Alphas in Disguise: A New Approach to Uncovering Them

Abstract

Fama-French (Carhart) alphas of passive indices should be zero, but recent evidence shows otherwise. Inaccuracies of factors in the performance measurement models have been put forward as the main reason for this. Some computationally intensive solutions to factor adjustment have been proposed, but are not applicable to all benchmark indices. We propose an optimisation algorithm that makes minor adjustments to the market, size, value and momentum factors to obtain zero alphas for any benchmark index. In the sample of 1281 active and 102 passive US equity mutual funds benchmarking against S&P500, our adjustment leads to augmentation of fund performance upwards in periods of index underperformance and downwards in periods of index outperformance. Overall, the adjusted alphas of both groups of funds are significantly negative, signalling poor performance. This is particularly pronounced for tracker funds, whose managers have not been successful in enhancing returns adequately to make-up for the costs involved in any of the sub-periods examined. JEL classification: G10, G11,C6

Review of new trends in the literature on factor models and mutual fund performance

“…In the UK, the FCA is making a step in the right direction in this area, where more disclosure from funds regarding benchmarks and objectives will be required. That will help re-assure investors that the prospectus benchmark is, indeed, the most appropriate one, prompting a more widespread use of benchmark-adjusted models such as Angelidis et al (2013) and Chinthalapati et al (2017) in performance evaluation. Alternatively, with the data sources getting more comprehensive and advances in computer technology, building characteristics-based benchmarks that change over time to capture style drifts in a fund is another way forward for more accurate benchmark-adjusted performance.…”

“…Similarly, Chinthalapati, Mateus and Todorovic (2017) tackle the issue of non-zero alphas in passive benchmark indices by proposing an optimisation algorithm 15 that calculates minor fixed adjustments that should be added to the time series of the Carhart's four factors. Adjusting the factors ensures a zero alpha for any chosen selfdesignated benchmark index of a mutual fund, without making any other change in the model parameters.…”

“…It is important to note here that a change in the value of alpha does not necessarily change the ranking of the fund within a group -an issue not discussed in the aforementioned studies -but of essence to investors trying to select the best fund within a group (say, the best small-cap fund). If the entire peer-group of funds benchmarks against the same index, then, adjusting fund alphas for benchmark alphas as in Angelidis et al (2013) or Chinthalapati et al (2017) will change the value of fund alpha (alphas for all funds decrease if a benchmark is outperforming and vice versa); but not the fund's ranking position in the peer-group. However, if such a benchmark does not match objectives of a fund, its peer group ranking is biased, simply because a fund may strategically choose a benchmark that is easy to beat.…”

“…The studies in this section differ from the ones in Section 3 as their purpose is to contribute to performance measurement, not to asset pricing. All the benchmarkadjusted and peer-group adjusted models assume that the most unbiased fund performance can be estimated based on the widely accepted standard Fama-French-Carhart risk factors if fund performance is accounted for the funds' self-declared benchmark (Angelidis et al, 2013, Chinthalapati et al, 2017 or peer-group benchmark (Hunter et al, 2014) as it eliminates the issues of arbitrary factor construction and a potentially mismatched reference benchmark. These new models are of use to investors aiming to identify funds relative performance, i.e.…”

“…Chinthalapati, Mateus and Todorovic (2017) Authors use an optimisation algorithm 20 to derive fixed (time invariant) factor adjustments (i =1,2,3,4), and adjusted factors , so that:…”