Econometrics and the Philosophy of Economics: Theory-Data Confrontations in Economics
792
Econometrics and the Philosophy of Economics: Theory-Data Confrontations in Economics
792eBook
Available on Compatible NOOK devices, the free NOOK App and in My Digital Library.
Related collections and offers
Overview
As most econometricians will readily agree, the data used in applied econometrics seldom provide accurate measurements for the pertinent theory's variables. Here, Bernt Stigum offers the first systematic and theoretically sound way of accounting for such inaccuracies. He and a distinguished group of contributors bridge econometrics and the philosophy of economics--two topics that seem worlds apart. They ask: How is a science of economics possible? The answer is elusive. Economic theory seems to be about abstract ideas or, it might be said, about toys in a toy community. How can a researcher with such tools learn anything about the social reality in which he or she lives?
This book shows that an econometrician with the proper understanding of economic theory and the right kind of questions can gain knowledge about characteristic features of the social world. It addresses varied topics in both classical and Bayesian econometrics, offering ample evidence that its answer to the fundamental question is sound.
The first book to comprehensively explore economic theory and econometrics simultaneously, Econometrics and the Philosophy of Economics represents an authoritative account of contemporary economic methodology. About a third of the chapters are authored or coauthored by Heather Anderson, Erik Biørn, Christophe Bontemps, Jeffrey A. Dubin, Harald E. Goldstein, Clive W.J. Granger, David F. Hendry, Herman Ruge-Jervell, Dale W. Jorgenson, Hans-Martin Krolzig, Nils Lid Hjort, Daniel L. McFadden, Grayham E. Mizon, Tore Schweder, Geir Storvik, and Herman K. van Dijk.
Product Details
| ISBN-13: | 9781400873234 |
|---|---|
| Publisher: | Princeton University Press |
| Publication date: | 12/29/2015 |
| Sold by: | Barnes & Noble |
| Format: | eBook |
| Pages: | 792 |
| File size: | 7 MB |
About the Author
Read an Excerpt
Econometrics and the Philosophy of Economics
Theory-Data Confrontations in Economics
By Bernt P. Stigum
PRINCETON UNIVERSITY PRESS
Copyright © 2003 Princeton University PressAll rights reserved.
ISBN: 978-1-4008-7323-4
CHAPTER 1
Introduction
This is a book about econometrics and the philosophy of economics, two topics that seem worlds apart. Econometrics is a study of good and bad ways to measure economic relations. Philosophy is a study of the nature of things and the principles governing human behavior. And the philosophy of economics is a study that searches for truth and knowledge about the part of reality that pertains to economic matters. This book will show that in economic theory-data confrontations the two topics are inextricably conjoined. Meaningful applied econometrics requires proper understanding of the purport of economic theory, and empirically relevant economic theorizing requires knowledge of the power of applied econometrics.
1.1 THEORY-DATA CONFRONTATIONS IN ECONOMICS
A theory-data confrontation is an empirical analysis in which theoretical arguments play an essential role. Economic theory-data confrontations occur in many different situations. In some of these confrontations economists try to establish the empirical relevance of a theory. In others econometricians search for theoretical explanations for observed regularities in their data. In still others economic researchers produce evaluations of performance and forecasts for business executives and government policy-makers.
1.1.1 A Unifying Framework
There is a unifying framework within which we can view the different activities in economic theory-data confrontations. All have a core structure consisting of three parts: two disjoint universes, one for theory and one for data, and a bridge between them. The theory universe is populated by theoretical objects that have all the features that the theory ascribes to them. The elements in the data universe are observations from which we create data for the theory-data confrontation. The bridge is built of assertions that describe the way that elements in the two universes are related to one another.
I think of an economic theory T as one of two abstract ideas. In one case T is a pair (ST, M), where ST denotes a finite set of assertions concerning some economic situation and M is a family of models of these assertions that delineates the relevant characteristics of the situation in question. The other T is a formal theory AT that is developed by the axiomatic method. It consists of the axioms and all the theorems that can be derived from them with the help of logical rules of inference. The intended interpretation of AT delineates the characteristics that its originator considered sufficient to describe the kind of economic situation about which he or she was theorizing.
