1 2011 11
(These facts are in line with conven-tional wisdom and easily confirmedusing aggregate census data.) There-fore, looking inside the “black box”of agglomeration economies oftenrequires creative research strategies.Recent work in this area, including myown, has been made possible by theincreasing availability of large data setsthat contain detailed information atthe plant, household, or worker level.Using micro-data, it is sometimes pos-sible to test predictions that are uniqueto one kind of agglomeration economyand not associated with another kind.In this way, it becomes possible tohighlight variables that should be of interest to policymakers.I will describe the evidence foragglomeration economies from jobsearch and matching using just sucha strategy. An important caveat isthat the research strategy describedhere does not rule out other sources of agglomeration economies. Instead, Ievaluate whether there is evidence forthis source of agglomeration economiesand then ask whether it may be largeenough to offer meaningful explana-tions for differences in productivityand density.
JOB SEARCH AND MATCHINGIN CITIES
In my recent working paper withHoyt Bleakley, we test for agglomera-tion economies from job search andmatching. The intuition for our testis as follows. Consider a worker in asmall city who loses her job. She hassome specialized skills (either innateor gained through experience) suitedto the activities she performed or theoutput she produced in her previousjob. If the separation from her previousjob is permanent, the worker now facesa choice: She could wait a long timebefore finding employment performingsimilar tasks but at a different firm. Or,because waiting is costly, it may makemore sense to accept a job elsewherein the local economy that is less suitedto her unique skill set. (Alternatively,she might choose to move to a locationwhere there is greater demand for herskills, but of course, moving is alsocostly.) Since her skills are less suitedto this job, some of her skills go un-used, and she may be less productive.This worker, in a small city,faces a “small numbers” problem: Shehappens to be without a job, but doesthere happen to be another firm thatneeds a worker with her skill set? Onthe other hand, workers in dense citiesbenefit from market thickness: Theyare less likely to be in a narrow labormarket at a moment in which theirskills are in excess supply. This poten-tial source of agglomeration economiesyields an interesting, and potentiallyunique, prediction: Workers shouldchoose to eschew their specializedskills
in large, dense cit-ies, where they are more likely to findjob openings suited to their talents.We evaluate this prediction byexamining the likelihood that workerschange
. Thesejob classifications, characterizing eitherthe tasks or activities performed orthe kinds of output produced, havebeen used in a number of labor-marketstudies on specific human capital.
Weexpect that in the presence of agglom-eration economies from job search andmatching, workers should choose tochange occupations and industries lessfrequently in denser labor markets.Further, this agglomeration econ-omy should also affect workers’ earlydecisions about skill specialization.In separate studies, economists KevinMurphy and Sunwoong Kim haveproposed how density might changethe market for specialized skills.In Kim’s model, sparsely populatedareas have fewer firms in each sector,and therefore, a worker might haveinvested less in narrow skills becauseshe anticipated that there would befewer potential employers in the eventof a separation.
Therefore, in largecities, workers choose to invest more inspecialized skills, making it even lesslikely that they would want to changeoccupations or industries in densecities and compounding density’s effecton productivity.
Using data from the decennialU.S. census and the monthly CurrentPopulation Survey (CPS), Bleakleyand I confirm this prediction. We findthat workers are less likely to changeoccupation or industry in metropolitanareas with high population density(Figure 1). The data are at the workerlevel, and the key outcome of interestis a change in each worker’s reportedoccupation or industry.
Respondentsto the 1970 census reported thesechanges for 1965 and 1970. TheCPS samples in the 1990s and 2000sreported these changes for individualworkers, both for the year of the surveyand up to three years earlier. The keyexplanatory variable is local populationdensity, measured for each worker’smetropolitan area of residence. Figure1 summarizes our main result. Here,each point represents a metropolitan
For example, see the study by Derek Neal andthe one by Daniel Parent on industry-specificskills; see Gueorgui Kambourov and IouriiManovskii’s recent paper on occupation-specificskills.
Alternatively, workers in small cities withspecialized skills might choose to move todenser cities.
For example, James Baumgardner foundthat doctors are more specialized in big cities;similarly, Luis Garicano and Thomas Hubbardfound more specialization among lawyers inlarger markets.
We obtain similar results whether our outcomeof interest measures a change in each worker’sreported occupation, a change in reportedindustry, or a change in either reported occupa-tion or reported industry.