Discussion Paper
No. 2015-60 | September 09, 2015
Josef Falkinger
The Order of Knowledge and Robust Action. How to Deal with Economic Uncertainty?
(Published in Radical Uncertainty and Its Implications for Economics)

Abstract

Economic uncertainty has to do with the consequences of actions under different circumstances. This raises two questions: First, how sensitive are the outcomes of actions to variations in the environment? Second, how clearly can we distinguish between environments? Robustness comes at the price of targeting actions less narrowly to specific conditions, so we lose gains from specialization. Need for robustness comes from our limited knowledge. Rational dealing with uncertainty requires to accord the degree of specialization to the reliability of knowledge about the relevant circumstances. In practical terms, under such an approach acting under uncertainty is related to guidelines for strategic thinking: Focus on priorities on a broader scale; the most refined set of actions is not always the best one.

JEL Classification:

D80, D81, D83

Links

Cite As

[Please cite the corresponding journal article] Josef Falkinger (2015). The Order of Knowledge and Robust Action. How to Deal with Economic Uncertainty? Economics Discussion Papers, No 2015-60, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2015-60


Comments and Questions



Anonymous - Invited Reader Comment
September 23, 2015 - 11:58
I would like to comment on Professor Falkinger’s “The Order of Knowledge and Robust Action. How to Deal with Economic Uncertainty?” from the perspective of someone who has used the model presented in that paper. For many years, I had an idea in mind, which I was unable to model satisfactorily. It was using Professor Falkinger’s model that I was, perhaps, finally able to do so (Habib, 2015). Going back at least to Edith Penrose seminal “The Theory of the Growth of the Firm” (1959), management theorists have been aware of the importance of firm resources and the distinction between general and specialized resources. Resources, both general and specialized, have been central to much management, organizational theory, and economic thinking. A number of models have been developed that have elegantly formalized the genesis of firm resources (Prescott and Visscher, 1980; Jovanovic and Rousseau, 2001) and insightfully explored their implications for firm size and scope (Mitchell, 2000) and for the shape of entire industries (Bresnahan and Trajtenberg, 1996; Jovanovic and Rousseau, 2005). It perhaps reflects poorly on me that I have not found these models easy to apply to settings beyond the ones for which they had originally been developed. Professor Falkinger provides a micro-founded model for thinking about general and specialized resources and investment, albeit not – yet – about these resources’ initial genesis. The beauty of the model for an applied theorist such as me is that, although micro-founded, the model generates extremely simple expressions for expected payoff and volatility, which are made to depend only on aggregate general and specialized investment, rather than on the myriad investments targeted at specific states. These very simple buildings blocks can easily be assembled into relatively complex structures. The basic dichotomy in the model as I understand it is that between two ‘meta-states,’ one of risk in which individual states can be distinguished, the other of uncertainty in which no such distinction is possible. These terms are of course Knight’s. In Arrow-Debreu-like fashion, investment can be targeted towards the individual states of risk, whereas it is at best possible to target investment towards the entire meta-state of uncertainty; thus does Professor Falkinger’s model generate specialized and generalized investment, specialized and generalized resources. A further virtue of the model is its recognition that neither the two meta-states nor the individual states of risk are entirely exogenous. To some extent at least, and at a cost, the meta-state of uncertainty can be decreased to the benefit of that of risk, and further ‘granularity’ may be obtained within that latter meta-state by subdividing individual states of risk into distinct sub-states. The informational environment – the relative importance of the two meta-states and the division into specialized states of risk – thereby becomes itself subject to optimizing behavior. In short, and reaching beyond management and economics, Professor Falkinger’s model holds the promise of helping us determine when we should be calculating (specialized investment) and when we should be principled (general investment); when we should expect to find bliss in ignorance (meta-state of uncertainty) and when we should get to the bottom of things (individual states within the meta-state of risk). It is a remarkable achievement.

