Page 187 - Theoretical and Practical Interpretation of Investment Attractiveness

investment volume, it is appropriate to use a fixed effects model. Fifth, the Hausman test 191
is used to select between fixed and random effects models . Under this test, ߠ ෠ the score is
ଵ
known ߠ ෠ to be consistent with the score, and ߠ ෠ the score is assumed to be effective. In the
ଶ
ଶ
case of investments, it is assumed that the parameters are calculated using random effects.
ߠ ෠ Then ߠ ෠ shows the vector of parameters estimated using the fixed effects model.
ଵ
ଶ
ܪ ǣ ߠ ෠ value ߠ is the effective value of the true parameter. If the null hypothesis is true,
ଶ
଴
ଶ
there is no systematic difference between the two estimates, and a random effects model is
used to estimate the regression parameters. If the null hypothesis is false, there is a systematic
difference between the two estimates and the hypothesis, that is, the efficiency of the
parameters estimated using random effects, is rejected and the parameters estimated using the
fixed effects model are considered reasonable. (effective) and this model is used. The
ଶ
Hausman statistic ߯ is distribution-based and is calculated as follows:
ିଵ
ᇱ
ܪൌ ሺߚ െߚ ሻ ሺܸ െܸ ሻ ሺߚ െߚ ሻ
௖
௖
௘
௖
௘
௘
In this case,
ߚ — vector of parameter coefficients (baseline estimate ), calculated using a fixed
௖
effects model
ߚ — vector of parameter coefficients (effective value ), calculated using a random
௘
effects model
ܸ - covariance matrix of the base estimate
௖
ܸ - effective covariance matrix of the estimate
௘
Due to the incorrect setting of the β coefficients calculated by the ECC method, due to
unobserved heterogeneity, it is not appropriate and effective.
In cases where the correlation coefficient between two regressors is high, even if there
is no multicollinearity problem, the high correlation coefficient will exaggerate the standard
errors of those regressors when estimating the econometric model. When selecting variables
for the model, the variable with the highest correlation is selected. For this reason, of the 13
variables analyzed in the model, 7 were selected as random variables. Our selection was based
on the information presented in the table below.

191 Houseman, J. A. 1978. Specification tests in econometrics. Econometrica 46:1251–1271.

184