# The research steps and methods of econometrics determine the relationship between variables and mathematics-model setting; Analyze the specific quantitative relationship between variables-estimate parameters; Conclusion The reliability of model test; Economic analysis and prediction-model application
# What is the difficulty in estimating the distributed lag model? A. degrees of freedom. If the degree of freedom is excessively lost, the estimation deviation will increase and the significance test will be invalid. B. Multiple * * * linear problems. There are usually many * * * linearities in lag variables. C. the lag length is difficult to determine.
# Tool variable method 1. It is highly related to the explanatory variable it replaces. 2. It has nothing to do with the random disturbance term. 3. It is not related to other explanatory variables to avoid multiple * * * linearity.
# The basic concept of virtual variables Virtual variables are artificially constructed variables, and the values of 0 and 1 are the representatives of attribute variables.
# The difference between simultaneous equation models A. The simultaneous equation model consists of several single equations. There is more than one variable to explain. B there are stochastic equations and deterministic equations in the model, but the stochastic equations must be included. C. There is not only one-way causality between the explained variables and the explained variables, but also mutual causality. Explanatory variables may be related to random disturbance terms.
# Incomplete multiple * * * Linear consequences:
The variance of parameter estimator increases.
2. When estimating the parameter interval, the confidence interval is often large.
3. In serious cases, hypothesis testing is easy to make wrong judgments.
4. In severe cases, r2 may be large, and F test is significant, but T test may not be significant, leading to wrong conclusions.
1. Simple correlation coefficient test 2. Variance expansion factor method 3. Intuitive judgment, such as large standard deviation of regression coefficient or deviation from economic theory 4. Stepwise regression method.
# Autocorrelation: Inertia of economic system. Lagging effect of economic activities. Correlation caused by data processing. Spider web phenomenon. Model setting error. Zero mean, underestimating the variance of parameter estimation, the influence on the model prediction, overestimating the unreliability of T, F and r2, the influence on the model, and the prediction accuracy is reduced.
# Heteroscedasticity: Some important explanatory variables are omitted from the model. Wrong model and specification. Variation of measurement error. Differences in overall units in cross-sectional data. Unbiased, consistent, invalid, exaggerating the statistical significance of the estimated parameters, affecting the prediction, and Y's prediction is invalid.