Introduction

Private equity (PE) funds are sophisticated financial intermediaries investing in relatively high-risk, illiquid securities in private companies on behalf of their investors, or limited partners (LPs). PE fund managers, or general partners (GPs), have the necessary expertise to successfully invest in those portfolio companies but they might be less transparent about their operations and disclosures to maintain their competitiveness. Such opacity vis-à-vis the general public and the less-regulated market status naturally results in two sources of agency problems: between GPs and their portfolio companies/entrepreneurs and between GPs and LPs (Morrell and Clark 2010; Kandel et al. 2011). Extant research in PE, more specifically on the agency problems and information asymmetries that arise in earlier-stage venture capital (VC), has concentrated on the relationship between GPs and portfolio companies/entrepreneurs (Jensen and Meckling 1976; Amit et al. 1998; Higashide and Birley 2002; Fried and Ganor 2006; Morsfield and Tan 2006; De Bettignies and Brander 2007; Yitshaki 2008; Bonnet and Wirtz 2012; Wongsunwai 2013). The GP-LP relationship between PE fund managers and their investors has been relatively understudied and thus is the main focus of this study.

In PE fund management, GP-LP agency problems are exacerbated by the arguably less rigorous regulation of PE funds as compared with other financial intermediaries such as mutual funds and hedge funds (Cumming 2005; Nielsen 2008; Cumming and Walz 2010; Ferran 2011). However, we also know that different types of investors in the PE market seem to enjoy different returns and are thus arguably better at mitigating agency problems and surmounting information asymmetries (Lerner et al. 2007). PE fund managers themselves tend to be quite opaque in regard to their investors. The limited partnership structure upon which funds are built promotes this opacity as PE fund investors risk losing limited partnership liability protection if they seek to involve themselves too much in the operation of the funds. Therefore, unlike PE fund managers who are skilled at negotiating terms of investment to ensure frequent and open interaction between themselves and their investee companies, LPs may not necessarily be incentivized to seek open and frequent interaction between themselves and their investee funds. In short, LPs may not necessarily seek to mitigate information asymmetries. However, prior research has found that open communication and frequent interaction between the investor and the investee to be a crucial element for the success of the investment (Sapienza 1992; Sapienza et al. 1996; Kanniainen and Keuschnigg 2003, 2004; Jääskeläinen et al. 2006; Katz 2009; Cumming and Dai 2010; Hain et al. 2016; Goktan and Muslu 2018). In this paper, we seek to add to the PE literature in determining the effect of open communication and frequent interaction between GPs and LPs, and the ensuing success of reduction in information asymmetries.

With reference to PE fund performance reporting, it has been well established that the main difficulty in determining PE fund success or failure rates over the life of the fund is that their performance measurements are relatively inaccessible to the public (Cumming and Walz 2010; Cumming et al. 2010). While mutual funds are required to make publicly available daily net asset values (NAVs), PE funds are not required, nor are expected to provide numerous reports of their holdings. Reporting frequency is agreed upon between GPs and each LP independently. Moreover, LPs in such funds, who are entitled to at least an annual performance report, are ultimately dependent on measurements of fund performance that are largely subject to PE fund managers’ discretion and more significantly, PE fund performance is determined by both realized and unrealized returns. Unlike mutual fund investors who can verify NAVs with publicly available information, LPs are dependent on relatively subjective valuation methods for active private investments (Cumming and Dai 2010; Cumming et al. 2013; DaRin and Phalippou 2017). Getmansky et al. (2004) suggested that permissiveness relating to relatively subjective valuations and reporting quality by hedge funds may result from the illiquid assets held by hedge fund managers, and as PE funds hold primarily illiquid assets, we believe this issue to be exacerbated in the PE market. Furthermore, there is no particular reporting standard that GPs must adhere to, just guidelines promoted by the respective national and regional PE associations. Thus, the way GPs report performance to various LPs in different countries around the world may differ significantly across funds. Although many national PE associations have encouraged the creation of more stringent guidelines for the reporting of unrealized returns, PE funds are not yet regulated as strictly as hedge funds or mutual funds and do not face any mandatory disclosure rules in any country regarding their performance.

In this paper, we seek to add to current research by providing evidence that PE fund reporting frequency is one of the possible explanations for the success of some type of LPs. Following Lerner et al. (2007), we also seek to address the “Limited Partner Performance Puzzle” that endowments’ returns from PE funds are nearly 14% greater than the average LPs. Although a recent study by Sensoy et al. (2014) has found that the puzzle is no longer the case because endowments no longer outperform other LPs during the 1996–2006 period, we believe our research is still highly relevant in determining the success of some LPs over others, especially endowments. Sensoy et al.’s (2014) research primarily analyzed performance measurements and they used a different sample period to attribute outperformance of endowments to greater access to top-performing VC partnerships. Our findings on PE fund managers’ reporting behaviors may provide further explanatory evidence of the superior performance of certain types of LPs over others, in this case endowments as institutional investors in the PE funds, from a more direct and governance-based point of view. We analyze PE fund reporting behavior of GPs and examine the theory of and empirical evidence on GPs’ reporting frequency. Institutional investors rank the quality of international disclosures from PE funds as one of the most important hurdles to be met before they make investment decisions. We therefore aim to obtain a specific picture of how LPs can better govern their investments and mitigate information asymmetries.

We believe that an examination of reporting behavior is particularly relevant considering its absence not only in PE research, but also in research related to other financial intermediaries such as hedge funds and mutual funds. We contribute to the debate in accounting literature about whether more or less frequent financial reporting is necessary and beneficial to investors while more frequent financial reporting might cause managerial myopia effects (Leuz and Verrecchia 2000; Nallareddy et al. 2017; Ernstberger et al. 2017; Kraft et al. 2018; D’Adduzio et al. 2018). We believe our study could have important implications for future research from financial intermediation managerial performance perspective. By using PE as a unique setting, we contribute to the business ethics literature as well about the ethical issues in the financial reporting quality to explore the possible biases within the performance reports from GPs to LPs (Stewart 1986; Staubus 2005; Frecka 2008; Morrell and Clark 2010; Kusnadi et al. 2016; Elayan et al. 2016). We further believe our study is the first one to use an international PE returns database to observe this type of reporting behavior.

