Evaluation of risk based portfolios : if we lever

Abstract

This thesis evaluates risk-based techniques by constructing five risk parity portfolios, Inverse Volatility, Equal Risk Contribution, Maximum Diversification, Diversified Risk Parity, and Factor Risk Parity. The empirical study focuses on the consequence after putting leverage on each strategy. Most of the risk-based portfolios are shown to be effective in improving portfolio performance over the Equally Weighted portfolio. However, monthly rebalancing and associated transaction costs can have a substantial drag on absolute returns.

This thesis also uses risk measurement methodologies, Value-at-Risk(VaR), Expected Shortfall, Drawdown, Principal Component, to examine the risk characters of risk-based strategies. Most of the risk parity strategies have lower VaR, Expected Shortfall and Drawdown than our benchmark, Equally Weighted Portfolio. The reallocation of risk contribution by assets and Principal Component factors are higher for Factor Risk Parity and Diversified Risk Parity.

However, this thesis finds that it is not possible to impose a positivity constraint Factor Risk Parity portfolio that matches the risk budget completely in this case. The existence of the Factor Risk Parity portfolio is not guaranteed when we have long-only constraints. The need to go short/long accelerates in times of crises, such as dot-com bubble and the financial crisis happened in 2008. Therefore, the flaw of optimal Factor Risk Parity method is that it can result in extreme weights.