COGNITIVE DETERMINANTS OF HUMAN CAPITAL INVESTMENTS AND PRODUCTIVITY

EMPIRICAL SIMULATIONS OF INVESTMENTS IN THREE COUNTRIES

THROUGH ARTIFICIALLY INTELLIGENT AGENTS

Ken Acks

TABLE OF CONTENTS

INTRODUCTION

THE MODEL

Overview
Formation of Desired Levels
Predicted Traits of Goods
Choice of Goods
Lock-in
1) Failure to Revise Predictions
2) Alteration of Desired Levels Due to Frustration or Satiation
3) Development of Propositions and Schemas
Summary of Model
The Direction of Change
Evolution
Hierarchy of Needs

ESTIMATION OF MODEL

Previous Work
General Considerations
Application to Educational Investments

SIMULATION OF CHANGES IN EDUCATIONAL INVESTMENTS

Description
Results
Future Modifications/Additional Research

CONCLUSION

REFERENCES

ABSTRACT

This paper attempts to show that nations and individuals fail to invest in human and physical capital due largely to cognitive factors generally ignored by economists. Cognitive forces influence human effort and its direction. We first develop a general theoretical model that integrates choice theories from several disciplines. It shows how expectations, and the rates at which individuals discount future earnings evolve. The evolutionary path is based upon the demand for characteristics of goods and/or activities through time. Expectations and discount rates can become locked-in at suboptimal levels. We then simulate the model with empirical data from three countries.

INTRODUCTION

This paper will attempt to show that nations and individuals fail to invest in human and physical capital due largely to cognitive factors generally ignored by economists. Cognitive forces influence human effort and its direction--as manifested by investments in education, in physical capital, in the search for profitable niches, in research and development, in hard work, or in overcoming obstacles. These investments, in turn, depend upon perceptions of expected returns, and utility functions with shifting values placed upon present vs. future consumption.

The plight of American ghettos highlights pervasive cognitive attitudes which are not conducive to investments in education, or in physical capital. Government programs that fail to account for these attitudes--as did many in the 1960's and 70's--are doomed to fail. Without hope, a sense of individual responsibility, respect for the rights of others, and a belief that one can influence his or her destiny, growth is not likely to occur. The values exhibited by certain individuals and groups are determined, in part, by the cognitive attitudes of others--notably leaders, parents, the media and other groups. Tolerance by dominant groups can determine whether social groups flourish or flounder, and resulting behavioral patterns.

Similarly, cognitive factors such as entrepreneurial vigor, the ability to accept and assimilate change, and the tolerance of other ethnic groups will play a greater role in determining the fate of democracy and capitalism in formerly Communist countries than such economic variables as income and price elasticities, rational expectations, or subgame perfect equilibria.

We will first develop a general theoretical model integrating choice theories from several disciplines. This model will analyze the formation of expectations, and the utility attached to present vs. future consumption. We will show how expectations, and the rates at which individuals discount future earnings evolve, and can become locked-in at suboptimal levels--causing some to reduce their searches for opportunities. The model will demonstrate how expectations, discount rates, and desired levels of investment depend upon the evolution of demand for characteristics of goods and/or activities through time. It will become evident that history can never be ignored in studying choices.

This paper will formalize the determination of what David Landes in The Unbound Prometheus called "the Faustian desire to master nature", what Harvey Leibenstein has deemed "X-Efficiency", what Max Weber labeled "The Protestant Ethic", and what David McLelland named "achievement motivation" (We are not implying here that these concepts are equivalent).

We will then adapt the model to educational investments, and simulate educational investments in three countries--a composite average country, high investment Japan, and low investment Burkina Faso.

Because attitudes and propositions change slowly and tend to move toward particular modes in homogeneous populations, and because our model involves a large number of economic and noneconomic factors, data from many different individuals, groups, and time periods must be brought to bear upon the estimation problem. To begin this task we will present a cross-section model using recent data from 70 nations, and simulations of three countries, an average country, high investment Japan, and low investment Burkina Faso.

THE MODEL

Overview

According to conventional utility maximization theory, individuals directly and instantaneously choose specific amounts of various goods based upon relative (but limitless) desires and prices of goods. This model cannot adequately account for large changes in utility attributed to educational investments over time and across individuals and countries.

We will attempt to model changing investment choices through an alternative dynamic three stage model. First, as in Hoch and Loewenstein (1991) individuals arrive at "desired levels" of consumption. However, desires are formed with respect to traits rather than goods. Desired levels differ from demanded levels because they are not completely constrained by prices and incomes. In the second stage agents predict the degree to which available goods contain desired characteristics. Third, consumers choose goods by minimizing the distance between predicted and desired characteristics, rather than by maximizing utility. Then, the process begins anew.

The three staged process is consistent with Prospect Theory, which distinguishes two phases in the choice process--an early editing phase and a subsequent phase of evaluation. In addition, marketing psychologists point to "evoked sets" in what are essentially "systems models" of decisionmaking. Evoked sets represent collections of characteristics that play a role in choice at a particular time period. Furthermore, Jon Elster (1990), in Nuts and Bolts for the Social Sciences explains actions through two successive filtering operations. According to Elster we begin with a large set of possible actions. The first filter is made up of physical, economic, legal, and psychological constraints. The actions consistent with these constraints form an agent's opportunity set. The second filter determines which action within the opportunity set will actually be chosen. Formation of Desired Levels

Many economists would argue, and conventional economic theory assumes, that our desires are infinite. However, this notion is inconsistent with common sense and introspection. At time t we have no desire for infinite amounts of food, or water (which often has a price of zero), or any other good--with the exception of money. Some may wish to consume, possess, or have access to, say 5 ounces of pasta, 6 ounces of vegetables, and 10 ounces of fish. Others may desire two or three times this amount--but no one would truly wish to possess, much less consume, an infinite amount of any good over a specified time period. Our notion of desired levels assumes that utility functions with respect to a single good or trait take the form displayed in the graph below. The utility functions increase initially, but decline at some point, or at least reach a maximum.

This maximum may result from time constraints, storage costs, social inhibitions, ethical values, or diminishing marginal utility.

As in Prospect Theory we utilize a subjective notion of utility closer to the original concept developed by Jeremy Bentham (1789), William Stanley Jevons (1911), and Leon Walras (1874). More recently, John Hicks (1976) and Thomas Juster (1992) have argued in favor of cardinal utility.

Perhaps many economists would feel more comfortable if we derived "desired levels" from constrained maximization, in which case utility is initially maximized with respect to time constraints, storage costs, and/or ethical values.

Conversion of the balance of the report to HTML is not feasible at this time.

To receive the full WordPerfect version of the report click here