Using comprehensive PE returns data from PitchBook, which comprise 4548 PE funds from 42 countries spanning the 2000–2012 period, we find that higher reporting frequency is associated with lower information asymmetries in the performance reports from GPs to LPs. We also find that endowments are systematically associated with less reported unrealized returns as a percentage of total returns generated from GPs. Moreover, endowments receive more performance reports from their PE funds, implying more stringent governance. These findings persist after controlling for various institutional and GP characteristics and are robust to several adjustments for endogeneity concerns. Our results provide proof that reporting frequency can serve to mitigate information asymmetries and agency problems, as well as serve to proxy for good corporate governance. We provide additional evidence to support prior studies on PE fund reporting behavior, such as Armstrong et al. (2005), Beuselinck et al. (2008), Cumming and Walz (2010), and Brown et al. (2019). However, we believe our study is innovative because those prior studies did not explore differences in institutional investor types or governance. Moreover, our robust results complement and support a recent study utilizing survey data by DaRin and Phalippou (2017) which suggest active governance effort is important for LPs in PE funds. While they find that the valuation of unrealized investments is a very important criterion to re-invest in seasoned funds, we have in this paper gone on to document, analyze and quantify such active governance by using reporting frequency. It is interesting to note that in their study, they also find that on average the LPs hold one advisory board seat for every three funds, and that they only attend 80% of the meetings. Put in another way, our results suggest that reporting frequency is especially important to LPs as investment advisory board members have no real operational powers (unlike investment committee members), so they rely more on reports issued by fund managers and if they do not attend meetings, as found by DaRin and Phalippou (2017), and prefer to use their own models to evaluate fund performance, they are especially reliant on reports issued by PE fund managers as benchmarks.

Overall, our analysis and results are consistent with the view that PE fund manager reporting behavior is directly linked to the information asymmetries reflected in PE performance reports. Some institutional investor types are systematically better at mitigating such information asymmetries, for example, endowments receive substantially more reports, and therefore experience substantially higher realized returns and lower unrealized returns from their PE investments.

The remainder of this paper is organized as follows. Section I discusses the prior literature and develops our hypotheses. Section II presents the data and our summary statistics, while section III covers the regression analyses and robustness checks. Section IV concludes and provides an outlook for future research.

Literature Review and Hypotheses

GPs act as financial intermediaries for LPs to invest in PE investments (Sahlman 1990; Norton 1995). The relationship between GPs and LPs, however, suffers the same principal–agent problems as in any corporation (Jensen and Meckling 1976; Fama 1980). Information asymmetry exists and is especially severe as LPs seek to select the best GP from numerous GPs available, to invest their funds (Burton and Scherschmidt 2004; Cumming et al. 2005). In this sense, two-way agency conflicts might emerge as LPs entrust more experienced GPs with investment decisions, and as GPs influence management of LP funds at the expense of LPs who eventually suffer from moral hazard (Akerlof 1970; Eisenhardt 1989). Although prior literature on agency problems focus on remedial tools such as screening to reduce the information asymmetries between GPs and LPs (Eisenhardt 1989; Norton 1995), we are hard pressed to find the economic mechanisms that include the role of reporting frequency for the reduction of information asymmetries between GPs and LPs in the performance reports and thus is worth investigating.

In the PE market, the determination of relative performance of investments within portfolios is critical for any investor, especially so for institutional investors such as pension funds, banks, insurance companies and endowments. These sophisticated investors essentially manage funds derived from their own clients, therefore, it is their responsibility to identify and invest in the best performing financial intermediaries that manage their portfolio of equities, bonds and the higher-risk alternative investments such as hedge funds and PE funds. Identifying the best performing financial intermediaries, however, is not straightforward. A seminal study of PE fund performance by Kaplan and Schoar (2005) analyzed data derived from voluntary reporting of fund returns by the PE fund managers to their investors. They reported that the performance of the 746 PE funds in their sample was close to that of the S&P 500 Index. Jones and Rhodes-Kropf (2003) also use the same data in their research but specifically qualify the self-reported data on NAVs due to the difficulty of accurately valuing privately held and illiquid investments and that different funds may also calculate NAVs in different ways. Subsequently, Gottschalg and Phalippou (2006) suggested that Kaplan and Schoar’s (2005) analysis of PE performance may be optimistic as a number of funds analyzed report high values for their unexited investments. The funds reported residual values of investments that, if written-off, would negatively affect the actual performance reported. Prior research has therefore established the relatively common practice whereby PE fund managers may report inflated valuations for securities held in private companies that are not yet exited. Even though such overvaluations can lead to reputational costs, international evidence provided by Cumming and Walz (2010) finds systematic biases in managers’ reporting of fund performance. Getmansky et al. (2004) also suggest that systematic biases in reporting of fund performance may also occur in the hedge fund industry. They argue that serial correlation in hedge fund returns may likely be the result of illiquid securities within fund portfolios that are contained in the fund as they are not actively traded and for which market prices are not always readily available.

Such intentional overvaluations by GPs are not only determined by the information asymmetries faced by LPs, but also by the expected marginal costs and benefits from those overvaluations. For example, the trade-off between the expected marginal benefit of receiving higher performance and management fees versus the expected marginal cost of reputational loss in the market may encourage GPs to misreport performance. Prior theoretical research along these lines has shown that if investors are patient enough, insiders will disclose their private information more truthfully, and the accounting disclosure environment is generally sufficient to assess the credibility of insider disclosures (Verrecchia 1983; Benabou and Laroque 1992; Healy and Palepu 2001; Stocken 2000; Neus and Walz 2005).

