Rational expectations tests on financial analysts’ cash flow forecasts

Purpose - This study investigates whether financial analysts process information efficiently when they make cash flow forecasts. Design/methodology/approach - Using a sample of 3,967 observations spanning 2004–2016, we perform empirical analyses by regressing actual cash flows on previous cash flow forecasts for the current period and a vector of information variables known to the analysts at the time of the forecasts. Findings - We find that analysts do not incorporate past stock returns or past cash flow information efficiently when they generate cash flow forecasts. We also find weak evidence that analysts do not incorporate past consensus cash flow forecasts or past accruals information when they generate cash flow forecasts. Practical implications - Our findings contribute to the analyst forecast efficiency literature and highlight the difference between analyst cash flow forecast efficiency and earnings forecast efficiency. Originality/value - While extant research on whether analysts use publicly available information efficiently when generating earnings forecasts documents mixed findings (e.g. DeBondt and Thaler, 1990; Abarbanell and Bernard, 1992; Basu and Markov, 2004; Evanset al., 2017), our results regarding analysts’ cash flow forecast efficiency are unambiguous.