Short communication
Does cash flow predict returns?

https://doi.org/10.1016/j.irfa.2014.10.001Get rights and content

Abstract

In this paper, we propose the hypothesis that cash flow and cash flow volatility predict returns. We categorize firms listed on the New York Stock Exchange into sectors, and apply tests for both in-sample and out-of-sample predictability. While we find strong evidence that cash flow volatility predicts returns for all sectors, the evidence obtained when using cash flow as a predictor is relatively weak. Estimated profits and utility gains also suggest that it is cash flow volatility that is more relevant as a source of information than cash flow.

Introduction

There is a body of literature that examines the relationship between cash flow and stock returns (Campbell, 1991, Campbell et al., 2010, Campbell and Shiller, 1988, Campbell and Vuolteenaho, 2004, Dechow et al., 2004, Lettau and Wachter, 2007, Santos and Veronesi, 2004). The main finding of this literature is that cash flow and cash flow volatility are determinants of returns, suggesting that they should be useful for forecasting. While limited effort has been devoted to testing whether or not cash flow predicts returns, quite surprisingly, there is presently no evidence to suggest whether cash flow volatility predicts returns. This is an important question, because if predictability can be ascertained, then it should be possible for investors to make use of this information to devise trading strategies with relatively high profits when compared to a naive strategy that ignores this information.

The contribution of the present study is to analyze whether cash flow and cash flow volatility predict returns, and to what extent investors can make use of this information to generate profits. We draw on a small and rich literature that establishes the relationship between cash flow (and its volatility) and returns. The main conclusion from this literature is that cash flow and cash flow volatility are significantly correlated with returns. A natural question that this literature has not addressed is whether cash flow volatility actually predicts returns, as correlation does not necessarily imply predictability. Because we consider both the first and second moments of cash flow as predictors it allows us to understand the relative importance of the two, not only in a statistical sense (predictability) but also in terms of how much economic gains each offers to an investor.

The data that we use consist of firms listed on the New York Stock Exchange (NYSE), which are grouped into sectors based on the global industry classification standard (GICS) (see Narayan & Sharma, 2011). To test the null hypothesis of no predictability, we apply a newly developed in-sample panel predictive test of Westerlund and Narayan (2014). From this exercise we discover strong evidence that cash flow volatility is in fact able to predict sectoral returns, a result that holds also out-of-sample. However, we find weak evidence that cash flow predicts returns. While in in-sample tests, results suggest predictability in five sectors, out-of-sample tests reveal even weaker evidence. We then undertake an extensive analysis of the economic significance of the predictability. Our main findings based on cash flow volatility are; (i) in all sectors dynamic trading strategies generate statistically significant profits, (ii) investors in all sectors are willing to pay more to hold dynamic trading strategies over the historical average, and (iii) profits and investor utilities are heterogeneous, in that they vary from sector-to-sector. On the other hand, when we consider cash flow as a predictor, while we find all sectors to be profitable these profits are significantly less than those obtained using cash flow volatility. The finding that sectors are profitable even in the absence of predictability corroborates the evidence reported by Cenesizoglu and Timmermann (2012).

The rest of the paper is organized as follows. In 2 Hypothesis development, 3 Empirical framework, we introduce the new hypothesis and the empirical framework that will be used to test it. In Section 4, we report the results on return predictability and its economic significance. Section 5 concludes.

Section snippets

Hypothesis development

The goal of this section is to motivate the hypothesis that cash flow and cash flow volatility predict returns. Most valuation theories, including the simple present value model, identify either the change in expected cash flows or discount rates or both as the key determinants of returns. Campbell and Vuolteenaho (2004) developed an intertemporal asset pricing model. They argued that returns on the market portfolio have two components; permanent shocks, which reflect news about future cash

Empirical framework

The panel data predictive regression model that we consider has the following form:yi,t=αi+βixi,t1+ϵi,t,where i = 1, …, N and t = 1, …, T indexes the cross-section and time series dimensions, respectively, xi,t = ρixi,t  1 + εi,t, and ϵi,t and εi,t are mean zero disturbances, which could potentially be correlated with each other, and serially and/or cross-section correlated. This is a panel extension of the prototypical predictive regression model that has been widely used in the time series

Data

The hypotheses that cash flow and cash flow volatility predict returns are tested for firms listed on the NYSE. We use monthly data covering the period August 1996 to August 2010.1

