A Puzzle of Vertical Integration and Segmentation In U.S. Long Distance Telephony
The provision of long distance telephone service requires three upstream inputs, one of whichâtransmission capacityâexhibits engineering economies of scale up to the size of the largest firm in the industry, AT&T. The theory of the technological determinants of firm and industry structure suggests two extreme equilibria for the organizational form of this industry's retail participants: a vertically integrated form where firms sell retail using owned network facilities and a vertically segmented form where retailers purchase all in- puts from wholesale providers. This paper explains a puzzle about the U.S. long distance industry: since 1989, the reduction in AT&T's national market share of presubscribed lines is largely accounted for by simultaneous growth among both highly integrated firms and completely segmented firms. Partially integrated firms have experienced little growth yet, with few exceptions, have not changed their organizational form during this period. Using an original panel which details the entry, exit and investment behavior of every long distance firm in every U.S. market between 1989 and 1996, we empirically analyze growth by modeling the profitability of the non-Big Three (i.e., other than AT&T, MCI, and Sprint) using discrete choice models. Our primary finding is that the growth among highly integrated firms results in part not from economies of scale from owned transmission but from economies of scope between inputs. We also find evidence consistent with segmented firms benefiting from diversifying their sources of input supply. Sketching a theory of dynamic network formation that is consistent with the data, we note that demand uncertainty in this industry increases the benefit of purchasing inputs through a network but also implies a large option value from waiting before changing one's organizational form. We postulate that this latter effect induces partially integrated firms to retain their facilities during times of low demand.