Why do some researchers include international oil prices in a VAR model for small open economies? My intuition is that international oil prices cannot be a endogenous variable in the context of a small open economy. For example, I don't see how one would justify a two-variable VAR model (say, for international oil price and GDP) for a country like Ghana. Because oil prices are not endogenous for small open economies, right? I mean, Ghana's GDP or its lags do not affect the international price of oil. However, I see in the literature that people still use a VAR model with oil prices as one of the endogenous variables in the model. I am finding it hard to understand how they can do that, and these papers do not state how.

Any help is appreciated.

  • $\begingroup$ Fair point but how would you define/check for economy to be small? In theoretical modelling small economy is an assumption that needs to be qualified empirically. Perhaps the studies that you are talking about help in that direction. $\endgroup$
    – Dayne
    Nov 9, 2020 at 8:49
  • $\begingroup$ The papers I am talking about studies the effect of international crude oil import price on Ghana's economy using a VAR model. Ghana's economy cannot influence that price obviously. Maybe, it can influence its export price a little bit (often not), but definitely not the import price of crude oil. The inclusion of domestic variables in the VAR model is fine because, in a general equilibrium, everything connects to everything, kind of. But for external variables, I have not seen the justification yet for its inclusion. In a DSGE modelling, GDP cannot affect oil prices which makes sense to me. $\endgroup$ Nov 9, 2020 at 21:21

1 Answer 1

  1. Output and oil price can be endogenous even for relatively small and open country if the country is oil producer. Actually, in international trade whether economy is small depends on whether in particular case country can affect world price by varying its output. For example Kuwait is generally small open economy, but when it comes to oil production they are (at the time of writing) ranked 10th in the world so in oil market they are not small economy. For Kuwait there would be definitely an endogeneity between oil prices and output.

    Ghana (at the time of writing) it is ranked 49th in world oil production which is somewhere in the middle between being significant and trivial so I can imagine someone arguing that there could be some endogeneity between oil prices and output. I suppose it is possible that if Ghana would shut down its oil production world price would move slightly.

  2. VAR model actually does not require for all variables to be endogenous. It is completely fine to estimate VAR model where there are also exogenous control variables. Doing so does not violate any assumption of VAR. VAR allows for endogeneity to be present, it does not require it. However, if you are completely sure those variables are exogenous, and you are only using two variables, running VAR would be equivalent to just running two ARDL models where one of those ARDL models will be superfluous. It would not be the most efficient modeling choice but I also can't think of reason why it would introduce bias.

  • $\begingroup$ @EmmanuelAmeyaw well you do not mention any information about the paper you are talking about - in fact you did not mention you are even talking about specific paper you only generally mentioned literature. Regardless the above will still apply VAR is valid even when variables are not endogenous it would not be usual modeling choice in such case but probably authors of the paper had a reason or maybe it was just low quality paper. Cant comment on that since I dont know what paper you are even talking about $\endgroup$
    – 1muflon1
    Nov 9, 2020 at 21:38
  • $\begingroup$ You are right, but Ghana started exporting oil only after 2010 and became a net exporter of oil only in 2017. The papers I am talking about studies the effect of crude oil import price on Ghana's economy using data before 2011. I know, one can include exogenous variables in a VAR model (like VARX), but then, we can't study the effect of shocks to these exogenous variables, right? Correct me if I am wrong on this. Because each equation has one shock component which directly affects the endogenous variable on the on the right hand side of that equation. Where would one put exogenous shocks? $\endgroup$ Nov 9, 2020 at 21:40
  • $\begingroup$ @EmmanuelAmeyaw you can still model the shocks going from the exogenous variable to the dependent variable but if there is no endogeneity there should only be shocks going one way. For example, if oil price is exogenous but still determines output shocks to oil price would have effects on GDP, but shocks to GDP should have no effect on prices $\endgroup$
    – 1muflon1
    Nov 9, 2020 at 21:43
  • $\begingroup$ Ok. Thanks, 1muflon1. I appreciate your comment. I mentioned literature because I see it not for the papers I am talking about, everywhere, actually. The papers I am talking about are just examples for Ghana's economy. But actually, such studies of analyzing oil price shocks on non-oil exporting economies (using a VAR model) can be easily found in the literature $\endgroup$ Nov 9, 2020 at 21:45
  • $\begingroup$ @EmmanuelAmeyaw oil prices are often included as controls because energy prices have an effect on output but I cant recall ever seeing paper which would only have GDP and oil prices. $\endgroup$
    – 1muflon1
    Nov 9, 2020 at 21:48

Your Answer

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

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