7

Illiquid Assets and Risk

Investors who seek higher returns generally have to accept higher risk. To the extent that the efficient market hypothesis holds, it is impossible for investors to find risk-adjusted excess returns, with risk being defined in the standard CAPM context. In practice, of course, “…no market prices assets precisely at fair value all of the time”. However, “most markets price most assets with reasonable efficiency most of the time, providing few opportunities for easy gains” (Swensen, 2009). Any mispricing of risk gives rise to relative-value trading strategies, which help ensure that excess returns are arbitraged away.

The high degree of efficiency in markets for marketable assets encourages investors to seek investment opportunities in less transparent and efficient markets. As Swensen (2009, p. 82) argues, these markets are typically illiquid, “…since rewarding investments tend to reside in dark corners, not in the glare of floodlights”. However, as long-term investors venture into illiquid markets, such as private equity and real assets, the question immediately arises as to how risk should be defined and measured, an issue we already discussed in Chapter 5. To begin with, market prices are not observable. Instead, quarterly returns reported by limited partnership funds are based on subjective NAVs, which poses substantial challenges in terms of estimating risk-adjusted returns. And even if researchers find ways to get around this issue, for example, by working with cash flow data, time series for returns tend to be very short in most market segments other than US buyouts and US venture capital.

Investors thus face an important dilemma. In traditional markets, there are few opportunities to enjoy excess returns. By contrast, non-traditional markets which are usually relatively illiquid are more likely to be subject to inefficiencies, which may provide room for excess returns. However, in such markets, it is difficult to employ standard allocation models, which rely on a particular concept of measurable risk that does not easily apply to illiquid assets.

Against this background, this chapter starts by distinguishing between risk and uncertainty as different types of probability situations. In this context, we emphasize the importance of subjective elements in assessing risk in illiquid investments as opposed to working purely with objective probabilities. Turning to managing risk, finally, we discuss the relation between risk management and due diligence, emphasizing that these are clearly distinct, but complementary functions of the investment process.

Notwithstanding the issue of how risk in illiquid assets is defined and measured, our discussion accepts the general principle that risks and rewards are positively related. This principle is important as it implies risk management should be unbiased in the sense that it focuses on both downside and upside risks. In fact, as Damodaran (2007) observes, the Chinese symbol for “risk” is a combination of danger and opportunity. However, while the basic principle of a two-sided risk management approach applies to illiquid assets as well, particular skills are required to address the specific challenges investors face in committing capital to private equity funds and similar structures.

7.1 RISK, UNCERTAINTY AND THEIR RELATIONSHIP WITH RETURNS

In a single-asset portfolio, finance theory usually defines risk as the variance of returns of a specific investment. In a multi-asset portfolio, as the MPT shows, risk can at least partly be diversified away to the extent that asset returns are less than perfectly correlated. In empirical applications of the CAPM, risk is measured on the basis of historic movements of asset prices, which by definition requires the availability of appropriate time series. In private equity and real assets, this condition is hardly satisfied. First of all, the very nature of these asset classes implies that there are no observable market prices, which poses substantial challenges in terms of calculating market risk in the context of the CAPM. There are different approaches in the literature to deal with this problem but, as we reviewed briefly in Chapter 5, empirical estimates vary significantly. Thus, some investors continue to rely on informed judgment in formulating their mean–variance assumptions (Swensen, 2009, pp. 118–119). Furthermore, to the extent that academic research has attempted to estimate risk-adjusted returns, such efforts have focused predominantly on already more established private markets, such as US buyouts and US venture capital. By contrast, for many other market segments (such as distressed investing, mezzanine or growth capital in emerging economies) there are virtually no data that could be used in empirical investigations.

The challenges that arise in applying the CAPM to illiquid assets, such as private equity funds and similar partnership structures, raise a fundamental issue. How do we assess risk in illiquid asset classes where market prices cannot be observed and hence are subject to substantial uncertainty?

7.1.1 Risk and uncertainty

In addressing this issue, it is useful to go back to Knight (1921), who distinguishes three different types of probability situation: (i) a priori probability; (ii) statistical probability; and (iii) estimated probability.

• A priori probabilities can be derived deductively from inherent symmetries. Examples are games of chance where outcomes have a defined universe and there is an exhaustive set of events. Instances are completely homogeneous, with equal probability for each event to materialize.
• In the case of statistical probability, instances are not homogeneous and, consequently, probabilities assigned to each event are not equal. Without a defined universe, statistical probabilities are derived from empirical classifications of instances (i.e., the tabulation of current and past data). This, however, is based implicitly on the assumption that the distribution found in the past will hold in the future.
• In the case of estimates, there is no valid basis for classifying instances. Outcomes are unique or so infrequent that it is meaningless to tabulate them as a measure of their probability.

