Simulation of Controlled Financial Statements

Many accounting and finance studies investigate the time-series properties of historical accounting records from corporate financial statements. Some of them have recognized the potential benefits of using disaggregated monthly accounting records. Disaggregated data are beneficial because one can use more data points within a relatively short period of time, thus reducing the chance of structural change. The added data points and reduction of the number of variables needed to accommodate potential structural changes can enhance the statistical power of any subsequent analysis. The use of disaggregated data may also improve the predictive ability of time-series analytic approaches. In order to systematically assess various financial indicators and investigate the effects of different organizational characteristics, a large number of monthly statements with certain predetermined characteristics are desirable. However, such statements are not readily available. At best, monthly statements can be obtained from a few volunteer companies. Under this circumstance, simulation of controlled financial statements seems to be a reasonable solution. This research explores a methodology for simulating complete monthly financial statements based on actual company quarterly financial statements. The methodology incorporates the interrelationships among accounting numbers and the effects of exogenous variables. To test the empirical validity and whether the monthly results derived from the quarterly data can accurately track the real monthly figures, we compare the results simulated by the proposed method and those generated by a naive random walk model. We test both complete financial statements for three companies and sales statistics from the retail industry. The results of both tests demonstrate the superiority of the method proposed by this study over a naive random walk model. The proposed simulation method provides an opportunity for researchers to examine the time-series properties of financial statement elements by using the monthly data of a large number of companies. In addition, the simulation approach allows researchers to perform cross sectional comparisons on companies with different characteristics (e.g., sales behavior patterns and degrees of stability) in their financial and economic activities. Moreover, it enables the researchers to manipulate some of these characteristics to test various hypotheses.