The Emergence of Triple Switching and the Rarity of Reswitching Explained

by
Robert Vienneau
My articles My siteMy booksMy videos
Follow on:

I have written up a series of post as a research paper: first post, second, third, fourth, fifth, sixth, seventh. Here I present the abstract and most of the introduction.

Abstract: Empirical research indicates that the reswitching of techniques, as well as multiple switching with more switch points, is rare. This article explores parameter spaces in the analysis of the choice of technique to suggest why reswitching and triple-switching might be hard to find in empirical data. An example illustrates that the emergence of triple-switching requires specific evolutions of coefficients of production. Further evolution of technology removes the possibility of triple-switching. The example also illustrates that the roundaboutness of a technique is independent of the capital-intensity of a technique.

Consider the analysis of the choice of technique in post-Sraffian price theory. Kurz & Salvadori (1995) is a standard textbook presentation. Switch points, in which two techniques are both cost-minimizing at a given wage or rate of profits, are found as the zeros of certain polynomials of high degree. These zeros can be complex and, if real, need not be positive and below the maximum rate of profits. Nevertheless, theory suggests that multiple switch points between techniques are common. Han & Schefold (2006) and Zambelli (2018) are the most comprehensive empirical works to date, looking at switch points in comparing techniques drawn from Leontief matrices constructed from actual national income and product accounts. Reswitching and capital reversing, never mind multiple switching with more switch points, seem to be rare in empirical data. How can this discrepancy between expectations from theory and empirical results be resolved?

Kurz (2020) points out some difficulties with the empirical results. Often fixed capital is not taken into account. Only circulating capital is assumed, and the production of heterogeneous commodities, with varying input coefficients, in each industry is abstracted from. Some of these heterogeneous processes in an industry can be expected to be obsolete in the year in which data is gathered. Obsolete plant is operated in an economy side-by-side with more recent vintages. Firms often produce multiple products, and accounting conventions may assign a firm to different industries in different years. Heterogeneity in labor, changes in labor mixes, and changes in relative wages over time, are also ignored in this empirical work. The empirical research to date, although impressive still suffers from limitations that ought to be taken into account when assessing how rare reswitching is likely to be.

Nevertheless, Schefold (2023) investigates the supposed rarity of certain capital-theoretic phenomena, found surprising by marginalist economists. He randomly generates coefficients of production for alternate techniques. The resulting wage curves are near linear, that is, nearly affine functions. A small number of techniques, only one or two, contribute their wage curves to the frontier, except near extremes for the rate of profits. The continuous variation in the cost-minimizing technique with distribution, as postulated in marginalist theory, does not seem defensible. The reswitching of techniques does not seem likely on the wage frontier.

Changes of techniques in practice seem not to be a matter of choosing a cost-minimizing technique from an existing and well-known book of blueprints, following price signals. Rather, as Joan Robinson frequently remarked, new techniques are a matter of technical innovation, with reduced coefficients of production and perhaps with processes using new capital goods, not previously produced.

This article explores parameter spaces for technology with a different method. An example of triple-switching from Schefold (1980), to illustrate roundabout production, is extended with technological change. This particular model of structural economic dynamics (Pasinetti 1993) is not claimed to be realistic. Rather, it provides a two-dimensional parameter space that is partitioned by fluke switch points. A switch point is a fluke if it is a knife edge case in which almost all perturbations of model parameters destroy its defining properties. This article identifies points in the parameter space that are double-fluke cases. For instance, the wage curves at such a point are tangent at a switch point that is also on the wage axis. Each double-fluke case occurs for parameters that are intersections of two partitions in the parameter space. A picture of how triple-reswitching can arise emerges from a synthesis of local perturbations around these double-fluke cases. This extension of the analysis of the choice of technique suggests why triple-switching, for example, might be hard to find in empirical data. The example illustrates that the emergence of triple-switching requires specific evolutions of coefficients of production. Further evolution of technology removes the possibility of triple switching.