Knight (1921) associates “risk” with the first two categories, whereas “uncertainty” is linked to the third. As far as uncertainty is concerned, strict reasoning needs to be complemented with judgment and intuition, which makes any assessment subject to a wide margin of error. In fact, even the probability of error cannot be determined as instances are more or less unique, preventing statistical techniques from being applied to calculate probabilities. Thus, dealing with uncertainty implies a “probability judgment”, in contrast to a priori and statistical probabilities. In this sense, investments in private equity and real assets are arguably subject to uncertainty rather than risk.

While the differentiation between risk and uncertainty is conceptually useful, in practice we often find a continuum from “measurable” risks to uncertainty (i.e., “immeasurable” risks). As far as illiquid assets are concerned, recent contributions to the literature may be interpreted as efforts to make risk measurable by extracting information from non-market (cash flow) data whose nature would classify such investments as uncertain in the Knightian sense. However, more research is needed to confirm that the recent academic findings provide reasonable estimates for practical applications in risk management. The crux is this, however: to the extent that a data-driven frequentist approach (Bénéplanc and Rochet, 2011, p. 29) is possible in measuring risk in illiquid assets, their potential to generate excess returns is set to diminish and eventually disappear. Until then, risk managers in illiquid assets will need to rely to a large extent on a subjective approach to deal with uncertainty in the absence of sufficient observations that can be used to derive objective probabilities.

7.1.2 How objective are probabilities anyway?

Since the seminal work by Knight (1921), the concept of risk is inextricably linked to how we understand probability (Rebonato, 2007). However, the Knightian view that probabilities can be employed objectively to estimate risk as long as such probabilities are based on a sufficiently large sample of observations has not remained undisputed. Holton (2004) refers to statisticians Leonard J. Savage and Bruno de Finetti as advocates of the subjective interpretation of probability. According to Savage, it “is unanimously agreed that statistics depends somehow on probability. But, as to what probability is and how it is connected with statistics, there has seldom been such complete disagreement and breakdown of communication since the Tower of Babel.”1

According to their objective interpretation, probabilities can be determined through the application of logic or estimated through statistical analysis. “Investors ‘just know’ the objective probabilities attached to the different possible future states of the world” (Rebonato, 2007, p. 26). This may not be entirely satisfactory, however: “If we want to describe how human beings actually make their choices in the presence of uncertainty, we had better make sure that we truly understand how probabilities are actually used in decision making – rather than how stylized hyperrational agents endowed with perfect God-given statistical information would reach these decisions” (Rebonato, 2007, p. 26; italics in original). Thus, according to the subjective interpretation, probabilities reflect human beliefs in the sense of investors' own views on uncertainty.

Nevertheless, risk managers tend to prefer quantitative analytics as the most effective approach to both measuring risk and implementing risk management programmes. Only the (apparently) objective quantification allows measuring risks accurately and comparing different asset classes in terms of their risk profile. Thus, according to this view, the ability to objectively quantify risk is a prerequisite for the credible involvement of the risk manager in the investment process. Qualitative techniques are often dismissed as “generalizations”, “hunches”, “intuition” and “gut feel” and some “war quants”, Hubbard (2009) even suggests that they are no better than having no risk management programme at all. Instead, the solution is usually sought in creating databases with detailed information on individual assets and extracting risk-relevant information from non-market data (Lorenz et al., 2006). This search for objectivity, however, may be “…an illusory attempt to materialize our true probabilistic beliefs” (de Finetti).

To be sure, objective probabilities require stationary environments. However, as we discussed in Chapter 5, the risk/return properties of individual assets may shift dramatically during periods of financial turmoil. Risk models based on data that reflected “normal” market conditions became largely obsolete, if not dangerous. Thus, the selection of the sample period, which is a subjective decision of the risk manager, has a profound impact on the estimated risk of an asset. In some sense, therefore, investors are confronted with a version of Heisenberg's “uncertainty principle”, aimed at measuring probabilities in a static market with a relatively high degree of precision or accepting imprecision when estimating probabilities in changing market conditions.

7.1.3 How useful are benchmark approximations?

In illiquid assets, even static market data are usually spurious, which raises the question as to whether certain assets can be used as an approximation. An obvious candidate for private equity funds seems to be indexes of publicly listed private equity, such as the LPX 50 index. According to Sanyal (2009), “…it (the LPX 50 index) is one of the best proxies since PE-VC funds are not regularly traded and have data limitations”. European regulators, such as EIOPA, have sympathized with this view, proposing the LPX 50 index in determining risk weights under the Solvency II standard approach. However, this view ignores that private equity funds are set up deliberately in a way to allow fund managers to harvest an illiquidity premium and hence possess fundamentally different risk profiles compared with publicly listed private equity firms or specific vehicles for which daily market prices are quoted.

