Lagged response or omitted predictor: The case of government popularity

By means of a simulation experiment, it is shown that the omission of a slow-changing predictor in a dynamic model often gives rise to the false inference of lagged response. Further, the misspecified dynamics tend to pass unnoticed when applying standard diagnostic tests. Additional tests and an alternative estimation procedure are suggested as remedies. To illustrate how the problem may manifest itself in empirical research, a study reporting strongly lagged effects of unemployment and inflation on government popularity is examined. After applying the suggested tests and estimation procedure, this result is called into question. There are no indications that a lag structure would exist. It is also shown how the omission of a relevant predictor might have produced the allegedly false finding.