Capital theory ‘paradoxes’ and paradoxical results: resolved or continued?

Abstract

Capital theory controversies and ‘paradoxes’ showed that, due to price-feedback effects, the wage–production price–profit rate curves may display shapes inconsistent with the requirements of the neoclassical theory of value and distribution. Subsequent findings on a number of quite diverse actual single-product economies suggested that the impact of those effects is of limited empirical significance. This paper argues that, by focusing on the distributions of the eigenvalues and singular values of the system matrices, we can further study these issues and derive some meaningful theoretical results consistent with the available empirical evidence. Consequently, the real paradox, in the sense of knowledge vacuum and, thus, requiring further research, is the distributions of the characteristic values and not really the ‘paradoxes in capital theory’.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12

Notes

  1. 1.

    Data limitations do not allow the treatment of the non-competitive imports and, therefore, our investigation is based on a closed economy model. Future research should fill in the gap by treating imported means of production as a primary input (see Metcalfe and Steedman 1981, pp. 9–11).

  2. 2.

    The transpose of a \(1 \times n\) vector \({\mathbf{y}} \equiv [y_{j} ]\) is denoted by \({\mathbf{y}}^{\text{T}} ,\) and the diagonal matrix formed from the elements of \({\mathbf{y}}\) is denoted by \({\hat{\mathbf{y}}}.\) Furthermore, \({\mathbf{A}}_{j}\) denotes the \(j\)th column of a semi-positive \(n \times n\) matrix \({\mathbf{A}} \equiv [a_{ij} ],\) \(\lambda_{{{\mathbf{A}}1}}\) the P-F eigenvalue of \({\mathbf{A}}\) and \(({\mathbf{x}}_{{{\mathbf{A}}1}}^{\text{T}} ,{\mathbf{y}}_{{{\mathbf{A}}1}} )\) the corresponding eigenvectors, while \(\lambda_{{{\mathbf{A}}k}} ,\) \(k = 2, \ldots ,n\) and \(\left| {\lambda_{{{\mathbf{A}}2}} } \right| \ge \left| {\lambda_{{{\mathbf{A}}3}} } \right| \ge \cdots \ge \left| {\lambda_{{{\mathbf{A}}n}} } \right|\), denote the non-dominant eigenvalues, and \(({\mathbf{x}}_{{{\mathbf{A}}k}}^{\text{T}} ,{\mathbf{y}}_{{{\mathbf{A}}k}} )\) the corresponding eigenvectors. Finally, \({\mathbf{I}}\) denotes the \(n \times n\) identity matrix, \({\mathbf{e}}\) the summation vector, i.e. \({\mathbf{e}} \equiv [1,1, \ldots ,1]\), and \({\mathbf{e}}_{j}\) the \(j\)th unit vector. It is also noted that, given any \({\mathbf{A}}\) and an arbitrary \(\varepsilon \ne 0,\) it is possible to perturb the entries of \({\mathbf{A}}\) by an amount less than \(\left| \varepsilon \right|\) so that the resulting matrix is diagonalizable (see, e.g. Aruka 1991, pp. 74–76).

  3. 3.

    If wages are paid ex ante, then \(\rho\) is no greater than the share of profits in the SSS.

  4. 4.

    One of the referees noted that the WPCs are “meromorphic functions […]. Such curves show strong curvatures in the neighbourhood of singularities, and these may be anywhere in the complex plane; they need not be real, let alone between zero and \(R\)”.

  5. 5.

    As we will see in the next sections of the present paper, this heuristic case is more geared towards reality than one at first sight might think.

  6. 6.

    For other heuristic cases, which are also interesting, both theoretically and empirically, and include more than one hyper-basic industry, see Mariolis and Tsoulfidis (2014, pp. 214–215, 2016, pp. 154–167).

  7. 7.

    See, e.g. Horn and Johnson (1991, Chap. 3) and take into account that \({\mathbf{J}}\) is similar to the column stochastic matrix \({\hat{\mathbf{y}}}_{{{\mathbf{J}}1}} {\mathbf{J}} {\hat{\mathbf{y}}}_{{{\mathbf{J}}1}}^{ - 1}\).

  8. 8.

    Nevertheless, C. Bidard, H. G. Ehrbar, U. Krause and I. Steedman have detected some ‘monotonicity (theoretical) laws’ for the relative prices (see Bidard and Ehrbar 2007, and the references therein).

