returns NIPA Table 825 line 2 See the text for

the choice of the factor 02

Note that the FFANIPA frequently update their data Our numbers are based on the latest available releases as of January1 2002

Further adjustments for the labor component of pro ts are described in the text

B Income Underreporting on Tax Forms

This subsection describes the ndings of the IRS Tax Compliance Measurement Program (TCMP) which motivates theincome underreporting adjustment in NIPA

Every third year between 1973 and 1988 a sample of about 55000 tax lers was subjected to extensive audits The TCMPprogram has since been discontinued TCMP audits differed from regular IRS audits in that only experienced IRS examinerswere used and in that examiners reviewed each item on the return line by line The TCMP studies include information aboutall components of income including income from proprietorships and partnerships These studies were supplemented byseparate studies of small corporation income tax returns for 1977 and 1980 For large corporations regular audit yields wereextrapolated by the IRS based on a regression using averages of data for 1984 1985 and 1986 to compute what audit yieldswould have been had all large corporations been audited The results of the studies up to 1982 are summarized in IRS (1988)

According to the TCMP results income underreporting on tax returns is very prevalent especially among small rms Forthe category ldquoOther Sole Proprietorshiprdquo which refers to nonfarm sole proprietors with the exception of informal suppliers(baby-sitters street vendors etc) the ratio of detected nonreported income to taxpayer reported income (accounting for bothunderstated income and overstated expenses) is 0219 for 1973 0229 for 1976 0299 for 1979 and 0419 for 1982 Forpartnerships the ratios are 0139 for 1973 0248 for 1976 and 0277 for 1979 (the 1982 ratio is less reliable since reportedpartnership pro ts are close to zero in that year) The reason NIPA uses larger tax adjustments for proprietors and partnershipsis that the TCMP conjectures that for every dollar detected in the TCMP audit an extra 234 dollars go undetected forproprietors (328 for partnerships) From what we were able to determine these ldquomultipliersrdquo are based on very littleinformation and one wonders whether the IRS has an incentive to in ate these numbers Nonetheless to be conservative weuse an income underreporting adjustment which re ects the use of such multipliers

REFERENCES

Antoniewicz Rochelle L ldquoA Comparison of theHousehold Sector from the Flow of FundsAccounts and the Survey of Consumer Fi-nancesrdquo Working paper Federal ReserveBoard 2000

Avery Robert B Elliehausen Gregory E andKennickell Arthur B ldquoMeasuring Wealthwith Survey Data An Evaluation of the 1983Survey of Consumer Financesrdquo Review ofIncome and Wealth December 1988 34(4)pp 339ndash69

Benartzi Shlomo ldquoExcessive Extrapolation and


the Allocation of 401(k) Accounts to Com-pany Stockrdquo Working paper UCLA 2000

Bernardo Antonio and Welch Ivo ldquoOn the Evo-lution of Overcon dence and EntrepreneursrdquoWorking paper UCLA 1998

Blanch ower David G and Oswald Andrew JldquoEntrepreneurship Happiness and Supernor-mal Returns Evidence From Britain and theUSrdquo National Bureau of Economic Re-search (Cambridge MA) Working Paper No4228 1992

Brennan Michael J and Torous Walter N ldquoIn-dividual Decision-Making and Investor Wel-farerdquo Economic Notes July 1999 28(2) pp119ndash43

Bureau of Economic Analysis Detailed data for xed assets and consumer durable goodsWashington DC US Department of Com-merce 1989ndash1998

Campbell John and Cochrane John ldquoBy Forceof Habit A Consumption-Based Explanationof Aggregate Stock Market Behaviorrdquo Jour-nal of Political Economy April 1999 107(2)pp 205ndash51

Campbell John Lettau Martin Malkiel Burtonand Xu Yexiao ldquoHave Individual Stocks Be-come More Volatile An Empirical Explora-tion of Idiosyncratic Riskrdquo Journal ofFinance February 2001 56(1) pp 1ndash44

Collins Michael Crowe David and CarlinerMichael ldquoExamining Supply-Side Constraintsto Low-Income Homeownershiprdquo Workingpaper Joint Center for Housing Studies Har-vard University 2001

Cooper Arnold C Woo Carolyn Y andDunkelberg William C ldquoEntrepreneursrsquo Per-ceived Chances for Successrdquo Journal ofBusiness Venturing Spring 1988 3(2) pp97ndash108

