0
$\begingroup$

I am reading this VAR model paper https://www.rieti.go.jp/en/publications/summary/15030006.html which does not report the regression results of the VAR model, but just IRFs and historical variance decomposition. This is just one of many such papers. Why is that the case? I agree IRFs are important for finding indirect effects. But why significance of lag variables are not reported in most VAR papers.

$\endgroup$
8
  • $\begingroup$ well this is not really about economics. There is always a lot of important stuff that goes into any research which ends up not being reported. Journals will often have strict maximum page limits of 30-40-50 pages. In addition, people prefer to read shorter papers so in most academic writing courses you will be advised to always think carefully about what needs to be in the paper and what not. For example, in standard OLS it is important to check for heteroskedasticity and people do that but wont report results of the test usually and just mention errors were adjusted. $\endgroup$
    – 1muflon1
    Nov 13, 2020 at 21:48
  • $\begingroup$ Similarly, while it might be important to check significance of lags in your VAR model this will often be omitted from. Not because it is unimportant but because that additional table competes for scarce space with other text, tables and graphs as well as readers attention. So it is just about economizing on the space in your paper. $\endgroup$
    – 1muflon1
    Nov 13, 2020 at 21:48
  • $\begingroup$ but how can we know their lag variables were significant? Or that is not really important and that is why it is ignored. I mean, in economizing, we ignore the less important stuff, right? $\endgroup$ Nov 14, 2020 at 0:14
  • $\begingroup$ you would have to ask them for those extra results. It is normal to message researchers to ask if these or other tests were done and if so what is the result. Also the issue is that what is important from econometric perspective might not be from economic one. Again it might be crucial for a mode to test for heteroskedasticity or unit root etc, but readers won’t be interested in those results so putting them in paper wastes space for economic interpretation policy discussion etc. Also nowadays good journals require people to submit code and data and make it publicly available $\endgroup$
    – 1muflon1
    Nov 14, 2020 at 0:18
  • $\begingroup$ So you can easily just replicate the result yourself and check $\endgroup$
    – 1muflon1
    Nov 14, 2020 at 0:19

0

Your Answer

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

Browse other questions tagged or ask your own question.