Our knowledge concerning equity factor investing has increased immensely over the last few years which has a great impact on how factor portfolios can be constructed. Clearly, we are much more aware of the time varying properties of these factors, but more importantly we have a much better idea of the macro factors and regimes that impact these risk premia. This work tells us that unconditional portfolio construction is risky. The correlations between equity factors change. The diversified portfolio over the long run or at a given point in time may not be the risk faced tomorrow or in a different macro environment.
This thesis on the macro risks associated with equity factors has been extensively researched by the folks from Scientific Beta and presented in the Journal of Portfolio Management paper, “Macroeconomic risks in Equity Factor Investing”.
The changing correlation relationships between equity factors is a sizable problem. While the correlation across equity factors seem low, there is a large variation between the correlations during good times and bad times. Not accounting for the changing correlation across macro environments creates portfolios with higher potential risk.
The idea of equal-weighted or equal risk weighted is not wrong, but we actually have more information on factor behaviors such that a portfolio that accounts for macro factors can be easily structured and can improve performance. The equal-weighted portfolio makes sense if investors are dealing with stable correlations across factors or there is no useful information on the time varying behavior of factors. Investors may not have skill to dynamically adjust macro risk, but they should be fully aware of these risks and account for macro surprises with portfolio construction.
Equity factors are sensitive to both bull and bear market conditions. Additionally, these factors are sensitive to other macro variables beyond the market environment. Investors should be able to diversify or tilt exposures based on knowledge of these macro sensitivities.
The table below shows the sensitivities of factor premia, size, value, momentum, low risk, profitability, and investments, to a set of seven different macro state variables. Some factors which have low pairwise correlation actually have similar sensitivities to a given state variable. For example, low risk and profitability are both affected similarly to shocks in dividend yield.
The idea of equal-weighted or equal risk weighted is not wrong, but we actually have more information on factor behaviors such that a portfolio that accounts for macro factors can be easily structured and can improve performance. The equal-weighted portfolio makes sense if investors are dealing with stable correlations across factors or there is no useful information on the time varying behavior of factors. Investors may not have skill to dynamically adjust macro risk, but they should be fully aware of these risks and account for macro surprises with portfolio construction.
Equity factors are sensitive to both bull and bear market conditions. Additionally, these factors are sensitive to other macro variables beyond the market environment. Investors should be able to diversify or tilt exposures based on knowledge of these macro sensitivities.
The table below shows the sensitivities of factor premia, size, value, momentum, low risk, profitability, and investments, to a set of seven different macro state variables. Some factors which have low pairwise correlation actually have similar sensitivities to a given state variable. For example, low risk and profitability are both affected similarly to shocks in dividend yield.
The research looks at the sensitivity of the premia versus different regime models and finds some surprising results. For example, the low risk factor premia will have some of the widest return spreads across regime indicators. Low risk is not the same as stable returns.
The impact of macro factors on diversification is significant. In the example below, two factor premia that seem uncorrelated can have a strong risk regime dependency that impacts risk. In this case, high profitability and momentum have a similar dependency during risk tolerance regimes while high profitability and value have the opposite dependency. The portfolio risks are vastly different.
While it is intuitive that some equity factor premia will be more highly correlated during some regimes and to specific macro risks, this research demonstrates the strong portfolio effects from not accounting for macro risks. What is even more surprising is their result that if an investor controls for one type of regime effect, a portfolio can still be very sensitive to a different regime effect.
There is a whack-a-mole problem with macro risks. If an investor controls one set of macro risks, another set will appear. However, an investor will always be worse off by not knowing the conditional risks that are present with seemingly uncorrelated factor premia.
No comments:
Post a Comment