Therefore, the information asymmetry level and the possible misvaluation in the PE performance reports might raise similar concerns about the ethical issues of GPs. Prior literature in business ethics have documented the ethical issues and failures in financial reporting in different ways. In one example, the intentional structuring of lease contracts to avoid disclosing leased asset and liability amounts has been found to have led to the accounting debacle in Enron (Frecka 2008). Research has also found that fraudulent financial reporting, the release of financial statements with errors so material as to require restatement and the biased reporting marred by defects, all lead to the conflicts between management’s interest in reporting its performance in a favorable light and investors’ interest in decision-useful financial information (Staubus 2005). Thus, the quality of financial reporting is closely related to the ethics of corporate managers, accountants, and audit committees (Stewart 1986). For example, more diversified audit committees will have higher-quality financial reporting (Kusnadi et al. 2016) and the disclosures of positive changes in firm ethical performance will positively affect firm value (Elayan et al. 2016). Ethical disclosure has meaningful impact to the business and public good.

Although PE funds are not required by law or any specific regulations to routinely report performance to investors, PE fund managers should still adhere to the minimum mandatory reporting requirements of the type of partnership or corporate structure they have chosen to administer their funds. Additional reporting obligations regarding frequency or quality would depend on the terms of the contract between GPs and LPs (and note that each separate investor may potentially agree to different terms). Prior studies in private equity also highlight how reporting quality and reputation matters in the PE industry and how reduction in information asymmetry ultimately improves disclosure quality. By comparing VC-backed IPOs with non-VC-backed ones, Morsfield and Tan (2006) found that earnings management would be less upward in those VC-backed IPOs, and Wongsunwai (2013) further provided evidence that firms backed by high-quality VCs would have less aggressive financial reporting and thus have lower probability linking to manipulation. Katz (2009) also found that the presence of and monitoring by PE sponsors restrains upward earnings management and induces a higher frequency of timely loss recognition. Therefore, financial reporting of PE-backed firms is seen as value-informative (Armstrong et al. 2005; Hand 2005). A recent study by Goktan and Muslu (2018) examines the reporting quality of portfolio companies that are backed by listed and unlisted PE firms. They found that the public reporting model of listed PE firms leads to greater capital market benefits than the private reporting model of unlisted ones. Although those studies are mainly focused on the relationships between GPs and their portfolio companies, their insights about the quality and frequency in financial reporting lead us to work toward understanding the GP-LP relationship by investigating the reporting frequency in the performance reports.

Reporting frequency is also a lively debate subject in the accounting literature given recent consultations by the Securities and Exchange Commission (SEC) in the U.S. that is considering changing reporting frequency from quarterly to biannually. One concern is that more frequent reporting may spur myopia in the financial markets. Although Leuz and Verrecchia (2000) and Fu et al. (2012) confirmed that increased disclosure and transparency is beneficial through improved liquidity and reduced cost of capital, other studies found competing results. By examining how mandatory quarterly reporting affects managers’ business decisions in terms of real activities manipulations, Ernstberger et al. (2017) found higher managerial short-termism resulting from increased reporting frequency requirements. Similar results of increased real earnings management following the adoption of quarterly interim management statements (IMS) have been found by Nallareddy et al. (2017). A recent study by D’Adduzio et al. (2018) added fuel to the debate by providing evidence that higher reporting frequency mitigates investor myopia by providing investors more information on future earnings.

Empirical evidence shows us that PE fund disclosure is increasingly important to investors (Beuselinck et al. 2009; Cumming and Johan 2007, 2013). We believe one of the most efficient ways to reduce information asymmetries between GPs and LPs is to increase the frequency of performance reports. Reporting frequency is viewed as a proxy for managerial expertise, operational efficiency, and good corporate governance. Moreover, fewer time lags between reports may mitigate managerial discretion and can mean fewer opportunities for GPs to continuously misreport unrealized returns and convincingly overstate performance. While we do not underestimate the difficulty of accurately valuing privately held and illiquid investments, we believe that more frequent reporting may increase accuracy levels, or at least, highlight potential inaccuracies and mitigate information asymmetry levels. More frequent reporting also enables institutional investors to make better informed cash flow decisions, to better benchmark fund performance against their own NAV models (DaRin and Phalippou 2017) and to better manage the relatively higher illiquidity risks associated with PE fund investments. PE fund managers will also find that more frequent reporting allows them to overcome the informational risk hurdles perceived by institutional investors, and to encourage future fund inflows. We therefore hypothesize the following:

H1

Higher reporting frequency is associated with lower information asymmetry in the performance reports from GPs to LPs.

As mentioned earlier, neither reporting frequency (beyond the annual minimum) nor reporting quality are predetermined by laws or regulations, and they are primary obligations resulting from previously agreed upon terms between a PE fund and its investors. In view of international differences across PE funds and investors (Schertler and Tykvova 2006), we seek to test whether the impact of reporting frequency on information asymmetry will persist across countries for different time periods and under different institutional environments. Our focus is on the legal environments. Prior studies suggest that investors are better protected and financial markets are more transparent in countries with stronger legal protections (La Porta et al. 1998, 2006). Cheng and Courtenay (2006) found that better regulations and accounting standards positively affect the quality of voluntary reporting across countries. Huang et al. (2019) in turn suggested that investors may find it less costly to verify information disclosed by firms in countries with better legal standards. With regard to PE disclosure, Cumming and Johan (2013) found that fair valuation clauses in PE fund contracts are more likely to be enforced in countries with stronger and better legal environments and therefore such clauses are more likely to be included by PE fund investors. It must be noted though that better legal environment may not necessarily remove all incentives or potential avenues for GPs to misreport portfolio (company) values and increase the information asymmetry level in their performance reports. Miller and Reisel (2012) suggested that investors in the bond market revert to contractual agreements to overcome weaknesses in country-level protections. As such, additional measures may be taken by LPs, such as contractually requesting more frequent reporting from GPs.