Dunne Timothy Roberts Mark J andSamuelson Larry ldquoPatterns of Firm Entryand Exit in US Manufacturing IndustriesrdquoRAND Journal of Economics Winter 198819(4) pp 495ndash515

Fama Eugene F and French Kenneth R ldquoCom-mon Risk Factors in the Returns on Stocksand Bondsrdquo Journal of Financial Econom-ics February 1993 33(1) pp 3ndash56

ldquoThe Equity Premium Puzzlerdquo Work-ing paper University of Chicago 2001

Flow of Funds Accounts Fourth Quarter 1952 to

1999 Washington DC Board of Governorsof the Federal Reserve System 1953ndash2000

Fenn George W Liang Nellie and ProwseStephen ldquoThe Economics of the Private Eq-uity Marketrdquo Working paper Board of Gov-ernors of the Federal Reserve System 1995

Gentry William M and Hubbard R Glenn ldquoEn-trepreneurship and Household Savingrdquo Na-tional Bureau of Economic Research(Cambridge MA) Working Paper No 78942001a

ldquoTax Policy and Entry into Entrepre-neurshiprdquo Working paper Columbia Univer-sity 2001b

Hamilton Barton H ldquoDoes EntrepreneurshipPay An Empirical Analysis of the Returns toSelf-Employmentrdquo Journal of PoliticalEconomy June 2000 108(3) pp 604ndash31

Hansen Lars P and Singleton Kenneth J ldquoSto-chastic Consumption Risk Aversion and theTemporal Behavior of Asset Returnsrdquo Jour-nal of Political Economy April 1983 91(2)pp 249ndash65

Harvey Campbell R and Siddique AkhtarldquoConditional Skewness in Asset PricingTestsrdquo Journal of Finance June 2000 55(3)pp 1263ndash95

Heaton John and Lucas Deborah ldquoPortfolioChoice and Asset Prices The Importance ofEntrepreneurial Riskrdquo Journal of FinanceJune 2000 55(3) pp 1163ndash98

ldquoCapital Structure Hurdle Rates andPortfolio ChoicemdashInteractions in an Entre-preneurial Firmrdquo Working paper Universityof Chicago 2001

Internal Revenue Service Income tax compli-ance research supporting appendices toPublication 7285 Publication 1415 Wash-ington DC US Government Printing Of- ce 1988

Johnson Barry W ldquoPersonal Wealth 1995rdquoSOI Bulletin Winter 2000 pp 59ndash84

Kennickell Arthur B and Starr-McCluerMartha ldquoChanges in Family Finances from1989 to 1992 Evidence from the Survey ofConsumer Financesrdquo Federal Reserve Bulle-tin October 1994 80(10) pp 861ndash82

Kennickell Arthur B Starr-McCluer Marthaand Sunden Annika E ldquoFamily Financesin the United States Recent Evidencefrom the Survey of Consumer Financesrdquo


Federal Reserve Bulletin January 199783(1) pp 1ndash24

Kennickell Arthur B Starr-McCluer Marthaand Surette Brian J ldquoRecent Changes in USFamily Finances Results from the 1998 Sur-vey of Consumer Financesrdquo Federal ReserveBulletin January 2000 86(1) pp 1ndash29

King Carol S and Ricketts Edward K ldquoEvalu-ation of the Use of Administrative RecordData in the Economic Censusesrdquo Workingpaper US Bureau of the Census (Washing-ton DC) 1980

Kraus Alan and Litzenberger Robert ldquoSkew-ness Preference and the Valuation of RiskAssetsrdquo Journal of Finance September1976 31(4) pp 1085ndash100

Mehra Rajnish and Prescott Edward C ldquoTheEquity Premium A Puzzlerdquo Journal of Mon-etary Economics March 1985 15(2) pp145ndash61

National Income and Product Accounts Washing-ton DC Board of Governors of the FederalReserve System various years

National Survey of Small Business FinancesWashington DC Board of Governors ofthem Federal Reserve System 1993

Of ce of Federal Housing Enterprise OversightHouse price index 1992 to 1998 Washing-

ton DC US Department of Housing andUrban Development various years

Parker Robert P ldquoImproved Adjustments forMisreporting of Tax Return Information usedto Estimate the National Income and ProductAccounts 1977rdquo Survey of Current Busi-ness June 1984 64(6) pp 17ndash25

