Abstract
This study examines the association between the political corruption of a local government and the readability of firms’ annual reports. Based on a sample of 12,742 firm-year observations during the 2006–2014 period, the study reveals that firms located in more corrupt regions tend to disclose less readable financial reports. Our additional analyses reveal that the level of annual report readability is lower for firms located in more corrupt regions, regardless of the firms’ level of return on assets. We also find that firms located in more corrupt regions and having more able managers are more likely to obfuscate information in annual reports. The results imply firms’ effort to minimise rent extraction from corrupt government officials. A further test shows that firms in more corrupt regions are more likely to report less readable Management Discussion and Analysis (MD&A) section of annual reports. This paper extends the prior literature on annual report readability and political corruption. The paper also provides additional evidence to the mixed results on the management's obfuscation behaviour related to the readability of financial disclosures. The findings may be of interest to regulators seeking out factors influencing firms’ readability of annual reports.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Glaeser and Saks (Citation2006) stated that more than 10,000 governmental officials involved in such corruption convictions as a conflict of interest, fraud, campaign-finance violations, and obstruction of justice during the 1990–2002 period.
2 We follow the same method used by Smith (Citation2016) to match a firm headquartered in a non-corrupt region to a firm headquartered in a corrupt region. That is, we convert ZIP code fields from Compustat and SEC filing to their respective Federal Information Processing Standard codes and then manually match each of these codes to a federal judicial district. All convictions are matched to the contemporaneous financial data.
3 We start our sample from 2006 because our control variables at the state level and county level are not fully disclosed by the US Census Bureau until the year 2006.
4 65.3% of the sample firms in Lo et al. (Citation2017) are incorporated in Delaware. Therefore, the mean of DLW in our sample is consistent and comparable to prior literature.
5 Prior literature suggests that propensity score matching avoids the model specification problems in Heckman selections models (e.g. Rosenbaum and Rubin Citation1983, Heckman et al. Citation1997).