Monday, April 30, 2018

"Are Private Equity Returns Too Good to Be True?"

From Columbia Law School's Blue Sky blog:
Investments in private equity are typically structured as 10 year limited partnerships in which fund managers act as general partners (GPs) and investors act as limited partners (LPs). Since the fund life is broken down into an investment and a liquidation period, GPs can only make new investments after the investment period has expired by raising a new fund. At that time, existing investments are not necessarily liquidated, so that current fund returns rely heavily on subjective performance estimates of their investments. This fact, stemming from a market setting of information asymmetries, has led many investors, industry observers, and academics to speculate that private equity firms distort their performance measurement around fundraising events. While this has been widely alleged, data limitations have made it difficult to draw sharp conclusions about this concern.
Various prior studies have focused on agency problems around fundraising, centering on the question of whether agents inflate portfolio values prior to fundraising. The common finding of these studies is that fundraising for a new fund coincides with times of high current interim valuations, especially for low-reputation funds, where cost of manipulation appears low (e.g., Barber and Yasuda 2017, Chakraborty and Ewens 2017 and Brown et al. 2017). This finding is open to two different interpretations. While GPs may advertise strong current fund performance by manipulating true estimates of current asset values as suggested by prior work, alternatively, GPs may time fundraising around true estimates of high current net asset values (NAVs).

Motivated by prior research, I argue that the data underlying these studies are aggregated too coarsely to attribute performance peaks to inflation of underlying asset values. To overcome data limitations of prior work, I have built a novel database of quarterly deal valuations in U.S. buyout funds that allows me to address the two hypotheses above.

What do we learn from deal level valuations about interim fund returns?
For a better understanding of why it is important to have data on individual portfolio companies, consider a fund with heterogeneous investment times. A peak of net asset values on the fund level, for instance, could be explained by inflated interim valuations of individual portfolio holdings. Alternatively, if deals made shortly before fundraising perform poorly so that their valuations after fundraising fall below their initial valuations at cost, these deals could reduce portfolio valuations after fundraising. If, in addition, deals made long before fundraising perform well, so that their initial valuations at cost increase around fundraising, the result would be a peak in aggregated values on the fund level just before a new fund is raised.

Indeed, I find strong evidence that funds with an especially low reputation have more successful investments well ahead of fundraising compared with deals made shortly before fundraising. Realized (or last observed) value multiples (VMs) of investments made long before the fundraising event exceed multiples of investments made shortly before the event, on average, by 130 percentage points. The difference stems from VMs of investments by low-reputation funds that were unrealized at the time of raising a new fund. I find no signs of systematically higher interim deal valuations.
Thus, the conjecture about manipulation is inconsistent with the deal-level evidence on fundraising. The finding of strong early deals increasing in value well ahead of fundraising, and weak later investments declining in value primarily after fundraising, is, instead, consistent with a timing of fundraising story.

Is fundraising timed to true performance estimates?
To test whether GPs are more likely to attempt to raise a new fund following estimates of truly good performance, I extend the results of Barber and Yasuda (2017). Their results suggest that NAV inflation predicts fundraising, and this effect is concentrated among those who most need to rely on interim NAVs as substitutes for reputation. By using hazard rate models, I find that fundraising is endogenously timed to true estimates of high current valuations, while their performance, especially in low-reputation funds, cannot be sustained in later deals. Consistent with this evidence, GPs appear to verify strong private performance with investments in publicly traded stocks (PIPEs or investments on the public stock exchange that are not taken private). I find that, on average, 10 percent of a private equity fund’s portfolio is not private equity but rather represents investments in publicly traded stocks that are not intended to be taken private. High positive returns in these public deals have a significant impact on the probability of fundraising. Backing NAVs of private portfolio companies with the verifiable performance of investments in publicly traded stocks appears to create less agency frictions than quickening of exits and “grandstanding” (Gompers 1996)....MORE