Overcoming Measurement Error Problems in the use of Survey Data on Expectations

Survey data frequently requires conversion from qualitative responses to quantitative series. A commonly cited criticism of the use of the survey data is that the conversion procedures incorporate measurement errors which render the series unusable. In this paper, we provide a novel contribution to the literature by paying attention to the nature of, and the treatment of, the measurement error that arises when such a conversion takes place. Such an error is frequently ignored or assumed to be orthogonal to known information. The contributions of this paper are three-fold. Firstly, simulation experiments show that the conversion error is systematic in it's relation to known information and show that need for the adequate treatment of conversion errors is essential. Secondly, this paper formally shows that the procedure proposed by Lee (1994), which employs a relatively weak assumption, is extremely successful in circumventing the problems. Finally, in contrast to previous findings, these results provide strong support for the view that expectations are formed rationally in the case of Australian total manufacturing output