  9. 9.

    Information on the sources of data and the construction of variables is available in the Appendix.

  10. 10.

    The price curve corresponding to industry 8 displays a maximum at \(\rho \cong 0.955,\), which is not visible in Fig. 2. It is deduced from more detailed data.

  11. 11.

    In effect, we found that, at \(\rho = 0,\) the vertically integrated capital-intensities in the circulating (fixed) capital case gave an arithmetic mean equal to 1.25 (to 3.87) and a standard deviation of 0.32 (of 2.52), with a coefficient of variation of 0.26 (of 0.65).

  12. 12.

    Only a single curve, corresponding to industry 2 in the circulating capital case, crosses the horizontal axis (at \(\rho \cong 0.639\)).

  13. 13.

    If \(p_{j}^{\text{S}}\) denotes the price of commodity \(j\) in terms of SSC, and \(w^{j}\) denotes the money wage rate corresponding to the normalization equation \(p_{j} = v_{j} ,\) then \((w^{j} )^{ - 1} p_{j} = (w^{\text{S}} )^{ - 1} p_{j}^{\text{S}}\) or \(w^{j} - w^{\text{S}} = w^{j} (1 - p_{j}^{\text{S}} v_{j}^{ - 1} )\) or, recalling Eq. (15), \(w^{j} - w^{\text{S}} = w^{j} \rho R(R^{ - 1} - k_{j} ).\) It then follows that (i) \(p_{j}^{\text{S}} v_{j}^{ - 1} = 1\) implies \(w^{j} = w^{\text{S}} ;\) and (ii) \(w^{j} - w^{\text{S}}\) is directly related to \(R^{ - 1} - k_{j}\) (compare the outer curves in Fig. 9 with the relevant differences \(R^{ - 1} - k_{j}\) in Figs. 3 and 6).

  14. 14.

    This statement per se refers to the economically relevant interval of \(\rho\) and relies on the detected monotonicity of the price curves and inflection points of the WPCs; not on the near-linearity of the latter curves. Even in \(2 \times 2\) corn-tractor systems the WPC may deviate considerably form the straight line (see Eqs. (9a–b)).

  15. 15.

    As is well known, the geometric mean is more appropriate for detecting the central tendency of an exponential set of numbers. In our case, it can be written as

    \(GM = \left| {\det {\mathbf{J}}} \right|^{{(n - 1)^{ - 1} }}\)

    We define the index of inseparability associated with the eigenvalues as: \(\varepsilon_{{{\mathbf{J}}1}}^{\lambda } \equiv 1 - \lambda_{{{\mathbf{J}}1}} \left( {\sum\nolimits_{i = 1}^{n} {\left| {\lambda_{{{\mathbf{J}}i}} } \right|} } \right)^{ - 1} = 1 - \left( {1 + \sum\nolimits_{k = 2}^{n} {\left| {\lambda_{{{\mathbf{J}}k}} } \right|} } \right)^{ - 1}\)

    The research to date suggests that there are statistically significant regressions, but not one-to-one relationships, between alternative measures of the characteristic value distributions and of the errors in approximate curves (see Mariolis 2015b; Mariolis and Tsoulfidis 2016, pp. 167–198). Thus, our analysis is rather phenomenological.

  16. 16.

    There is, however, a caveat for the SIOTs as adequate representations of actual economies. It is well known that these tables can be derived from the ‘System of National Accounts’ framework of Supply and Use Tables (SUTs), where the latter tables could be considered as the counterpart of joint production systems à la v. Neumann (1937) and Sraffa (1960). The hitherto evidence from the SUTs indicates that actual economies do not necessarily have the usual properties of single-product systems (Mariolis and Soklis 2010; Soklis 2011, 2015); for instance, there are cases in which elements in the (positive) vector of ‘labour-commanded prices’, \(w^{ - 1} {\mathbf{p}},\) decrease with the profit rate and, therefore, the monotonicity of the WPC depends on the numeraire choice.

References

  1. Angeloussis A (2006) An empirical investigation of the reswitching of techniques phenomenon for the Greek economy, 1988–1992. Master’s Thesis, Department of Public Administration, Panteion University, Athens, Greece (in Greek)

  2. Aruka Y (1991) Generalized Goodwin’s theorems on general coordinates. Struct Change Econ Dyn 2(1):69–91. Reprinted in Aruka Y (Ed) (2011) Complexities of production and interacting human behaviour (pp 39–66), Heidelberg: Physica-Verlag

  3. Bidard C, Ehrbar HG (2007) Relative prices in the classical theory: facts and figures. Bull Polit Econ 1(2):161–211

    Google Scholar 

  4. Bienenfeld M (1988) Regularity in price changes as an effect of changes in distribution. Camb J Econ 12(2):247–255

    Google Scholar 

  5. Bródy A (1970) Proportions, prices and planning: a mathematical restatement of the labor theory of value. Amsterdam, North-Holland, Budapest, Akadémiai Kiadó

    Google Scholar 

  6. Garegnani P (1970) Heterogeneous capital, the production function and the theory of distribution. Rev Econ Stud 37(3):407–436

    Article  Google Scholar 

  7. Han Z, Schefold B (2006) An empirical investigation of paradoxes: reswitching and reverse capital deepening in capital theory. Camb J Econ 30(5):737–765

    Article  Google Scholar 

  8. Horn RA, Johnson CR (1991) Topics in matrix analysis. Cambridge University Press, Cambridge

    Google Scholar 

  9. Iliadi F, Mariolis T, Soklis G, Tsoulfidis L (2014) Bienenfeld’s approximation of production prices and eigenvalue distribution: further evidence from five European economies. Contrib Polit Econ 33(1):35–54

    Article  Google Scholar 

  10. Krelle W (1977) Basic facts in capital theory: some lessons from the controversy in capital theory. Revue d’Économie Politique 87(2):282–329

    Google Scholar 

  11. Kurz HD, Salvadori N (1995) Theory of production: a long-period analysis. Cambridge University Press, Cambridge

    Google Scholar 

  12. Leontief W (1953) Studies in the structure of the American economy. Oxford University Press, New York

    Google Scholar 

  13. Mariolis T (2013) Applying the mean absolute eigen-deviation of labour commanded prices from labour values to actual economies. Appl Math Sci 7(104):5193–5204

    Article  Google Scholar 

  14. Mariolis T (2015a) Norm bounds and a homographic approximation for the wage-profit curve. Metroeconomica 66(2):263–283

    Article  Google Scholar 

  15. Mariolis T (2015b) Testing Bienenfeld’s second-order approximation for the wage-profit curve. Bull Polit Econ 9(2) (forthcoming)

  16. Mariolis T, Soklis G (2010) Additive labour values and prices of production: evidence from the supply and use tables of the French, German and Greek economies. Econ Issue 15(2):87–107

    Google Scholar 

  17. Mariolis T, Tsoulfidis L (2009) Decomposing the changes in production prices into ‘capital-intensity’ and ‘price’ effects: theory and evidence from the Chinese economy. Contrib Polit Econ 28(1):1–22

    Article  Google Scholar 

  18. Mariolis T, Tsoulfidis L (2011) Eigenvalue distribution and the production price-profit rate relationship: theory and empirical evidence. Evolut Inst Econ Rev 8(1):87–122

    Article  Google Scholar 

  19. Mariolis T, Tsoulfidis L (2014) On Bródy’s conjecture: theory, facts and figures about instability of the US economy. Econ Syst Res 26(2):209–223

    Article  Google Scholar 

  20. Mariolis T, Tsoulfidis L (2016) Modern classical economics and reality: a spectral analysis of the theory of value and distribution. Springer, Tokyo

    Google Scholar 

  21. Mariolis T, Soklis G, Zouvela E (2013) Testing Böhm-Bawerk’s theory of capital: some evidence from the Finnish economy. Rev Aust Econ 26(2):207–220

    Article  Google Scholar 

  22. Metcalfe JS, Steedman I (1981) Some long-run theory of employment, income distribution and the exchange rate. Manch School 49(1):1–20

    Article  Google Scholar 

  23. Meyer CD (2001) Matrix analysis and applied linear algebra. Society for Industrial and Applied Mathematics, New York

    Google Scholar 

  24. Neumann JV (1937) Über ein ökonomisches Gleichungssystem und eine Verallgemeinerung des Brouwerschen Fixpunktsatzes. In: Menger K (ed) Ergebnisse eines Mathematischen Kolloquiums, 8. Deuticke, Leipzig, pp 73–83

