Forecasting sales and generation of obsolete computers in the U.S.

Our goal is to characterize future trends in the generation of obsolete computers in the U.S. Our definition of obsolete is based on a user purchasing a replacement computer, after which the old one is stored, resold, or sent for waste management. Starting from historical sales data on new computers and assuming a plausible first life span distribution of desktop and laptop computers, the historical penetration rate is expressed in terms of the number of computers owned per capita. We then work to extrapolate the current trend to the future by using a logistic model, an S-shaped curve based on a rate constant and a carrying capacity, the latter expressing the asymptotic limit of adoption of the technology. The major challenge met is that the personal computer is still in an early stage of adoption from a technology life cycle perspective. It is thus still too early for statistical fits to yield a reasonable estimation of carrying capacity. Our approach is to use a bounding analysis, the guiding principle of which is characterization of a range of possibility as opposed to a best guess value. We determine plausible upper and lower bounds on the future carrying capacity to be 1.3 and 0.71 computers per capita respectively. Under our assumptions for computer lifetime, these bounds correspond to an equilibrium generation of 132 million obsolete computers as an upper bound and 76 million as a lower bound. Evolution towards these bounds is estimated using the logistic model. One result of the analysis is the generation of obsolete computers by 2020 will be at most 20% away from the long-term carrying capacity. At the lower bound of adoption, generation of obsolete computers will plateau in 2021 and in 2047 for the upper bound.