Knowledge that Transforms

In this paper, I show the need for and then derive a model in which the number of vacancies contacted by a job searcher depends on the intensity of search, with diminishing returns to search intensity. The usual assumption of the search literature (for example, see Dale Mortensen, 1970a) is that the number of vacancies contacted per period by a searcher is fixed. Such fixity is unsatisfactory for two obvious reasons. First, participants can vary the intensity of search and consequently can vary the number of vacancies contacted per period. Second, the number of existing vacancies changes over the business cycle so that, for a given search intensity, the number of vacancies contacted should change, too. Besides these flaws, the assumption of a constant vacancy contact rate turns out to imply that a participant never chooses to work and search simultaneously. Simultaneous work and search requires that the function describing a searcher's vacancy contact rate exhibit diminishing returns to search intensity, a property I call diminishing physical returns to search. The requirement for diminishing physical returns holds even when the individual maximizes a concave utility function; diminishing utility returns to search are insufficient to allow simultaneous work and search. Thus, for example, Mortensen's (1977) discussion of simultaneous work and search is inappropriate in his utility-maximizing model in which the vacancy contact rate is a linear function of search intensity. Even though simultaneous work and search requires diminishing physical returns to search, it is not immediately apparent what the source of such diminishing return would be.' At first glance, it may seem that if a searcher doubles his search intensity, he should double the number of vacancies he contacts. However, under the assumption that search requires travelling to firms to seek vacancies (rather than calling on the telephone, for example), it turns out that an increase in search intensity leads to a less than proportional increase in vacancy contacts. The reason is essentially that increasing the number of vacancies contacted requires an increase in the area over which search is conducted and the travel time required to cover this area increases faster than the area to be covered. Thus diminishing returns arise for spatial reasons. The model developed in this paper proves the foregoing propositions and also provides some other interesting insights, the most important being a nullification of the criticism by James Tobin, Robert J. Gordon, and others that search theory produces incorrect predictions of the cyclical behavior of quits.

Our 1972 paper originally proposed that discriminatory outputs are necessarily greater than nondiscriminatory outputs under conditions of spatial monopoly. This proposition was proved under a linear demand function and in a sense generalized in the same paper by intuitive speculation which indicated that the result would hold for non-linear demand cases as well. That speculation was confirmed later in our 1975 book, where we proved that it holds for all non-linear demands of the particular form given by

The major difference between segmented labor market and human capital theories about the labor force behavior of women lies in the attention paid to micro vs. market-wide or macro variables. In a study such as James Heckman's (1976) which includes no market variables, the demand for the labor of married women in any given education-experience (and hence offered wage) class is implicitly assumed to be infinitely elastic. Thus the observed differences in the labor force behavior of individual women are attributed entirely to differences in supply characteristics such as education and child status. On the other hand in segmented labor force analyses, such as Barbara Bergmann's and lrma Adelman's study, the macro phenomenon of occupational segregation by sex is seen as the major factor affecting the participation, wage rates, and hours of work of women. These two types of studies lead to different explanations of why the labor force participation of women has increased in recent years. Different sets of government policies aimed at improving the labor force situation of women are also implied. In this paper we present a model of the labor force behavior of married women in which both individual and family decision making, and macro labor market conditions are found to play important roles. An unemployment variable and an index summarizing the ratio of expected available local job slots for women to the potential female labor force population are incorporated into a marginal utility analysis of the labor force behavior of married women in Canada. The inclusion of the local opportunity for jobs variable is supported by detailed evidence on the labor force segregation of women in Canada. Consistent estimation results are presented for eleven age groups in a probit analysis of whether or not a married woman works, and for eight age groups in equations estimating the offered wage rates and annual hours of work of married women who do work. One unexpected finding is that working wives in Canada tend to work fewer hours per year when paid more per hour. This is contrary to the findings of other researchers for the United States, and has important policy implications. Although it is possible that our results differ from those of other researchers solely because we have analyzed data for another country, we argue in Section V of this paper that the difference in results is more likely due to differences in the form in which the labor supply function for wives is estimated and the choice of the variables which are used to control for child status. Our resulting uncompensated wage elasticities of hours of work are shown to be very similar to those reported by other researchers for men. The data base used in this study is the Family File of the first Public Use Sample to be made available from a Canadian census. Combined grouped R2s are presented showing the extent to which our equations explain the observed macro variations in the labor force behavior of married women classified by various characteristics. Finally we use our estimated model to see what changes we would expect in the labor force behavior of a hypothetical 41-year-old wife living in a small city in New Brunswick given a variety of changes *Faculty of Business Adminstration and Commerce, University of Alberta. For further computational results and theoretical arguments supporting various statements in this paper, see our book. The work for this paper was supported in part by the Statistics Canada-SSRCC Programme of 1971 Census Analytical Studies, and by the Faculties of Graduate Studies and Research and of Business Administration and Commerce of the University of Alberta. The empirical results in this paper are primarily based on Public Use Sample Data derived from the 1971 Canadian Census of Population supplied by Statistics Canada. The responsibility for the use and interpretation of these data is entirely ours. We would like to thank T. Daniel, K. Gupta, anonymous referees, and the managing editor for their helpful comments, and James Heckman for making available to us some of his work which had not yet been published.

