We first examine the target market for hedge funds by using the HFR family of indexes, now widely used as standard benchmarks. These returns are net of all fees and are constructed live since 1994. The broad index (fund composite) is subdivided into four strategies: event-driven, equity hedge, macro, and relative value. Table 2 provides summary statistics for these five Hedge Fund Research Index (HFRI) series for the 2010–20 period. During this period, as Table 2 shows, hedge funds experienced, in aggregate, an average arithmetic return of 4.7%, with a volatility of 6.2%, both annualized from monthly data. Subtracting the one-month US T-bill rate translates into an excess return of 4.2%. This return is not all alpha, however, because all HFRI is directional, as indicated by the high S&P correlations. Even so, the question of interest is whether alternative risk premia can replicate some of these hedge fund returns. Next, Table 3 reports for the 2010–20 period excess returns on three major market factors—the S&P 500, the Barclays–Bloomberg (BB) Treasury Bond Index, and the BB High-Yield Bond Index. These series are essentially buy-and-hold long-only indexes that can be implemented in practice at a low cost. Table 3 also shows the performance of the Fung–Hsieh trend-following factors for the same period. Fung and Hsieh (2004) developed the classic approach to hedge fund replication. Their seven-factor model includes the same three market factors plus a “small-cap” size effect and three trend-following returns—on bond, currency, and commodity futures—later extended to short-term interest rates and equity indexes.15 The returns on these five trend-following strategies do not include transaction costs and are essentially unfunded—that is, represent excess returns. As Table 3 shows, these trend factors are extremely volatile, reflecting their embedded leverage.16 For comparison purposes, excess returns have been rescaled to a 10% volatility. Most trend factors did poorly over this period, calling into question their usefulness as sources of positive risk premia.
Table 4 describes the performance of the HFR bank risk premia indexes—19 indexes covering various asset classes and styles, with continuous data for the 2010–20 period. Volatility numbers in Table 4 are generally high and widely different across series, which reflects the inverse volatility weighting process used in the construction of the indexes. As a result, I focus again on the average excess return for a 10% target volatility, as well as one-factor alphas. Consider, first, the equity strategies’ performance in Panel A of Table 4. On the scaled measure, average returns are positive for the equity value and multi-style strategies. The scaled return for equity value is 6.9%, which is significant. Some of this return is attributable to directional SPX risk, but the remaining alpha is still positive. Second, consider the rates category. The overall rates carry rates, momentum rates, and volatility rates risk premia all have positive scaled returns, ranging from 3% to 6%. This performance is unlike the FH bond and STIR trend factors in Table 3, which are both negative. Note that the overall, carry, and momentum factors have high positive correlations with Treasury bonds. This result is as expected because rates carry strategies tend to invest in longer-term maturities as a result of generally positively sloped yield curves, and so they are long duration (see Martens, Beekhuizen, Duyvesteyn, and Zomerdijk 2019).
Alternative risk premia products are the next logical step in the progression toward automation of active management. Indeed, such trading algorithms aim to capture compensated risk factors, or risk premia, that may be embedded in returns provided by hedge funds. The development of this new market, now estimated at around $360 billion, is a response to investors’ search for lower fees. This article has summarized the performance of the ARP market as provided by banks’ total return swaps. In this approach, investors outsource both the execution and design of the trading strategies to broker/ dealers. Unlike conventional academic risk premia, however, these products are fully investable because they account for all transaction costs. As a result, these bank products provide realistic estimates of returns on risk premia strategies. Using indexes provided by HFR for the 2010–20 period, this article reports several new results. Even over this short period, I found that several risk premia provide significantly positive returns, especially within equities, rates, and credit. Currency products, in contrast, have not performed well. The performance of commodity products has been mixed. Admittedly, much like hedge funds, many of these bank products have directional exposure to equities. Armed with the information about the performance of these products, I turned to the question of whether hedge fund performance can be replicated by traditional market factors (equities, rates, and credit) augmented by bank risk premia. I found a significant increase in explanatory power from the BRP products. Interestingly, this model dominates the traditional workhorse of hedge fund replication, the famous Fung–Hsieh (2004) seven-factor model. So, this new framework can provide improved replication of hedge fund index returns. This improvement is especially the case for quantitative hedge funds, which are more likely to follow systematic trading rules than other funds, but less the case for security-focused hedge funds. Finally, I found that any excess return provided by these hedge fund indexes is largely eaten away after accounting for market factors and a selection of bank risk premia. These results should help inform investments in alternative risk premia and hedge funds. In particular, the BRP products can help guide the manager search process by splitting the track record into contributions from ARP exposures and alphas. A high R2 with low alpha implies that investors should allocate to ARP. In contrast, a low R2 and high alpha is an indication of manager skill.