Info-Gap Economics

Info-gap theory is described without using equations. The idea of an information-gap is introduced. The three components of an info-gap analysis are discussed: uncertainty model, system model, and performance requirement. The robustness and opportuneness functions are introduced and their use in decision making is discussed. Robustsatisficing and opportune-windfalling strategies are described. Implications of info-gap theory are outlined. Applications of info-gap theory to a wide range of disciplines are mentioned.

A very simplistic example gives a preliminary feel for info-gap analysis: how it is done, what one can learn, and how it supports decision making. The system and uncertainty models and the satisficing and windfalling performance requirements are formulated. The robustness function is derived, the robust-satisficing decision strategy is discussed, and three properties are studied: trade-off between robustness and performance, zeroing of the robustness at the anticipated outcome, and crossing of robustness curves leading to reversal of preferences between alternative actions. The opportuneness function is formulated and derived, and how it supplements the robustness function is demonstrated. The opportune-windfalling decision strategy is discussed.

We consider a sequence of monetary policy analyses with progressively more and different aspects. We start in section 3.1 with selecting an interest rate based on a Taylor rule, given uncertainty in the parameters of the macro-economic model. In section 3.2 we include public expectations about inflation and output gap, and consider both the choice of the interest rate and the choice of the degree of credibility of the central bank’s announced goals. In section 3.3 we add shocks to inflation and output gap, while the tails of the probability distributions of these shocks may differ substantially from what is anticipated. In section 3.4 we consider an uncertain economic model subject to shocks, in an economy with heterogeneous sectors whose beliefs about central bank credibility are uncertain to the bank.

We consider the analysis of decisions in four different aspects of financial stability. In section 4.1 we look at a simple example of structured securities as collateralized debt obligations. We explore the robustness to uncertain correlation between the underlying assets. In section 4.2 we study the robustness of the value at risk of a portfolio, when the probability distribution has uncertain fat tails. Stress testing is discussed conceptually in section 4.3 based on a suite of models. In section 4.4 we study the strategic allocation of assets using both the robustness and the opportuneness functions for exploring the risks and windfalls of uncertain payoffs.

We consider three different examples of public policy selection. Section 5.1 develops an info-gap robustness analysis of welfare considerations in the choice between imposing a limit on pollution emission or imposing a tax on emissions. The marginal costs and benefits are uncertain to the regulatory agency, and emissions quotas are traded in a competitive market. In section 5.2 we study the policy choices of a regulator responsible for enforcement of pollution limits. The regulator chooses the number of emission permits traded in a competitive market, the audit frequencies of the firms, and the allocation of enforcement resources. Firms are distinguished according to their costs of abatement, which are uncertain to the regulator. Non-proportional “profiling” audit and allocation strategies are examined. Section 5.3 studies policy choices regarding long-range economic impacts of climate change which may result from industrial emission of greenhouse gases. The potential for drastic economic impact of global warming motivates strong abatement measurements, while the great uncertainties in the processes motivates extensive research. We study the robustness implications of the policy maker’s choices between abatement and research.

We illustrate the combination of statistical estimation and forecasting with info-gap robust-satisficing in three examples. In all cases the data are not only statistically random, but also subject to nonrandom info-gaps. Info-gaps in economic data may arise due to delay between the initiation of a systemic change and its manifestation in data. In addition, current best-estimates of economic data are infogap-uncertain because these data may be substantially revised in the future as new information becomes available and new definitions and classifications are adopted. Section 6.1 considers a regression of serial data and the use of that regression to predict the next outcome. Section 6.2 discusses an auto-regression, and section 6.3 studies the confidence interval of the least squares estimate of a variable.

We discuss a range of info-gap models for non-probabilistic quantification of uncertainty. We consider info-gap models for a single uncertain parameter and for an uncertain function. We describe how an info-gap model is formulated from limited information.

We consider Friedman’s positive economics, and Samuelson’s scientific response, in the light of info-gap robust-satisficing.