ABSTRACT Skewness is one of the strongest predictors of commodity futures returns. The popular behavior‐based explanation is that investors have a preference for positive skewness, which causes positively skewed commodity futures to become overpriced and subsequently earn lower expected returns. We regress skewness on estimated betas from a conditional seven‐factor model that allows betas to vary with characteristics. The fitted values from these regressions are referred to as beta components , while the residuals represent non‐beta components . Standard Fama–MacBeth regressions and portfolio‐sorting analyses show that the predictive power of skewness is exclusively driven by its beta components, suggesting that skewness primarily captures systematic risk.

In this study, we define unexpected gross profit (GAP) as the difference between the current level of gross profit and its expected value, estimated using a moving average over the preceding eight quarters. With Fama-MacBeth t-statistics of 8.32 under ordinary least squares and 2.72 under weighted least squares, GAP emerges as one of the strongest predictors of individual stock returns, comparable to standardized unexpected earnings (SUE) measures. Moreover, GAP exhibits low correlations with both SUE and the revenue surprise measure, while providing incremental predictive power. Using the newly developed instrumented principal component analysis (IPCA) framework, we find that the predictive ability of GAP is primarily attributable to its role as a proxy for systematic risk (beta) rather than mispricing (alpha).

We estimate the Instrumented Principal Component Analysis (IPCA) latent factor model of Kelly, Pruitt, and Su (2019) and find that it explains a substantial share of the cross-sectional variation in monthly returns of oil stocks, with an R² of 0.536. This explanatory power markedly exceeds that of two popular observable factor models with characteristic-driven time-varying betas, which yield R²s of 0.221 and 0.233, respectively. Four firm characteristics, size, book-to-market, short-term momentum, and 12-month momentum, are statistically significant instruments for the latent-factor betas. Moreover, oil-related factors, including commodity futures index returns, monthly changes in Brent crude oil price, and monthly changes in crude oil imports, provide significant incremental explanatory power even after controlling for five latent factors and two sets of observable risk factors.

In this paper, we construct ex-ante measures of skewness from ten major commodity futures contract characteristics, including lagged skewness. We first employ monthly cross-sectional regressions of skewness on several lagged contract characteristics. Second, we follow a forecast combination approach and run monthly cross-sectional regressions of skewness on individual contract characteristics. Both approaches generate expected skewness that is significantly and negatively correlated with commodity futures contract returns, even when we construct expected skewness without using lagged skewness. Our empirical evidence, therefore, provides strong support for the key prediction of Barberis and Huang's (2008) model relating asset return skewness to asset returns.

Using the measurement error variance (MEV) criterion introduced by Lee, So, and Wang (2020), we evaluate the performance of expected return proxies (ERPs) constructed from asset pricing models that allow factor loadings to vary with firm characteristics, encompassing both observable and latent factor frameworks. We find that the best‑performing ERPs are derived from observable risk factor models, specifically those of Fama and French (2015), Hou, Xue, and Zhang (2015), Barillas and Shanken (2018), Daniel, Hirshleifer, and Sun (2020), and Stambaugh and Yuan (2017), when betas are allowed to depend on firm characteristics. ERPs from each of these five models consistently dominate those generated by latent factor approaches with characteristic‑dependent betas, including models based on instrumented and projected principal component analysis (IPCA and PPCA).

Using employee job-level data, we empirically test the equilibrium matching between a firm’s debt usage and its employee job risk aversion (“clientele effect”), as predicted by the existing theories. We measure job risk aversion for a firm’s employees using their labor income concentration in the firm, calculated as the fraction of the employees’ total personal labor income or total household labor income that is accounted for by their income from this particular firm. Using a sample of about 1,400 U.S. public firms from 1990-2008, we find a robust negative relation between leverage and employee job risk aversion, which is consistent with the clientele effect. Specifically, when a firm’s existing employees have higher labor income concentration in it, the firm tends to have lower contemporaneous and future leverage. Moreover, in terms of new hires, firms with lower leverage are more likely to recruit employees with less alternative labor income. Our results continue to hold after we control for firm fixed effects, other employee characteristics such as wages, gender, age, race, and education, and managerial risk attitudes. Further, the matching between a firm’s leverage and its workers’ labor income concentration in it is more pronounced for firms with higher labor intensity and those in financial distress.

