In an efficient capital market, firms with better future growth options usually have higher equity valuation. To exercise these growth options, firms with a higher market valuation should have a lower payout ratio and invest more on projects with positive net present value (NPV). However, one study indicates a negative correlation between capital expenditures and industry Tobin’s Q since the middle of the 1990s. Furthermore, previous studies document an investment growth anomaly that there is a negative relation between firm-level capital investment and future stock returns. In the capital budgeting context, some studies argue that given expected cash flows, lower costs of capital lead to higher NPV of new projects and higher firm investment. Since lower costs of capital is also associated with lower expected stock returns, researchers observe a negative investment-return relation. A mispricing-based explanation indicates that if firm-level investment is mispriced by the market due to investor expectational errors or limits to arbitrage, subsequent realized stock returns largely reflect the corrections of market expectations. The oddity of firm investment documented in these studies inspires us to closely investigate the cross-sectional relation between firm investment and subsequent stock returns.
Unlike the previous literature on the relation between investment and stock returns, we focus on firm abnormal investment, which is the gap between actual and predicted investment levels. All information on changes in future firm cash flows, including firm investment decisions, will be instantaneously transferred into a firm’s stock prices in an efficient market. Therefore, abnormal investment may reflect shocks to a firm’s long-run growth opportunities and carry new information about the firm’s fundamentals in the future. For instance, Chen et al. (2017) and Bakke and Whited (2010) show that managers use private information when making their investment decisions. If market investors fully incorporate such new information into stock prices contemporaneously, we should not observe an empirical association between abnormal investment and future stock returns. However, if market investors react to such new information and update their expectations on a firm’s future growth with a delay, the current abnormal investment may exhibit certain predictability of future stock returns.
A firm’s abnormal investment may also be a proxy for agency costs due to conflicts of interests. On the one hand, the managers of a firm with poor investment opportunities and high free cash flow have an incentive to over-invest for their own benefits, e.g., empire building, rather than for the benefits of shareholders. Fairfield et al. (2003) and Titman et al. (2004) provide empirical evidence that over-investment may generate inefficiency and impair firms’ stock performance. On the other hand, agency issues may also be associated with firm under-investment. Due to the conflict of interest between shareholders and bondholders, overhang debts prevent shareholders from capturing the benefits of positive NPV investment opportunities, giving rise to firm under-investment. The conflict of interest between managers and shareholders may also lead to firm under-investment. Hart (1983) and Bertrand and Mullainathan (2003) propose the “lazy manager” hypothesis that managers prefer a quiet life and choose not to spend effort on firm investment. Guerrieri and Kondor (2012) and Aghion et al. (2013) offer the “career concern” hypothesis that managers forgo positive NPV projects because the risk associated with new investment may cost them their jobs. Besides the delayed market reaction explanation, the empirical relation between abnormal investment and future stock returns may reflect the agency cost reduction in firm market value.
Using a large sample of U.S. public firms during 1974–2017, we adopt an accounting-based investment model to decompose firm investment into predicted and abnormal components. “Abnormal investment” is defined as the absolute value of the difference between actual and predicted investment, which measures the degree of a firm’s investment deviating from its predicted level. We also define over-investment (under-investment) as the absolute value of the abnormal investment which is greater (less) than zero. Next, we sort firms into decile portfolios at the end of June over our sample period, based on the ranks of most recent estimated abnormal investment, under-investment, and over-investment. The decile portfolios are rebalanced every year. After adjusting for common systematic risk factors (Fama-French three factors, Carhart’s momentum factor, and Pastor and Stambaugh’s liquidity factor), we find that both abnormal investment and under-investment are negatively related to the performance of the decile portfolios. However, we do not find any evidence that over-investment affects the performance of the decile portfolios. A portfolio taking a long position on the firms with bottom decile under-investment and a short position on the firms with top decile under-investment generates a positive and statistically significant five-factor model alpha. The long-short portfolio’s annualized five-factor model alpha is 5.04%, which is also economically significant.
We then employ the Fama-MacBeth regressions to examine the empirical association between abnormal investment and future stock returns, controlling for firm characteristics. We find that abnormal investment is negatively correlated with future stock returns. When we include both investment and abnormal investment in the same regression, we find that the variation in abnormal investment retains the power of explaining future stock returns, while the coefficient of investment is statistically insignificant. Consistent with the portfolio analysis results, our multivariate regression shows a negative relation between under-investment and future stock returns. However, we cannot find a similar relation between over-investment and stock returns. Taken together, our results suggest that it is the under-investment that mainly drives the negative relation between abnormal investment and future stock returns.
We next examine the two potential mechanisms (discussed above) through which under-investment leads to a decrease in future stock returns: (1) the market delayed reaction channel and (2) the agency cost channel. With respect to the first mechanism, we first investigate whether under-investment conveys information about future profitability, asset growth, and financial distress. After controlling for firm characteristics, we find that under-investment is negatively associated with the change in earnings and the change in assets over the next year. With one standard deviation increase in under-investment, a firm’s earnings growth rate over the next year will decrease by $0.06%$, which is about 60% of the sample mean of earnings growth rates. A one standard deviation increase in under-investment will also be associated with a $0.63%$ decrease in a firm’s asset growth rate over the next year, which is 5.73% of the sample mean of asset growth rates. Using a bankruptcy prediction model, we find that firms with under-investment are more likely to experience future financial distress. With one standard deviation increase in under-investment, the probability of financial distress will increase 0.30%, which is 5.77% of the sample mean of unconditional financial distress probabilities.