Economic significance results

Investor utility is computed as described in Section 3.3. The risk aversion parameter is set to γ = 6 and we use the US three-month Treasury bill rate as a proxy for the risk-free rate of return.2 The results reported in Table 4 show strong support for the forecast based on the unrestricted predictive regression model in the case of volatility predictors. Take, as an example, the case when CFP volatility is measured by RV,

An explanation of the results

In this section, we have two goals. First, since we discover that cash flow volatility is a better predictor of returns than cash flow, we begin by explaining why this is the case. Second, we focus on the volatility-based predictor which offers strong results in support of predictability, both statistically and economically. We clearly observe that profits and indeed investor utilities are different depending on the sector in which one invests. What this implies is that the role played by CFP

Concluding remarks

This paper is motivated by the lack of empirical evidence on whether or not CFP and CFP volatility predict returns. We form panels of firms listed on the NYSE based on sectors and apply tests for both in-sample and out-of-sample predictability. Two measures of CFP volatility are considered, SD and RV. We discover strong evidence that CFP volatility predicts sectoral returns for at least 14 of the 15 sectors, but weak evidence that CFP predicts sectoral returns. We further show that the

References (42)

  • J. Westerlund et al.

    Does the choice of estimator matter when forecasting returns

    Journal of Banking and Finance

    (2012)
  • A. Ali

    The incremental information content of earnings, working capital from operations, and cash flows

    Journal of Accounting Research

    (1994)
  • A. Ang et al.

    Downside risk

    Review of Financial Studies

    (2006)
  • R. Bansal et al.

    Consumption, dividends, and the cross-section of stock returns

    Journal of Finance

    (2005)
  • R. Bansal et al.

    Risk for the long run: A potential resolution of asset pricing puzzles

    Journal of Finance

    (2004)
  • M. Botshekan et al.

    Cash flow and discount rate risk in up and down markets: What is actually priced?

    Journal of Financial and Quantitative Analysis

    (2012)
  • J. Campbell

    A variance decomposition for stock returns

    Economic Journal

    (1991)
  • J. Campbell et al.

    Growth or glamor? Fundamentals and systematic risk in stock returns

    Review of Financial Studies

    (2010)
  • J. Campbell et al.

    The dividend–price ratio and expectations of future dividends and discount factors

    Review of Financial Studies

    (1988)
  • J.Y. Campbell et al.

    Predicting excess stock returns out of sample: Can anything beat the historical average

    Review of Financial Studies

    (2008)
  • J. Campbell et al.

    Good beta, bad beta

    American Economic Review

    (2004)
  • Cited by (7)

    • The relationship between cash flow uncertainty and extreme risk: International evidence

      2023, Pacific Basin Finance Journal
      Citation Excerpt :

      The results remain in line with our assumptions, with all the interaction terms being insignificant except for column [2]. EPU, quantified by a measure (EPU index) created by Baker et al. (2016), has been proven to affect several aspects of financial and economic markets, including inflation and unemployment (Jones and Olson, 2013; Caggiano et al., 2017), stock market volatility (Bannigidadmath and Narayan, 2021; Narayan and Westerlund, 2014), and corporate governance (Wang et al., 2014). Specifically, this policy-related uncertainty might easily change firms' normal cash flow patterns (Drobetz et al., 2018), completely altering their operating performance.

    • Sudden changes in extreme value volatility estimator: Modeling and forecasting with economic significance analysis

      2015, Economic Modelling
      Citation Excerpt :

      We undertake the similar out-of-sample forecast evaluation exercises (what we have done in the previous subsection) using different in-sample and out-of-sample period. We have considered three cases: (1) 50%/50%, which means that the first 50% of the data have been taken for in-sample analysis and the remaining 50% of the data have been taken for out-of-sample evaluation; (2) 25%/75%, which means that the first 25% of the data have been taken for in-sample analysis and the remaining 75% of the data have been taken for out-of-sample evaluation; and (3) 75%/25%, which means that the first 75% of the data have been taken for in-sample analysis and the remaining 25% of the data have been taken for out-of-sample evaluation (Narayan and Westerlund, 2014; Narayan et al., 2013, 2014; Westerlund and Narayan, 2012). In case of fractional sample length, a value to the nearest integer has been chosen.

    • Robustness of cash flow value: Investment in ASEAN

      2019, Journal of Asian Finance, Economics and Business
    View all citing articles on Scopus
    View full text