PurposeWe aim to examine two issues. First, we intend to identify the best performing expected return proxies. Second, we investigate whether the expected return proxies for individual stocks can track the corresponding realized returns during extremely good or extremely bad times of the economic environment related to business conditions, stock market valuation and broad market performance.Design/methodology/approachWe construct four sets of expected return proxies, including: (1) characteristic-based proxies; (2) standard risk-factor-based proxies; (3) risk-factor-based proxies that allow betas to vary with firm characteristics and (4) macroeconomic-variable-based proxies. First, we estimate expected returns for individual stocks using newly developed methods and evaluate the performance of these expected return proxies based on the minimum variance criterion of Lee et al. (2020). Second, we regress expected return proxies and realized returns on indicator variables that capture the extreme phases of the economic environment. Then we compare the estimated coefficients from these two sets of regressions and see if they are similar in magnitude via formal hypothesis testing.FindingsWe find that characteristic-based proxies and risk-factor-based proxies that allow betas to vary with firm characteristics are the two best performing proxies. Therefore, it is important to allow betas to vary with firm characteristics in constructing expected return proxies. We also find that model-based expected return proxies do a reasonably good job capturing actual returns during extremely bad and extremely good phases of business cycles measured by leading economic indicators, consumer confidence and business confidence. However, there is a large gap between the adjustment of model-based expected returns and realized returns during extreme episodes of stock market valuation or broad market performance.Originality/valueWe examine four types of expected return proxies and use the newly developed methodology as in Lee et al. (2020) to see which one is the best. In addition, we document whether model-based expected returns from individual stocks adjust partially or fully to keep pace with actual returns in response to changing economic conditions. No prior studies have examined these two issues.

We evaluate the performance of expected return proxies during extreme credit market conditions and extreme phases of business cycles when realized returns on banks stocks are large in absolute value. We construct three sets of expected return proxies for individual bank stocks: (i) characteristic-based proxies; (ii) standard risk-factor-based proxies; and (iii) risk-factor-based proxies in which betas depend on firm characteristics. Based on the newly developed minimum error variance (MEV) criterion (Lee et al., 2020), the best performing expected return proxy is the risk-factor-based model that allows betas to vary with firm characteristics. We also examine whether these three expected return proxies can capture actual returns during either extreme credit market or extreme business-cycle conditions. We find that both risk-factor-based proxies explain returns better than characteristic-based proxies during these periods.

In this study, we investigate the tail dependency between bank stocks in China and 35 common risk factors. We measure univariate and multivariate conditional tail risk probabilities. The evidence indicates that tail events from risk factors in the banking, security trading, real estate, and energy industries have the largest effects on the realization of extreme returns from Chinese bank stocks. The univariate conditional tail risk is considerably higher than the unconditional tail risk. The impact of multiple tail events from several risk factors occurring simultaneously is much stronger than tail events from one single risk factor. In general, there is a stronger cross-market tail linkage between emerging market risk factors and bank stocks in China when compared with developed market risk factors. However, the cross-market tail linkage between developed market risk factors and bank stocks in China rose sharply during the 2008 financial crisis.

Pension discount rates have a powerful effect on the size of reported defined benefit corporate pension liabilities because of the long-term nature of projected benefit obligations. Firms often choose pension discount rates that are above the guideline long-term Treasury, AAA-grade, and AA-grade corporate bond yields. We assess the sizes of understated pension liabilities relative to these benchmark interest rates and relate them to individual firms’ implied cost of equity. We find that firms with large understated pension liabilities have a higher implied cost of equity after taking into account standard control variables and other pension information such as funded status and mandatory contributions.

Purpose The question is whether debt market investors see through managers' attempts to hide their pension obligations. The authors establish a robust relation between understated pension liabilities and corporate bond yield spreads after controlling for factors that have been previously identified as having a significant impact on firms' cost of borrowing. The results support the idea that bond market investors are not being misled by the use of high pension liability discount rates by some companies to lower their reported pension obligations. For a small fraction of debt issuers, the reported pension liabilities are larger than the pension liabilities valued at the stipulated interest rate benchmarks. For these issuers with overstated pension liabilities, bond investors adjust their borrowing costs downward. Design/methodology/approach The authors investigate the relation between corporate bond yield spreads and understated pension liabilities relative to long-term Treasury and high-grade corporate bond yields. They aim to answer two questions. First, what are the sizes of over or understated pension liabilities relative to guideline benchmarks? Second, do debt market investors see through the potential management manipulation of pension discount rates? The authors find that firms with large understated pension liabilities face higher marginal borrowing costs after taking into account issue-specific features, firm characteristics, macroeconomic conditions and other pension information such as funded status and mandatory contributions. Findings The average understated projected benefit obligations (PBOs) are understated by $394.3 and $335.6, equivalent to 3.5 and 3.0% of the beginning of the fiscal year market value, respectively. The average understated accumulated benefit obligations (ABOs) are understated by $359.3 and $305.3 million, equivalent to 3.1 and 2.6%, of the beginning of the fiscal year market value, respectively. Relative to AA-grade corporate bond yields, the average difference between firm pension discount rates and benchmark yields becomes much smaller; the percentage of firm pension discount rates higher than benchmark yields is also much smaller. As a result, understated pension liabilities become negligible. The authors establish a robust relation between corporate bond yield spreads and measures of understated pension liabilities after controlling for issue-specific features, firm characteristics, other pension information (funded status and mandatory contributions), macroeconomic conditions, calendar effects and industry effects. Originality/value S&P Rating Services recognizes the issue that there is considerably more variability in discount rate assumptions among companies than in workforce demographics or the interest rate environment in which firms operate (Standard and Poor's, 2006). S&P also indicates that it would be desirable to normalize different discount rate assumptions but acknowledges that it is difficult to do so. In practice, S&P Rating Services conducts periodic surveys to see whether firms' assumed discount rates conform to the normal standard. The paper makes an initial attempt to quantify the size of understated pension liabilities and their impact on corporate bond yield spreads. This approach can be extended to study firms' costs of equity capital, the pricing of seasoned equity offerings and the pricing of merger and acquisition transaction deals, among other questions.