These measures have both costs and benefits to GPs and LPs. Prior research related to firms vis-à-vis their investors found that increased reporting may lead to higher quality of proprietary information and more accurate market valuation of the firm assets (Verrecchia 1983, 1990; Darrough and Stoughton 1990; Hayes and Lundholm 1996; Admati and Pfleiderer 2000). In addition, some recent empirical studies also confirmed such benefits in terms of reducing the firm's cost of capital and improving liquidity in the stock market (Leuz and Verrecchia 2000; Fu et al. 2012). Gigler et al. (2014) also highlighted the benefits of increased financial reporting frequency to serve as more effective discipline tools of market prices and deter negative NPV projects. Mensah and Werner (2008) found that more frequent interim reports could lead to more efficient security pricing. Van Buskirk (2012) found that more frequent disclosure will provide an incentive for increased private information acquisition by sophisticated investors which will lead to reduced information asymmetry and accelerate the rate where information is impounded into market prices. Although Van Buskirk (2012) covers the retail sector only, we believe his findings are highly relevant to our study of PE funds where private information disclosure by GPs are relied heavily upon by LPs to determine the value of their investments and also to mitigate managerial discretion. We expect LPs to seek increased disclosure from GPs due to potential benefits listed above.

There are of course costs to these additional measures. Prior theoretical research has found that increased voluntary disclosure could potentially affect PE fund investment efficiency and might also be costly in terms of additional compliance costs (Stein 1989; Gigler et al. 2014; Roychowdhury et al. 2019). These costs lead us to seek to determine the persistence of voluntary reporting frequency by PE funds across different legal jurisdictions. We therefore refer to Renders and Gaeremynck (2007) who looked at early (voluntary) IFRS adoption across seven European countries. They argued that the adoption of this accounting standard that is associated with increased disclosure by firms depends on the level of investor protection and the ensuing cost of adoption. In countries with weaker investor protection, firms that have higher private benefits are less likely to adopt IFRS due to the potential loss of these benefits, or higher costs to insiders. They contrasted this with countries that have stronger laws protecting investors that limit private benefits of control, and thus lowering the costs of implementation. Following their study, we believe that GPs will increase disclosure in better legal environments to take advantage of the benefits vis-à-vis their investors listed above, but also because GPs have lower private benefits in these better institutional environments and thus the cost of additional disclosure is lower.

With regard to different types of institutional investors, Lerner et al. (2007) found that endowments perform relatively better in PE investments than other LPs. As one of the most active capital market players, endowments prudently invest their assets with specific long-term perspective attempting to maximize returns while controlling for risks to reserve capital for future generation as well as to meet the current missions. Many prior studies in the literature have been focused on the performances and strategies of endowments trying to shed more lights on the secrets of their successes. In Lerner et al.’s (2008) follow-up paper, they used College Board and NACUBOFootnote 1 survey data, and found that the superior investment performance of endowments is attributable to their higher-quality student body, more sophisticated use of alternative investment tools, and larger investment size. By using more updated NACUBO data, Dimmock (2012) also found that those wealthier and highly selective universities will hold much riskier portfolios in their endowments. Brown et al. (2014) also showed that endowments would change payout policies in response to financial market shocks to better mitigate downside risks.

Although some other studies shared different voices on such outstanding performances no longer existed with evidences that endowments, on average, did not outperform the market index and have negative alphas in their sample (Barber and Wang 2013; Dahiya and Yermack 2019), Barber and Wang (2013) still found that strategic allocations to alternative investments were a source of excess returns for a handful group of elite endowments. Such an outcome coincides with a recent study by Lo et al. (2019) who used U.S. IRS tax filing data to investigate the risk, return and asset allocations of endowments. They concluded that the larger the size of endowments, the more they will invest in riskier and higher-returning assets. Given the fact that over the past two decades, endowments shifted their asset allocation weights more and more towards alternative assets such as hedge funds, private equity, venture capital, private real estate, and illiquid natural resources (Dimmock et al. 2019), this strategic change contributed to the outstanding performance generated from those endowments and has been widely utilized by other institutional investors. Brown et al. (2010) also used asset allocation data from university endowment funds to show that actively managed funds outperform passively managed ones.

Allocation to alternative asset classes such as private equity requires better asset selection skills, longer gestation periods and higher minimum investments, endowments as pioneers and major investors in the PE market reaped high returns. Although we analyze different periods in our sample, we believe endowments will still perform relatively better than other LPs as previous research finds that they are arguably more sophisticated LPs. In Lo et al.’s (2019) study, they also determined that other characteristics including expenses and governance structures might also impact investment outcomes as independent boards were linked to higher returns. Therefore, we expect that for endowments, higher reporting frequency will be the chosen contractual substitute to overcome weaker investor protection and this will jointly affect the information asymmetry level in the performance reports of those PE funds residing in better legal environments. Our second hypothesis is summarized as:

H2

The negative association between reporting frequency and information asymmetry in the performance reports from GPs will be strongest when LPs are endowments and legal environments are better as GPs have lower private benefits of control to relinquish.

Data and Summary Statistics

Our study takes advantage of PitchBookFootnote 2’s comprehensive PE returns database. The PitchBook data not only include voluntarily self-reported fund-level performance multiples, but also a rich trove of information on related LPs, fund managers, and fund characteristics. Our sample data span 2000 through 2012, with a total of 4548 PE funds from 42 countries. The data comprise 491 LPs, with a wide range of institutional investors, where approximately 70% are pension funds (a combination of public and corporate), nearly 22% are insurance companies, and about 7% are endowments. The remaining 1% cover funds of funds, sovereign wealth funds and other LPs.

Our cross-jurisdictional analysis across investor and PE fund type adds to the existing literature on single country analysis or single LP type. Moreover, to obtain the maximum amount of information on fund performance and reporting behavior, we analyze various fund types as well as various fund styles. This differentiates our study from recent work that has focused primarily on buyouts or venture capital funds (Harris et al. 2014; Robinson and Sensoy 2016).

Table 1 summarizes the main variables in our dataset. We investigate the ex-post examined reporting frequency from each GP to various LPs per year and the RVPI as a percentage of TVPI reported as dependent variables. The explanatory variables include the PE valuation variables DPI, RVPI, the GDP growth rate for each country’s economic development, the MSCI returns for each country’s stock market conditions, legality index, Hofstede’s long-term orientation variable as a cultural dimension, as well as a variety of control variables to capture LP and GP characteristics.

