ABSTRACT
Regression models easily lend themselves to making predictions. This chapter explains how to make forecasts with regression models. Point and interval forecasts are possible, as are ex ante, ex poste, in-sample, and out-of-sample forecasts. The emphasis is on projecting trend lines with regression models. Ex ante forecast made before the fact, so its accuracy cannot be checked. Ex poste forecast is made after the fact, so that its accuracy can be checked. The chapter also shows how to project trend lines into the future. It uses regression analysis where the dependent variable is the forecast variable and the independent variable is time. Time is a ticker that increases by 1 unit each period. The lin-log trend line fit assuming a diminishing growth rate. The log-lin trend line fit assuming an increasing growth rate. Random walk is a series of sequential movements in which the magnitude of each move is determined randomly.