Pension discount rates have a powerful effect on the size of reported defined benefit corporate pension liabilities because of the long-term nature of projected benefit obligations. Firms often choose pension discount rates that are above the guideline long-term Treasury, AAA-grade, and AA-grade corporate bond yields. We assess the sizes of understated pension liabilities relative to these benchmark interest rates and relate them to individual firms’ implied cost of equity. We find that firms with large, understated pension liabilities have a higher implied cost of equity after taking into account standard control variables and other pension information such as funded status and mandatory contributions.

This paper analyzes the impact of shareholder-creditor conflicts on corporate risk-taking. Specifically, I examine the role played by institutional dual-holders (i.e., those simultaneously holding the same firm's debt and equity) in corporate innovation. Baseline results show that firms held by dual-holders generate fewer but more valuable patents. To alleviate endogeneity concerns, I use a difference-in-differences approach based on financial institution mergers. Further analysis suggests that decreased sensitivity of managerial compensation to firm risk might be a possible channel. Overall, I provide new evidence that shareholder-creditor conflicts indeed exist and lead to risk-shifting, and that dual ownership can partially mitigate this problem.

We examine how the tail behavior of risk factors affects the tail behavior of individual bank stock returns in the United States. Using 26 common risk factors, we construct univariate and multivariate conditional exceedance measures. We find that returns on banking industry, security-trading industry, and broad market portfolios have the largest impact on the probability of observing high positive tail returns on bank stocks. A small-minus-big bank return factor, market volatility, and a profitability risk factor have the largest impacts on the probability of lower tail returns. Bank capital ratios and total allowances for loan losses are notably related to tail risk.

Costs are sticky on average, that is, they fall less for sales decreases than they rise for equivalent sales increases. We examine the effect of this asymmetric cost behavior on a firm's dividend policy. Given investors' aversion to dividend cuts, we predict that firms with higher resource adjustment costs and stickier costs pay lower dividends than their peers because they are less able to sustain any higher level of dividend payouts in the future. We find evidence consistent with this prediction. Further, using a regression discontinuity design that exploits variation in labor adjustment costs generated by close-call union elections, we provide evidence suggesting that the negative relation between cost stickiness and dividend payouts is driven by resource adjustment costs. Our paper sheds new light on the determinants of dividend policy and demonstrates the role of cost behavior in corporate decisions.

Considering turnover intention and dismissal tendency,managers' two-stage dynamic incentive model was designed in which managers' expected utility and enterprises' expected return were determined according to the three possibilities and probabilities of re-election,compensation turnover and non-compensation turnover.Five conclusions were drawn as follows.Firstly,the greater a manager's turnover intention is,the lower his effort is and the higher his requirement for the performance share ratio will be.Secondly,when a manager gets more dismissal payment than the increase of basic salary in the next period while he provides the same effort,the more the dismissal intention is,and the greater the manager's effort is,the lower his requirement for the performance share ratio will be.Thirdly,the dynamic basic salary system can help to improve managers' effort.Fourthly,the higher the proportion of monopolistic surplus is,the lower the manager's effort will be.Finally,the greater the performance share ratio and the non-monetary utility sensitivity are,the higher the manager's effort will be.

We study labor power as an important but largely under-explored determinant of payout policy. Using a regression discontinuity design that exploits locally exogenous variation in labor’s collective bargaining power, we find that an increase in labor power generated by close-call union elections leads to a lower level of dividend and total payout in subsequent years. Operating flexibility appears to be a plausible underlying mechanism through which labor power influences corporate payout. Firms use the saved earnings from reductions in payout to invest in net working capital rather than paying off debt or increasing cash holdings. Our paper sheds new light on the determinants of payout policy and the role of labor power in corporate finance decisions.

Using proprietary employer-employee matched data of the U.S. Census Bureau, we measure individual workers’ risk tolerance towards their jobs by their family labor income diversification, and formally test the firm-level equilibrium matching between capital structure and employees’ job risk attitudes (the “clientele effect”). Consistent with theories, we find a robust, positive association between a firm’s debt usage and its employees’ family labor income diversification. This relation is stronger for firms with higher labor intensity and those with greater distress risk. Further, higher-leverage firms recruit new employees with greater labor income diversification. For identification, we exploit the California Paid Family Leave Legislation as a shock to family income diversification, and the SFAS 123(r) rule change as a shock to firm leverage.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI