Loss Function Assumptions in Rational Expectations Tests on Financial Analysts Earnings Forecasts
ABSTRACT The inefficiency of the Financial Analysts in using the public information whilst making their Earnings Forecasts has been documented by numerous previous researches. The previous researches assume that analysts aim to minimize their mean squared forecast errors i.e. analyst face a quadrati...
| Main Author: | |
|---|---|
| Format: | Dissertation (University of Nottingham only) |
| Language: | English |
| Published: |
2008
|
| Subjects: | |
| Online Access: | https://eprints.nottingham.ac.uk/22344/ |
| Summary: | ABSTRACT
The inefficiency of the Financial Analysts in using the public information whilst making their Earnings Forecasts has been documented by numerous previous researches. The previous researches assume that analysts aim to minimize their mean squared forecast errors i.e. analyst face a quadratic loss function. These researches thereby use the
Ordinary Least Squares regression tests to test the efficiency of the analysts. On the contrary, I assume that the analysts face a linear loss function in that they seek to minimize their absolute forecast errors. I use the Least Absolute Deviations (LAD) regression-based tests in my empirical studies. The focal point of my empirical study is to conduct and contrast the Rational Expectations Hypothesis under the assumption of these two different loss functions. My findings are similar to the earlier researches under the quadratic loss function assumption but offer no evidence of inefficiency on part of financial analysts under the linear loss function. The LAD regression-based test shows that financial analysts are rational in their expectations . |
|---|