Graph convolutional neural networks (GCNNs) have been successfully applied to a wide range of problems, including low-dimensional Euclidean structural domains representing images, videos, and speech and high-dimensional non-Euclidean domains, such as social networks and chemical molecular structures. However, in computer vision, the existing GCNNs are not provided with positional information to distinguish between graphs of new structures; therefore, the performance of the image classification domain represented by arbitrary graphs is significantly poor. In this work, we introduce how to initialize the positional information through a random walk algorithm and continuously learn the additional position-embedded information of various graph structures represented over the superpixel images we choose for efficiency. We call this method the graph convolutional network with learnable positional embedding applied on images (IMGCN-LPE). We apply IMGCN-LPE to three graph convolutional models (the Chebyshev graph convolutional network, graph convolutional network, and graph attention network) to validate performance on various benchmark image datasets. As a result, although not as impressive as convolutional neural networks, the proposed method outperforms various other conventional convolutional methods and demonstrates its effectiveness among the same tasks in the field of GCNNs.

We discuss how researchers in finance and accounting can take advantage of textual analysis methods using examples from recent work. We propose that accounting researchers can contribute to a broader debate in economics and related fields using their deep knowledge of corporate disclosures as well as their training in measurement. We underline the importance of thinking about the trade-off between simple and sophisticated tools in the choice of textual analysis methods. Finally, we offer some guidance on how to report on text-based empirical work effectively to build credibility for impact. We conclude by suggesting some paths for future research.

The business media disseminates managers’ earnings guidance news more broadly or creates new information content on the guidance (Drake et al. 2014). We hypothesize that the media’s information dissemination encourages managers to continue issuing earnings guidance because their intended messages in earnings guidance can be transmitted more broadly and free of cost to managers. We also hypothesize that the media’s information creation induces managers to stop issuing earnings guidance because the media might interfere with managers’ intended guidance outcomes. Using a comprehensive dataset of media articles covering management guidance from 2007 to 2012, we find evidence supporting these hypotheses. Additionally, we find evidence that the negative impact of the media’s information creation on future guidance issuance is amplified with bad news guidance and attenuated with high analyst following. Our findings provide fresh insight into the information flow in financial markets by documenting not only positive but also negative influences that the media, through its dual roles, exerts on managers’ guidance behavior.

We argue that managers strategically issue earnings warnings based on whether evidence exists to verify misstatements in their prior disclosures. Using securities class action lawsuits from 1996 to 2019 pertaining to forward-looking earnings disclosures, we find that earnings warnings are positively associated with the verifiability of misstatements in prior forward-looking earnings disclosures. The results survive with entropy balancing and firm-fixed effects to mitigate endogeneity concerns. The positive relation between earnings warnings and misstatement verifiability is more pronounced for firms 1) with a general counsel in the top management team and 2) that face higher ex-ante litigation risk. The positive relation is weaker for firms whose managers engaged in insider trading during the class action lawsuit period. We also show that earnings warnings reduce the likelihood of a lawsuit settlement—as opposed to dismissal—only when the lawsuit involves misstatements that are more (rather than less) verifiable. Taken together, our findings suggest that managers strategically issue earnings warnings only when it helps to reduce litigation cost.

This paper examines how general counsels (GCs) in top management influence the use and timing of changes in accounting estimates (CAEs), an important but underexplored area in financial reporting. We provide evidence that when pre-CAE earnings narrowly miss analyst forecasts, firms with a GC in top management (GC firms) are more likely than non-GC firms to use income-increasing CAEs to meet or beat earnings expectations. These results are more pronounced for firms with weak integrity cultures, low analyst following, and non-industry-specialist auditors. We also find that in GC firms, post-CAE earnings are less predictive of future cash flows, suggesting a decline in earnings informativeness. Overall, our findings are consistent with GCs shaping the exercise of reporting discretion within GAAP rather than solely serving a gatekeeping function, highlighting how legal expertise at the executive level influences both the credibility and informativeness of reported earnings.