Nicolas P.B. Bollen

Owen Graduate School of Management
401 21st Avenue South
Nashville, TN 37203
office: 615.343.5029
fax: 615.343.7177

Research Story 3: Hedge fund misreporting: Conditional smoothing

In 2002 I was preparing a lecture on hedge funds for my course Equities Markets. I had come across the following graph in an advertising brochure from Standard and Poors. It shows the pro-forma performance of their investable hedge fund index (S&P HFI) product relative to the S&P 500 Index and the Lehman Aggregate Bond Index:

I stared at this with utter amazement and admiration. How could these guys do it! The value of the index marched steadily upwards. High returns, such little variation, so smooth...Eventually it dawned on me that maybe there was something amiss. Perhaps the reported returns did not reflect all the ups and downs of the investments within the funds. For example, if the fund is invested in illiquid securities, perhaps the manager reports a moving average of the true economic returns. From my recollections of a PhD course I took, I had a hunch that this would lower the variance and leave the average growth rate of the fund unaffected.

Unfortunately, I quickly found a paper by Andrew Lo and his students Mila Getmansky and Igor Makarov at MIT on the same subject. I thought about this for a while, and then decided that simple smoothing was not quite how I envisioned a manager acting. On the upside, when a manager delivers high returns, he or she has a big incentive to fully report. Their compensation would be higher, and they would look better relative to their peers. However, on the downside, when the returns are low, you expect the smoothing to occur. The experience at National Australia Bank's (NAB) currency options trading desk illustrates this pattern perfectly. Stephen Brown, editor of the Journal of Financial and Quantitative Analysis alerted me to this example. In January 2004, NAB announced cumulative losses totaling AUD 360 million generated by the currency options trading desk. The traders had discovered a way of entering false transactions into NABís computerized record-keeping system in order to conceal losses, thereby meeting profit targets to which the tradersí compensation was tied. A detailed consulting report prepared by PricewaterhouseCoopers, Investigation into Foreign Exchange Losses at the National Australia Bank chronicles the sequence of the currency deskís realized monthly profits and losses, and the profits and losses reported to NABís record-keeping system, as reproduced here:

According to the report, the currency desk began October 2002 with a cumulative overstatement of its portfolio of AUD 7,972,000. In October, the desk earned AUD 8,946,000 but reported only AUD 974,000 in order to eliminate the existing cumulative overstatement. The desk underreported income in three other months (March, July, and October 2003) in order to reduce existing cumulative overstatements. In November and December 2002, the desk fully reported its profits. In the remaining nine months, the desk overreported profits. In five of these months, the desk reported positive profits when in fact the desk realized losses. In summary, the currency traders often underreported losses but only underreported profits when there was an existing overstatement of the deskís portfolio.

With my then-PhD student Veronika Krepely, I worked on a model of asymmetric or conditional smoothing. This resulted in our forthcoming Journal of Financial and Quantitative Analysis paper paper Conditional Return Smoothing in the Hedge Fund Industry.