Journal of Business Finance & Accounting最新文献

We use machine learning methods to predict firm-specific stock price crashes and evaluate the out-of-sample prediction performance of various methods, compared to traditional regression approaches. Using financial and textual data from 10-K filings, our results show that a logistic regression with financial data inputs performs reasonably well and sometimes outperforms newer classifiers such as random forests and neural networks. However, we find that a stochastic gradient boosting model systematically outperforms the logistic regression, and forecasts using suitable combinations of financial and textual data inputs yield significantly higher prediction performance. Overall, the evidence suggests that machine learning methods can help predict stock price crashes.

Existing research based on the US setting suggests that high-quality audits impede firm innovation by inducing managerial myopia that leads managers to sacrifice innovative activities. We extend this literature by examining the impact of audit quality on firm innovation in a major emerging market, China. Given the weak institutional environments in China, we posit that increases in audit quality enhance firm innovation by facilitating firms’ external financing abilities and promoting managers’ innovation efforts. Consistent with our predictions, we find that increased audit quality due to “switching” to BigN auditors after auditor mergers significantly improves firms’ innovation quantity and quality. We further find that increases in audit quality impact firm innovation by increasing external financing and motivating both internal and external investment in innovation. Overall, our results indicate that audit quality has a positive governance effect on firms’ real decisions in a major economy with weak legal environments.

We investigate whether heterogeneity in pay-performance sensitivity (PPS) among top management team (TMT) members affects management earnings forecasts (MEFs). Extant studies find that dispersion of PPS among TMT members results in a lack of coordination among TMT members and has a significant consequence for the firm. Since the issuance of forecasts is jointly determined by TMT managers, we argue that heterogeneity in PPS among TMT members will affect forecast issuance. We find that firms are less likely to issue MEFs when PPS dispersion among TMT members is high. Moreover, the forecasts of these firms are less optimistic. Finally, our analyses reveal that the effect of dispersion of PPS among TMT members on MEFs is more pronounced when the difference in PPS between the chief executive officer and other executives is high or team tenure is short. Overall, our results suggest that heterogeneity in PPS among TMT members affects a firm's voluntary disclosures.

Investment-related social media platforms are transforming the traditional intermediary landscape by rapidly disseminating user-empowered opinions and recommendations, raising concerns about the credibility of information on such platforms. We examine the differential content verifiability of SeekingAlpha.com articles with sell recommendations (short articles) versus articles with buy recommendations (long articles). We find that short articles contain greater verifiable content than long articles, and verifiable content in short articles generates greater market reactions and better mitigates return reversals than that in long articles. This asymmetry contrasts prior research evidence that greater content verifiability accompanies traditional sell-side analyst reports with buy recommendations. Our results are robust to various confounding factors, including author effects, among others. Our results become more pronounced in the presence of greater retail ownership. Taken together, our results provide new evidence on investors’ assessment of the credibility of information on social media platforms.

We analyze a unique set of Chinese firms to isolate the impact of accounting standards on financial reporting comparability and capital markets. From 2001 to 2006, these companies simultaneously maintained two sets of financial statements because of their dual-class share structure: statements for A-shares followed Chinese Generally Accepted Accounting Principles (GAAP), whereas those for B-shares followed International Financial Reporting Standards (IFRS). We find a disparity in financial reporting comparability between these two accounting standards: IFRS produce less comparable information than Chinese GAAP. We also find that the disparity in comparability is related to fair value accounting, especially when corporate governance is weak.

Analyzing the launch of high-speed rail (HSR) services in China as a natural experiment, we identify a positive externality stemming from lower information acquisition costs: the reduction in firms’ overinvestment in tax avoidance. Specifically, we find that outsiders undertake more corporate site visits and firms engage in less tax avoidance after the opening of HSR lines in the cities where these firms are located, leading to enhanced firm value. In another result consistent with expectations, we document that the impact of the introduction of HSR lines on tax avoidance is concentrated in firms in which insiders exhibit a high propensity to extract rents through aggressive tax strategies. Our results imply that more efficient transportation facilitates site visits and the acquisition of firm-specific information, particularly soft information. This improvement strengthens external monitoring, thereby limiting the ability of insiders to accumulate private benefits under the guise of tax avoidance that benefits all shareholders as the residual claimants.

This paper examines the association between firms’ political connections and their voluntary disclosure of information related to government subsidies, using a sample of Chinese non-state-owned enterprises. We find that politically connected subsidy recipients disclose less information about their subsidies (source and policy basis) in their annual reports than do unconnected firms. This association is mainly driven by connected firms whose subsidies are difficult to justify and are stronger for firms registered in more corrupt provinces and firms with higher media attention. In addition, connected firms disclosing more subsidy information receive fewer future subsidies than do other connected firms. These findings suggest that politically connected subsidy recipients tend to withhold subsidy information to reduce the costs that accompany public scrutiny of subsidies granted through relationships.

Information economics predicts that a principal will prefer a finer performance evaluation system to a coarser one. Nevertheless, coarse performance evaluation is often used in practice. To explain this seeming contradiction, I construct a model of a principal and envious agents. I show that a coarse evaluation system can do as well as a finer one if agents are sufficiently envious, that is, if they incur large utility loss when they are paid less than their peers. This result supports the use of coarse performance evaluation that aggregates signals of good performance, in the form of fixed wages or inflated ratings.

We examine how expositional quality (measured by discretionary readability and lexical focus—new measures) in the textual content of prospectuses preceding seasoned equity offerings affects the gross spread after controlling for the quality of numerical information. We measure the gross spread as the difference between the offer price and the underwriter's purchase price per share, expressed as a percentage of the offer price. We find that the gross spread is higher in firms whose prospectuses have lower discretionary readability, while the discretionary readability of the preceding 10-K has little effect. Further, when the prospectus readability is lower than that of the preceding 10-K, the gross spread increases more than it decreases when readability is higher by the same amount. Low focus (lexical focus) in the Use of Proceeds section of the prospectus results in a higher gross spread. However, low prospectus focus is associated with higher gross spread only in the bad-news subsample. Overall, our results show that the gross spread is incrementally associated with textual obfuscation over the numerical quality of earnings. Moreover, our findings indicate that investors get distinctly different and useful information from different measures of textual quality in different parts of the text. Finally, we present novel insights showing that investors respond differently to different measures of textual quality, for example, readability and lexical focus, depending on whether there is good or bad news.

Accounting standards require firms to write down their inventory when its market value (or net realizable value) drops below cost. Recognizing inventory write-downs decreases gross margins and increases scrutiny of managers’ inventory management practices. Further, it is difficult for market participants to assess inventory balances. Therefore, managers often delay recognizing inventory losses. Compared with other analysts, analysts who issue disaggregated gross margin (GM) forecasts evaluate a firm's inventories more closely, which should motivate managers to recognize expected inventory losses. Consistent with this prediction, managers’ likelihood of recognizing an expected inventory loss increases as more analysts provide GM forecasts. However, the number of analysts issuing standalone earnings per share (EPS) or EPS and other disaggregated forecasts is not associated with inventory loss recognition. We address endogeneity concerns by documenting that firms are less likely to recognize an expected inventory loss after exogenous factors cause analysts to stop issuing GM forecasts for the firm. We contribute significantly to the literature by showing that analysts' GM forecasts are closely associated with managers’ timely inventory loss recognition. This insight bridges the gap between anecdotal evidence, which suggests analysts assess inventory carefully, and scholarly research indicating that analysts' forecasts may not fully capture the effects of inventory changes. Our findings suggest that managers act as if inventory management practices are more strictly monitored only when analysts issue GM forecasts.