Read an Excerpt
CHAPTER 1
"Job Search" and Economic Theory
There is some overlap between the subject matter of the present study and that of labor-market theory in economics. In the classical conception, labor is a commodity, like wheat or shoes, and is hence subject to market analysis: employers are the buyers, and employees the sellers of labor. Wages (or, in more refined formulations, the total benefits accruing to a worker by virtue of holding a given job) are analogized to price. Supply and demand operate in the usual way to establish equilibrium: the price of labor fluctuates in the short run until that single price is arrived at which clears the market. For homogeneous work, wage dispersion and unemployment are not possible; firms paying more than the equilibrium price for labor will thereby attract workers from firms paying less. This excess of supply over demand will drive down the price. Firms losing employees will similarly be constrained to raise wages. Workers unemployed in the short run may bid for work, driving down wages to the point where they, and those currently working, will all be employed at the new, lower equilibrium wage. This elegant package ties together wages, unemployment, and labor mobility.
Like perfect commodity markets, however, perfect labor markets exist only in textbooks. Unemployment, obviously, persists. On wage dispersion, a recent text on labor economics summarizes a number of empirical studies by saying that even "in the absence of collective bargaining, employers will continue indefinitely to pay diverse rates for the same grade of labor in the same locality under strictly comparable job conditions ... There is no wage which will clear the market" (Bloom and Northrup, 1969:232). Reynolds, in a detailed empirical study of New Haven, concluded that labor mobility and wage determination are more or less independent; the movement of labor has little effect on wages, and "voluntary movement of labor ... seems to depend more largely on differences in availability of jobs than on differences in wage levels" (1951:230, 233). Brown, in his study of college professors, divided disciplines into those with excess supply and excess demand for teachers. He naturally assumed that job-changers in excess-demand disciplines would have received more job offers than those in fields with excess supply. Though there was some tendency in this direction, he reports that it was "not decisive," and that "repeated attempts to explain the differentials in market behavior by dividing the disciplines into excess supply and excess demand have not produced any conclusive evidence of the expected, usual relationships" (1965b: 117, 118n; for the basis of the excess supply and demand index see pages 87–91, 354, 361).
Several factors militate against perfect labor markets. Inertia as well as social and institutional pressures exert constraints on the free movement of labor contemplated in economic theory (cf. Kerr 1954; Parnes, 1954). Union agreements and community restraints discourage employers from adjusting wages to meet supply and demand (cf. Reynolds, chs. 7–9). The factor most relevant to the present discussion is imperfection of information.
The neoclassical theory of commodity markets generally takes the possession of complete information by market participants as one requirement of a perfect market. In Stigler's widely cited treatment of price theory, this is described as a sufficient condition (1952:56). But it is not simple to say exactly what complete information means, as Shubik noticed (1959), in trying to construct game-theoretic models for the behavior of actors in commodity markets. Alfred Marshall, a founder of the neoclassical synthesis, hedged on the issue, saying that it was not "necessary for our argument that any dealers [buyers or sellers] should have a thorough knowledge of the circumstances of the market" (1930:334). He felt that so long as each participant behaved strictly in accord with his supply or demand schedule, the equilibrium price would ultimately be reached. He does seem, in this argument, to assume that buyers and sellers are at least aware of the identities of all those they might transact business with and even their current bid (or price) — only not necessarily their entire supply or demand schedules. Stigler comments that the "New York City market for domestic service is imperfect because some maids are working at wages less than some prospective employers would be willing to pay, and some maids are receiving more than unemployed maids would be willing to work for" (Stigler, 1952:56). This is caused primarily, one would guess, because the underpaid maids do not know the identity of very many potential employers, nor do the overpaying employers know who is available at a lower wage. Clearly, knowledge of the identities of these people is prerequisite to determining under what conditions they will offer (or purchase) services.
While there is disagreement on just how much information actually is possessed by workers in various labor markets, it seems clear that there is considerable ignorance. Reynolds holds that workers' "knowledge of wage and nonwage terms of employment in other companies [than their own] is very meager ... much of what workers purport to know about other companies is inaccurate" (1951:213). Even less is known about the general state of knowledge of employers; all would agree, however, that few employers know of all or most individuals who could potentially fill vacancies they have open.
