1
$\begingroup$

Say I have a structural equation that is forecasting Y. Y is a nominal financial variable that grows with time due to inflation. I have an economic determinant that is a rate and therefore does not increase with time. Is it wise or necessary to deflate Y so that the variables behave the same way? I do not particularly want Y deflated as this regression will be forecasted and this just introduces another hard to forecast variable.

$\endgroup$
  • 1
    $\begingroup$ If you think the rate is a good predictor of the real future value of Y, but Y also grows with inflation and you are interested in the nominal value of Y, then you will not get around at least implicitly forecasting inflation. $\endgroup$ – Tobias Mar 29 '17 at 12:35
  • $\begingroup$ Right, but implicitly forecasting it introduces less error (from a practical standpoint) than explicitly adding the variable. The other variables will do that work for me, but introducing another variable adds more error. I do not NEED to deflate Y, but I am wondering if doing so to mimic the behavior of the rate (unemployment) adds any value econometrically. A colleague suggested this to me, but I had never heard of that convention before. $\endgroup$ – user12679 Mar 29 '17 at 15:40
  • $\begingroup$ What about inflating all your past values up to today's one. Your forecasting would then predict intrinsically today-related nominal/real values. $\endgroup$ – keepAlive Mar 30 '17 at 16:40
-1
$\begingroup$

In macro, the standard approach to forecast variables which are intrinsically nominal like GDP seems to be to separate the forecast. This is, you want to have a model for real GDP and for inflation. The final forecast for nominal GDP is the "combination" of both. This is for instance the standard practice of the Fed (of Atlanta) and the Bank of England (pages 39 to 41).

In your case of a financial variable, you could model the real (deflated) variable, and then, as inflation is given by the economy, just take the official inflation forecast of your country from the central bank. They are the real experts on this so is a safe thing to do in my opinion. Also, the market expectations of inflation, crucial in finance, are very much informed by central bank estimates. This approach gives you a cleaner estimation of your model too. As you say, your determinant is a rate, and thus inflation is just molesting in the left hand side.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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