Table 1 Variable definitions and summary statistics
Full size table

Our first primary variable of interest is reporting frequency, which is calculated as the actual ex-post examined reporting times from a GP to different LPs. Note that it is common for fund managers to exhibit some differences in their reporting behaviors, because reporting frequency and quality depend on each LP’s requirements. For example, if a GP has three LPs, it may be providing an annual performance report to LP #1, a quarterly performance report to LP #2, and a monthly performance report to LP #3.Footnote 3

From our data, we can empirically investigate such behavior across PE funds in different countries and over our sample period. This provides us with a unique opportunity to explore all the possible reasons behind. For example, in addition to the different contract terms between GPs and LPs, there may also be other governance or monitoring reasons. Moreover, the PitchBook database provides detailed information on testing international differences across countries and over time, which can further explain the reasons behind such reporting behavior.

We examine the data in a panel setting across countries and years. Our first step is to present some of the highlighted details from our PitchBook data and provide preliminary means difference test results. We divide our entire sample by LP types (for a total of 491, which includes 9 endowments, 163 pension funds, 296 insurance companies, 10 funds of funds, 2 sovereign funds, and 11 "other" LPs which the data classify simply as "other"). Our sample thus covers a wide range of LPs and provides an interesting opportunity to explore PE fund characteristics across different LP types. The LPs in our sample began investing in the PE sector from 1976; the latest began in 1999. In later sections, we will discuss the relatively superior performances by endowments. But here we just provide additional evidence to support the early-mover advantages documented by Lerner et al. (2007), because, in our sample, endowments were among the first to begin investing in PE.

In panel A of Table 2, we find that, as compared with other LPs, endowments are relatively smaller with fewer total assets under management, but endowments almost always occupy the top two positions in terms of annual commitments, contributions, and distributions to their PE funds, which implies that they are among the most active players in the PE sector. The last column also confirms the activeness in the PE market of endowments with about 183 PE funds invested every year.

Table 2 Mean descriptive statistics by main characteristics
Full size table

Panel B of Table 2 divides our sample by LP types and presents all the sample mean characteristics, including total number of funds invested, average fund size, four main performance measurements, reporting behavior, information asymmetry proxy, and the number of LPs associated with each fund per year. Public pension funds invest in the highest number of PE funds, followed by insurance companies and endowments. In the second column, we summarize average fund sizes, with all types of LPs investing significantly differently from each other where funds of funds are generally the largest, followed by insurance companies and public pension funds. And the last column shows that the number of LPs per PE fund for each year in our sample. It is somewhat puzzling that PE funds invested by insurance companies have an average of about nine LPs investing in them, which implies those funds are relatively popular. But we find that they exhibit rather poor performance over time, so the level of popularity is quite surprising.

The columns 3–8 of Panel B in Table 2 are what we want to discuss more as those involved the main variables of interest of this study. We summarize the four most commonly used performance measurements. Besides the “Other LP Type” category, endowments exhibit the highest IRR, the highest DPIs, the second lowest RVPIs, and the second highest total TVPIs.Footnote 4 We find similar evidence as in Lerner et al. (2007), and we demonstrate that endowments are relatively the best investors in the PE market.Footnote 5 On average, they exhibit significantly higher realized returns and significantly lower unrealized returns than all other LPs. Due to a lack of observations of IRR measurements, we focus on those three valuation multiples in later sections. In the next column, we also summarize the reporting behavior from different LP types in terms of reporting frequency. We find new evidence that may help explain why endowments perform better than other LP types, which may also help address the so-called “Limited Partner Performance Puzzle” raised by Lerner et al. (2007). Endowments enjoy the highest reporting frequencies from their PE funds, an average of about 3 times over the course of a calendar year. This is significantly higher than all other LPs. We believe that higher performance reporting frequency can help reduce information asymmetries between GPs and LPs and serve as a proxy for better governance procedures between them.

We discuss this notion in more detail in later sections, and we provide robust results to confirm our findings. The next column summarizes our second main variable of interest to proxy for the information asymmetries in the performance reports. We created this variable under the notion that within RVPI reported, the residual value is an estimated fair value of the underlying portfolio investments and it is largely subject to PE fund managers’ discretions even not manipulations. If such residual value makes up most of the component in the total value reported, then the inaccurate estimates drive the performance of GPs. Given GPs have relatively more information advantage about the portfolio firms invested and their actual performances with strategic exit plans, the higher the RVPI as a percentage of TVPI reported, the more information asymmetries are between GPs and LPs and thus more variability those valuation multiples may experience over time. We note that endowments have the lowest percentage reported among all other LPs which means the information asymmetries in the performance reports are the lowest between endowments and their GPs. This is also a proxy for good governance between endowments and their GPs and such facts may explain the higher level of monitoring found by endowments, as well as another potential secret of their success.

Panel C of Table 2 presents the preliminary results regarding the implementation of FAS 157 and the associated possible impact will have on our main variables. In September 2006, the Financial Accounting Standards Board (FASB) changed the rules about NAV valuations, part of FAS 157 (ASC 820) requires fair-value reporting of balance sheet assets and all U.S. funds should comply by the end of 2008. In this way, it is important to take this regulation change as a consideration for our tests. Here we mainly present the means difference tests regarding reporting frequency and the RVPI as a percentage of TVPI reported. We constructed a dummy variable of post-2008 which equals one when the FAS 157 should have been implemented after year 2008 and equals zero otherwise. We restrict our subsample in the U.S. context by different LP types to further explore how such a regulation will actually change the reporting behaviors from PE fund managers given this regulation change. If all else equal, such a change should improve the disclosure quality and will have relatively bigger impact for those who have lower reporting frequencies. From our tests, we have found that the reporting frequency is significantly more after the FAS 157 implementation. But for the impact across different LP types, we did find that fund of funds, who seek much lower reporting frequency before the FAS 157 will have significant improvement, but for those LPs who had negotiated better reporting behavior, some of them decreased the reporting frequency, although such decreases are in trivial numbers. For the information asymmetries variable, we also find the similar results as presented for the reporting behavior, for example, insurance company will receive significantly improved-quality performance reports from fund managers. We also run several robustness checks regarding the FAS 157 implementation in the following regression analysis section and confirm that it will improve the disclosure quality from PE fund managers.