    Google Scholar 

  25. Ochoa E (1989) Value, prices and wage-profit curves in the U.S. economy. Camb J Econ 13(3):413–429

    Google Scholar 

  26. Pasinetti L (1977) Lectures on the theory of production. Columbia University Press, New York

    Google Scholar 

  27. Petrović P (1991) Shape of a wage-profit curve, some methodology and empirical evidence. Metroeconomica 42(2):93–112

    Article  Google Scholar 

  28. Schefold B (1971) Mr. Sraffa on joint production. Ph.D. Thesis, University of Basle, Basle, Switzerland

  29. Schefold B (2008) Families of strongly curved and of nearly linear wage curves: a contribution to the debate about the surrogate production function. Bull Polit Econ 2(1):1–24

    Google Scholar 

  30. Sekerka B, Kyn O, Hejl L (1970) Price system computable from input–output coefficients. In: Carter AP, Bródy A (eds) Contributions to input–output analysis. North-Holland, Amsterdam, pp 183–203

    Google Scholar 

  31. Shaikh AM (1998) The empirical strength of the labour theory of value. In: Bellofiore R (ed) Marxian economics: a reappraisal, vol 2. St. Martin’s Press, New York, pp 225–251

    Google Scholar 

  32. Shaikh AM (2012) The empirical linearity of Sraffa’s critical output-capital ratios. In: Gehrke C, Salvadori N, Steedman I, Sturn R (eds) Classical political economy and modern theory: essays in honour of Heinz Kurz. Routledge, London, New York, pp 89–101

    Google Scholar 

  33. Soklis G (2011) Shape of wage-profit curves in joint production systems: evidence from the supply and use tables of the Finnish economy. Metroeconomica 62(4):548–560

    Article  Google Scholar 

  34. Soklis G (2015) Labour versus alternative value bases in actual joint production systems. Bull Polit Econ 9(1):1–31

    Google Scholar 

  35. Spaventa L (1970) Rate of profit, rate of growth, and capital intensity in a simple production model. Oxf Econ Papers 22(2):129–147

    Google Scholar 

  36. Sraffa P (1960) Production of commodities by means of commodities. Prelude to a critique of economic theory. Cambridge University Press, Cambridge

    Google Scholar 

  37. Steedman I (1999) Vertical integration and ‘reduction to dated quantities of labour’. In: Mongiovi G, Petri F (eds) Value distribution and capital: essays in honour of Pierangelo Garegnani. Routledge, London, New York, pp 314–318

    Google Scholar 

  38. Treitel S, Shanks JL (1971) The design of multistage separable planar filters. Inst Electr Electron Eng Trans Geosci Electron 9(1):10–27

    Google Scholar 

  39. Tsoulfidis L, Mariolis T (2007) Labour values, prices of production and the effects of income distribution: evidence from the Greek economy. Econ Syst Res 19(4):425–437

    Article  Google Scholar 

Download references

Acknowledgments

The material in the manuscript has been acquired according to modern ethical standards and does not contain material copied from anyone else without their written permission and there is no conflict of interest. We are indebted to two anonymous referees of this journal for helpful remarks and hints. An earlier version of this paper was presented at the 19th European Society for the History of Economic Thought Conference, 14–16 May 2015, Roma Tre University, Italy. We thank the participants of this conference and especially Heinz D. Kurz, Fabio Petri, Bertram Schefold, Persefoni Tsaliki and Stefano Zambelli for their comments, criticisms and discussions. The usual caveat applies.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Theodore Mariolis.

Appendix: Data sources and construction of variables

Appendix: Data sources and construction of variables

The input–output table of UK for the year 1990 is available from the OECD STAN database (http://www.oecd.org), and the degree of disaggregation is such that 33 product/industry groups are identified (see Table 1).

Table 1 Industry classification

The market prices of all products are taken to be equal to one; that is to say, the physical unit of measurement of each product is that unit which is worth of a monetary unit (in the SIOT of the UK economy, the unit is set to one million pounds). The various variables used in our estimations were constructed as follows:

  1. 1.

    The matrix of direct technical coefficients, \({\mathbf{A}}\), is obtained from the SIOT by dividing each industry’s inputs by their respective gross output.

  2. 2.

