Balance Sheet Information And Future Stock Returns via CSInvesting
Wharton School
University of Pennsylvania
1314 Steinberg Hall – Dietrich Hall
Philadelphia, PA 19104-6365
University of Michigan Business School
701 Tappan St
Ann Arbor, MI 48109
Wharton School
University of Pennsylvania
1312 Steinberg Hall – Dietrich Hall
Philadelphia, PA 19104-6365
May 2006
Abstract
Numerous studies have documented that the most recent annual change in net operating assets is negatively related to future stock returns. In recent work, Hirshleifer, Hou, Teoh and Zhang (2004) show that the level of net operating assets scaled by the previous year’s total assets is also negatively related to future returns. They argue that their levels variable is superior to the change variable used in prior research because it picks up cumulative past changes, rather than just the most recent annual change. We point out that deflation of a level by a lagged level produces a change. As such, their level variable is similar to the change variable used in prior research, and their claim that it picks up cumulative past changes in net operating assets is misleading.
Balance Sheet Information And Future Stock Returns – Introduction
Prior research documents that the most recent annual change in net operating assets predicts future earnings and stock returns. For example, Sloan (1996) shows that the most recent annual change in non-cash working capital is negatively related to future accounting rates of return and future stock returns. Subsequent research by Fairfield, Whisenant and Yohn (2003) and Richardson, Sloan, Soliman and Tuna (2005) shows that Sloan’s results extend to the most recent annual change in net operating assets.
A recent paper by Hirshleifer, Hou, Teoh and Zhang (2004) (HHTZ hereafter) argues that the level of net operating assets (NOA hereafter) is a superior predictor of future earnings and stock returns. In an intuitively appealing argument, HHTZ suggest that the level of NOA picks up all cumulative past difference between operating income and free cash flow. They argue that the most recent annual change in NOA is only a fragmentary indicator of these differences, and that the level of NOA provides a more comprehensive measure (HHTZ, p. 300). They further ague that:
“A stock measure is also simpler, as it derives from the current year balance sheet, whereas a flow measure is calculated as a difference across years in balance sheet numbers.” (HHTZ, p.300, footnote 5)
In support of their argument, they estimate a number of regressions using the level of NOA deflated by lagged total assets as their measure of the level of NOA. The empirical results support their prediction that this variable is a robust predictor of future earnings performance and stock returns.
Our primary objective in this note is to make a simple point. A level deflated by a lagged level is equivalent to a change. As such, HHTZ’s claim that their measure is more comprehensive because it picks up the cumulative amount of past changes is incorrect. Similar to previous research, their variable essentially just measures the most recent annual change in net operating assets.
We show that HHTZ’s measure differs in subtle ways from the change in net operating variable used in previous research. As such, its ability to predict future earnings and stock returns differs somewhat from the measures used in previous research. The signs and the magnitudes of these differences depend on the sample period and research design. We emphasize that these differences cannot be attributed to the ability of HHTZ’s variable to capture cumulative past changes in NOA. Rather, it is just an alternative way of capturing the most recent annual change in NOA.
We close by briefly presenting some empirical results that illustrate our point. First, we show that most of the explanatory power of HHTZ’s levels variable with respect to future returns comes from the most recent annual change in NOA. Second, we show that most of the variation in HHTZ’s levels variable is attributable to the most recent annual change in NOA. Third, we show that contrary to HHTZ’s conclusion, lagged changes in NOA contribute relatively little to the prediction of future stock returns.
See full PDF below.