On the integration of multiple criteria decision aiding and forecasting: Does it create value in portfolio selection?

A systemic integration of multiple criteria decision aiding (MCDA) and forecasting is presented for enhancing the quality of investment decisions. First, we provide a new methodological approach for combining forward-looking information, as an input to further MCDA analysis and we illustrate that the informational content of the MCDA scores is distinct from that of the original data used as input to the MCDA. Moreover, we provide very compelling evidence that the produced MCDA scores and rankings have incremental information, over and above that being contained in either the original data or the forecasts themselves. Therefore, it can be used successfully in providing methods which will have a high, a priori, chance of resulting in financial performance enhancements; to this end, we have illustrated this finding with a number of different approaches, including asset rotation and portfolio construction. Next, we show that the MCDA ranked variables can be used for both in-sample explaining and out-of-sample forecasting of critical economic variables, such as the real GDP growth or the unemployment rate, so that MCDA might be able to assist in identifying leading indicator properties for economic forecasting. The evaluation of the suggested approach entails extensive empirical testing on the various US economy market sectors, as expressed by the corresponding S&P 500 Exchange Traded Funds (ETFs), for a period of 22 years, documenting specific economic stylized facts, reported for first time in the relevant literature.