When Cheaper Software Reshapes the Market
The model can now be used for a counterfactual: what happens when custom software becomes much cheaper? The important comparison is not between a world with software and a world without it, but between firms that start in different adoption states and between different ways software can become more important. This chapter follows non-adopters, firms that always adopt, and switchers through the modeled price change, then compares the price shock with lower adoption costs and software-biased technical change.
The price shock sorts firms by starting position
The authors calibrate the model to the current software era and then raise the rental rate of custom software to its late-1980s level. In the model, this corresponds to a 63% decline in the rental rate between 1987 and 2018. The counterfactual holds other structural parameters fixed and asks how firms, wages, scope, and aggregate outcomes respond. The result is not a historical experiment. It is a disciplined scenario built on the mechanism already established.
The first lesson is heterogeneity. A lower software price does not deliver the same benefit to every firm because firms begin with different productivity, scope, and adoption status. The paper separates three groups: firms that never adopt, firms that adopt both before and after the change, and firms that switch from non-adoption to adoption once software becomes more affordable.
Non-adopters receive no direct reduction in the price of their software input because they do not use that input. They are nevertheless affected through the labor market. The software shock increases labor demand among firms that expand and adopt, which raises the equilibrium wage in the model. Non-adopters then face a higher labor cost without receiving the software-related offset. Their unit costs rise by about 6%, and their optimal scope falls; the exit threshold also rises, reducing the profitability of small firms.
This is a general-equilibrium effect. It is easy to imagine the price shock as a private discount available only to software buyers, but the model does not hold wages fixed. When the expanding part of the economy demands more labor, firms that do not participate in the software transition still face the changed price of labor. Their position can worsen even though they never purchase the cheaper input.
Always-adopters do receive the direct price reduction, but their average unit-cost response is close to zero—about a 0.5% decline—because the software saving is largely offset by the increase in wages. The average conceals variation by scope. Smaller adopters devote a larger share of their costs to software and can benefit more from its price decline. Larger adopters devote a smaller software share and are more exposed to the wage increase. The model's non-homothetic factor mix therefore makes the same price change unequal even within the adopter group.
Always-adopters expand their scope and sales because the reusable input becomes cheaper to spread across establishments. Yet their gain is not the only force shaping concentration. Firms that were previously below the adoption threshold may enter the technology regime and grow faster than the established adopters. The top of the firm distribution can gain from a cheaper shared input while also facing new competition from switchers.
Switchers are the most directly affected group. Before the software improvement, they remain non-adopters because their scope or productivity is not sufficient to justify the cost. After the improvement, the software plan clears the adoption threshold.