Regime-investing sector model performance in benchmark tilting; evidence in the US

Abstract

It is generally accepted that business cycles phases have tendencies, allowing for a level of predictability. In addition, it is widely noted that differing assets, including equity sectors, exhibit differing characteristics through these cycle phases. This paper attempts to study whether benchmark-tilting investors could potentially utilize quantitative sector rotation to exploit these observations. This is done by first analysing sectoral behaviour through the cycle, and second by constructing a model, in which probabilities for the upcoming business cycle phase are formed, which in turn are used to dynamically optimise and reallocate sectors in a portfolio based on the sectors’ previous characteristics in the business cycle phases. The study is done from the perspective of institutional investors, attempting to cover practical issues related to the strategy. Business cycle probabilities are formed utilising the OECD CLI, a well-established composite leading indicator of economic activity. The study is conducted utilising the S&P 500 index and its sector indexes, with robustness added through similar analysis with European data.

The results obtained in the paper point to business cycle-based sectoral rotation being a potentially useful additional tool for enhanced indexes. Clear behavioural differences between sectors were observed through the business cycle, and although no fully countercyclical sector was found, sectors exhibited cyclical and defensive characteristics. Regime optimised models managed to improve performance over both, the benchmark, as well as non-regime optimised portfolios. Performance improvements when taking into account transaction costs were significant in the US, over 100 basis points annually for the best performing models, with slightly weaker improvements in Europe. The overperformance mostly arrived through contractionary periods, with no clear improvement observed otherwise. The model also proved potentially useful, being robust to smaller delays in reallocation, and exhibiting a level of flexibility. Some practical concerns do remain, due to periods of poor sustained performance.