Resumo:
Propsito do Trabalho:
Our work aims to assess if different disclosure practices of numerical information (text, table, column graph and line graph) affect the financial analysts perception. Incidentally, we also investigate if personal characteristics (overconfidence and gender) also affect their perception ability.
The study of graph interpretation ability by accountants and financial analysts is relevant in the accounting context because most modern annual reports contain graphs. For example, the IASB recently amended IFRS 7 Financial Instruments: Disclosures requiring that if the quantitative data disclosed as at the end of the reporting period are unrepresentative of an entitys exposure to risk during the period, an entity shall provide further information that is representative (IFRS 7.35). Whether this is the case, the implementation guidance exhorts the presentation of graphs: [] if an entity typically has a large exposure to a particular currency, but at year-end unwinds the position, the entity might disclose a graph that shows the exposure at various times during the period [] (IFRS 7.IG20).
Additionally, impression management literature suggests that graphs are much vulnerable to manipulation because graphs are non-audited and not prescribed (Jones, 2011).
Indeed, because standard-setters are exhorting entities to present graphs in the notes, the use of graphs in financial reporting might increase significantly. Therefore, the knowledge of individuals ability to interpret graphs and the impact of their personal characteristics in performing such a task becomes very important, either to prepares and auditors of financial reports, standard-setters and accounting professors; and may enhance impression management literature in many venues.
Base da plataforma terica:
Different types of accounting information disclosure by companies change decision-making of both financial analysts (Ghosh & Wu, 2012) and investors (Dilla, Janvrin, & Jeffrey, 2013). It also affects the confidence interval of forecasts (Lawrence & OConnor, 1993).
In an experiment with undergraduate students, Beattie & Jones (2002) showed that students perceived a company whose graphic had received a measurement distortion as better than the same company if such graphs had not being distorted.
Different types of presenting financial information also influence the accuracy of accounting tasks (Davis, 1989; Tang et al., 2014). Both researches achieve with experimental empirical evidence the same conclusion by Vessey (1991): different types of information disclosure have different effects in the understanding performance in different situations. However, Davis study was performed with a small sample size (30) of MBA students and the paper by Tang et al. also uses students, but with a larger sample (157). Our study attempts to provide replication with a sample comprised by financial analysts, instead of students. A second advancement is that no study used textual disclosure as a basement measure to provide a comparison between the different types of graphical information disclosure.
H1: Different forms of information representation (text, table, column graph and line graph) differently affect the informational perception and the decision-making of financial analysts.
This study also attempts to investigate if intrinsic characteristics of the individual alter the interpretation of the different ways of disclosing information. Previous research (Cardoso, Barcellos, Aquino, & Sales, 2014) provided evidence that intrinsic characteristics of an accountant can alter their decision-making process. This study will focus on overconfidence trait, which can be divided in three subtypes: overestimation, over-placement and over-precision (Mannes & Moore, 2013). In this study, we use the over-placement sub dimension, that can be described as the over-placement of ones performance relative to others (Moore & Healy, 2008). The overconfidence trap can lead to errors in judgment and, in turn, bad decisions (Hammond et al., 2006).
There is empirical evidence that overconfident individuals will commit more errors and that gender has a role in this intrinsic characteristic: males are more overconfident (and commit more errors) than females. Economics undergraduate male students are more overconfident than females (Bengtsson, Persson, & Willenhag, 2005) and male traders trade with more overconfidence than females (Barber & Odean, 2001). A cross-cultural study in the US, Germany, Italy and Thailand confirmed that women financial analysts were more risk averse than male financial analysts (Beckmann & Menkhoff, 2008). Therefore, we postulate two additional hypotheses:
H2: Overconfident financial analysts commit more errors than non-overconfident ones.
H3: Male financial analysts are more overconfident than females.
Mtodo de investigao:
To test those hypotheses we applied an online survey experiment through Surveymonkey. It can be a way to improve both internal validity and external validity (Brandon, Long, Loraas, Mueller-Phillips, & Vansant, 2014), since it is a randomized experiment with professional financial analysts that actually make decisions in their day-to-day basis.
