Measurement Errors and Tests for Rationality
指出变量误差会影响理性预期检验，使用利率、股价和汇率调查数据，通过多源预期修正误差后仍拒绝理性假设。
The traditional tests for rationality, the regression and volatility tests, have often rejected the hypothesis of rationality for survey data on expectations. It has been argued that these tests are not valid in the presence of unit roots and hence cointegration tests should be applied. The cointegration tests have often failed to reject the hypothesis of rationality. The present article argues that errors in variables affect tests of rationality. We use multiple sources of expectations to correct for the errors-in-variables bias but find that the hypothesis of rationality is rejected even after this correction. The article uses survey data on interest rates, stock prices, and exchange rates.