Coming to me as it did after almost a decade's absence from the academic profession, I accepted this invitation only with trepidation. While the profession has been extending the frontiers of economics, I have been operating deep within its margin, first discovering how dismal our science really can be as it applies to the finances of private universities, and then, during the past four years, applying to the real world economic principles that Alfred Marshall would have had no difficulty recognizing. I have no particular interest in describing the first of these experiences. Its only lessons were that' the laws of economics are truly made of iron; and that any organization that hopes to make the best use of its limited resources had better be organized more hierarchically than a university. The experience of being a practitioner of regulation, in contrast, has been immensely satisfying, because it has afforded almost unlimited opportunity for the application of simple micro-economic principles to the real world. The applicable principles are easy to characterize: that economic efficiency calls for prices equated to marginal social opportunity costs; and that, whenever it is technologically feasible, competition is the best institutional mechanism for achieving that result, as well as for minimizing X-inefficiency and ensuring the optimum rate of innovation. What has been especially intriguing about my experience is that it has embraced two quite different regulatory situations-one, the traditional public utilities, where competition seems for the most part infeasible, and the economistregulator is moved to play an active role in trying to produce efficient results; the other, airlines, in which it appears the prime obstacle to efficiency has been regulation itself, and the most creative thing a regulator can do is remove his (and her) body from the market entryway. But the process of applying these principles-even of simply getting out of the wayhas been far from simple. The slate on which the economist-regulator writes is scribbled with the scratchings of lawyers, jurists, and politicians; the world to which he would apply his principles is excruciatingly imperfect and resistant; and the compass he needs is one that would help him thread his way through the thickets of second best. The really challenging job is deciding not what the ultimate economically rational equilibrium should look like, but what is economically rational in an irrational world, and how best to get from here to there. That, too, turns out to be a kind of frontier; and life on it is full of excitement.

The substitution of other factors, such as capital, labor, and materials for energy is a central issue in formulating energy policies. Models have been designed to determine the elasticity of substitution between these factors and how much government intervention is needed to commercialize new substitutions. Pooled cross-national time-series data for seven countries are used to estimate a translog cost function and price elasticities of substitution and demand. The intercountry differences are found to be small for the U.S., West Germany, and Japan. This indicates that pricing policies could play a major role in promoting energy conservation as U.S. industries shift to the technologies already employed by the other countries. 5 references.

A theoretical model is presented whereby the relationships among fertility child quality womens wage rates and labor supply can be studied. The model utilizes a disaggregate multivariate household approach. The theoretical underpinnings of the multiple equation model are described. The data base and variables that were used are discussed. The focus of the model is a wifes lifetime labor supply as influenced by fertility decisions price and income variables instead of current labor force participation. The model shows that the number of children desired responds negatively to their cost and positively to family income. Desired family size feeds back negatively to mothers market earning power. Contrasting to previous model evaluations this model shows a negligible influence on the labor force participation of married women by the number of children in the family. Future research should focus on exogenous forces impinging on the decision to have children.