In economic theory-data confrontations the "theory" (IT) is either the M of a pertinent ST or an interpretation of some AT. For example, let T denote the theory of consumer choice under certainty. In one theory-data confrontation of T , IT might present the way Franco Modigliani and Richard Brumberg's (1955) life-cycle hypothesis views consumer allocation of resources over time. In another confrontation, IT might delineate Kenneth Arrow's (1965) ideas of how consumers allocate their net worth to safe and risky assets. There are many possibilities. Note, therefore, that regardless of what IT supposedly describes, the elements in the corresponding theory universe remain theoretical objects.
The characteristics of theoretical objects are not interesting per se. Hence in a theory-data confrontation, econometricians do not question the validity of IT in its own universe. Instead they ask whether, when the objects in the theory universe have been interpreted by certain bridge principles, they can use IT to deduce true assertions about elements in the data universe. These principles constitute the bridge between the theory universe and the data universe.
In a given theory-data confrontation, the data universe consists of a collection of observations and data. The nature of the observations depends both on IT and on the original purposes for which those observations were collected. The purpose of an empirical analysis need not be the same as the purposes for which the observations were collected. The observations are used to create data for the theory-data confrontation. The makeup of the data depends on IT and the design of the particular empirical analysis.
The theory-data confrontation that I have described above is pictured in Fig. 1.1. On the right-hand side of the figure we see at the bottom the sample population on whose characteristics observations are based. Econometricians use their observations to create data that they feed into the data universe on top. On the left-hand side of the figure we find at the bottom the theory, that is, (ST , M) or AT as the case may be. Thereupon follows IT, the relevant parts of which are fed into the theory universe on top. The bridge between the two universes contains all the bridge principles and nothing else, and the arrows describe the flow of information in the system. Finally, the numbered ellipses are nodes in which researchers receive and send information and decide on what to feed into the pertinent boxes.
1.1.2 A Disturbing Riddle
Economic theory is developed and econometrics is used in the theory-data confrontation to obtain knowledge concerning relations that exist in the social reality. Generating such knowledge is problematic. We have seen that the references of the variables in the theory universe are theoretical objects, for example, toys in a toy economy. It appears, therefore, that meaningful econometric work stands and falls with the references of observations and data in the data universe belonging to the social reality. In Chapter 3, I demonstrate that the references of most variables in contemporary econometric data universes live and function in a socially constructed world of ideas. This world of ideas has little in common with the true social reality. That fact raises a serious question concerning the relevance of contemporary econometrics: How is it possible to gain insight into the social reality with data concerning a socially constructed world of ideas?
Figure 1.2 illustrates the gravity of the situation in which econometricians find themselves. At the top of that figure we discover the top of Fig. 1.1, that is, the theory and data universes and the bridge between them. Below them on the left-hand side is the toy economy in which reside the references of all the variables in the theory universe. On the right-hand side we observe the socially constructed world of ideas that contains the references of all the variables in the data universe. Finally, at the bottom we find the elements that constitute the social reality. The arrows and the question mark underscore the fact that it is uncertain how combining elements from a toy economy with elements from a socially constructed world of ideas enables econometricians to learn interesting things about the social reality.
1.1.3 The Resolution of the Riddle
The question I posed above amounts to asking, how is a science of economics possible? A long time ago, Immanuel Kant (1781, 1787) asked a similar question: "Wie ist reine Naturwissenschaft möglich?" ["How is pure natural science possible?"] The answer he gave to his question differs from the answer that I give to mine. Since both the differences and the similarities are interesting to us, I shall recount Kant's answer and the reasons he posited in support of that answer.
