Research

We examine how mandatory country-by-country reporting (CbCR) affects the investment and capital allocation efficiency of multinational entities (MNEs). We posit that the introduction of U.S. CbCR induces affected MNEs to collect and process standardized country-level information, which enhances managerial information sets and enables them to make better investment decisions. Using a difference-in-differences design, we find that firms subject to the regulation choose more efficient levels of capital investments and allocate more capital toward high-opportunity segments relative to unaffected MNEs. We further observe that investment efficiency improves faster than internal capital allocation efficiency, consistent with greater information frictions in allocation than in overall budgeting decisions. Our findings are robust to various placebo tests and alternative research designs, including regression discontinuity analyses. Additional analyses show that improvements in aggregate investment efficiency are more pronounced in firms with weaker pre-existing information environments and that efficiency gains coincide with improvements in operational outcomes closely related to the mandated disclosures. Overall, our results are consistent with U.S. CbCR improving the efficiency of corporate investment decisions by alleviating managerial information frictions.

We examine how individual accounting employee movements across firms impact the transfer of corporate disclosure practices. We employ a large dataset of employee movements across firms through employee disclosures sourced from a professional networking platform and apply GPT-4 to identify employees in accounting roles. We use narrative disclosures in 10-K filings to measure corporate disclosure practices, which offer rich and nuanced discretion within a disclosure. We first identify an increase in similarity between the qualitative disclosures of firms experiencing an employee movement relative to other firm pairs. This increase is observable for both rank-and-file and executive-level employees and varies in magnitude by employee tenure and the similarity of the old and new roles. Moreover, the increase in similarity is more substantial in firm pairs that are industry peers. In subsequent tests, we use a difference-in-differences design at the firm-year level to show that firms hiring employees with previous exposure to highly informative disclosures increase their disclosure informativeness. Collectively, we report evidence that individual accounting employees at all levels transfer disclosure practices across firms.

We examine how narrative text in financial statements contextualizes firms' tax outcomes, which narrative topics are most informative, and whether the contextual information is useful to financial statement users. We build on prior research, which broadly finds that text is informative, by explicitly modeling context as an interaction between text and numbers that improves th mapping to firm outcomes. Taxes are an ideal setting for this analysis because the link between an accounting number (pre-tax income) and a firm outcome (tax expense) is both conceptually grounded and systematically distorted by reporting and regulatory rules that narrative disclosures can help explain. Using embeddings derived from management discussion and analysis (MD&A) sections of 10-K filings, we train deep neural networks that learn how textual context alters the relation between pre-tax income and tax expense. We show that context from the MD&A has significant explanatory power, improving the mapping betwen book income and tax outcomes by 17.7% to 21.3%. In contrast, income tax footnote narratives often obscure rather than clarify this mapping. Moreover, we find that disclosures referring to M&A activities, firms' external environments, and forward-looking and strategic considerations add the greatest informational value, whereas discussions relating to the application of accounting standards even reduce the informational value of context in explaining tax outcomes. Finally, we show that MD&A context is useful to financial statement users – e.g., by improving analysts' effective tax rate forecasts. Collectively, out findings demonstrate the value of contextual information in understanding distortions between book and tax numbers.