Cross‐autocorrelations among asset classes: Evidence from the mutual fund industry

Abstract

Purpose

The purpose of this paper is to examine asset class cross‐autocorrelations at the macro‐level by exploring the return associations among mutual fund asset classes. The low transactions costs of trading mutual funds make this extension important since informed traders can potentially use mutual funds to exploit asset class return cross‐autocorrelations that were not exploitable with individual securities.

Design/methodology/approach

The Granger causality tests and correlation results are employed to ascertain whether significant relationships exist among asset classes. Using a time series of 2,739 daily returns for 641 mutual funds comprising 20 asset classes, trading strategies are developed using the initial sample and evaluated out‐of‐sample on a risk‐adjusted basis.

Findings

Both the cross‐autocorrelations and Granger causality tests suggest that most of the domestic equity asset class returns can predict future global and international equity returns. Further, the trading‐rule portfolios provide a greater return per unit of risk (Sharpe and Treynor ratios) thus dominating all buy‐and‐hold portfolios. Risk‐adjusted excess returns (Jensen's α) of the trading rules are positive and statistically significant at the 1 per cent level. The results of trading strategies also reveal that there are no statistically significant return differences between load and no‐load funds.

Research limitations/implications

Redemption fees seem to be standard practice now, except for money market funds and funds specially designed for market timers. Thus, the trading strategy returns of this paper overestimate actual returns. However, investors may still find the proposed trading strategies beneficial because redemptions fees can be avoided if investors get the opportunities to trade in mutual fund supermarkets. The trading strategies may have implications for other international markets where the sizes and styles of the mutual funds' assets are increasing enormously with a few trading restrictions.

Originality/value

A noteworthy and original contribution of this study is the two‐day Granger causality test. This paper documents that the duration of mutual funds' return predictability extends beyond a one‐day horizon. The duration of daily mutual fund return predictability is believed to be unexplored and should be of considerable relevance to practitioners and regulators.

Citation

Miller, E.M., Prather, L.J. and Imtiaz Mazumder, M. (2008), "Cross‐autocorrelations among asset classes: Evidence from the mutual fund industry", Managerial Finance, Vol. 34 No. 11, pp. 756-771. https://doi.org/10.1108/03074350810900488