Kant's (1787) book, Kritik der reinen Vernunft, which in F. Max Müller's translation (Kant, 1966) became Critique of Pure Reason, is an analysis of the powers of human reason in gaining knowledge about the world independently of all experience. He argued that knowledge begins with experience but insisted that all knowledge need not arise from experience. Experience suffices to establish that "snow is melting in the streets today." A priori reasoning is needed to ascertain that "every change has its cause." Knowledge gained from experience alone Kant called empirical knowledge. All other knowledge he referred to as knowledge a priori. Some knowledge a priori can be obtained independently of experience. For example, it can be known from a priori reasoning alone that, "all instances of seven added to five result in twelve." Such knowledge Kant referred to as pure knowledge (p. 3).
There are two sources of human knowledge in Kant's theory: sensibility and understanding. The first is the faculty by which objects are received as data. The other is the faculty of judging. Knowledge of objects requires the cooperation of both faculties. "Without sensibility objects would not be given to us; without understanding they would not be thought by us" (p. 45).
Kant distinguished between two kinds of judgments: the analytic and the synthetic. A judgment is an operation of thought that connects a subject and a predicate. In an analytic judgment the concept of the subject contains the idea of the predicate. Examples are "all bachelors are unmarried" and "all bodies are extended." In both cases the idea of the predicate is contained in the idea of the subject. The validity of an analytic judgment can be established by a priori arguments that appeal to the logical relation of subject and predicate.
A synthetic judgment is one in which the idea of the subject does not contain the idea of the predicate. Examples are "the object in my hands is heavy" and "seven plus five equals twelve." In either case the predicate adds something to the concept of the subject. Most synthetic judgments are judgments a posteriori in the sense that they arise after an experience. There are also synthetic judgments that are judgments a priori, and their validity can be established by a priori reasoning alone. Of the two examples above, the first is a synthetic judgment a posteriori and the second is a synthetic judgment a priori (pp. 7–10).
Kant believed that necessity and strict universality were characteristic features of synthetical judgments a priori (p. 3). He also insisted that such judgments permeated the sciences. For example, mathematical propositions, such as 5 + 7 = 12, "are always synthetic judgments a priori, and not empirical, because they carry along with them necessity, which can never be deduced from experience" (p. 10). Similarly, in geometry a proposition such as "between any two points the straight line is the shortest line that connects them" is a synthetic judgment. It is also a priori because it carries with it the notion of universality (p. 11). Finally, natural science (Physica) contains synthetical judgments a priori as principles. Examples are "in all changes of the material world the quantity of matter always remains unchanged" and "in all communication of motion, action and reaction must always be the same." Both judgments are obviously synthetical. They are also a priori since they carry with them the idea of necessity (pp. 10–12).
To solve his original problem Kant had to figure out how synthetical judgments a priori were possible. The best way to do that was to study the functioning of the human mind, and he thought of that functioning as occurring in three stages. In the first, the mind places in space and time the manifold of experiences that humans receive through their senses. Space and time are not empirical concepts. They are pure forms of sensuous intuition (Anschauung) that are necessary representations a priori of all intuitions (pp. 23–35). In the second stage, the mind organizes the manifold matter of sensuous intuitions in forms of sensibility that constitute concepts of the understanding that humans use to judge the meaning of their experiences. Kant believed that the forms exist in the mind a priori, and he attributed the synthesizing act of arranging different representations in forms to twelve basic categories of thought (pp. 54–66). In the third stage, the mind combines categories of thought and established concepts to give a unifying account of the manifold world of sense impressions that humans face. He ascribed the mind's ability to accomplish that to what he deemed the highest principle of cognition: the existence of an a priori synthetical unity of apperception that is constitutive of the synthetical unity of consciousness and hence the self (pp. 76–82).
In this context there are two especially interesting aspects of Kant's view of the functioning of the human mind. First, he distinguished between two kinds of reality: the world of phenomena that human beings experience and the world of things as they are in themselves independently of human observation. One can have knowledge about phenomena but not about things-in-themselves (Dinge-an-sich). Secondly, knowledge of phenomena is limited by the way human faculties of perception and understanding synthesize experiences. In doing that, the two faculties with application of the ideas of space and time and the twelve categories of thought structure the world of phenomena in accordance with their own way of knowing.