Popkin Joel and Kirchoff Bruce A ldquoBusinessSurvival Rates by Age Cohort of BusinessrdquoWorking paper US Small Business Admin-istration 1991

Russo J Edward and Schoemaker Paul J HldquoManaging Overcon dencerdquo Sloan Manage-ment Review Winter 1992 33(2) pp 7ndash17

Survey of Consumer Finances Washington DCBoard of Governors of the Federal ReserveSystem 1989 1992 1995 1998

US Bureau of the Census Department of Com-merce New Home Sales 1993 to 1998Washington DC US Bureau of the Censusvarious years

US Small Business Administration Small Busi-ness Indicators 1998 Washington DC USSmall Business Administration 2000

Vissing-Joslashrgensen Annette ldquoComment onHeaton J and D Lucas Stock Prices andFundamentalsrdquo NBER Macroeconomics An-nual 1999 14(1) pp 242ndash53


Using data for all private equity from both theSurvey of Consumer Finances (SCF) and theFlow of Funds Accounts and the National In-come and Product Accounts (FFANIPA) overthe period 1989 to 1998 as well as proprietorand partnership data from the FFANIPA overthe longer period 1952 to 1999 we nd that theaverage return to all private equity is similar tothat of the public market equity index This issurprising since investing in the equity of asingle private company is likely to be muchriskier than investing in the public equity indexFirst survival rates of private rms are onlyaround 34 percent over the rst ten years of the rmrsquos life Second even conditional on sur-vival the distribution of equity returns acrossentrepreneurs is wide Third the average entre-preneur holds most of his investment in thesame private rm in which he works makinghis equity return highly correlated with his hu-man capital return Fourth while it is dif cult toprecisely estimate the overall risk of privateequity our estimates suggest that the index ofprivate equity is likely as volatile as the publicequity index and that aggregate private equityreturns are highly correlated with the publicequity market Finally the amount of idiosyn-cratic risk of a single private rm implies thatthe aggregate (index) return is likely an overes-timate of the average of the returns to eachindividual entrepreneur further strengtheningthe conclusion that private equity returns arelow Our results are robust to a variety of ad-justments for the labor component of entrepre-neurial income retained earnings in the rm rm births and deaths initial public offeringsand acquisitions and potential income underre-porting due to tax evasion Overall the diversi- ed portfolio of public equity seems to offer afar more attractive risk-return trade-off than thatobtained by the typical entrepreneur

To put our results into perspective considerwhat theory suggests the expected private eq-uity return to be The higher risk from lack ofdiversi cation of private equity should lead to ahigher private equity premium than that on pub-lic equity How much higher than the averagepublic equity return would we expect the aver-age private equity return to be John Heaton andDeborah Lucas (2001) model and calibrate thehurdle rate which would make a household in-

different between investing in a portfolio of asingle private rm a public equity index andT-bills or a portfolio of just the public equityindex and T-bills For an investor with a relativerisk-aversion coef cient of 2 (as well as reason-able assumptions about the debt-to-asset ratio ofthe private rm and the fraction of entrepreneur-ial wealth invested in private equity) purelyidiosyncratic private equity risk generates a hur-dle rate of about 10 percent above the publicequity return2 Michael J Brennan and WalterN Torous (1999) estimate a certainty equiva-lent wealth loss of investing in a single (public) rm of about 64 percent over a ten-year horizonfor an investor with a relative risk-aversion co-ef cient of 2 This loss increases to 95 percentfor an investor with a relative risk-aversion co-ef cient of 3 Shlomo Benartzi (2000) nds thata return premium of 20 percent is needed for anindividual (with relative risk-aversion coef -cient of 4) to invest 45 percent of his portfolioin a single publicly traded stock when the totalportfolio is restricted to have 40 percent bondsand 60 percent stock When allowing the totalportfolio to contain 100 percent equity andreducing risk aversion to 2 the required excessreturn of the individual stock over the publicequity index return declines to 5 percent There-fore the premium required to induce investorsto hold equity in a single private rm wouldalso have to be large For simplicity we willcite 10 percent as the required premium foruse in some of our ldquoback-of-the-enveloperdquocalculations

