models, forecastsI’m currently trying to assess the usefulness of using google insight for search to predict initial claim for unemployment in the US. Choi and Varian have written an article where they compare out-sample forecast of a AR(1) model and the same model where a google search variable is added. I asked this question because I suppose that there is a better benchmark than simple AR(1) model.
There were no answers to this question.
All content is licensed under CC BY-SA 3.0.