FMOLS/DOLS sufficient to answer research question? Or VECM needed?

For econometric discussions not necessarily related to EViews.

Moderators: EViews Gareth, EViews Moderator

Posts: 1
Joined: Sun Mar 18, 2018 3:08 pm

FMOLS/DOLS sufficient to answer research question? Or VECM needed?

Postby jdlewis5293 » Sun Mar 18, 2018 4:17 pm

I am currently doing a thesis on the historical effect of tax size and tax progressivity on income redistribution from taxes. The tax policy literature has established that income redistribution by taxes = tax size * tax progressivity. That is, they are inseparable. Changes in redistribution cannot occur without changes in tax size and tax progressivity. Even if policy makers wanted to increase redistribution, they would need to create new tax rates which directly effect tax size, progressivity, then redistribution. In other words, the causal direction is Size+Progressivity --> Redistribution. Furthermore, all of my variables are constructed from the same income data.

As of now, I have conducted FMOLS and DOLS on this cointegrated relationship. The dependent variable is Redistribution. I believe that the coefficients from the results of FMOLS and DOLS is enough to answer my research question, but I don't see many journal articles that stop simply at this step. I typically see articles proceed to VECM and run a Granger causality test. I ran VECM on these variables and got the results, but the granger causality test is giving me strange causal flows that do not make sense. I feel that the VECM is complicating my analysis more that it needs to be.

My question for you all: Would FMOLS/DOLS be enough to test this hypothesis below given the stated background information?

Hypothesis: Tax size accounts for more of the change in the redistributive effect of tax policy during the last century than tax progressivity.

Return to “Econometric Discussions”

Who is online

Users browsing this forum: Google [Bot] and 2 guests