We have compared the realized performance of our quantitative emerging markets equities (QEME) strategies with the hypothetical performance of recently launched generic factor indices. Although our QEME strategies exhibit statistically significant exposures to these generic factor indices, their outperformance remains largely unexplained. We argue that this can be explained by the fact that our strategies are designed to benefit from factor premiums in a much more efficient way than factor indices.
Robeco was one of the first to show that well-known stock selection variables for developed markets, such as value and momentum, are also highly effective in emerging markets. Robeco’s QEME strategies are designed to consistently outperform the capitalization-weighted emerging markets index, using a proprietary quantitative stock selection model and a proprietary portfolio optimization algorithm. The stock selection model ranks each stock in the investable universe on a combination of various valuation and sentiment factors, as illustrated in Figure 1. The valuation theme prefers stocks with a low price compared with their fundamentals, such as book value. The sentiment theme consists of both price momentum variables, which measure investor sentiment, and analyst revisions, which measure analyst sentiment.
In recent years, index providers such as MSCI, FTSE, Russell and S&P have introduced various new indices that are also designed to benefit from established factor premiums. Providers of passive investment solutions, such as index funds and ETFs, may use these indices to offer cheap and transparent exposure to factor premiums. This raises the question how the performance of our actively managed QEME strategies compares with such potential passive alternatives. In this note we empirically investigate this issue, focusing on the question if, and to which extent, the performance of our QEME strategies can be explained by generic factor indices.
We first compare the return statistics of the Robeco Quant Emerging Markets Equities products with those of the MSCI Value Weighted and Momentum Tilt indices over the live periods of the Robeco funds. The indices ignore transaction costs, which can be quite significant in emerging markets, especially for high-turnover strategies such as momentum. The realized returns of the Robeco strategies are, of course, after trading costs.
The results in Table 1 show that the performance of the factor indices was relatively weak over this more recent period. Specifically, the outperformance of the MSCI Value Weighted and Momentum Tilt indices amounted to less than 1% per annum, so well below their long-term average returns. If we compare our Core QEME strategy with a simple 50/50 combination of the two indices we observe that the outperformance is more than three times higher for the Core QEME strategy, with only a slightly higher tracking error. The Active QEME has 2.5 times the tracking error, but over five times the outperformance of the simple combination, thereby more than doubling the information ratio. The finding that the Robeco QEME funds were able to deliver a solid performance during a period of relatively weak performance for these generic strategies confirms that our more sophisticated approach paid off.
We next conduct a regression analysis in order to examine to which extent the outperformance of our QEME strategies can be explained by generic factor indices. Specifically, we regress the monthly benchmark-relative returns of our funds on the monthly benchmark-relative returns of the two MSCI factor indices discussed above. The results of this analysis are reported in Table 2.
The regression results show that although the Robeco QEME funds exhibit statistically significant exposures to the generic value and momentum indices (high T-statistics), most of their outperformance remains unexplained. Specifically, after adjusting for exposures to the MSCI value and momentum indices, the outperformance (alpha) of the Core fund becomes 1.85% (versus 2.14% raw), while the outperformance of the Active fund becomes 3.02% (versus 3.34% raw). It is safe to conclude that the generic factor indices are not a good substitute for these Robeco funds.
Robeco’s added value over the generic factor indices comes from various aspects, of which we will give several examples. One explanation is that we do not only consider generic value and momentum factors, but also more sophisticated (and more effective) factor definitions. A second element is that we apply our integrated risk management philosophy, meaning that we eliminate unrewarded risks in the definition of our factors. Yet another explanation is that it is more effective to integrate value and momentum factors in one model instead of running separate value and momentum strategies, as with a 50/50 combination of two MSCI factor indices. The reason for this is that generic value and momentum strategies tend to partly offset each other, as a generic value strategy tends to prefer stocks with poor momentum, while a generic momentum strategy tends to prefer stocks with unattractive valuations. A final source of added value is that our portfolio construction process is much more sophisticated than the straightforward rules-based weights used by indices, e.g. in controlling risk.
Note in addition that the MSCI indices ignore costs of trading and associated risks (such as counterparty risks). The Robeco strategies already have a proven track record with regard to issues such as dealing with multiple listings, ADRs and GDRs versus local listings, corporate actions and data quality.