2
$\begingroup$

For a paper I was using the micEconCES package to estimate the CES production function for a country at the aggregate. For a two-input function with capital and labour I used for the variables the Perpetual Inventory Method to construct aggregate capital, labour hours and GDP for output. I used methods provided in the package to do the estimation. But now I am kind of unsure, what I have to check the data for f.e. autocorrelation, multicollinearity etc. or do I even have to check for that?

$\endgroup$

1 Answer 1

2
$\begingroup$

You should always validate statistically your model, running various model diagnostics. It appears you work with time series (single country?).

Here, certainly autocorrelation may be an issue. After all, your series for capital is by construction an autoregressive equation with a forcing factor (investment).

Software packages usually warn for signs of multicollinearity.

Also, a serious issue could be that of endogeneity of regressors.

$\endgroup$
3
  • $\begingroup$ Yea I'm only examining the US with time series data for aggregate capital, labour and output (GDP). Would I have to run tests for all of these variables. Are there any must do tests (it's a paper at undergraduate level)? $\endgroup$
    – macro123
    Apr 20, 2018 at 19:27
  • $\begingroup$ @macro123 Autocorrelation for sure. The issue of endogeneity is not testable, but potentially serious. It is usually treated using Instrumental Variables. You should look up related literature. $\endgroup$ Apr 20, 2018 at 20:00
  • $\begingroup$ what do you mean by endogenity being serious in this case or why do you think its especially serious here? And couldn't I do a Hausman test for it? $\endgroup$
    – macro123
    Apr 21, 2018 at 17:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.