The Bank for International Settlements in Basel, Switzerland is likely abuzz this week watching history unfold before its eyes. After all, one of the lynch pins of the organization’s Basel II Accord was the requirement for banks to mark-to-market all assets – including less liquid ones. And it appears that doing so in a leveraged environment has put several banks into a death spiral in recent weeks (see featured post below).
But the BIS is also keeping an eye on hedge fund leverage. The organization just released a working paper called “Estimating Hedge Fund Leverage” that proposes a new method of calculating the level of leverage used by hedge funds and, it is hoped, a way to measure any resulting systemic risks to the financial system. Regular readers may remember that this topic was also covered by the Fed’s Tobias Adrian last year (see previous story).
As the authors of this report point out, leverage comes in two basic forms: funding leverage – where you literally borrow money to goose returns (or losses) and instrument leverage – where the securities themselves have leverage baked in (such as a futures contract, option or swap). But at the end of the day, if a fund rises twice as much as the market on “up” days and falls twice as much on “down” days, then the source of leverage is less relevant. In fact, divining leverage based on historical returns will also capture the leverage implicit in the balance sheets or business models of individual securities.
The study uses a regression technique developed by William Sharpe back in the 1990s and recently used by people like professors Bill Fung and David Hsieh to “replicate” hedge fund performance. The idea is simple. Just add up the various beta coefficients in a regression of the hedge funds returns against a series of factors.
The authors regress hedge funds from the HFR database against a variety of linear and non-linear variables below:
When these coefficients were added up, here are the “leverage” numbers for the funds of funds (top) and equity hedge fund (bottom) indices – the broadest indices analyzed by this paper:
In order to check their results against funds’ stated use of leverage, the authors use a thick line to show funds that told HFR they DO use leverage and a thin line to show funds that said they DID NOT use leverage. As you can see, the leverage implied by historical performance regressions of both strategies’ returns is close to 1.0 (i.e. no leverage) and fell precipitously in 2007.
Interestingly, fund of funds (the top chart) that reported using leverage had returns-implied leverage that was only slightly more than those who reported NOT using leverage. Also, leverage-happy equity hedge funds delivered returns that were in keeping with the use of leverage from 2004 to 2007, but less so in the previous years.
While this method of inferring leverage has the benefit of requiring only publicly-available information, the authors of the report acknowledge that it still needs some work:
“Overall, the results of this paper suggest that extracting reasonable estimates of hedge funds’ use of leverage from publicly available data is not a straightforward exercise, leaving room for future improvements in the empirical technique. The complexity of hedge funds’ actual positions, and the difficulty in tracking their exposures, significantly complicates any empirical application…”
Maybe something to consider down the road for, say, Basel XXIV.