    The matrix of capital stocks, \({\mathbf{S}}\), is constructed by using the capital flows table, which allocates the gross investment flows, \(I_{ij}\), of each industry to itself and to the other industries. With the aid of the capital flows table we form a matrix of weights by dividing the elements of each column by the respective column sum, i.e. \({\mathbf{I}}_{\text{w}} \equiv [I_{ij} (\sum\nolimits_{i = 1}^{n} {I_{ij} )^{ - 1} } ]\). By assuming that the capital stocks are allocated amongst industries in a way similar to that of investment flows, we can write \({\mathbf{S}} \equiv {\mathbf{I}}_{\text{w}} {\hat{\mathbf{\varvec\kappa }}}\), where \({\hat{\mathbf{\varvec\kappa }}}\) denotes the diagonal matrix formed from the vector of net capital stocks, \({\varvec{\upkappa}}\) (both the capital flows table for the year 1990 and \({\varvec{\upkappa}}\) are available in the OECD STAN data base). The so derived matrix of capital stocks is subsequently divided by the actual gross output vector, \({\bar{\mathbf{x}}}\), to obtain the matrix of capital stock coefficients, i.e. \({\mathbf{K}} \equiv {\mathbf{S}}{\hat{\bar{\mathbf{x}}}}^{ - 1}\). In similar fashion we could construct the matrix of depreciation coefficients, but the lack of depreciation data at the required industry detail did not allow the construction of such matrix.

  3. 3.

    The vector of direct labour coefficients, \({\mathbf{l}}\), is estimated using the wage bill (the product of annual wage times the number of employees) of each of our 33 industries. The problem with this estimation is that the self-employed population is not accounted for. Fortunately, the OECD data base provides information on both the total employment and the number of employees for each of our 33 industries. From the available data, we estimate the average industry wage and we divide it by the economy-wide minimum wage, the so-estimated relative industry wages are subsequently multiplied by the total employment (employed plus self-employed) and so we derive the homogenized industry employment. This reduction, of course, is only meaningful when the relative wages express with sufficient precision the differences in skills and intensity of labour. The adjusted for skills total employment is divided by the industry total output to obtain the vector of direct labour coefficients.

  4. 4.

    By assuming that all wages are consumed and that consumption out of wages has the same composition as the vector of the final consumption expenditures of the household sector, \({\mathbf{c}}^{\text{T}}\), directly available in the SIOT, the commodity vector defining the ‘actual’ real wage rate is estimated as \({\mathbf{b}}^{\text{T}} = [\mathop {\hbox{min} }\nolimits_{j} \{ w_{{{\text{m}}j}} \} ({\mathbf{p}}_{\text{m}} {\mathbf{c}}^{\text{T}} )^{ - 1} ]{\mathbf{c}}^{\text{T}}\), where \(\mathop {\hbox{min} }\nolimits_{j} \{ w_{{{\text{m}}j}} \}\) denotes the economy-wide minimum money wage rate in terms of market prices, and \({\mathbf{p}}_{\text{m}}\) the vector of market prices, which is identified with \({\mathbf{e}}\).

  5. 5.

    Substituting \(w = {\mathbf{pb}}^{\text{T}}\) in \({\mathbf{p}} = w{\mathbf{v}} + \rho {\mathbf{pJ}}\) yields \({\mathbf{p}} = \rho {\mathbf{pJ}}[{\mathbf{I}} - {\mathbf{b}}^{\text{T}} {\mathbf{v}}]^{ - 1}\). Thus the ‘actual’ relative profit rates for both models are estimated as the reciprocals of the P-F eigenvalues of matrices \({\mathbf{J}}[{\mathbf{I}} - {\mathbf{b}}^{\text{T}} {\mathbf{v}}]^{ - 1}\).

About this article

Verify currency and authenticity via CrossMark

Cite this article

Mariolis, T., Tsoulfidis, L. Capital theory ‘paradoxes’ and paradoxical results: resolved or continued?. Evolut Inst Econ Rev 13, 297–322 (2016). https://doi.org/10.1007/s40844-016-0043-4

Download citation

Keywords

  • Capital theory
  • Characteristic value distributions
  • Hyper-basic industry
  • Spectral decompositions
  • Wage–price–profit rate curves

JEL Classification

  • B21
  • B51
  • C67
  • D57