I originally posted my question in the statistical procedures but maybe its better placed in this forum:
I'd like to perform Mincer-Zarnowitz regressions to test whether a forecast is associated with a forecast error that is unpredictable.
My equation is a simple OLS: Realized value = b0 + b1*Forecast.
Since my sample consists of yearly data, I run regressions annually, i.e. for each year I observe the coefficients for b0 and b1. In a next step, I build time-series averages for the coeffiicents, i.e. I build the mean over the yearly coefficients.
To see if there are systematic errors, I need to test whether the time-series average coefificent for the Forecast is statistically different from 1 using Newey West T statistics. For similar situations where I test if a sample mean of a series is statistically different from 0, I use the following procedure:
proc model data=two;
instruments / intonly;
fit Forecast / gmm kernel=(bart,5,0) vardef=n;
Is it possible to change this to 1? Otherwise, what would be the best way to get Newey West T-stats to test whether the coefficient for Forecast is statistically different from 1?
I already tried to add a test statement (e.g. test Forecast=1). However, I am afraid this delivers the Wald-test but not Newey West T stats.
Registration is open! SAS is returning to Vegas for an AI and analytics experience like no other! Whether you're an executive, manager, end user or SAS partner, SAS Innovate is designed for everyone on your team. Register for just $495 by 12/31/2023.
If you are interested in speaking, there is still time to submit a session idea. More details are posted on the website.