We collected data via an electronic questionnaire applied by the Brazilian Accounting Association (BAA, CFC). The electronic message containing the web link to the questionnaire was sent to all Brazilian accountants regularly registered with the association at August 2012. Professionals were exhorted to access the BAAs webpage to answer the survey. Based on respondents expertise, they were required to answer a determined batch of questions; actually, those that presented themselves as financial analysts were required to answers questions related to graph interpretation. In total, 295 professional accountants which main duties are related to financial analysis comprise the sample for this research. This sample was randomly classified among four subsamples to each we presented the same informational content concerning the amount of people going in and out of a store during a 12 minutes time period.
Hence, it is a 42 mixed-design experiment (four types of information presentation: textual information, table, line graph and column graph between subjects, and two questions: in which minute is there the highest amount of people entering the store, and in which minute is there the highest amount of people exiting the store within subjects). The subjects were randomly assigned to one of the four types of information presentation with a probability of .40 of being assigned to the line graph and a probability of .60 of being assigned to one of the other 3 conditions.
After the experiment, under the same electronic questionnaire, respondents were asked if they thought that they would be in the 10% group that responded right both questions. Those that answered yes were coded as overconfident, since overconfidence can be defined as the over-placement of ones performance relative to others (Mannes & Moore, 2013; Moore & Healy, 2008).
Resultados, concluses e suas implicaes:
First, the two questions asked to the respondents were grouped in one variable that could assume the value of 0 (no correct answer), 1 (one correct answer) or 2 (both questions correctly answered), for the results to be estimated in one statistical test. A chi-square model was estimated on the effect of the different ways of information disclosure on the performance in getting the right answers.
The result from the chi-squared test shows that different types of information disclosure affect the perception and decision-making of financial analysts, providing support for H1. However, the chi-squared test cannot test if a specific type of information presentation is better than another. Thus, we estimated an ordered logit model. Supporting H2 the coefficient for the overconfidence was negative and significant, showing that overconfident financial analysts tend to commit more errors than non-overconfident ones. The logistic model also shows that the column graph and the line graph do improve the financial analysts perception when compared to textual disclosure, but table disclosure does not have a significant result.
Aiming to perform a robustness check, we identified that the type of information disclosure did not influence overconfidence, showing evidence that different types of disclosure with similar difficulty does not impact in the overconfidence of financial analysts.
To test H3, a chi-squared test was performed with the distribution of males and females in the overconfidence distribution, and the result was non-significant. This may be an effect of self-selection. Maybe only the overconfident females choose to be a financial analyst, or this profession changed their overconfidence, making both genders homogeneous in this trait.
Thus we found evidence for H1 (different forms of information representation (text, table, column graph and line graph) differently affect the informational perception and the decision-making of financial analysts) and H2 (overconfident financial analysts commit more errors than non-overconfident ones), but not for H3 (male financial analysts are more overconfident than females).
There are multiple further implications of these results for practitioners and researchers. Prepares and auditors of financial reports should identify which type of numerical disclosure better and faithfully present what they purport to represent; for example: text, table, column graph or line graph. Additionally, auditing firms might develop auditors ability to audit graphs; i.e., not only if data presented in graphs are reliable, moreover, they should develop auditors ability to assure if the graph type chosen by companies are the best appropriate for disclosed data interpretation by users.
Investment banks and rating agencies might assess personal traits of their financial analysts, such as overconfidence. As presented in literature and reinforced in this paper, overconfident individuals commit more errors than non-overconfident ones. Indeed, investment banks and rating agencies could also develop financial analysts ability to detect impression management and other graph distortions, and judge if the graph type chosen by companies are the best appropriate for disclosed data interpretation.
Accounting professors and lectures should also develop such ability in undergraduate students and in professional education programs.
