Closed economy data correspondance

Dear members,

I would like to know if I should substract net exports (NX) from my GDP (Y) data if I model a closed economy, so that data corresponds exactly to my modelisation. Or I could just leave that out.

Thank you in advance.


I think this is ultimately a question you will need to answer for yourself, although I do not believe it is typical for authors to remove net exports when dealing with US data (see e.g. Smets and Wouters (2007) or Schmitt-Grohe and Uribe (2012)). My guess is this convention exists because the current account is a comparatively small component of aggregate output, and hence the bias introduced by failing to account for it is “small”. If you were considering a country where the CA was a large percent of GDP, omitting the rest of the world would be an increasingly difficult assumption to defend.

Of course much of this depends on your goal: are you trying to determine causal relationships between key macroeconomic variables, or are you just trying to generate accurate forecasts? The econometric issue with omitted variables is no different in time series than in cross-sectional data. Your model is a supposition for the data generating process (DGP) of the macroeconomic phenomena you care about. If you assume a closed economy you are assuming the DGP for these series is not influenced by the current account. Of course we know that the actual data is in fact constructed using current account data, so from an econometric point of view you are omitting a relevant explanatory variable which implies its impact is soaked up by the residuals. This might matter for causal interpretations due to omitted variable bias.

Edit: See the following discussion Open vs closed economy

1 Like