It is remarkable that two abstract ideas, space and time, and just twelve a priori categories of thought should enable humans to relate meaningfully to the world of phenomena and to make synthetic judgments about its constituents that have the marks of necessity and universality. To understand how that is possible requires a closer look at the categories of thought. Kant arranged the categories in four groups with names, quantity, quality, relation, and modality. In the quantity group one finds three categories: unity, plurality, and totality. In the quality group reside three other categories: reality, negation, and limitation. In the relation group one finds the categories inherence and subsistence, causality and dependence, and community. Finally, in the modality group resides a fourth triple of categories: possibility, existence, and necessity (pp. 62– 66). Kant believed that the categories enabled the mind to delineate characteristic features of objects in the world of phenomena. Examples of how the mind accomplishes that can be found in E 1.1. In reading the examples, note that even the simplest judgments make use of a combination of several categories.
(Continues...)
Excerpted from Econometrics and the Philosophy of Economics by Bernt P. Stigum. Copyright © 2003 Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
Table of Contents
List of Figures ixList of Tables xi
Preface xv
Chapter 1: Introduction 1
PART I. FACTS AND FICTION IN ECONOMETRICS 31
Chapter 2: The Construction of Social Reality 33
Chapter 3: The Social Construction of Reality 52
Chapter 4: Facts and Fiction in Econometrics 71
PART II. THEORIZING IN ECONOMICS 93
Chapter 5: Theories and Models 95
Chapter 6: The Purport of an Economic Theory 117
Chapter 7: Rationality in Economics 143
Chapter 8: Topological Artifacts and Life in Large Economies 168
PART III. THEORY-DATA CONFRONTATIONS IN ECONOMICS 191
Chapter 9: Rational Animals and Sample Populations 193
Chapter 10: The Theory Universe 216
Chapter 11: The Data Universe 237
Chapter 12: The Bridge Principles 262
PART IV. DATA ANALYSES 283
Chapter 13: Frequentist Analogues of Priors and Posteriors Tore Schweder and Nils Lid Hjort 285
Chapter 14: On the COLS and CGMM Moment Estimation Methods for Frontier Production Models Harald E. Goldstein 318
Chapter 15: Congruence and Encompassing Christophe Bontemps and Grayham E. Mizon 354
Chapter 16: New Developments in Automatic General-to-Specific Modeling David F. Hendry and Hans-Martin Krolzig 379
PART V. EMPIRICAL RELEVANCE 421
Chapter 17: Conjectures, Theories, and Their Empirical Relevance 423
Chapter 18: Probability versus Capacity in Choice under Uncertainty 458
Chapter 19: Evaluation of Theories and Models Clive W. J. Granger 480
PART VI. DIAGNOSTICS AND SCIENTIFIC EXPLANATION 497
Chapter 20: Diagnoses and Defaults in Artificial Intelligence and Economics 499
Appendix: Section 20.3 from a Logical Point of View by Herman Ruge Jervell 520
Chapter 21: Explanations of an Empirical Puzzle: What Can Be Learned from a Test of the Rational Expectations Hypothesis? Heather M. Anderson 525
Chapter 22: Scientific Explanation in Economics 558
Chapter 23: Scientific Explanation in Econometrics: A Case Study Heather M. Anderson, Bernt P. Stigum, and Geir Olve Storvik 578
PART VII. CONTEMPORARY ECONOMETRIC ANALYSES 611
Chapter 24: Handling the Measurement Error Problem by Means of Panel Data: Moment Methods Applied on Firm Data Erik Biørn 613
Chapter 25: On Bayesian Structural Inference in a Simultaneous Equation Model Herman K. van Dijk 642
Chapter 26: An Econometric Analysis of Residential Electric Appliance Holdings and Consumption Jeffrey A. Dubin and Daniel L. McFadden 683
Chapter 27: Econometric Methods for Applied General Equilibrium Analysis Dale W. Jorgenson 702
Index 755