Similarly, several researchers have employed the NASDAQ as an approximation for (US) venture capital and small-cap indexes, such as the Russell 2000, for US buyouts. However, there are similar reservations against such approaches. While they are easy to implement, there are important doubts as to whether they provide meaningful insights in the risk/return profile of private equity, an asset class that is usually accessed through fundamentally different vehicles. The same question can be raised, for example, with regard to the applicability of REITs to investments in real estate partnerships.

As far as private equity investments in emerging markets are concerned, historical returns data are almost entirely missing. On what basis do limited partners determine their decision as to how much capital they want to commit to partnerships in such markets? Essentially, there are three approaches, none of which is satisfactory. First, investors simply take a market-neutral position in the sense that their exposure to private equity and real assets in emerging markets resembles the relative investment volume in these markets. For instance, an investor would invest 20% of his allocation to private equity in funds targeting emerging markets, if such markets absorbed 20% of private equity commitments worldwide. While this allocation may be considered as a market equilibrium portfolio in the Black–Litterman (1992) sense, it does not say anything about the relative attractiveness of emerging markets as a destination for private equity capital. Investors may deviate from the market view, and the Black–Litterman approach presents an appropriate framework for deriving portfolio weights based on investors' proprietary risk-adjusted return expectations and the degree of confidence they have in their stated views. However, this leaves open the question of how such views are formulated.

Alternatively, investors may use public equity indexes for emerging markets, such as the MSCI EM index or national market indexes. However, in so doing they run into the same issues as we have mentioned above. Finally, investors may base their risk-adjusted return expectations on data from more mature markets, notably the USA, adjusted for a risk premium. This approach fails to recognize, however, that private investments in emerging markets and more advanced markets may follow fundamentally different dynamics. In private equity, for example, the majority of transactions in emerging economies are growth capital deals, with GPs frequently taking a minority position. Leverage is hardly used. Furthermore, the universe of fund managers tends to be significantly more diverse, with international private equity firms competing with a rapidly growing number of local GPs. In many markets, exit markets are still embryonic and country risk is perceived to play a comparatively more important role. Sudden swings in international investors' risk appetite can lead to significant fluctuations in emerging markets' exchange rates. For all these reasons, it is doubtful if data from more advanced markets can provide any meaningful guidance.

Generally, the more the benchmark resembles the specific characteristics of investments in limited partnerships, the more meaningful the results will be. However, researchers looking for close benchmarks are likely to face comparable problems in terms of data availability and their reliability. Thus, there is an important trade-off investors are confronted with. To the extent that investors decide to invest in illiquid asset classes – the dark corners of the market, as Swensen (2009) has put it – they will need to accept that measuring and managing risk will have to rely to a considerable extent on subjective assessments.

7.1.4 Subjective probabilities and emerging assets

These considerations suggest complementing the standard treatment of risk in the CAPM, which is predicated on the frequentist approach in the Knightian sense, with a subjective approach that emphasizes the risk investors accept when venturing into new markets where CAPM-type risk measures do not exist. In this sense, risk has to do with the “newness” of the investment, with investment decisions being taken in the absence of information about historical returns, distributions and correlations. As Knight (1921) observes, business decisions deal “…with situations which are far too unique, generally speaking, for any sort of statistical tabulation to have any value for guidance”.

An obvious example is venture capital, where venture capitalists back companies at a very early stage of their lifecycle. Typically, a significant percentage of companies backed by a fund will lose money, with the fund's returns driven by a small number of very successful deals, sometimes called “home runs”. As Kaplan et al. (2009) find, failures and successes are predominantly determined by a venture capitalist's ability to pick the right business (the market for a particular product or service and the technology) rather than the management team, which can – and often is – replaced in case of underperformance. However, whereas management teams usually have a track record, which is an important variable in the venture capitalist's decision process, the market potential for a new product or new technology is essentially unknown and hence fraught with substantial investment risk.

Investors in limited partnership funds face similar challenges. As we discussed in Chapter 2, there were very few investors in private equity funds as recent as the early 1980s. Their total exposure was just about USD 2 billion, which is roughly equivalent to what a single fund may raise today, adjusted for inflation. Limited partners committing to this new asset class were venturing into unknown territory, without any guidance by historical returns and their variance. Thirty years later, private equity is an established asset class, and many large institutional investors are exposed to it to varying degrees. However, within the asset class new niches emerge, such as private equity in developing economies. It is the newness of investing in private equity funds or infrastructure funds targeting such markets that is inevitably associated with risk, which is deliberately sought by investors chasing excess returns.