My view in the early 1970's of Keynesian, non-market-clearing-type models was that the soundness of their theoretical structure hinged on an yet absent theory of the stickiness of wages or prices. The application of contracting theory to macro analysis seemed promising in this respect. The presence of employee risk aversion or of transaction costs associated with market arrangements which could include elements of capital that were specific to employment or other aspects of production and exchangeseemed to motivate some long-term, implicit or explicit agreements about wages or prices. In particular, a sluggish adjustment of wages to current economic conditions could be rationalized by this approach. Further consideration of the contracting model suggests that its rationale for wages and prices-as far it goes does not explain the key features of Keynesian analysis with regard to the determination of employment and output. For example, long-term labor agreements do not imply a failure of employment to increase when all parties to the agreements perceive that they could be made better off by such a change. The socalled involuntary unemployment of Keynesian models that is, a situation where everyone perceives accurately that the marginal product of labor exceeds the marginal value that potential workers place on their time-is not compatible with efficient labor agreements. Even in contracts that specify, ex ante, the value of nominal wages over some interval of time, it would be mutually advantageous for workers and firms to determine levels of employment in an efficient manner. The contracting approach may rationalize some departures of real wages from the marginal product of labor and/or the marginal value of worker time, but it does not imply that levels of employment would differ significantly from the (efficient) values that would have been attained under flexible wages. Rather than rationalizing the non-marketclearing model a useful as if approach, contracting analysis suggests that-despite the possible existence of sticky wages-the continuous market-clearing model may provide a satisfactory framework for the analysis of employment and output. Notably, the approach suggests that such market features wages or the apparent non-price, quantity rationing associated with layoffs would be of secondary interest in analyses of business cycles. Since the prevailing wage need not represent the marginal product of labor, the presence of excess labor supply at this wage need not signal involuntary unemployment in any economic sense. The conclusions derived from the contracting model can be generalized by observing that the key assumption of Keynesian analysis is the inefficiency of some aspects of private sector activity in comparison to corresponding activities carried out by the government. This central feature is, of course, the underlying basis for the policy activism that typifies Keynesian thinking. In some simple disequilibrium macro models, relative private sector inefficiency is represented by wages or prices, in contrast to the flexibility of such government policy instruments the money supply, taxes, or expenditures. Technical limitations of the private market in the coordination of production and exchange-as reflected in wage-price stickiness and the associated determination of employment and output through a non-price rationing process-are remedied through the superior coordinating *University of Rochester. I have benefited from comments by Herschel Grossman, Bob Hall, and Ben McCallum. The National Science Foundation has supported this research.

A common characteristic of a large class of markets is that one side of the market is more informed than the other about the properties of one of the goods being traded. In some instances, this presents no serious problem. If the informed agents deal on a regular basis with the less-informed agents (for example, local grocers, barbers), there may be little incentive for the informed agents to take advantage of their superior information. In other cases, the problem may be avoided if it is profitable for specialists (or some government agency) to provide the information at a relatively low cost (for example, credit agencies, Consumer Reports). Frequently, however, these kinds of market responses provide at best a partial reduction in the informational asymmetry. There may still be substantial benefits to the less-informed agents from acquiring more information. How the market will respond under these circumstances has been the focus of much recent research. Most of the attention, however, has been directed at examining the possibility that a signalling convention will emerge. The essential idea is that sellers of high quality products may choose contracts or invest in observable characteristics which distinguish their products from those of lower quality. Although I believe that signalling is an important and pervasive phenomenon, the conditions necessary for effective signalling to emerge may not always be satisfied. It is important, therefore, that we understand how the allocation of goods is affected in the absence of signalling, when the only variable that agents may use to distinguish quality is the price. This paper provides an overview of some of my recent research on this question. My investigation begins with a welfare analysis of the Walrasian equilibrium. Specifically, the question is whether or not it is necessarily desirable for trade to take place at a price which clears the market. My analysis indicates that it is not. Under some conditions, it may be possible to make every agent in the market better off simply by raising the price. Besides generating some obvious policy implications, this result also suggests that the Walrasian equilibrium may not always be the appropriate equilibrium concept for this model. In a market with homogeneous goods, it is generally argued that independently of how the prices are set, as long as there is a large number of buyers and sellers, competitive pressures will force the price toward a stable Walrasian equilibrium. When an adverse selection problem appears, however, the possibility that some buyers may prefer a price higher than the one which clears the market casts some doubt as to whether such pressures will still be present. It is no longer obvious that the market will clear or even that all trade will take place at a single price. These points can be conveniently illustrated using George Akerlof's model of the used car market. There is a set of cars of varying quality q distributed over an interval [ql, q2] with densityf (q). Each agent in the economy has an identical utility function u(c, q; t) = c + tq where c is consumption of other goods, q is the quality of car he consumes, and t is a parameter equal to his marginal rate of substitution of car quality for consumption. (If an agent does not consume a car, q may be set equal to zero.) The set of agents can be divided into two subsets, those that initially own exactly one car and those that own none. Each owner has the same utility parameter, t = 1; for the nonowners, however, t is distributed continuously over some interval [tl, t2] with density h(t). As long as each owner can directly identify the quality of his own car, the supply curve will have the usual positive slope. A utility maximizing owner with a car of quality q will sell at price p if and only if q _ p. As the price rises, therefore, more cars will be supplied. If *Department of economics, University of Wisconsin. This research was supported by the National Science Foundation under Grant SOC-77-08568.