Table 3 presents simple correlations to highlight the relationships across the major variables. The correlations in Table 3 confirm the hypothetical view about the information asymmetries and reporting behavior at the fund level. It is clear that, when reporting frequency is higher, the information asymmetries decrease, and the same negative relationship also applies to endowments and legal environment. However, note that our correlations also highlight some potential collinearity issues across different explanatory variables, which we explore in our multivariate empirical tests in the next section.

Table 3 Correlations
Full size table

Regression Analyses

Now that we have laid out some of the unique features and consistent findings from prior studies, we perform our regression analyses using standard OLS methods. We report double clustering model results by controlling individual GP funds, fund types, countries and years effects estimates in Tables 4 and 5. To address the potential endogeneity issues, we detailed our instrumental variable strategy in Table 5. We also perform several interaction and subsample tests as robustness checks before drawing our conclusions. Those reports are in Tables 6, 7, and 8.

Table 4 Regression models for the determinants of report frequency
Full size table
Table 5 Regression models for reporting frequency impact on information asymmetries in PE
Full size table
Table 6 Interaction models for reporting frequency impact on information asymmetries in PE
Full size table
Table 7 Regression models for reporting frequency impact on information asymmetries in PE—subsamples by fund age
Full size table
Table 8 DiD regression models for FAS157 impact on information asymmetries in PE
Full size table

What Factors Determine the Reporting Frequency?

In our main regression analyses, we first consider what factors drive the reporting frequency across PE funds to their investors. As there is no mandatory rule similar to the public equity market that listed firms should report quarterly or biannually as requested by the SEC or stock exchanges like NYSE, the reporting rules among GPs and LPs are typically agreed by the limited partnership contracts and LPs are entitled to at least one annual report. Thus, it is important to get the idea of the causality direction and relationship of the reporting frequency and performance measurements. The main regression models in Table 4 use the following specification:

$${{\text{Report}} {\text{Frequency}}}_{\text{it}}=\alpha +{\beta }_{1}{\text{DPI}}_{\text{it}}+{\beta }_{2}{\text{RVPI}}_{\text{it}}+\theta {X}_{\text{it}}+{\varepsilon }_{\text{it}}$$

Xit denotes control variables including GDP growth rates, stock market conditions, legal environments, fund and GP characteristics, LP types as well as one Hofstede’s cultural dimension of LTO and ε is the error term. Most of the major variables are defined in Table 1. Note that there are many explanatory variables that we could have included but chose to exclude. The primary reasons for our parsimonious specification are as follows. First, the selected variables are plausibly pertinent to PE fund reporting frequency. Second, note that the excluded variables are highly collinear. Hence, any additional control variables for the available sets of countries and years would not be perfectly suitable without potentially introducing spurious results into the regressions. Examples include some other cultural dimensions measured by Hofstede, as well as other legal and institutional variables. Our selection and reporting of variables were conducted to assess the factors that directly impact information asymmetries and reporting behaviors in PE funds.

Table 4 presents our test results with reporting frequency as a dependent variable and Models (1) and (2) are using full sample while Models (3) and (4) are using subsamples for all PE funds report more than once in a year. Our purpose here is to investigate what factors determine the reporting frequency and whether it is mainly required by different LPs. For those PE funds report more than once in a year, what other factors might drive such behaviors. All models in Table 4 return consistent and statistically significant coefficients on different LP types where only endowments are positively associated with reporting frequency implying that they will require relatively more performance reports from their GPs. The two performance variables of DPI and RVPI do not report significant results which alleviate some concerns regarding the driving force of reporting frequency is more performance related but not set by the limited partnership contracts. Besides these main discoveries we present in Table 4, we also find some other interesting findings: (1) for GPs residing in better developed economies and in better legal environments, LPs tend to require less frequent performance reports; (2) for GPs residing in countries sharing more long-term oriented cultures, their LPs seems to be more patient investors and request relatively less performance reports. In our subsample tests of Models (3) and (4), we expect the impact from our full sample tests will be more prominent and we find consistent results. We also find that more active GPs with more investments on hand will tend to provide more frequent performance reports which intuitively correct.

Will Higher Reporting Frequency Mitigate Information Asymmetries Between GPs and LPs?

In order to test our main hypothesis that higher reporting frequency is associated with lower information asymmetries in the performance reports, we first run a preliminary test following the below specification without introducing other control variables:

$${\frac{\text{RVPI}}{{\text{TVPI}}}}_{\text{it}}=\alpha +{\beta }_{1}{{\text{Report}}\; {\text{Frequency}}}_{\text{it}}+{\beta }_{2}{\text{Endowments}}\; {\text{Dummy}}_{\text{it}}+{\beta }_{3}{\text{Legality}}\; {\text{Index}}_{\text{it}}+{\varepsilon }_{\text{it}}$$

We use double clustering OLS models by controlling fund types, individual funds, countries, and years fixed effects but without other facets of control variables. In this way, we can get a better picture what factors will affect the information asymmetry levels presented in the performance reports. Model (1) in Table 5 indicates that higher reporting frequency will reduce information asymmetries significantly with a pronounced economic effect: a one standard deviation increase in reporting frequency will reduce the information asymmetries in performance reports by 14.7% relative to the mean value of all country-years. In addition, better legal environments and endowments will also be associated with significantly lower information asymmetries. Combined the results from Table 4, we can find that when GPs report more frequently, their performance reports will suffer less information asymmetries. Moreover, as endowments will require more performance reports, they tend to enjoy less information asymmetries and such results could help us confirm that reporting frequency is a good proxy for high-quality governance and monitoring efforts by endowments to request more frequent reporting in their contract terms. Better legal environments will also be an important factor which we cannot ignore.