The ability to harvest excess returns depends on the investor's success in identifying new market opportunities and learning how to succeed in new markets. This is true for fund managers as well as the limited partners in their funds. Importantly, the decision not to pursue new investment strategies in order to avoid risk is no guarantee of stable, if unexciting, returns. As markets mature and become more competitive and transparent, the potential for achieving excess returns diminishes. In this context, some observers (e.g., Fraser-Sampson, 2006, p. 92) have pointed, for example, to the outperformance of buyout funds in the European market relative to the United States for most of the 1990s and early 2000s. A key factor, in Fraser-Sampson's view, were market imperfections in the former. Whereas GPs' “…ability to source deals proactively and be able to transact it on an exclusive basis was still very much alive in Europe …”, at that time it had already largely disappeared in the USA. However, investors backing fund managers targeting European companies had little, if any, objective information they could have used to formulate a view on the risk they would take.

Similar considerations motivate limited partners to continue to adjust their investment strategies in favour of emerging markets, notwithstanding the absence of historical return data and standard risk metrics, such as the variance and correlation of asset prices. Instead, they rely mostly on subjective considerations and qualitative indicators in assessing the risk of venturing into new markets. However, in so doing, they depart from standard risk management practices, which are generally understood to be a purely quantitative discipline consistent with regulatory frameworks, such as Basel II/III or Solvency II.

7.2 RISK MANAGEMENT, DUE DILIGENCE AND MONITORING

7.2.1 Hedging and financial vs. non-financial risks

Mirroring the particular concept of risk applied in the CAPM and in MPT, the focus of risk management is often narrowly defined. Managing financial risks is often equated with hedging risks, a view that is not easily applicable to investments in illiquid investments. Damodaran (2007) identifies several reasons why hedging is frequently viewed synonymously with managing risks. First of all, the majority of risk management products have been developed for risk hedging, including insurance instruments, derivatives or swaps. Risk hedging typically generates substantial revenues and, not surprisingly, is therefore considered as the “centrepiece of the risk management story”. Second, Damodaran (2007) argues that human nature tends to remember losses (the downside of risk) more easily than profits (the upside of risk). As a result, extreme losses, for example, caused by market meltdowns, whet investors' appetite for risk-hedging products focusing on downside risks. Third, he points to the well-known principal/agent problem. While managers tend to prefer to hedge risks, the principal may actually prefer to take a risk. We will revisit this issue in greater detail in Chapter 16, where we discuss the role of the risk manager.

Apart from the reasons that Damodaran (2007) gives why risk management is often narrowed down to hedging, one may argue that regulation is likely to have played a role as well. Financial regulation often favours risk-hedging solutions through which regulated investors may enjoy lower capital requirements. Regulatory tightening, which typically occurs in the aftermath of financial crises, provides additional incentives for risk hedging. However, to the extent that risks are non-quantifiable, their treatment becomes vague and even murky. Frequently, this is the case with non-financial risks. For example, assessing risks related to environmental, social and governance issues (ESG) relies to a considerable extent on information whose nature is essentially qualitative. This poses important challenges not least from a regulatory standpoint, explaining why non-quantifiable (sometimes referred to synonymously as non-financial) risks are often treated as an afterthought. The narrow definition of risk management is also caused by the apparent preference of auditors and regulators for tangible and precisely quantifiable information.

7.2.2 Distinguishing risk management and due diligence

Another clarification is in order. Risk management must not be confused with due diligence, although the latter is often regarded as a major risk management tool in the alternative asset industry. In fact, discussions around this subject show significant confusion regarding risk management (that builds on risk measurement), due diligence and monitoring.

Generally speaking, due diligence covers all activities associated with evaluating an individual investment proposal. Meyer and Mathonet (2005) and Talmor and Vasvari (2011) provide a detailed overview of the fund due diligence process. This process entails investigating and evaluating the investment premise of specific partnerships, aiming “…to arrive at better investment decisions by following a rigorous stepwise investigation of specific investment opportunities” (Talmor and Vasvari, 2011, p. 81). Importantly, due diligence is both quantitative and qualitative, focusing, among other things, on the proposed investment strategy the fund in question intends to follow; the organization of the management company; the specific skills and attributes of the team, and its track record; and the legal terms and conditions.

