What Economic Modelling Hypotheses Should Underlie Regulation & Policy Advice? Towards a Probabilistic Approach

Abstract

The recurrence of financial crises in the last fifty years reveals a market coordination failure that regulation and economic policy were unable to prevent. While several explanations exist for the emergence of financial and economic crises, none seems to be fully satisfactory or conclusive. This text develops the idea that the theoretical assumptions that underlie policy and regulation are inadequate to fully map economic reality and that this explains the observed policy and regulatory failures. The paper suggests to assess the adherence of standard economic assumptions to reality from a probability viewpoint. It notes in particular that the no-arbitrage condition is the key assumption retained by modern finance for the pricing of financial assets. It is supported by the same set of logical premises that are associated to competitive markets in general equilibrium theory. Still today, this group of normative requirements remains central in policy advice and regulation, be it through Modern Finance or other applied fields such as Public Choice. Based on simple probabilistic arguments, the paper reminds that the domain of relevance of the above standard normative conditions is necessarily very limited. The inadequacy of the assumptions retained by regulators and policy makers to map actual economic reality can thus explain the regulatory and policy failures that allowed financial crises to develop. Public action should use more wisely normative conditions. A closer look at the adherence of the relevant assumptions to reality may be a way for improving policy and regulation effectiveness in the future.

Keywords:

• Regulation
• modern finance
• no-arbitrage
• general equilibrium
• post-keynesian

JEL CODES:

• A130 Relation of Economics to Social Values
• B410 Economic Methodology
• P000 Economic Systems: General

Acknowledgements

The author is Managerial Adviser in the Operations Directorate at the European Investment Bank. Opinions expressed are personal. Previous versions of this text were presented at the Annual Conference of the Eastern Economic Association, New York, May 9–11 2013, at the MAF 2016 conference on Mathematical and statistical methods for actuarial sciences and finance, University of Paris—Dauphine, March 30–31 and April 1, 2016 and at the EAEPE conference on Industrialisation, socio-economic transformation and institutions, Manchester Metropolitan University Business School, 3–5 November, 2016. The author is grateful for comments received on previous versions and for discussions and encouragement, in particular to William Black, Stefano Bragoli, Eugenia Correa, Marcella Corsi, Marie-Paule Donsimoni, Alessandro Fortunelli, Pier Luigi Gilibert, Rebecca Graziani, Alan Hawkes, Alessandro Roncaglia, Andrea Salanti, Domenico Sartore, Mario Tonveronachi, Jan Toporowski as well as two anonymous referees. The usual disclaimer applies.

Notes

1 In formal (deterministic) logic it is known that an implication between proposition A (say: it is good to follow ‘mainstream’ policies) and proposition B (say: there will be no crisis) is always true except when A is true and B is false. Hence, if proposition B is false, either the implication itself or the proposition A are false. The word ‘mainstream’ refers here to that body of core policy propositions (such as ‘get the prices right’, ‘the market self-regulate’, etc.) that have taken a hegemonic position amongst the Western elites in the last forty years.

2 The assumptions are those of the static and competitive market model, which are retained by the Public Choice literature, for instance in the third edition of Forte (1993), where they are taken as the dominant organizing principle in Public Finance.

3 The EMH descends from Bachelier (1900). Mandelbrot and Hudson (2004, p. 13) note that taking the daily percentage movements of the Dow Jones Industrial Average between 1916 and 2003, the normality retained by Bachelier and Fama that, together with independence, is the main statistical ingredient of the Market Efficiency Hypothesis, implies that only 58 daily price changes should have exceeded 3.4%, against an actual 1001. Similarly, the daily percentage increases in excess of 4.5% should have been 6, against 366, and daily price changes in excess of 7% should have occurred only once every 300,000 years, against 48 times in the period considered.

4 See for instance Allais (1947/1998, p. 184, n. 69 e 1962, p. 701; Malinvaud (1953, p. 263) and McKenzie (2002, p. 192, condition II). Mc Kenzie (1959/2009, ft. 9, p. 120) also quotes Koopmans (1951).

5 When translated into the policy debate, the ‘heuristic assumptions’ of experts become axioms taken so universally for granted that that their limiting character is generally forgotten of.

6 de Finetti (2005, p. 54) showed that that, with n agents when all individual utilities are simultaneously maximised, there are ∞n-1 Pareto equilibria.

7 Many explanations of economic crises can be found in 24 papers produced by the Work Package n. 3 on of the five-year EU financed research project on Financialisation, Economy, Society and Sustainable Development (FESSUD): ‘a multidisciplinary, pluralistic project which aims to forge alliances across the social sciences, so as to understand how finance can better serve economic, social and environmental needs’. These 24 papers are available on line at http://fessud.eu/working-papers/#WP3 and a synthesis is provided in Sawyer (2017, pp. 82–97). Eichengreen (2015, p. 5) offers a detailed comparison of the ‘Great Depression’ and the ‘Great Recession’ and reminds us of Queen Elisabeth II’s 2008 question visiting the London School of Economics: ‘Why did no one see it coming?’. For a neo-classical real business cycle perspective on crises one can refer to Ohanian (2010); Weidmann (2017) provides an ordo-liberal viewpoint; Congdon (2014) defended the monetarist interpretation whereas, from a neo-Keynesian perspective, Reinhart and Rogoff (2009, 2010) emphasize leverage and particularly public debt, a work which was later criticized by Herndon, Ash and Pollin (2014). Sinn (2010) provides a mainstream critique of ‘Casino Economics’. A critical review of several explanations of the crisis is provided in Hossein-zadeh (2014), whereas Neal (2011) offers the perspective of an economic historian and Engelen et al. (2012) discuss the political issues of controlling finance. Viewpoints from authors working in International Financial Institutions are: Kelley and Love (2010), Borio and Disyatat (2011), Lin (2013), Claessens et al. (2014) and Bayoumi (2017). Guesnerie (2013) emphasizes the factor he calls expectational market failure, an approach under which the crisis is defined as the collapse of a previously shared common view of the future. Post-Keynesian explanations of the crises are offered by Hein (2012), Florio et al. (2011–2012) and Bellofiore and Vertova (2014). Tokunaga amd Epstein (2018) highlight the role of the endogenous interaction of Large Complex Financial Institutions with the US shadow banking system, where the dollar plays a dominant role for financial innovation. For a Marxian viewpoint, one can refer to Foster and Magdoff (2009). Most of these explanations have points in common, for instance in the important role attributed to financialisation, defined by Epstein (2005) as ‘the increasing importance of financial markets, institutions and motives in the world economy’ (see also: Epstein, 2018). In this respect, Toporowski (2015) discusses in particular financial asset price inflation. The approach presented in the third section of this paper can be made consistent with many of the above explanations by combining in different ways the assumptions that are discussed there.

8 Papandreou (1959) argued that, in terms of philosophy of science, economics is not a pure deductive system (like logic or mathematics), nor an applied one, because even if one makes reference to empirical evidence, its theoretical constituents are capable or confirmation but not of rejection, unless this rejection is confirmed for all the possible ‘social spaces’ (say institutional contexts), something which he considers impossible in the present stage of our knowledge. For this reason, instead of ‘theory’, for economics Papanderou suggests to use the word ‘model’. However, for the purposes of this text, the words ‘theory’ and ‘model’ can be used as synonymous (see also footnote n. 27 and n. 29). Hence ‘theoretical’ assumptions can be understood also as ‘modelling assumption’.

9 Beliefs are necessarily subjective. As pointed out by Hume, quoted by Dow (2015, p. 33): ‘Reason and evidence [… are …] necessary for knowledge, but not sufficient: they rely […] on belief’.

10 In this review of Keynes’ ‘Treatise on Probability’ and Jeffreys’s ‘Scientific Inference’ books, de Finetti contrasts the view of probability seen as a branch of logic dealing with uncertain statements, associated to the names of Hume, Keynes, Jeffreys and himself, to the view of probability seen as a branch of mathematics, associated with the viewpoint of Von Mises and the frequentists.

11 Instead of ‘induction’, Dow (2015) uses the narrower concept of ‘abduction’, which as noted by Douven (2017), was introduced by Peirce as: ‘the process of forming explanatory hypothesis’ and which then became: ‘what the logical empiricists called the ‘context of justification’—the stage of scientific inquiry in which we are concerned with the assessment of theories’. However, since de Finetti saw induction as the basic process for learning, a much broader concept than the frequentist one of Peirce and the American empiricists, here induction is intended to cover also what Dow calls ‘abduction’.

12 Once it is accepted that any probability judgment is necessarily subjective, ‘objective probability’ simply means ‘convergence of subjective probability views’ amongst different people and thus the subjective and objective approaches to probability are not exclusive of each other (de Finetti, 1951). In any case, the view retained here is that only the subjective approach can support a rational policy debate amongst different viewpoints (or ‘beliefs’) that, in an intellectually honest approach, should all be considered equally plausible ex ante.

13 De Finetti (1958, pp. 1–9) distinguished 4 main approaches to probability: i) the axiomatic approach, which is purely mathematical; ii) the subjective approach which takes probability as a measure of ignorance at individual level; iii) the classical approach, in which events are taken as equally probable; and, iv) the statistical or empirical approach which applies to events that can be repeated and that defines probability in terms of frequency. In all approaches the same theorems hold and probability calculations are the same: it is only their interpretation that changes, being ‘objective’ in the case of iii) and iv).

14 Some of the endogenous variables can be policy targets in the logic of Tinbergen (1952). Roncaglia (2010), Roncaglia and Tonveronachi (2014), and Tonveronachi (2018) reach somewhat convergent conclusions starting from a different starting point.

15 For France, whose administrative tradition has strongly influenced that of the EU institutions, the root of these ideas can be found in a report that Jacques Rueff drafted in 1958 at the request of Antoine Pinay, then Minister of Finance of de Gaulle; see Parguez (1993). This principle is also somewhat embedded in the Amsterdam Treaty, where the zero-deficit should be targeted over the period of a business cycle. Domar (1944) defined his celebrated debt stability condition by the requirement that the growth rate of the economy should exceed the interest rate paid on debt. A discussion of this principle in relation to European debt stability problems is provided by Sardoni (2011). Parguez (2014) distinguishes ‘good’ and ‘bad’ public deficits depending on whether they result from active development policies or from austerity policies. His 2010 ‘monetary stability requirement’ foresees that public debt should grow in line with private debt to avoid financial stability problems, particularly in the banking sector (see Cingolani, 2013).

16 This is also true in Marx, whose ‘reserve army’ is the buffer stock regulating the business cycle. Suitably re-interpreted, his reproduction schemes raised the issue of effective demand before Keynes (see for instance Łaski & Walther, 2015). Relying on the heuristic assumption of the absence of capital, Pasinetti (1993) has elegantly illustrated why, when there are several sectors that have different rates of growth due to learning and technical progress, even if the economy starts from full employment, structural dynamics will generate unemployment in the absence of a corrective macroeconomic intervention.

17 The debate on austerity policies is of course not yet closed. For a recent contribution, sympathetic to the thesis of ‘expansionary austerity’ see Alesina, Favero and Giavazzi (2019) and, for a critique see Botta and Tori (2018), Arestis et al. (2018), as well as Nuti (2015).

18 Government expenditures, being a policy instrument, are by definition exogenous to the private sector economy in the short-term. Beyond this, the discussion is still not closed on whether or not net Government expenditures can be considered fully exogenous in the medium to long-term. In the short-term in a monetary economy the arguments of the Modern Monetary Theory are hard to refute for countries having a sovereign currency (Mosler, 1997–98; Wray, 2003; see also footnote 39). The monetary circuit view (Graziani, 2003; Parguez, 1996) also supports the same interpretation, which is also consistent with the endogenous money approach prevailing in modern central bank practice (Arestis & Sawyer, 2006; Rochon, 2006; Rochon & Olawoye, 2012; Rochon & Rossi, 2017) and none of the latter are restricted only to the short-term. For a definition of reduced form and a discussion of exogeneity see: Artus, Delau, Malgrange (1986), chapter 6, pp. 117–139.

19 For a recent critical review of the former, see Botta and Tori (2018). As far as the critique of the latter is concerned, there is a very wide literature, but most of the arguments were already made in Bresciani Turroni (1939) and Hayek's (1944).

20 The author is grateful to Stefano Bragoli of the European Investment Fund for help in building-up the argument of this example and for several discussions.

21 Strictly speaking, the decision rule only works when the wary policy maker’s advisers have opposite views.

22 These are respectively that: $P(EH)=P(E)*P(H/E)$ and $P(H/E)=P(H)\frac{P(E/H)}{P(E)}$ (see De Finetti (1979/2008), p. 37).

23 This refers to chapter 1: ‘Market equilibrium: a first approach’ of Arrow and Hahn (1971), pp. 19–22.

24 A market is ‘cleared’ by prices when at that price supply equals demand.

25 See in particular Koopmans (1951, theorem 4.7, p. 66)

26 For a criticism of this hypothesis from the viewpoint of chaos theory, see McCauley (2011).

27 Lunghini (1968) and Papandreou (1963) argued that a taxonomy of ‘social spaces’ is necessary in any economic deductive approach, an argument they developed mainly for the deterministic case (see also footnotes 8 and 29).

28 The author is grateful for and agrees with Prof. Salanti's comment that this aspect should be re-examined in future work, where one should consider that not all permutations are possible at each node. But for the modest illustrative purposes of this text, the simplification retained seems acceptable, as it affects mainly the total number of assumptions considered (the denominator), rather than the number of assumptions covered by alternative theories (numerator). In other words, it affects more the calculation of what is retained here as the ‘absolute probability’ of a theory rather than the ‘relative probability’ of competing theories.

29 Some of the issues and typically sectors, money, production, openness in distribution and technology are also of relevance in terms of institutional and/or social economics. Indeed, without pushing the analogy too far, all the 10 issues retained can be seen as related to Papandreou’s (1959, 1963) concept of ‘social space’ against which he suggested economic models should be tested.

30 Like Keynes, de Finetti denies the relevance of the distinction between risk and uncertainty but retains the somewhat equivalent difference between homogeneity and heterogeneity of probability assessments by different individuals, the latter case being non-convergent to frequency if the heterogeneity of opinions is strong. The post-Keynesian viewpoint on Keynes and uncertainty is presented by Kregel (1976).

31 Montesano (1970, p. 712) notes that Pareto in his famous manual retains ergodicity, which he calls 'closed cycle property’ precisely because it is the only hypothesis compatible with the static general equilibrium. See also Montesano (2006).

32 In a social decision process, one should not be surprised to find non-ergodic environments, as the elementary particles of the analysis (the individuals) interact socially. Contrary to molecules, which interact passively by following the laws of nature, individuals have their own opinions and follow them. For this reason, Markov processes, which are an essential building block of the Efficient Markets Hypothesis, should not be expected to be relevant for analysing interactions in financial markets, as the latter are strongly influenced by psychology, mimetic behaviour and other socially-related factors.

33 Pasinetti (1986) stressed the essential distinction between the alternative value paradigms of production, in which time is embedded as an essential element, and exchange, in which it can be ignored.

34 For an account of this period of Italian economic thought, see Faucci (2000, chapter VII) and Pomini (2017). de Finetti (1935a, 1935b) provided an early critique of the Paretian approach that is devastating for its practical applications, although De Finetti was in fine quite sympathetic to Pareto’s original goals.

35 The English translation was published under the title: ‘The Betrayal of the Intellectuals’. In this respect see: Carrick-Hagenbarth and Epstein (2012).

36 See also Hahn (1970, pp. 1–2) and Pasinetti (2000, p. 412). See Schefold (2018) for a recent contribution dealing with one of the main technical aspect of the “Cambridge capital controversies” of the seventies.

37 Polanyi-Levitt (2013, p. 239) noted: ‘The neoliberal counter-revolution of the 1980s was prepared in think-tanks and universities. It was signalled by the 1974 award of a Nobel Prize to Hayek’. On the latter, see Offer and Söderberg (2016), in particular pp. 127–131.

38 If one does not have at least two goods: consumption and investment, the multiplier concept is meaningless, whereas in a barter economy one needs at least three goods in order to start discussing about money as we usually intend it, i.e. performing the three monetary functions of numéraire, means of payment and store of value, for all past present and future transactions on goods and services (see Rosenstein-Rodan, 1936).

39 If one adheres to the idea that economic theory can be looked at with the glasses of system analysis (Mockers, 2016; Alfred Eichner, 1991), a basic distinction to draw is that between open and closed models (Chick, 2004). In closed models all variables are endogenous, i.e. determined by the model itself, like it is the case for money in the fourth option above, whereas in open models, external factors important for the determination of the endogenous variables are determined outside the model. For open models, it makes sense to ask whether a variable is endogenous or it is exogenous. In the first case, the variable can be a target or an objective of economic policy, whereas in the second case it can be a policy instrument. Obviously, this implies implicitly prima facie causality of the exogenous variables on the endogenous variables. In the endogenous money’s court, economists do not all agree on the merits of Modern Monetary Theory, despite the fact that it offers in principle a simple conceptualisation of the unconventional monetary policies followed by leading central banks to get out of the crisis (see the contributions in Fullbrook & Morgan, 2019 as well as footnote 18).

40 Some references are: Lefebvre (2000), Isard et al. (1998/2017), Leontief et al. (1953) and Leontief and Strout (1963).

41 This allegedly ‘heuristic’ hypothesis should be relaxed in future work.

42 Contrary to Figure 4, in this Figure 5 and in Figures A1, A2 in the Appendix, on the issue of dynamics (second circle) the assumption of temporary equilibrium has been included under the mainstream approach and depicted in black, while the assumptions of traverse and path dependency have been excluded and depicted in grey, as per Table 8. All issues under uncertainty have been taken as excluded from the mainstream and depicted in grey. The same conventions have been applied at lower level.

43 Alternative arrangements of issues and assumptions, carried possibly at a lower level (with more issues), would produce different results, but are unlikely to entail a much higher probability for the mainstream case. Of course, it is not meant here that mainstream assumptions are irrelevant, as the example of the market for fresh eggs in New York shows, where 10 million eggs are sold daily at exactly the same price (the author is in debt with Prof. Marie Paule Donsimoni for this anecdote, which was popular amongst graduates of top American universities in the 1970s).

44 Introduction by Ignazio Visco to the workshop on the Governor’s Concluding Remarks on Bank of Italy Annual Report for 2019, organized on 22.06.2020 by Accademia dei Lincei, Centro Linceo Interdisciplinare ‘Beniamino Segre’ and Società Italiana degli economisti. See also Visco (2020).