GRAMMATICAL VIOLATION AND LEGIBILITY OF FINANCIAL STATEMENTS IN BRAZIL
Name: PAULO HENRIQUE AMARAL RODY
Publication date: 22/09/2023
Examining board:
Name |
Role |
|---|---|
| JOSE ELIAS FERES DE ALMEIDA | Advisor |
Summary: The objective of this thesis is to analyze the association between the degree of readability and
the level of grammatical violation of financial statements, considering the effects of
grammatical violations on the textual quality of information from publicly traded companies
listed on Brasil Bolsa Balcão - [B]³. The literature evidence lacks complementation regarding
the textual and narrative part of explanatory notes and management reports, respectively, when
compared with accounting numbers, considering that both documents are integral parts of the
financial statements. In this thesis, the Grammatical Violation proxy is presented as a proxy
capable of indirectly measuring the quality of accounting information, within an environment
of information asymmetry and conflicts of interest between companies and investors.
Grammatical Violation is defined as the presence of violations of grammatical norms, such as
errors in punctuation, spelling, agreement and other writing errors in a text. In this thesis, the
Grammatical Violation proxy was measured automatically via Python, both for document
collection and for processing to generate variables. A low level of Grammatical Violation
suggests a high level of quality of the financial statements and this can reduce the risks of
misinterpretations by users, which can increase investor confidence in the capital allocation
process. The Readability proxy is measured and applied to the same documents and also
analyzed via Python. The sample comprises non-financial companies listed in [B]³, from the
period 2010 to 2021 (twelve years). According to the numerical properties of the dependent
variables, four econometric regression models were estimated using different estimators to
increase the robustness of the results. The first model tested the relationship between
Readability and Grammatical Violation, the second model tested the relationship between
Grammatical Violation and Shareholder Return, the third model analyzed the relationship
between Grammatical Violation and Republication of financial statements and the fourth model
evaluated the relationship between Grammatical Violation and Results Management. The main
results found stand out: the higher the level of Readability, on average, the lower the degree of
Grammatical Violation of the companies; the larger the size of the company, on average, the
lower the degree of Grammatical Violation of the financial statements; the greater the debt, on
average, the greater the degree of Grammatical Violation of the companies' financial
statements; companies audited by Big Four have a lower degree of Grammatical Violation of
financial statements; the higher the level of Grammatical Violation, on average, the lower the
companies' Share Return; companies with a higher level of Grammatical Violation are less
likely to republish financial statements; and the higher the level of Grammatical Violation, on
average, the higher the companies' level of Earnings Management. This thesis contributes to
the literature as one of the few Brazilian studies in the area to use text processing and automation
tools to interpret language patterns within the accounting environment. It also contributes by
signaling that the level of Grammatical Violation of the financial statements can be used as an
indicator of reliable accounting information and credibility, due to the reduction of
informational asymmetry, capable of being obtained through this proxy.