Due diligence is undertaken by the LP's deal teams. To the extent that risk management is involved in the process, this will typically happen relatively late (i.e., close to the final investment decision), for example, to provide a second opinion on the proposal put forward. However, by focusing on a concrete investment proposal, it is outside the scope of the deal teams to examine the potential impact of the investment from a portfolio standpoint. This is precisely where risk management comes in (Table 7.1). More specifically, it is the responsibility of the risk manager to evaluate the expected risk-adjusted return profile of the portfolio and the extent to which this profile is likely to be affected by new investments. Unlike due diligence, which is concerned with a concrete investment proposal, risk management is an ongoing process, covering all aspects of financial risks, including capital risk and liquidity/funding risk (see Chapter 8). As far as co-investment and secondary investment decisions are concerned, risk management typically also focuses on the potential impact of the investment opportunity on the sectoral concentration and foreign exchange exposure of the portfolio.

Table 7.1 Main conceptual differences

Investment decisions reflect the set of information that was available at the time when the due diligence and risk management work was undertaken. However, even the most thorough due diligence process cannot fully eliminate investment uncertainty, especially taking into account the long life of limited partnerships spanning 10 years or more. Much can happen during this period, which is difficult – if not impossible – to anticipate. A significant number of partnerships in private equity and real assets in each vintage year fail to meet investors' return expectations and a non-trivial share of them even fail to return the invested capital (Chapter 10). It is unlikely that such funds include only LPs whose due diligence was sloppy in the first place.

As important as due diligence is, important factors may be overlooked or misjudged, and the fund's original characteristics may change over time, for instance due to “style shift”. This necessitates continuous monitoring, which could be seen as “in between” risk management and due diligence, with shared responsibilities between the LP's risk management and its investment managers.

Monitoring is part of a control system for the investment as well as the risk management process (see Figure 7.1). Monitoring of a fund should not be confused with the management of portfolio companies or projects, an activity which is exclusively the responsibility of the GP. By contrast, LPs manage their portfolio of funds and monitor the fund managers (Meyer and Mathonet, 2005). This entails identifying performance-relevant issues by engaging with the GP and other LPs in a partnership, for instance through annual meetings or advisory boards. From the perspective of LPs, monitoring may be described as “ongoing due diligence” to gather information that can be used in the decision process as to whether to invest in one of the GP's follow-on funds. This information also provides important input for risk measurement purposes. Risk managers primarily monitor the development of the entire portfolio of funds and coordinate possible corrective actions with the investment managers.

Figure 7.1 Monitoring as part of control system.

7.3 CONCLUSIONS

This chapter has started from one of the most basic observations in finance, namely that higher returns are inevitably associated with higher risk. Even if the efficient market hypothesis may not always hold, it will be difficult for investors in traditional markets to find risk-adjusted excess returns. This experience has motivated an increasing number of investors to venture into alternative asset markets, which are less transparent and efficient and hence provide a greater potential for generating excess returns.

However, this immediately raises the question as to how returns and risk in these markets are measured. As far as investments in illiquid limited partnership funds are concerned, market prices cannot be observed, and in the absence of market prices standard risk measures cannot easily be calculated. Thus, investors face a dilemma. While excess returns in traditional markets hardly exist, the calculation of risk-adjusted returns in non-traditional, illiquid markets is subject to substantial challenges. As a result, traditional risk management and portfolio construction techniques, which are based on the frequentist approach in the Knightian sense, are difficult to apply.

What to do? Academic research has focused on developing methods to extract information from non-observable valuation data that can be used in the traditional CAPM framework. However, such studies have remained rare, and their results suggest that there remains considerable uncertainty about the riskiness of illiquid investments. Furthermore, constrained by the availability of data, much of the available research has focused on market segments that already have a relatively long history, such as US buyouts and US venture. By contrast, there is very little, if any, research on the risk/return profile of investments in partnerships targeting emerging markets, distressed assets or mezzanine. Inevitably, therefore, investors who venture into the dark corners of financial markets have to accept that their investment decisions will need to rely to a significant degree on their subjective qualitative risk assessments. This raises important questions not only from the perspective of the investor, for example, with regard to the treatment of illiquid investments within broader asset allocation and risk modelling. It also raises important issues from a regulatory viewpoint, with financial regulation being embedded typically in quantitative risk models.

In this context, finally, we clarified that risk management should not be confused with due diligence, with the latter focusing on individual investment opportunities rather than examining the entire portfolio, which falls under the auspices of risk management. While the two functions are separate, they complement each other, including during the ongoing monitoring process.

1 Quoted from Savage, L.J. (1954) The Foundations of Statistics. John Wiley & Sons, New York.