I also add this article from the Financial Times, which cites a study that estimates that actual annualized return for private equity in the UK as far lower than the returns PE firms provide and even far less than the stock market:

Private equity returns are not all they seem

Here is what the data from PE firms suggest:

quote:
Take for instance the British Private Equity and Venture Capital Association’s 2017 performance measurement survey. This revealed that over the previous decade, UK private equity generated returns of 11 per cent a year, far outpacing the 6.3 per cent on the FTSE All-Share index. Over five years, the figure was a no less impressive 17.8 per cent as against 10.3 per cent for quoted stocks.

But here is the estimates from the study (using US firms as well as European firms):

quote:
Interestingly, PME figures are less flattering for private equity. A large study conducted in 2015 by three academics looked at nearly 800 US buyout funds between 1984 and 2014. They found that before 2006, these funds delivered an excess return of about 3 per cent per annum, net of fees, relative to the S&P 500 index. In subsequent years though, returns have been about the same as on the stock markets. A study of 300 European funds produced similar results.

And here is the actual study (a working paper):

How Do Private Equity Investments Perform Compared to Public Equity?

More specifically, it seems that PE used to generate alpha (3% annually) in the 1980s through 2000s:

quote:
Our estimates imply that
each dollar invested in the average buyout fund returned at least 20% more than a dollar invested
in the S&P 500. This works out to an outperformance of at least 3% per year.

But since 2005, PEs have performed similar to the S&P 500, so not really generated alpha, but with the liquidity risk that should require a premium:

quote:
k. For the more recent and less fully realized post-2005 vintage funds, however,
performance has been roughly equal to public markets.

They also discuss the potential for selection bias in the available PE data, but they think it's unlikely since their data sources are consistent with other sources. But I'm not convinced since the same issue happens with all sorts of data sources, including meta-analyses for academic studies and the file-drawer problem (i.e., only significant results are usually published), and given the lack of transparency of PE data and the incentives to publish favorable data, my guess is there is a clear risk that there is some survivorship bias and/or those that severely underperform are more likely to downplay, if not, outright omit it.

So their data source may not have selection bias, compared to the available data, but the available data may be biased. In other words, I think the results are best-case scenario for PE, especially since given those incentives, even the published data is probably as rosy of a picture as they could present.