Finally, our results suggest further researches in the following venues. We only analyzed one personal trait: overconfidence. However, other traps may exist, such as self-consistency (persisting in a wrong decision only because one cannot admit that his/her decision was wrong) and confirming-evidence seekers (one only seeks evidence that confirms his/her perception and beliefs). Both traps can harm the perception and decision-making of financial analysts regarding graphical information. Additionally, companies are increasing the usage of videos and webcasts to present their reports. Hence, the impact of disclosure medium (e.g., printed versus video) on graphical perception and interpretation was not investigated so far.
Referncias bibliogrficas:
Barber, B. M., & Odean, T. (2001). Boys will be boys: gender, overconfidence and common stock investment. The Quarterly Journal of Economics, 116(1), 261292.
Beattie, V. & Jones, M. (1993). Information design and manipulation: financial graphs in corporate annual reports. Information Design Journal, 7(3), 211-226.
Beattie, V. & Jones, M. (2000). Changing graph use in corporate annual reports: a time series analysis. Contemporary Accounting Research,17(2), 23-26.
Beattie, V. & Jones, M. J. (2002). Measurement distortion of graphs in corporate reports: an experimental study. Accounting, Auditing & Accountability Journal, 15(4), 546564.
Beckmann, D. & Menkhoff, L. (2008). Will Women Be Women? Analyzing the Gender Difference among Financial Experts. Kyklos, 61: 364384.
Bengtsson, C., Persson, M., & Willenhag, P. (2005). Gender and overconfidence. Economics Letters, 86(2), 199203.
Brandon, D. M., Long, J. H., Loraas, T. M., Mueller-Phillips, J., & Vansant, B. (2014). Online instrument delivery and participant recruitment services: Emerging opportunities for behavioral accounting research. Behavioral Research in Accounting, 26(1), 123.
Cardoso, R., Barcellos, L., Aquino, A., & Sales, P. (2014). An Assessment of Professional Accountants Cognitive Reflection Ability. Social Science Research Network. Retrieved from http://ssrn.com/abstract=2423730
Davis, L. R. (1989). Report format and the decision makers task: An experimental investigation. Accounting, Organizations and Society, 14(5), 495508.
Desanctis, G., & Jarvenpaa, S. L. (1989). Graphical presentation of accounting data for financial forecasting: An experimental investigation. Accounting, Organizations and Society, 14(5), 509525.
Dilla, W. N., Janvrin, D. J., & Jeffrey, C. (2013). The Impact of Graphical Displays of Pro Forma Earnings Information on Professional and Nonprofessional Investors Earnings Judgments. Behavioral Research in Accounting, 25(1), 3760.
Ghosh, D., & Wu, A. (2012). The Effect of Positive and Negative Financial and Nonfinancial Performance Measures on Analysts Recommendations. Behavioral Research in Accounting, 24(2), 4764.
Hammond, J. S., Keeney, R. L., & Raiffa, H. (2006). The hidden traps in decision making. Harvard Business Review, 1998(Sptember-December 1998), 210.
Jones, M. (2011). Creative accounting, fraud and international accounting scandals. New Jersey: Wiley. ISBN: 978-0-470-05765-0
Lawrence, M. & OConnor, M. (1993). Scale, variability, and the calibration of judgmental prediction intervals. Organizational Behavior and Human Decision Processes, 56(3), 441458.
Mannes, A. E., & Moore, D. a. (2013). A behavioral demonstration of overconfidence in judgment. Psychological Science, 24(May), 11907.
Mannes, A., & Moore, D. (2013). I know im right! A behavioural view of overconfidence. Significance, 10(4), 1014.
Moore, D. a, & Healy, P. J. (2008). The trouble with overconfidence. Psychological Review, 115(2), 502517.
Moriarty, S. (1979). Communicating Financial Information Through Multidimensional Graphics. Journal of Accounting Research, 17(1), 205224.
Tang, F., Hess, T. J., Valacich, J. S., & Sweeney, J. T. (2014). The effects of visualization and interactivity on calibration in financial decision-making. Behavioral Research in Accounting, 26(1), 2558.
Vessey, I. (1991). Cognitive Fit: A Theory-Based Analysis of the Graphs Versus Tables Literature. Decision Sciences, 22(2), 219240.
|