Anonymous - Referee Report
September 23, 2015 - 12:04
The paper is a potentially interesting way of tackling the topic which might have much value but it is a very difficult, and also a rather disconcerting read, in several respects. 1. It barely relates at all to existing literature, except a few key (and mostly rather, or very, old) papers. It, therefore, simply ignores a lot that has been done more recently in similar areas. It is then hard to tell how (if at all) it relates to those. A further implication is that it is then difficult to determine its contribution. 2. It is extremely abstract, both mathematically and in its use of metaphors. In this respect the paper might be significantly improved and more likely to make impact if the author was able to help the reader by producing a simple toy model or example that would illustrate the distinctions he is drawing. 3. The paper has a number of definitions and assumptions, but no actual results that arise from these. This is odd, to say the least, at least in terms of the standard model of a paper, in which you'd expect the analysis to culminate in at least a proposition or two. 4. All of this really makes the paper far less effective than it might be. It is difficult to engage and overall I struggled to make sense of the arguments at depth and found it hard to concentrate. Indeed for my own part I must confess that I was challenged to get to the end of the formal analysis. It does occur to me that my behaviour may indeed be an example of the kind of process the author is talking about but at the same time it reinforces my view that for an economic journal this is not an effective communication!

Anonymous - Invited Reader Comment
October 15, 2015 - 11:55
When I received an invitation to give my view on “The Order of Knowledge and Robust Action. How to Deal with Economic Uncertainty?” by Josef Falkinger, I thought it would be one of those papers on decisions in uncertainty/ambiguity where authors begin with some [experimental] evidence violating the expected utility approach, for which reason there is a need in something new, and voila, here is a theory that explains the violations and/or is confirmed by another dozen of experiments. Professor Falkinger’s writing does not fit in this scheme. There is no particular model to violate, no experiments, not even a well-­‐rounded theory with testable implications. Instead, there is an attempt to create a whole new structure of thinking about uncertainty. The main objective of the paper, as I understood it, is to give a set of definitions related to the potentially imprecise knowledge and to introduce some key assumptions that outline the new frame of thinking about the interaction of human decisions and the realizations of the states of nature. I would see this as a first step in a rather big research agenda. A first step is never easy. The more important it would be to make it easier for the reader. Yet, in my view, the paper is quite difficult to follow. Perhaps, the introduction does not do a good job of introducing properly into the problem. It throws in that “having knowledge about a set of distributions rather than a specific distribution, is usually addressed as Knightian uncertainty” (p.2), a claim with which I cannot fully agree, as there is indeed a class of ambiguity models that assume that there is a set of distributions, yet not all ambiguity models would operate with probabilities at all, not to mention the assumption of “having knowledge” of a set of them. The paper never explains what exactly it means, to have knowledge about distributions. Instead, knowledge is first introduced implicitly as “under limited knowledge about the measure pi on A, a cruder frame in which possible future events are distinguished in a less differentiated way, may be a more reliable guide for actions” (p.4) and then as a “stock of general knowledge G, how future outcomes can be generated by actions today” (p.6). This G is a positive [real?] number used as a coefficient in the assumption that suggests that the expected outcome of an action is a linear function of the resources involved. Why? The beginning of Section 2 is not very helpful either. The metaphor of a map did not work for me (although it might work for somebody else). Instead, I wanted to understand what makes “instruments” so different from “acts” and “environments” so different from events that one would not be able to use, for example, David Schmeidler’s (1989) framework? Or, maybe even that of Leonard Savage (1954) where probability distributions, strictly speaking, are not known either, everybody can have his own one. Or, is the idea of a “cruder frame” of knowledge any similar to rational inattention? Answers to these questions, would help me absorb further ideas advanced in the paper. As I said before, I appreciate the attempt of constructing a new framework of thinking, which is never easy. But I would want to know why other frameworks do not work, and how can I benefit from using the new framework. This may sound pragmatic, but if the new framework is of no use, do we really need it?

Josef Falkinger - Reply to comments
December 10, 2015 - 12:18
see attached file