inferring competitor's cost Ralph is eonsidering entering the custom keyboard market for personal eomputers. The keyboard is eustomizable and will operate aeross a variety of systems. Before proeeeding, Ralph decides to appraise the eompetition. The elosest is Enterprise Produets (or EP). EP markets a simiIar produet, though it laeks the versatility of Ralph' s design. A eonsultant gathers reeent data from EP's finaneiaIs and reports the following regression that relates reported eost of goods soId (egs) to sales for the EP produet:

egs = 938,248 + 59(units soId)

(960,960) (8.2) r. = .79.

Specifieation analysis indieates no autocorrelation, heteroscedastieity or whatever.

RaIph is eneouraged, sinee the estimated variable eost of 59 per unit is weIl above Ralph' s variable eost of 42 per unit.

The next day Ralph accidentally sees a confidential eost analysis prepared by the accounting group at EP. Their report ineludes the following regression oftotal production eost (tpe) on units produeed:

tpc = 2,230,207 + 43(units produced)
(389,545) (3.6)
r; = .91.
CarefuIly explain the differenee between the two regressions.AppendixLO1