A critical assessment of the time-series literature in accounting pertaining to quarterly accounting numbers

Abstract

This paper summarizes, critiques, and synthesizes the time-series literature in accounting pertaining to quarterly accounting numbers. It reviews work on quarterly earnings, quarterly balance sheet and income statement subcomponents, and quarterly cash-flows from operations (CFOs). Several salient findings emerge. First, the premier ARIMA models attributed to Foster (1977), Griffin (1977) and Brown and Rozeff (1979) were identified on relatively small samples dominated by “old economy” firms. It appears that the descriptive validity of these ARIMA structures must be called into question when analyzing more current databases replete with high-technology, regulated, and financial-service firms (Lorek & Willinger, 2007). Second, the use of ARIMA-based analytical review procedures in audit settings is not cost effective. Third, recent evidence (Lorek & Willinger, 2008, 2011) supports the univariate Brown & Rozeff (100) × (011) ARIMA model as the best statistically-based prediction model for quarterly CFO, a finding of considerable import to analysts, investors, and researchers.