Inference for Simulation Models
Simulation of the Garbage Can ModelStill to be written
Raffaello Seri
Professor of Econometrics
My research interests include statistics, numerical analysis, operations research, psychology, economics and management.
Publications
Randomness, Emergence and Causation: A Historical Perspective of Simulation in the Social Sciences
This chapter is a review of some simulation models, with special reference to social sciences. Three critical aspects are identified, i.e. randomness, emergence and causation, that may help understand the evolution and the main characteristics of these simulation models. Several examples illustrate the concepts of the paper.
Asymptotic Properties of the Plug-in Estimator of the Discrete Entropy under Dependence
We consider the estimation of the entropy of a discretely-supported time series through a plug-in estimator. We provide a correction of the bias and we study the asymptotic properties of the estimator. We show that the widely-used correction proposed by Roulston (1999) is incorrect as it does not remove the $O\left(N^{-1}\right)$ part of the bias while ours does. We provide the asymptotic distribution and we show that it differs when the values taken by the marginal distribution of the process are equiprobable (a situation that we call degeneracy) and when they are not. We introduce estimators of the bias, the variance and the distribution under degeneracy and we study the estimation error. Finally, we propose a goodness-of-fit test based on entropy and give two motivations for it. The theoretical results are supported by specific numerical examples.
Model Calibration and Validation via Confidence Sets
The issues of calibrating and validating a theoretical model are considered, when it is required to select the parameters that better approximate the data among a finite number of alternatives. Based on a user-defined loss function, Model Confidence Sets are proposed as a tool to restrict the number of plausible alternatives, and measure the uncertainty associated to the preferred model. Furthermore, an asymptotically exact logarithmic approximation of the probability of choosing a model via a multivariate rate function is suggested. A simple numerical procedure is outlined for the computation of the latter and it is shown that the procedure yields results consistent with Model Confidence Sets. The illustration and implementation of the proposed approach is showcased in a model of inquisitiveness in ad hoc teams, relevant for bounded rationality and organizational research.
How many times should one run a computational simulation?
This chapter is an attempt to answer the question “how many runs of a computational simulation should one do,” and it gives an answer by means of statistical analysis. After defining the nature of the problem and which types of simulation are mostly affected by it, the chapter introduces statistical power analysis as a way to determine the appropriate number of runs. Two examples are then produced using results from an agent-based model. The reader is then guided through the application of this statistical technique and exposed to its limits and potentials.
Analytical Approaches to Agent-Based Models
The aim of this article is to present an approach to the analysis of simple systems composed of a large number of units in interaction. Suppose to have a large number of agents belonging to a finite number of different groups: as the agents randomly interact with each other, they move from a group to another as a result of the interaction. The object of interest is the stochastic process describing the number of agents in each group. As this is generally intractable, it has been proposed in the literature to approximate it in several ways. We review these approximations and we illustrate them with reference to a version of the epidemic model. The tools presented in the paper should be considered as a complement rather than as a substitute of the classical analysis of ABMs through simulation.
