The post is about Remedial Measures of Heteroscedasticity.
Heteroscedasticity is a condition in which the variance of the residual term, or error term, in a regression model, varies widely.
The heteroscedasticity does not destroy the unbiasedness and consistency properties of the OLS estimator (as OLS estimators remain unbiased and consistent in the presence of heteroscedasticity), but they are no longer efficient, not even asymptotically. The lack of efficiency makes the usual hypothesis testing procedure dubious (مشکوک، غیر معتبر). Therefore, there should be some remedial measures for heteroscedasticity.
Remedial Measures of Heteroscedasticity
For remedial measures of heteroscedasticity, there are two approaches: (i) when $\sigma_i^2$ is known, and (ii) when $\sigma_i^2$ is unknown.
(i) $\sigma_i^2$ is known
Consider the simple linear regression model $Y_i=\alpha + \beta X_i + u_i$.
If $V(u_i)=\sigma_i^2$ then heteroscedasticity is present. Given the values of $\sigma_i^2$, heteroscedasticity can be corrected by using weighted least squares (WLS) as a special case of Generalized Least Squares (GLS). Weighted least squares is the OLS method of estimation applied to the transformed model.
When heteroscedasticity is detected by any appropriate statistical test, then the appropriate solution is to transform the original model in such a way that the transformed disturbance term has a constant variance. The transformed model reduces the adjustment of the original data. The transformed error term $u_i$ has a constant variance i.e. homoscedastic. Mathematically
\begin{eqnarray*}
V(u_i^*)&=&V\left(\frac{u_i}{\sigma_i}\right)\\
&=&\frac{1}{\sigma_i^2}Var(u_i)\\
&=&\frac{1}{\sigma_i^2}\sigma_i^2=1
\end{eqnarray*}
This approach has limited use as the individual error variances are not always known a priori. In case of significant sample information, reasonable guesses of the true error variances can be made and be used for $\sigma_i^2$.
Let us discuss the second remedy of heteroscedasticity from remedial measures of heteroscedasticity.
(ii) $\sigma_i^2$ is unknown
If $\sigma_i^2$ is not known a priori, then heteroscedasticity is corrected by hypothesizing a relationship between the error variance and one of the explanatory variables. There can be several versions of the hypothesized relationship. Suppose the hypothesized relationship is $Var(u)=\sigma^2 X_i^2$ (error variance is proportional to $X_i^2$). For this hypothesized relation we will use the following transformation to correct for heteroscedasticity for the following simple linear regression model $Y_i =\alpha + \beta X_i +u_i$.
\begin{eqnarray*}
\frac{Y_i}{X_i}&=&\frac{\alpha}{X_i}+\beta+\frac{u_i}{X_i}\\
\Rightarrow \quad Y_i^*&=&\beta +\alpha_i^*+u_i^*\\
\mbox{where } Y_i^*&=&\frac{Y_i}{X_i}, \alpha_I^*=\frac{1}{X_i} \mbox{and } u_i^*=\frac{u}{X_i}
\end{eqnarray*}
Now the OLS estimation of the above transformed model will yield the efficient parameter estimates as $u_i^*$’s have constant variance. i.e.
\begin{eqnarray*}
V(u_i^*)&=&V(\frac{u_i}{X_i})\\
&=&\frac{1}{X_i^2} V(u_i^2)\\
&=&\frac{1}{X_i^2}\sigma^2X_i^2\\
&=&\sigma^2=\mbox{ Constant}
\end{eqnarray*}
For remedial measures of heteroscedasticity, some other hypothesized relations are:
- Error variance is proportional to $X_i$ (Square root transformation) i.e $E(u_i^2)=\sigma^2X_i$
The transformed model is
\[\frac{Y_i}{\sqrt{X_i}}=\frac{\alpha}{\sqrt{X_i}}+\beta\sqrt{X_i}+\frac{u_i}{\sqrt{X_i}}\]
It (the transformed model) has no intercept term. Therefore we have to use the regression through the origin model to estimate $\alpha$ and $\beta$. To get the original model, multiply $\sqrt{X_i}$ with the transformed model. - Error Variance is proportional to the square of the mean value of $Y$. i.e. $E(u_i^2)=\sigma^2[E(Y_i)]^2$
Here the variance of $u_i$ is proportional to the square of the expected value of $Y$, and $E(Y_i)$ = \alpha + \beta X_i$.
The transformed model will be
\[\frac{Y_i}{E(Y_i)}=\frac{\alpha}{E(Y_i)}+\beta\frac{X_i}{E(Y_i)}+\frac{u_i}{E(Y_i)}\]
This transformation is not appropriate because $E(Y_i)$ depends upon $\alpha$ and $\beta$ which are unknown parameters. $\hat{Y_i}=\hat{\alpha}+\hat{\beta}$ is an estimator of $E(Y_i)$, so we will proceed in two steps:
- We run the usual OLS regression dis-regarding the heteroscedasticity problem and obtain $\hat{Y_i}$
- We will transform the model by using estimated $\hat{Y_i}$ i.e. $\frac{Y_i}{\hat{Y_i}}=\alpha\frac{1}{\hat{Y_i}}+\beta_1\frac{X_i}{\hat{Y_i}}+\frac{u_i}{\hat{Y_i}}$ and run the regression on transformed model.
This transformation will perform satisfactory results only if the sample size is reasonably large.
- Log transformation such as $ln\, Y_i = \alpha + \beta\, ln\, X_i + u_i$.
Log transformation compresses the scales in which the variables are measured. However, this transformation is not applicable in some of the $Y$ and $X$ values that are zero or negative.
Visit: R Language Frequently Asked Questions