However, our setting will also raise a natural question about endogeneity issues. Such an issue might arise from the selection criteria of endowments that are driving the demand for higher reporting frequency and thus reduce the information asymmetries in the performance reports. It is also possible that some omitted variables could simultaneously drive endowments’ choices and information asymmetries. In addition, when adding new control variables into the model specification, it will also raise the chance for interacting or mitigating factors from some of the variables as well as the possibility of collinearity issues. Therefore, we designed a two-stage regression model with an instrumental variable to further address this potential issue.

For the instrumental variable (IV), we first tracked the voluntary switches of reporting frequency within each funding commitment contract by each GP and created a dummy variable which highlights those unchanged ones where coded as one and otherwise zero. The idea is to identify those PE fund managers who will voluntarily switch their reporting frequency as a treatment group to compare those who do not switch. If it is a good instrument then it should be correlated with the endowments’ selection criteria but uncorrelated with our information asymmetry variable (Angrist and Krueger 2001), because if they do not switch, the information content in their reports might be less likely to change. Our following models in Table 5 use the below specification:

$$1{\text{st}}\; {\text{Stage}}:\Pr\left({\text{Endowments}}\; {\text{Dummy}}=1\right)=\alpha +{\beta }_{1}{{\text{Report}}\; {\text{Frequency}}}_{\text{it}}+{\beta }_{2}{IV}_{\text{it}}+\theta {X}_{\text{it}}+{\varepsilon }_{\text{it}}\;2{\text{nd}} \;{\text{Stage}}: {\frac{\text{RVPI}}\;{{\text{TVPI}}}}_{\text{it}}=\alpha +{\beta }_{1}{{\text{Report}} \;{\text{Frequency}}}_{\text{it}}+{\beta }_{2}{{\text{Endowments}}\; {\text{Dummy}}}_{\text{it}}+\theta {X}_{\text{it}}+{\varepsilon }_{\text{it}}$$

Our 1st stage model (Model (2)) confirms our proposition and reveals the selection criteria by endowments as compared with other LPs which might shed some more light on their superior performance in the PE market. They tend to invest more in relatively smaller-sized mature PE funds with more active managers and report more frequently, and those funds are more likely to reside in countries with better economic development and legal environments. More specifically, the instrumental variable returns significant positive coefficients indicating that those PE funds invested by endowments are less likely to switch their reporting frequency within each funding commitment contract.

In our 2nd stage regression models, we use the instrumented results from our 1st stage model and still find both reporting frequency and endowments return significant negative results which confirm our H1 (Model (3)) and minimize the possible endogeneity concerns. When we add more control variables in the 2nd stage model as presented in Model (4), we still find that reporting frequency is a good indicator to reduce the information asymmetries with sizable economic significance: a one standard deviation increase in reporting frequency will reduce the information asymmetries in the performance reports by 3.4% (Model (3)) and 1.8% (Model (4)), respectively. We also find that larger-sized PE funds are associated with less information asymmetries and fund age is another critical factor to consider as younger funds are just established, and their information asymmetries levels tend to be the highest and will reduce when portfolios are continuously divested.

Can Reporting Frequency, Legal Environment and Endowments Jointly Mitigate Information Asymmetries Between GPs and LPs?

After using instrumental variable tests to alleviate potential concerns about endogeneity and selection biases in our models, we continue to perform several additional tests to further support our propositions. As discussed in the previous subsections, as we introduce more control variables in the model, there could be some interacting and mitigating effect from those variables of each other that reduce the model explanatory power. In our following tests of Table 6, we use a triple interaction model to specifically focus on the mitigating effect from endowments with reporting frequency and legal environments on the level of information asymmetries in the performance reports. Our model specification is as follows:

$${\frac{\text{RVPI}}\;{{\text{TVPI}}}}_{\text{it}}=\alpha +{\beta }_{1}\;{{\text{Report}} \;{\text{Frequency}}}_{\text{it}}+{\beta }_{2}\;{\text{Endowments}}\; {\text{Dummy}}_{\text{it}}+{\beta }_{3}\;{{\text{Legality}}\; {\text{Index}}}_{\text{it}}+{\beta }_{4}\;{{\text{Report}}\; {\text{Frequency}}}_{\text{it}}\,*\,{{\text{Endowments}}\; {\text{Dummy}}}_{\text{it}}+{\beta }_{5}{\;{\text{Legality}} \;{\text{Index}}}_{\text{it}}*{{\text{Endowments}} \;{\text{Dummy}}}_{\text{it}}+{\beta }_{6}\;{{\text{Report}}\; {\text{Frequency}}}_{\text{it}}{*{\text{Legality}}\; {\text{Index}}}_{\text{it}}+{\beta }_{7}\;{{\text{Report}}\; {\text{Frequency}}}_{\text{it}}\,*\,{{\text{Endowments}} \;{\text{Dummy}}}_{\text{it}}\,*\,{{\text{Legality}}\; {\text{Index}}}_{\text{it}}+\theta {X}_{\text{it}}+{\varepsilon }_{\text{it}}$$

Such model specification can help us understand the mitigating effects jointly affect our information asymmetry variable and we will focus on the sign of β7 and expect the sign to be significantly negative.

Our models in Table 6 confirm this proposition as well as our H2, that when endowments invest in PE funds residing in better legal environments and request more performance reports, the information asymmetry level in those reports will be significantly lower. By analyzing both the full sample, and those subsamples excluding PE funds only reporting once a year, we find consistently robust results. And by controlling the fund age effect in every model specification which could capture not only the maturity of the fund but also their vintage year and investment timing effect, we provide relatively robust results to support our proposition.