Only in the last ten years have economists begun to suggest how information is obtained and diffused in markets. Most of the models presented deal with "search" behavior, the active attempts of buyers and sellers to determine each others' identities and offers; maximization of utility by rational actors using marginal principles pervades these models. Stigler, the first to present such analyses, asserted that if "the cost of search is equated to its expected marginal return, the optimum amount of search will be found" (1961:216). In his conception, cost of search, for a consumer, is approximately measured by the number of sellers approached, "for the chief cost is time" (1961:216). He does not consider how buyers and sellers determine one anothers' identities, an issue of particular importance in labor markets, but also with some applicability to general commodity markets. Indeed, one might want to make a clear distinction between two stages of search: 1) finding the buyers (or sellers), and 2) determining their offers. Adapting some comments of Rees (1966), we might call the first part the extensive aspect of search, and the second, the intensive. This dichotomy applies less to markets in highly standardized commodities, where the nature of an offer is as straightforward as the identity of the person making it. But in, for example, labor markets, there are many subtleties in the nature of a bid to employ or to offer services, and gathering information about each such bid may be much more time-consuming than finding out who is making bids. In practice, a job-searcher must make a tradeoff between the two aspects: the more people he discovers who are bidding, the less he will be able to find out about each bid.
Of course, some intensive activity will precede hiring, especially in higher level jobs. In commodity markets, the knowledge that a given person is offering wheat at K dollars per bushel ordinarily suffices to be sure that you will be able to buy it from him. At worst, he might sell you less than you want (wheat being entirely divisible, as classical commodities should be). But having specified an employer offering an acceptable wage, or having found an employee offering his labor at a price one is willing to pay, does not by any means guarantee the consummation of the transaction. Especially in higher-level jobs, a direct inquiry is generally felt to be necessary, in which prospective employer and employee learn more about each other and decide whether a job should be offered, and if offered, accepted.
In general, measuring the costs and benefits of search poses quite difficult problems. The proposal to consider time as the main cost (see also McCall, 1970) is more appropriate for blue-collar workers who cannot easily search during 9to-5 jobs, than for PTM workers. A very important cost for them, on the other hand, involves their frequent use of personal contacts. In my sample, more than 80 percent of the personal contacts used not only told the respondent about his new job, but also "put in a good word" for him. Contacts cannot be asked to do this too often without the respondents' using up their "credit" with them, straining the relationship. There are, moreover, as Brown points out, opportunity costs in searching for a particular job (or employee): one may necessarily forego searching for others (Brown, 1965b: 187). Brown's model, similar to Stigler's in its assumptions of optimal search behavior, allows for these other costs (1965b: 185–198). His formulation, however, could be operationalized only via extensive survey data.
Similar difficulties arise on the benefit side; all models found measured benefits in money terms (Stigler, 1962; Brown, 1965b; McCall, 1965, 1970). Stigler and McCall recognize that the future benefits of the present search must be taken into account. McCall interprets this primarily in terms of expected length of employment in a job; Stigler's more general formulation points out that if current price offers are correlated with future ones, information now found also has future benefits. Each suggests appropriate discounting procedures. No attempts are made, however, to specify the value in the future of holding a particularly prestigious job now. More relevant, even, from the point of view of my study, no attempt is made to assess the value of contacts acquired in a particular position. This may be psychologically a minor factor; many of my respondents had never realized, until the time of interview, how much of their career was mediated by personal contacts acquired in previous jobs; but the actual benefits may be considerable, as discussed later on in Chapters 5 and 6.
The primary contribution of the present study to this discussion lies in an analysis of the notion of "search," as viewed from the supply side of labor markets. Of the authors surveyed, only Brown tries to work into his theory the idea that different methods of search yield different amounts of information. Brown's idea is that job-seekers, in effect, compute costs and benefits for each method they might use. That method with the highest expected net benefit is used first; then all calculations are repeated, and the searcher chooses a method for his second try. Search continues until marginal benefits equal marginal cost. Time, in this model, is effectively omitted as a parameter, since after any method is used, the next time period is assumed to begin (1965b: 191–198).
While this account is closer than others to my empirical findings on labor-market behavior, it is nevertheless inadequate. One of my original motives in choosing a PTM sample for study was, in fact, my interest in observing sophisticated search procedures; I assumed that if anyone would be likely to search in a careful, effective way, it would be people in PTM jobs. My results, however, lead me to doubt that information in labor markets, at least in PTM markets, is diffused primarily by "search."
