Ln density 0048 0051 0033 0040 0042 0026 0040 0044

This preview shows page 13 - 19 out of 61 pages.

ln(density)0.048 0.051 0.033 0.040 0.042 0.026 0.040 0.044 0.027(0.002)a(0.002)a(0.001)a(0.003)a(0.002)a(0.002)a(0.003)a(0.002)a(0.002)aInstruments used:ln(1831 density)---YYYYYYln(1881 density)------YYYFirst stage statistics---395.7 395.7 395.7 518.7 518.7 518.7Over-id testp-value------0.990.190.21R-squared0.560.720.65------servations for each regression.W1= raw wage,W2= wage net of sector and individual observables (but noteducation),W3= wage also net of individual effectsIFrom this literature, typically get that agglo. externalities elasticity0.02to 0.05INote: elasticities tend to be higher when micro-event are studied: 0.1-0.3Econ 280D. Spring 2019. C. GaubertLecture 1Agglomeration: Evidence12 / 59
Quasi-experimental evidence approachIGreenstone, Hornbeck and Moretti (2010)IGHM look at the effect that “million dollar plants” (huge industrial plants)have on incumbent firms in the vicinity of the new plantIConsider the following example (from paper):IBMW did worldwide search for new plant location in 1991. 250 locationsnarrowed to 20 US counties. Then announced 2 finalists: Omaha, NB andGreenville-Spartanburg, SC. Finally, chose Greenville- Spartanburg.IGHM obtain list of the winner and loser counties for 82 MDP openings andcompare winners to losersIrather than comparing winners to all 3,000 other counties, or to counties thatlook similar on observablesEcon 280D. Spring 2019. C. GaubertLecture 1Agglomeration: Evidence13 / 59
Greenstone, Hornbeck and Moretti (JPE 2010)IData: Annual Survey of Manufactures (ASM) and the Census ofManufactures (CM) from 1973 - 98.Iemployment, capital stocks, materials, total value of shipments, firmidentifiersISIC and locationIFollows firms in winning and loosing counties over timeI+ use measure of closeness between industries, to detect heterogenous effects(IO linkages, workers rotation, RD/patent linkages)IStrategy: TFP regressions (OLS residual), with pre-post MDP opening leveland trend effectsEcon 280D. Spring 2019. C. GaubertLecture 1Agglomeration: Evidence14 / 59
ResultsIFive years after the new plant opened, incumbent plants in winning countiesexperienced a sharp relative increase in TFPFig.1.—All incumbent plants’ productivity in winning versus losing counties, relativeto the year of an MDP opening. These figures accompany table 4.Econ 280D. Spring 2019. C. GaubertLecture 1Agglomeration: Evidence15 / 59
ResultsIFive years after the new plant opened, incumbent plants in winning countiesexperienced a 5-12% relative increase in TFPIChannels? spillovers stronger between firms in industries that share workersand use similar technologies. (interaction term w/ measure of industrydistance)ILabor pooling, knowledge transfer rather than input/output linkagesIIncrease in skill-adjusted wages (+2.7%) – signals increase in local labordemandIIncrease in firm entry (+12%) – signals increase in profitability, at least inthe short runEcon 280D. Spring 2019. C. GaubertLecture 1Agglomeration: Evidence16 / 59
Agglomeration and persistenceIBleakley and Lin (QJE, 2012) provide indirect evidence of agglomerationexternalitiesIThey look for an event that removed a location’s natural (i.e. exogenous)productivity advantage/amenity.

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture