Epistemic Economics and Organization

Anna Grandori

• Publisher: Routledge
• Year: 2013
• ISBN:
  9781136679117
• Edition: 1
• Topic: Betriebswirtschaft
• Subtopic: Business allgemein

This book proposes a new approach to economics, management and organization that should help in making economic organization 'wise', 'innovative' and 'robust' in an uncertain and risky world. Although the modern economy and society is 'knowledge intensive', Anna Grandori argues that the dominant economic, organizational and behavioural models neglect to a large extent the problem of valid knowledge construction and effective knowledge governance.

The book integrates inputs from economics and behavioural science with insights from the philosophy of knowledge to define new micro-foundations: neither a calculative, deductive and omniscient 'rational actor'; nor an experiential, adaptive and biased 'behavioural actor'; but a knowledgeable and imaginative 'epistemic actor'.

The implications for contracts and organizations, sustained also by insights from law, are shown to be far reaching, including a new view of the nature of the firm as an entity-establishing agreement under which to discover uses of resources under uncertainty, and as a democratic institution.

Part I

Micro foundations

From bounded to epistemic rationality

1‘Models of man’ and the ‘rationality divide’

2Savage and Simon revisited: how both ‘maximizing’ and ‘satisficing’ simplify problems

3Endogenizing assumptions: contingent rationality

4The ‘psychology’ versus the ‘logic’ of judgement and discovery

5The logic of economic discovery: an epistemic decision model

6Conclusions

7Summary

The term rationality, (from the Latin ‘ratio, rationis’) according to its encyclopaedic, etymological and philosophical definitions, means ‘the principle governing the activity of knowing’: something is rational if ‘it proceeds from reason’, if it ‘is founded on logically sound procedures, on scientific method’. The meaning of rationality has been reduced to a much narrower meaning in both economic and behavioural sciences, albeit in different ways. In this chapter it is argued that a reduction in the wisdom and validity of possible behaviours is the result of following either of the two prevailing economic and behavioural paradigms on rationality. Furthermore, it is argued and shown that those wiser behaviours are not only logically possible, but indeed are followed by a variety of innovative and discovery-based decision processes. Therefore, the proposed enrichment of extant decision models with models of the logic of economic discovery is also bringing superior predictive power to the decision making theory employed in economic sciences; in particular the capacity of predicting and guiding the innovative yet rational behaviours that are so important in modern knowledge intensive economies, but very poorly acknowledged and modelled in economic sciences. The questions rarely posed in the analysis of the micro-foundations of our disciplines, and central stage here, are of the type: if a problem is not ‘given’, are there any (logically sound) ways of formulating it? If so, knowing them may help in escaping from addressing ‘wrong’ or badly defined problems. Are there any (logically sound) methods for ‘having good ideas’, i.e. for generating and testing novel action alternatives? If so, knowing them may prevent remaining trapped in conventional and past-based courses of action. These questions are logically prior to the questions that are typically addressed in extant decision making models, where problems are defined to start with, and alternatives are at best ‘searched’, not ‘researched’.

‘Logically sound procedures’ can be defined with respect to various logical operations relevant in the activity of knowing and deciding. In particular, two distinctions are particularly relevant for discussing and renewing how the notion of rationality is employed in economic and behavioural sciences, and will be widely used in this book. First, knowing activity involves both a ‘logic of justification’ and a ‘logic of discovery’ of any proposition (a central distinction in philosophy of science, as is well known) (e.g. Lakatos and Musgrave 1970; Simon 1977). Second, decision making involves two differing broad categories of judgements, both calling for logically sound procedures: rationality in ‘instrumental’ judgements (what are the best means to ends) and rationality in ‘epistemic’ judgements (is it true that those means have those effects? Is it true that those ends are desired?) (Foley 1987).

In economics, the meaning of rationality has been restricted to refer only to the logically sound procedures for ordering alternatives according to preference and to the logic of justification of choice among them (Shackle 1972). Hence, it can be said that economics has been concerned with instrumental rationality – a set of logically sound procedures for consistency in utility judgements and means to end choices – not with epistemic rationality – a set of logically sound procedures for knowing what the preferences, alternatives and consequences are in the first place. Second, a judgement expressed by Lakatos (1976) about mathematics could be applied equally well to economics: it has been concerned only with the ‘deductive’ rather than with the ‘heuristic’ component of rationality, that is only with the ‘logic of justification’, not also with the ‘logic of discovery’.

This use of the term ‘heuristics’ also throws light on another, related, mismatch of meaning between the philosophy of science and knowledge on one side, and economic and social sciences on the other. ‘Heuristics’ in the philosophy of science is a branch or sub-discipline concerned with ‘the logic of discovery’: the set of ‘logically sound procedures for researching and finding’, for generating conjectures and controlling them, the very method of empirical science. ‘Heuristics is neither psychology nor logics, but an independent discipline, the logic of discovery’ (Lakatos 1976, appendix II, note 4). As such (some) ‘heuristics’ can be qualified as ‘rational’, in the same sense as the methods of scientific empirical research can be defined to be rational: they are logically correct methods for discovery (Grandori 2010a). Hence, the concept of ‘heuristics’ has also been restricted, and ‘downgraded’, in behavioural and economic sciences usage: it is used to refer to psychological devices, substituting cognitive ‘shortcuts’ and ‘fast and frugal’ thinking to deductive rationality and statistical decision theory. As such, ‘heuristics’ have been associated with quick and low effort choice (Gigerenzer et al. 1999) or with ‘biases’ (Kahneman et al. 1982) rather than with rational discovery.

