The Swedish industrial support program of the 1970s revisited

Abstract

The economy-wide dynamic cost-benefit study of the Swedish industrial subsidy program 1976 through 1984 (Carlsson et al. Res Policy 10(43):336–354 1981; Carlsson J Ind Econ 32(1):9–14, 1983a, b) is revisited in light of later economic development. Since the Swedish Micro to Macro model (Eliasson Am Econ Rev 67(1):277–281 1977a, 2017a) was used for quantification, this article is both (1) a study on the calibration of high dimensional micro-based and nonlinear economic systems models, and (2) a post inquiry into the empirical credibility of the cost-benefit calculations performed. We find that the Micro-based Macro model represents the minimum of detailed resolution necessary for the dynamic cost benefit calculations of the micro interventions in the Swedish economy we study. Even though the increased model complexity meant significant parameter calibration difficulties, a thoroughly researched model specification with exactly defined policy interfaces (with the markets of the economy) should take priority over parameter estimation problems, and always be preferred to estimating the parameters of a wrongly specified model perfectly. The oil price shocks of the 1970s caused radical market disorder in the western economies, bankrupting some 35% of Swedish manufacturing and threatening the Swedish government with massive unemployment. We confirm the earlier results that the government choice of a radical employment rescue policy came at enormous social cost in the form of economic stagnation, and still did not prevent the unemployment of the rest of OECD Europe from hitting Sweden a decade later, and persisting well into the next millennium. According to an alternative simulated policy scenario on the model, had the subsidies been replaced with a general lowering of the payroll tax of the same magnitude and the consequent increase in unemployment taken immediately during 1976–1980, production structures would have been radically and rapidly reorganized, normal employment would have been rapidly restored, and neither the stagnation nor the radical increase in unemployment of the early 1990s would have occurred. In retrospect we see no reason to worry about the empirical credibility of this computed dynamic trade off between Keynesian demand and Schumpeterian supply effects (caused by resource reallocations and endogenous structural change due to the price change), as we did then. We conclude with certainty that this trade-off would not even have been discovered as a possibility had we used a traditional model that did not embody these micro-macro linkages.