For blue-collar markets, Reynolds holds that "the core of the effective labor supply at any time ... consists of ... people who are entering the market for the first time, who have been discharged, who have quit their previous jobs because of dissatisfaction, or who have been unemployed long enough for their benefit rights to be exhausted" (1951:106). It is reasonable that all such people will be searching for jobs, whereas those employed full-time at the blue-collar level would do so only with difficulty. Yet, Reynolds reports that 25 percent of those in his sample who changed jobs had lined up a new job before quitting, and moved to better jobs than those who quit or were discharged without having set up a new job first (1951:215). Given the 9-to-5 character of most blue-collar work it seems doubtful that those who found a new job before leaving the old one did much searching.
While PTM workers probably have more time to search, evidence indicates that this is often not the route to job-change. Brown reports that 26 percent of his sample of college professors, when asked how they found their present job, said that they "did nothing and were recruited" (1967:119). In my PTM sample, 29 percent of the respondents answered "no" to the question of whether there was a "period of time when you were actively searching for a new job" (before finding the current one).
Moreover, whether a respondent actively searched is systematically related to the method by which he found his job, and to the nature and quality of job obtained. Stage of career is of some importance in predicting search behavior. Table 7 indicates that those in their first job, or in sixth or subsequent jobs were most likely to have searched; the latter cases are individuals changing jobs frequently on account of dissatisfaction. The tendency to search decreases with age; those under 34 searched in 78.6 percent of the cases, those 34 or over in 63.9 percent (p = 0.01).
Amount of education is unrelated to search behavior, but a measure of the prestige and quality of college attended for bachelor's degree shows interesting results. Table 8 shows that far fewer of those who attended colleges in the "top" 40 percent searched for their jobs than those in lower prestige colleges. A very similar table could be produced for the ranking of institutions granting graduate degrees to these respondents.
A first temptation is to conclude that the quality of education received in better colleges makes one more desirable afterwards; it is also possible is that those attending better schools are preselected, regardless of educational quality of the schools, so that they would be more likely to be sought out later. While there may be some truth in each of these ideas, I would also suggest that contacts acquired at higher prestige colleges are generally better placed in the occupational structure and will ultimately be of more help to their protegés — more likely to be in a position from which they can seek them out to offer or inform them of a job.
It is of special interest that, as shown in Table 9, higher-income jobs are less likely to go to those searching for them. Moreover, in the PTM group (see Table 10), managers are least likely to find their jobs through search, and technical workers most likely. This is a logical result of the greater reliance of the latter on formal means and direct application.
How one's job was found is closely related to search behavior. The use of formal means or direct application implies active search; in such cases, the search does lead to the information that yields a new job. For those finding a new job through contacts, however, the situation is more complex. First, only 57.4 percent of these individuals report having actively searched. Moreover, in many cases, the job taken was not found as a result of this search. When a contact was the source of job information, therefore, it is interesting to ask on whose initiative — respondent's or contact's — the job-information was passed. If the respondent was not searching, his contact is likely to have taken the initiative; but this is often true also when the respondent was searching, as when the contact heard through the "grapevine" of the search, or passed the information without knowing that any search was in progress.
For 57.9 percent of the individuals finding their new job through contacts (N = 157), the interaction during which job information was passed was, in fact, initiated by the contact. In about half these cases, he knew that the respondent was looking for a new job; this means that a little over a quarter of the time, initiative came from someone who had not been approached and did not know whether his friend would even be interested. In another 20.9 percent of the instances, the respondent contacted his friend, asked him if he knew of anything, and was told about the job he subsequently took; 8.3 percent of the respondents were contacted by someone they did not know and were told that they had been recommended for a job. The person doing the recommending turned out to be a personal contact of the respondent. In 13.4 percent of the cases, the respondent and his personal contact were interacting for some purpose unrelated to job information; in the course of this meeting, the information happened to be transmitted. Some of these were instances of "bumping into" friends on the street or at professional meetings; others involved prearranged meetings, but for other purposes. (The "interaction unrelated to job information" category is probably underestimated since it was not available as a choice in the mail sample — a few cases would be coded that way from written-in comments. Over 24 percent of interview cases fall in this classification, but only 5.5 percent of mail surveys.)
(Continues…)
Excerpted from "Getting a Job"
by .
Copyright © 1995 Mark Granovetter.
Excerpted by permission of The University of Chicago 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.