Building on those observations, and on a long lasting research program oriented by them (Grandori 1984, 2001a, 2010a), Part I of this book aims at providing renewed and enlarged micro-foundations to economic behaviour. Theoretically, then, the view proposed in this first part of the book originates from the observation of the mismatches between the broader philosophical meaning of rationality and the narrower meanings attributed to the term in economic and behavioural sciences; and will show how the latter reduced meanings are at the origin of many puzzles and problems in important areas such as contract theory and innovation management. Empirically, it originates from the observation that actual behaviours inspired by instrumental rationality only, on calculating costs and benefits, all too often lead to proceeding in quite a distorted manner, disregarding all the important considerations about the validity of action hypotheses that are considered, and on ‘simplifying’ problems too much for making those calculations possible (rather than implying ‘wide’ or even ‘complete’ knowledge of the world, as they are commonly said to imply). On the other side, actual behaviours saving on effort and based on limited search and limited understanding of the problem at hand, all too often ‘leave resources on the table’. This expression, commonly employed in negotiation analysis is evocative of the real problems I am pointing at: suppose that a negotiator or a designer of a new product adopts the rule of stopping at the first satisfactory alternative found, or lowers their aspirations in the face of difficulty … Even in problems with no closed boundaries, with infinite possible solutions, much better can be done, and is often done.

The analysis will feature the following aspects and be developed through the following analytical steps:

•revisiting and criticizing the standard ‘rationality divide’ between the ‘rational actor paradigm’ and the ‘bounded rationality paradigm’, offering a different account or interpretation of the two models, allowing to interpret them not as incomparable or rival ‘paradigms’ but as comparable and contingently rational decision strategies;

•specifying a further, missing, ‘epistemic’ model of decision behaviour that, in the course of being heuristic (aimed at discovery) is also ‘rational’ (logically correct); capable of explaining and guiding behaviour in the rather uncharted and neglected territory of the rational discovery of economic action.

1 ‘Models of man’ and the ‘rationality divide’

The model of rational decision making employed in economic sciences typically includes the statement that the set of possible actions A and the set of possible ‘states of the world’ S (or the ‘contingencies’ under which actions generate their consequences) are ‘known’ (the elements in the set are enumerable or a criterion of belongingness to the set is given). An expected utility can be defined ‘over’ the combinations of a and s belonging to A and S respectively: by ordering the values of outcomes according to preference and by assigning probabilities when their value is not known for certain. The only rational learning method considered is the Bayesian updating of probabilities, if they can be assigned, upon the observation of new information on a and s.

If analysed from an epistemological standpoint, those standard assumptions on rationality in economics should be characterized as a set of ‘logically sound procedures’ for the calculative component of decision only (Shackle 1972).

If analysed from an epistemological standpoint, it may also be noticed, however, that these assumptions allow two different interpretations – a conventionalist and a realist version.

In a conventionalist interpretation, they can be read as follows: let’s assume that actors behave ‘as if’ they have knowledge of the inputs and are calculating and choosing the option with the maximum expected value in a given set of options, and explore and test the predictions (Friedman 1953). The prediction, as known, is that, under those restrictive conditions approximating perfect competition, if actors behave ‘as if’ they were value maximizers, then equilibrium prices will be reached and resources allocated efficiently at their best use.

In the most common interpretation, though, the conventionalist assumption that inputs are known and calculations are performed according to value maximizing calculation procedures, has been transformed into the descriptive and realist assumption that actors do in fact possess a complete and infallible knowledge of the possible states of the world and their possible actions. Economists have apparently started to believe that the admission that knowledge is fallible and that actors do not have perfect foresight, would amount to abandoning the rational actor paradigm for accepting a ‘bounded rationality’ assumption: e.g. ‘with rational agents contingencies are never unforeseen, they are at worst indescribable’ (Tirole 1999); ‘If parties do not foresee even relatively obvious events, it would seem necessary to assume that they are boundedly rational’ (Hart and Moore 1999).

This realist version of the rational actor paradigm entails some epistemological problems that a conventionalist version does not.

