Non-discretionary FX strategies: From the portfolio perspective of a Danish institutional investor

Kim Nydahl Grønlund & Rune Linneboe Pedersen

Student thesis: Master thesis


Some financial institutions have begun marketing FX indices to institutional investors based on non-discretionary trading strategies that are unaffected by the biases of individual traders (Hafeez and Folkerts-Landau, 2006; Merril Lynch, 2007). They argue strong performance from historical back-testing and one of the institutions even name the title of their paper, “Currencies: Pensions Saviour” (Hafeez and Folkerts-Landau, 2006). The ulterior motives of these institutions, however, give rise to scepticism regarding these results, in particular in regards to data mining issues and selection biases. In this paper we investigate whether investing in non-discretionary FX-strategies would in fact improve the portfolio performance of Danish institutional investors. To overcome the potential problems with the existing FX indices, the authors create an index of FX returns, based on three non-discretionary investment strategies with the roots in a review of the literature on the returns in the foreign exchange markets: 1) A PPP strategy, based on the empirical evidence that exchange rates indeed revert towards a PPP equilibrium. 2) A Carry strategy, based on the empirical evidence that uncovered interest parity does not hold. 3) A Technical strategy, based on the empirical evidence that exchange rates trend and that past exchange rate changes are useful in predicting future exchange rate changes. The strategies are conducted for the G10 currencies for reasons of liquidity and economic stability. The currency positions are readjusted monthly or quarterly to be able to respond to changing market conditions. The analysis is conducted from January 1980 to March 2008. For the portfolio analysis the primary asset classes for the Danish institutional investors were defined as 1) a Danish bond portfolio 2) a Danish stock portfolio 3) a global stock portfolio and 4) a global bond portfolio. The portfolio analysis was conducted in a modern portfolio theory framework. The initial results from the main analysis were very positive. The FX index provided superior risk adjusted returns to and exhibited very little correlation with the returns of the traditional asset classes considered by Danish institutional investors. Also the FX index as well as its underlying strategies generated positive and significant Jensen‟s Alpha (at the 1% level for the FX index). The FX index should have played a significant part of the optimal portfolio (over 27%) even after constraining the analysis to better reflect the investment pattern of the Danish institutional investor. Conducting the sensitivity analysis revealed a different picture. Generally positive was the findings that 1) portfolio weights assigned to the FX-index were largely stable over time 2) the favorable correlations of the FX index did not change in bull and bear stock markets. Applying estimated transaction costs of 0,16% per round trip to the analysis, changed the positive results achieved earlier considerably. The risk adjusted return of the FX index was now no longer superior to the indices of a Danish bond and a Danish stock portfolio. The Jensen‟s Alpha generated by the individual FX strategies was no longer significant. The Jensen‟s Alpha of the FX index was, however, still significant at the 10% level. In spite of the above the FX index still managed to hold its own in portfolio optimization. After incorporating the authors estimate of transaction costs the FX index still should have constituted 14,0-17,1% of the optimal portfolio more than the holding in both Danish stocks and global stocks. With the transaction cost estimates of other authors, 0,1% (Lebaron,1999) and 0,05% (Cheung and Osler, 1999), the weights to assign to FX would have been between the 19,2% and 25,2%. The authors attribute the positive result from portfolio optimization after incorporating transaction costs to the key strength of investments in non-discretionary FX strategies: Risk diversification. While the FX index was able to hold its own in portfolio optimization even after incorporating transaction costs, the sensitivity of the results to changes in transaction costs are considerable. For this reason transaction costs are the main threat to the performance of non-discretionary FX strategies. Another caveat of the returns from non-discretionary FX strategies is that they do not appear to be normally distributed. The returns from the FX index and the underlying strategies exhibited high excess kurtosis, and for the carry strategy, negatively skewed return Based on the analysis we conclude that investing in non-discretionary FX strategies would improve the portfolio performance of Danish institutional investors. While risk-adjusted may not be necessarily superior, the diversification benefits appear substantial, also in times of bear stock markets. We find that a portfolio holding of approximately 15% in a set of non-discretionary FX strategies, potentially more depending on transaction cost assumptions, would optimize the portfolio performance of Danish institutional investors. Careful analysis, however, of the transaction costs and the impact of non-normally distributed risk should be conducted before incorporating a set of non-discretionary FX strategies into the portfolio.

EducationsMSc in Applied Economics and Finance, (Graduate Programme) Final Thesis
Publication date2008
Number of pages106