Risk Premium Principal Components in factor risk parity portfolios

Abstract

Factors, and risk are two words which have gained popularity among academics and practitioners alike recently. The popularity of factors is shown in their many applications in pricing and risk-based asset allocation. A recently developed method for latent factor extraction called Risk Premium Principal Component Analysis (RP-PCA), has been shown to outperform the standard PCA in terms of estimating latent factors in the Arbitrage Pricing Theory factor model. I take the idea of better estimation of APT factor model by RP-PCA and test empirically whether RP-PC factor risk parity portfolios outperform those of PCs and Independent Components (ICs) which have been previously used in factor risk parity strategies. In this thesis I find some support that RP-PC factor risk parity portfolios do perform better in terms of risk, but results are unconvincing and further research should be allocated to the topic.