These results confirm that under-investment contains information about firm fundamentals in the future. In an efficient market, investors should promptly incorporate the information carried by abnormal investment into stock prices. To show that the negative relation between under-investment and future stock returns is partly due to the delayed market reaction to under-investment, we employ an empirical test which is similar to the research design used by Abarbanell and Bernard (1992) and Shane and Brous (2001) in their examinations of the post-earnings announcement drift. We show that after controlling for the future change in earnings, the future change in assets, and the likelihood of future financial distress, the relation between under-investment and future stock returns is not statistically significant. About 47.06% of the negative association between under-investment and future stock returns is due to the association between under-investment and future firm fundamentals, supporting the market delayed reaction channel.
To explore the second mechanism, the agency cost channel, we investigate whether the negative relation between under-investment and future stock returns is more pronounced for firms with weaker external monitoring or higher agency costs. If under-investment is driven by potential agency issues, then market investors will adjust firm value according to under-investment related agency costs, leading to lower stock returns. We first classify firm–year observations with under-investment into two sub-samples using the annual industry medians of blockholder ownership, the ownership of a firm’s blockholders who hold more than 5% of the firm’s outstanding shares. Firms with blockholder ownership above the median are classified as those with stronger external monitoring and lower agency costs. We find that the negative relationship between under-investment and future stock returns is only statistically significant in the low blockholder ownership sub-sample. We next divide firm–year observations with under-investment into two sub-samples based on two direct proxies of agency costs: expense ratio and asset utilization ratio. Higher expense ratios are associated with less efficiency and higher agency costs, while higher asset utilization ratios are associated with greater efficiency and lower agency costs. We find that although the negative relation between under-investment and future stock returns is statistically significant in both partitions, the economic impact of under-investment on future stock returns is larger for firms with higher agency costs. Combined, these findings support the agency cost channel that agency conflicts may lead to firm under-investment and hurt firm value. [1]
Finally, we conduct a set of robustness tests to validate our main findings. First of all, we re-estimate the impact of abnormal investment, under-investment, and over-investment on future stock returns using a panel regression with the year and industry fixed effects. To mitigate the concern about econometric issues in estimating the investment model, we reconstruct our three abnormal investment proxy variables using a single panel regression between 1974 and 2017 and rolling panel regressions with five-year estimation windows. To mitigate any concern on potential model misspecification, we estimate the predicted investment using the two alternative investment models developed by Harvey et al. (2004) and Titman et al. (2004). These robustness tests generally support our main findings that there is a negative relation between abnormal investment and future stock returns and that the negative relation is mainly driven by under-investment, not over-investment. In our supplementary tests, we examine whether the negative relation between under-investment and future stock returns is due to the firm-specific information carried by under-investment or the potential positive association between under-investment and the systematic financial distress risk factor. We do not find evidence supporting the systematic financial distress risk exposure explanation. We also investigate the impact of market recessions on the negative relation between abnormal investment and future stock returns. We find that the negative relation between abnormal investment and future stock returns is much weaker during market recession periods than non-recession periods, suggesting that, during market recession periods, market investors are more likely to react to the negative information carried by under-investment without a delay.
Our paper is closely related to Titman et al. (2004), which also investigates the association between abnormal capital investment and subsequent stock performance. Titman et al. (2004) find that firms with the most over-investment are likely to under-perform during the following five years. This empirical relation is stronger for firms with more cash flows or fewer debts. Our paper differs from their work in two dimensions. First, they measure the abnormal capital investment as the deviation of a firm’s capital expenditures from its average capital expenditures over the past three years, whereas our abnormal investment is estimated based on an accounting-based framework which controls for the cross-sectional and time-series variations of firms’ growth opportunity, leverage, cash holding, age, size, stock returns, and historical investment. Second, they find that firms with the least abnormal capital investment tend to out-perform firms with the highest abnormal capital investment in terms of stock returns. Our paper shows that after adjusting for the cross-sectional and time-series variations in firm characteristics, it is under-investment that drives the negative relation between abnormal investment and future stock returns, not over-investment.
Our paper contributes to the earlier investment literature in three important ways. First, our paper sheds light on the investment growth puzzle by showing that abnormal investment, to a certain degree, drives the negative relation between investment and future stock returns. Second, we show that stock markets react differently to firm under-investment and over-investment. The negative relation between abnormal investment and future stock returns is mainly due to under-investment. Third, we provide evidence on both the market delayed reaction channel and the agency cost channel through which under-investment may have a negative impact on future stock returns.
Siqi Liu is an Assistant Professor in Finance at Queen’s University Belfast.
Chao Yin is an Associate Professor in Finance at the University of Edinburgh Business School.
Yeqin Zeng is an Associate Professor in Economics and Finance at Durham University Business School.
This post is adapted from their paper, “Abnormal Investment and Firm Performance,” published on the International Review of Financial Analysis in 2021 and available on SSRN.
[1] If stock markets are efficient, agency costs associated with under-investment may lead to a contemporaneous change in stock prices and should not be associated with lower future stock returns. We acknowledge that in an efficient market, the agency cost channel would also require that investors underreact to the implications of agency costs for firm investment decisions.