However, another concern that might be raised is that younger-aged funds usually have much more unrealized or unexited investments as compared with more mature funds that approach the end of their limited partnership term. In this regard, the information asymmetry level is naturally high during the earlier life of the fund and will be consistently reduced over the life of funds. We thus consider whether the mitigating effect would be different for mature funds and younger funds. We further divide our sample by the median fund age of four and perform our robustness checks. In our Table 7, we use the same model specification as we used in Table 6 and we find the mitigating effects on information asymmetries are more pronounced for funds aged below four or who are much younger in our sample and we cannot find such mitigating effects for the other group. By controlling the effect from fund age, we still find that when endowments invest in PE funds residing in better legal environments and request more performance reports, the information asymmetries level in those reports will be significantly lower, especially for those much younger funds. This result also echoes our view that reporting frequency could be a good governance tool to address the potential agency problems between GPs and LPs. Endowments is a good example for other LPs to learn from.

To further clarify these explanations, Fig. 1a–d present the information asymmetries variable, reporting frequency, and performance measurements trends over the 10-year fund life period. Compared with the two other major LP types, public pension funds and insurance companies, endowments consistently perform the best over fund life across those four main characteristics.

Fig. 1
figure 1

a RVPI as % of TVPI over fund life. b Reporting frequency over fund life. c DPI over fund life. d RVPI over fund life

Full size image

Figure 1a shows that endowments and public pension funds tend to receive performance reports with consistently lower information asymmetries over the years while insurance companies do not present such a trend. In Fig. 1b, we note that, over the life of a fund, endowments’ reporting frequencies are consistently the highest, followed by public pension funds and insurance companies. The better governance role between endowments and their PE fund managers is therefore obvious.

In Fig. 1c, d, which present the trends on the two main performance measurements, we observe that, in terms of the realized returns of DPI multiples, over the fund life, endowments and public pension funds increase their investment efficiency in PE, with endowments regaining a relatively better position. In terms of unrealized returns of RVPI multiples, we find that, over the fund life, the public pension funds and endowments all exit their PE investments efficiently with a declining slope. Insurance companies perform much worse in PE investments than the other two LP types.

Will FAS 157 have any Impact on the Reporting Frequency and Information Asymmetries?

In Table 2 Panel C, we have shown that after the FAS 157 implementation in the U.S., the overall reporting frequency increased, and the information asymmetry level decreased. However, we still find such changes suffered some variability across different LPs. We thus designed a new difference-in-differences (DiD) test to find out whether FAS 157 will have any impact on the information asymmetry level given the changes in those PE funds on reporting behaviors. Our model specification is presented below:

$${\frac{\text{RVPI}}\;{{\text{TVPI}}}}_{\text{it}}={\alpha +\beta }_{1}\;{{\text{FAS}}157 \;{\text{Dummy}}}_{\text{it}}+{\beta }_{2}\;{{\text{Frequency}}\; {\text{Change}} \;{\text{Dummy}}}_{\text{it}}+{\beta }_{3}\;{{{\text{FAS}}157 \;{\text{Dummy}}}_{\text{it}}\,*\,{\text{Frequency}} \;{\text{Change}} \;{\text{Dummy}}}_{\text{it}}+\theta {X}_{\text{it}}+{\varepsilon }_{\text{it}}$$

Similar to the framework of Agrawal (2013), the coefficient β3 is a DiD estimate of the effect of FAS 157 implementation on the dependent variable. Following our instrumental variable creation, the Frequency Change Dummy is coded as one when any fund voluntarily switches to increase the reporting frequency and equals zero when any fund voluntarily switches to decrease the reporting frequency. Please note that within each funding commitment contract, funds barely switch their reporting frequency, therefore any changes could be attributed to exogeneous event like regulation change and are subject to some more attention. We thus expect to find significantly negative coefficient on β3 which helps us to understand the impact from FAS 157 implementation on information asymmetries in the performance reports from GPs to LPs.

Our results in Table 8 confirm our proposition that for those funds voluntarily switching to report more after the FAS 157 implementation, the information asymmetry level will be lower. This is not surprising and intuitively makes sense: if a fund manager is willing to switch and increase his performance reporting frequency as a result of the regulation change like FAS 157 to encourage more market-based fair valuations on illiquid assets remaining in their portfolio, those LPs would receive relatively more accurate valuation and the performance multiples from those reports thus reduce the information asymmetries between those two parties. Our results might also contribute to the debate about FAS 157 as a controversial rule in the PE market. But we have to admit that our FAS 157 dummy variable construction could be coincide with the recent global financial crisis period which might reduce some explanatory power of our models. Since we only compare those funds with positive or negative changes in their reporting frequency and teasing out those unchanged ones, it will alleviate some concerns, but we would still accept it as a limitation for our study.

Conclusions

Using comprehensive PE returns data from PitchBook, which comprise 4548 PE funds from 42 countries spanning the 2000 to 2012 period, we find that higher reporting frequency is associated with lower information asymmetries in the performance reports from PE fund managers to institutional investors. We also find that endowments are systematically associated with less reported unrealized returns as a percentage of total returns generated from PE funds. Moreover, endowments receive more performance reports from their PE funds, implying more stringent governance even when investing in countries with strong legal environments. These findings persist after controlling for various institutional and GP characteristics and are robust to several adjustments for endogeneity concerns. We believe our analysis of financial reporting behaviors from an arguably less-regulated set of financial intermediaries (as compared to mutual fund and even hedge fund managers) is an interesting contribution to the finance, accounting and business ethics literature on financial reporting and disclosure quality.

However, our study is not without limitations. Perhaps the most notable one is that the PitchBook dataset and our empirical tests do not enable a perfect assessment of causality as we do not have a natural experiment that affords a clearly exogenous test. This database also has some potential shortcomings such as backfill bias that those poorly performing funds might simply not report at all and the pension funds pressure can also create a bias against non-disclosure. Having all those obstacles overcome would facilitate future research on this topic.

Overall, we believe our findings are useful for policymakers and for market monitors. Although the PE market is not subject to mandatory disclosure regulation yet, the contract terms are nevertheless subject to corporate laws. Reporting frequency can act as a good governance and monitoring effort from LPs by implementing higher frequency terms into their contracts. We believe that, by requiring or recommending stronger standards, the efficiency of capital investments will increase because possible performance overstatement and information asymmetry will be reduced, and the negative impact on reputation will be reduced as well.