Rational Expectations and Survey Data
Eng, Yoke Kee (2002) Rational Expectations and Survey Data. Masters thesis, Universiti Putra Malaysia.
How economic agents form their expectations of future economic events has been an importance issue in macroeconomics for many years. Indeed, business firm's expectations has played a central role in the business cycle theories of both Pigou (1927) and Keynes (1936). Acknowledging that the behavioral assumption of the rational expectation is important to modem economic theory and econometric modeling, this study is undertaken for the purpose of investigating whether the Malaysian Business Expectation Survey For Limited Companies, provides the basis for prediction which, satisfy the rational expectations hypothesis (REH) and property in the sense of Muth (1961). The business survey-based expectations, drawn from Business Expectations Survey of Limited Companies (BESLC), conducted biannually by the Malaysia' s Department of Statistic, offer a unique opportunity to accumulate empirical evidence on expectation formation and decision-making at micro level. Four criteria of 'rationality' is examined in the study namely, unbiasedness, serial correlation of forecasts error, weak-form efficiency and orthogonality. Essentially, this study utilizes business survey data in a manner that is different from prior study by testing the rationality of firm's expectations at different level of aggregation. Accordingly, the sectoral subdivisions are as follows: the aggregated respondents of the BESLC survey data are group into three divisions of significant sectoral in Malaysia: that is, primary sector, industrials sector and service sector. At an even higher disaggregated level, the manufacturing sub-sector under the industrial sector, which can be further segmented into consumer goods industry, capital goods industry, as well as light and heavy intermediate goods industry. Evidently, the significance of the use of disaggregated data is noted in this study. Apparently, REH are rejected comprehensively when directs test were performed on the sectoral segmentation level. At a higher disaggregated level, as the direct tests are applied to the data from most of the constituent industries of manufacturing sectors, these test provide at least some amount of direct evidence in favor of the hypothesis, which is often simply assume to be valid, that expectations are rational as defined by Muth (1961). Hence, this implies that the prior investigations of the rationality of survey expectational data have overlooked relevant data by disregarding the potential of aggregation bias encompassed by the aggregated survey data. Although the presence of the aggregation bias has not been formally tested, the results of this study here suggest that this potential bias may be of importance.
Repository Staff Only: Edit item detail