Ernst Nagel offers an interesting starting point. He observed (Nagel 1963: 214) that a ‘trivial way in which any statement can be said to be unrealistic is that it can never give a complete description of all the infinite aspects of any real objects or situation’. This criticism applies to the realist version of economic rationality, according to which the rational actor actually has a complete knowledge of the world, and actually foresees all possible contingencies. The criticism is consistent with and rooted in the tradition of the philosophy of knowledge and science, to which Nagel himself contributed. According to that tradition (after Russell and Popper), in fact, any assertion or assumption about knowledge being ‘objective’, ‘complete’, ‘perfect’, or assertions like ‘there are no unforeseen contingencies’, would be considered naïve and even illogical statements. A core argument, due to Popper, is that one can never prove, according to logically sound procedures, that a statement about reality is true or to have considered ‘all the possible contingencies’ because they are infinite not only in number but also in kind. Hence, one can provocatively say that the assumptions of classic economic rationality in their realist version … are not rational, in the sense of applying ‘logically sound procedures’ (Grandori 2010a). In fact, Nagel was actually defending the legitimacy of making ‘unrealistic assumptions’ in economic modelling, but interestingly concluded that this methodological approach can be defended precisely because there can be no ‘complete’ description of reality and no ‘perfect foresight’ of the world!

The alternative notion of ‘bounded rationality’ (Simon 1955), has been defined in strong reference with the economic model intended in its realist version, and developed by relaxing some of its core assumptions about the knowledge of decision inputs. In Simon’s words, in the ‘classical’ (economic) concept of rationality,

the organism must be able to attach definite payoffs (or at least a definite range of payoffs) to each possible outcome. This, of course involves also the ability to specify the exact nature of the outcomes – there is no room in the scheme for ‘unanticipated consequences’ The payoffs must be completely ordered – it must always be possible to specify in a consistent way, that one outcome is better than, or as good as, or worse than any other. And, if the certainty or probabilitic rules are employed, either the outcomes of particular alternatives must be known with certainty, or at least it must be possible to attach definite probabilities to outcomes. My first empirical proposition is that there is a complete lack of evidence that, in actual human choice of any complexity, these computations can be, or are in fact, performed.

(Simon 1955: 104)

The alternative ‘satisficing’ model was then defined in terms of ‘partial’ and ‘sequential’ exploration of problem spaces, guided by ‘aspiration levels’ (ALs) or ‘thresholds’ on the expected utility of the sought alternatives, and a choice rule stated in terms of ‘acceptability’ – an alternative is acceptable if its payoff is superior to the threshold.

In this way, the main alternative to classic economic rationality was born as a ‘daughter of a minor godness’, a form of rationality ‘trying’ to approximate the classic economic template, but in almost all situations falling short of accomplishing the task and ending up in a ‘reduced’, more ‘limited’ form of rationality, a form of ‘weaker thought’.

The two paradigms have been fighting for (more than 50) years, but the type of objections they raised toward each other remained only on a ground that could be seen through their respective lenses. What has remained out of sight is a possible epistemological criticism to both.

On one side, as said, the economic notion of rational actor can be used in a conventionalist interpretation, whereby the use of ‘unrealistic assumptions’ can be defended, but not in a realist descriptive version. Toward a realist version in fact, the charge of ‘unrealistic assumptions’ (one of Simon’s main charges) becomes relevant. In addition, an even stronger objection of logical impossibility of complete description and knowledge of the world may be raised from an epistemic viewpoint, as argued above.

On the other side, it seems that Simon’s criticism, based on the ‘empirical proposition’ that humans ‘are not able’ to perform all the calculations involved in economic models was in a sense too weak and in a sense too strong. It was too weak because it criticized the classic assumptions interpreted in their realist and descriptive version, with an empirical and descriptive argument: human behaviour, descriptively and empirically, is commonly ‘less’ knowledgeable and calculative. It did not criticize whether, normatively and logically, the way of being knowledgeable and calculative of economic classic rationality was actually ‘the’ model toward which to ‘tend’ in our intended rationality. The acceptance of the equation between ‘complete knowledge’ – or ‘omniscience’ – and value maximizing rationality, present in the realist version of the rational actor model, produced too strong a criticism of value maximizing as an empirically inapplicable decision process or strategy; a behaviour for which there is ‘no evidence’.

The observation I advance here and develop in the following paragraphs is that, while there is no evidence (and actually no logical foundation) for complete knowledge, there is plenty of evidence (and strong logical foundations) for the applicability of value maximizing strategies to specified classes of structured and stylized problems (Savage 1954). It can be noticed here that Simon himself provided an outstanding example of the applicability of value maximizing reasoning with incomplete knowledge, in his analysis of the employment relationship (1951). The model reconstructs the authority relation as a Pareto-efficient contract among two actors maximizing their respective utility, which is specified as a function of the benefit or cost of performing the task x, and of the wage w to be paid or gained: for the employer (e), max U_e = U_e(x) – U_e(w); for the labour provider (l) max U_l = U_l(x) + U_l(w). The attractiveness of a contract giving the employer the right to choose the task is conditional to the circumstance that there is uncertainty about which task will be better (contingencies cannot be foreseen), and the utility of having option rights is high for the employer and low for the worker.

Hence, a utility maximizing logic does not imply or require perfect knowledge and complete foresight. Actually, an important contribution by Simon himself has been to demonstrate that the authority rela...