Of course obtaining a precise measure of themean return to private equity is extremely dif- cult The notoriously dif cult exercise of esti-mating the mean on a highly volatile returnseries over a relatively short time period is wellknown This dif culty is exacerbated when us-ing fairly imprecise data on estimates of private

2 This number is based on the average of line 2 and 3 inTable 4 of Heaton and Lucas (2001) This case correspondsmost closely to the ratio of private equity to net worthdocumented below based on the SCF and the ratio of debt toassets for proprietors and partnerships in the FFA Theauthorsrsquo calculation assumes a zero correlation betweenprivate and public equity (the private investment project intheir model has no aggregate risk) A positive correlationwould increase the private equity hurdle rate












I Data Sources



















Other 204













































1989 1992 1995 1998

















Market value 3292 4376 6734 13217











































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES



























































I Data Sources



















Other 204













































1989 1992 1995 1998

















Market value 3292 4376 6734 13217











































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES



















































I Data Sources



















Other 204













































1989 1992 1995 1998

















Market value 3292 4376 6734 13217











































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES

























































Other 204













































1989 1992 1995 1998

















Market value 3292 4376 6734 13217











































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES





















































































1989 1992 1995 1998

















Market value 3292 4376 6734 13217











































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES










































































1989 1992 1995 1998

















Market value 3292 4376 6734 13217











































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES


















































1989 1992 1995 1998

















Market value 3292 4376 6734 13217











































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES


















































































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES




































































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES





























































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES



















































Data from the SCF



Taxevasion





SampC PampP



Source





























(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES






































































(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES






























































(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES






















































(1) R tt 1 3 5AMV t 1 3 1 AE tt 1 3

AMV t

(2) AMV t 1 3 5 MV t 1 3 1 IPO tt 1 3



3 ~1 2 rRE 2 LC tt 1 3




~040 for C Corps





Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES


















































Acquirer




wprice2718 857 1657 5088 1312 2522 8943 1993 2477

Scaled value $133847 $43741 $85275 $211678 $85410 $106895 $610613 $196099 $158987


wprice1367 268 1021 2960 378 1516 5126 558 1367

Scaled value $ 63720 $20799 $33824 $131533 $36593 $ 31261 $407889 $ 77468 $ 58073


wprice136 30 61 352 59 48 424 46 77

Scaled value $ 7342 $ 5238 $ 9250 $ 23413 $ 9756 $ 5533 $ 60403 $ 13371 $ 19198


wprice265 87 520 268 133 892 454 161 760

Scaled value $ 13242 $10439 $14002 $ 15186 $14902 $ 12937 $ 37734 $ 32293 $ 23073


wprice114 15 52 220 28 77 518 50 66

Scaled value $ 3869 $ 295 $10909 $ 11690 $ 1317 $ 11628 $ 36187 $ 3626 $ 5083


wprice1094 175 510 2590 235 667 4801 414 623

Scaled value $ 47951 $ 8483 $24306 $106954 $16085 $ 25938 $351533 $ 41536 $ 28861

D Total valuea $ 63720 $15381 $24306 $131533 $23341 $ 25938 $407889 $ 42038 $ 28861(1) (2) (3) (1) (2) (3) (1) (2) (3)

























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES





































































































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES


























































































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES















































































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES




































































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES


























































Returns

Annualized returns

Arithmeticaverage

Geometricaverage

Standarddeviation



131 128 69


132 128 77



140 127 170


156 156 37


242 182 411






Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES



















































Single Private Firm













5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES























































5K1N 1 K2N 1 1 K

1 1 K1N 1 K2N 1 1 K ~N 2 1N 5 K1N




1 1 ~N 2 1K K1N




N K1N




















(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES


































































(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES

























































(4)

1Value at the

time of the survey



2 1




















































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES


































































































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES



























































































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES














































































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES



































































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES
























































ESTATE TAX RETURNS









SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES





















































SDC DATABASE







PampP~Equity t

where



FU136000105 1 FU133180005)






where











REFERENCES





















































where











REFERENCES














































































































Documents

The Returns to Entrepreneurial Investment: A Private Equity Premium Puzzle?faculty.haas.berkeley.edu/vissing/tmav_aer.pdf · 2003-04-08 · The Returns to Entrepreneurial Investment: