Factor Construction: Beta vs $-Neutrality

Is Beta Driving Factor Performance?

June 2017. Reading Time: 10 Minutes. Author: Nicolas Rabener.

SUMMARY

  • Factors constructed $-neutral didn’t benefit much from beta-exposure from 2000 to 2017
  • Beta-neutrality is only a must for Low Volatility
  • However, beta-neutral factors offer lower correlation to long-only indices

INTRODUCTION

Calculating factor performance is, unfortunately, as much art as science as numerous assumptions need to be made. Amongst these are the definitions of the stock universe, the portfolio size, the factor, the ranking methodology, the rebalancing period, the transaction costs, etc. All of these decisions have a significant impact on the factor performance and there are no clearly established industry standards. One of the finer points is how the long and short portfolios should be constructed, where the choice is between $-neutral and beta-neutral. Pondering about this brings up the question how much factor performance is driven by beta, if they are constructed $-neutral like the well-known Fama-French factors. In this short note we’re going to analyse the difference in factor performance between $-neutral and beta-neutral portfolios.

METHODOLOGY

The factors used in the analysis are Value, Size, Momentum, Low Volatility, Quality and Growth. Factor data is available for all developed markets. Portfolios select the top and bottom 10% of the stock universes in the US, Europe, and Japan compared to 20% for the UK, Australia, Hong Kong and Singapore given smaller stock universes.

The $-neutral portfolios are model-free as the long portfolios receive the same allocation in $-terms as short portfolios, e.g. $100 each. For the avoidance of doubt, $ simply refers to a unit of currency and not to the US$. Beta-neutral portfolios are constructed by increasing the weights of stocks which have a one-year beta to the local index of below 1 and decreasing the weights of stocks with a beta above 1, i.e. as a result the long and short portfolios will both have a beta of 1.

VALUE FACTOR (2000 – 2017)

The chart below shows the net beta, which is the beta of the long portfolio minus the beta of the short portfolio, for the Value factor from 2000 to 2017 (read Value US Sectoral Ana