Purpose
The purpose of this paper is to revisit the question of whether analysts anticipate accruals’ predicted reversals (or persistence) of future earnings. Prior evidence documents that analysts who provide information to investors are over optimistic about firms with high working capital (WC) accruals. The authors propose that empirical models using WC accruals alone may be incomplete and hence not entirely appropriate to assess the level of analysts’ understanding of accruals. The authors argue that analysts’ optimism about WC accruals might not be due to their lack of sophistication, but rather driven by incomplete accrual information embedded in forecast accuracy tests.
Design/methodology/approach
The authors use non-financial US firms for the period between 1976 and 2013. The authors define earnings forecast errors as the analysts’ consensus earnings forecasts minus the actual earnings provided by IBES deflated by share price from CRSP. The authors carry out forecast error regressions on individual accrual components by decomposing total accruals into categories. The authors perform the tests across 12 months starting from the initial analysts’ forecasts, which are generally issued in the first month after the prior period earnings announcement date. The final sample contains 48,142 firm–year observations per month.
Findings
The empirical tests show no correlation between analysts’ forecast errors and revised total accruals. The findings are robust to different samples, periods, model specifications, decile ranked accruals, high accruals, absolute forecast errors, controlling for cash flows (CF) and high accounting conservatism. The findings imply that if analysts are to achieve more accurate forecasts, they should be considering all rather than some accrual components. The authors interpret this evidence as an indication of analysts’ relative sophistication with respect to accruals.
Research limitations/implications
The authors recognise that analysts’ correct anticipation of accruals’ persistence does not mean that their earnings forecasts are entirely free of bias. Analysts can make forecast errors for various reasons including strategic biases. For instance, the tests show pessimistic forecast errors with respect to CF, which is in line with similar findings in prior research (Drake and Myers, 2011). Hence, the authors suggest that future research examine this correlation in greater depth as CF components are with the highest level of persistence, and hence should be predicted most accurately.
Practical implications
The results imply that the argument about analysts’ lack of sophistication with respect to accruals’ persistence is not warranted. The results imply that forecasts appear to contribute to market efficiency. Another implication is that analysts seem to utilise all relevant accrual information in their forecasts, hence traditional accrual definition should be revised in future studies. Key inferences of the paper imply that the growing use of analysts’ reports by institutional investors and money managers in their decision-making processes is justified despite the debate in the prior literature on the role and the reputation of analysts as surrogates of market expectations.
Originality/value
The research sheds a new light on the question whether sell-side security analysts are able to anticipate the persistence of accruals in future earnings.
Earnings Predictability And Broker-Analysts’ Earnings Forecast Bias
