Are There Urbanization Economies in a Post-Socialist City? Evidence from Ukrainian Firm-Level Data Volodymyr Vakhitov Saint Petersburg October 11, 2012
2 Outline Motivation Data description Model and Estimation Issues Preliminary results
3 Agglomeration in the Nutshell ? Common labor pool? Relationships between managers and/or owners? Common market?
4 Motivation: Objective Agglomeration economies: External (to the firm) economies of scale Localization: Internal to the industry Internal to the location Urbanization: External to the industry Internal to the location
5 Motivation: Urbanization Economies Urbanization economies: external to the firm and the industry as a whole, internal to the location Jacobs (1969): spillovers matter! Innovations Knowledge sharing Borrowing and developing new product [Schumpeterian] churning
6 Motivation: Important Issues How to measure ( Research Policy, 2009 )? Size (employment, # of plants) Diversity (similarity and concentration indices) Are measures comparable (robustness)? Cluster boundaries (aggregation): What is “the same location”? What is “the same industry”?
7 Motivation: Post-Socialist City Land “price” and previous allocation “Lock-in effect” within city boundaries Ownership issues Outdated capital stock
8 Outline Motivation Data description Model and Estimation Issues Preliminary results
9 Data Description Registry: “Official” data (National Statistics Office) Panel ( ), submitted by firms Firm level (plant level on the way) Excludes budgetary sector and banks
10 Data Description Ownership: “Official” data (State Property Fund and FDI statements) Panel ( ), collected by SPF Firm level only
11 Data: Sample Composition YearUrban firms, Nonzero Output, K, LManufacturingSample* SmallBigTotal ,58714,736126,32321,34619, ,08514,334137,41922,76220, ,66814,008144,67623,19421, ,81014,644150,45423,75921, ,85715,333155,19024,14821, ,28715,543158,83024,66821, ,51115,985164,49624,13821,496 Total932,805104,5831,037,388164,015147,857
12 Data Description: Raw Variables Territory and industry codes Output, employment, capital, materials Ownership (private: domestic / foreign) Output assortment Innovation activity FDI Purchases from other sectors Exports and imports
13 Data Description: Constructed Variables Size: Other employment in the city ln(empl) = ln(# firm) + ln(empl/# firms) (Henderson, 2003) Diversity: Market: employment or firm count based HHI Internal: output composition HHI Share of imports in the inputs Use of products from other sectors produced in the same city (!!!)
14 Outline Motivation Data description Model and Estimation Issues Preliminary results
15 Model TFP Model (Rosenthal and Strange, 2004): Econometric Specification (Henderson, 2003):
16 Estimation issues One-stage and two-stage estimation Restricted to manufacturing in cities Area and industry-year fixed effects Robust estimation 2-stage: TFP by Olley-Pakes, then regress on agglomeration variables + controls
17 Outline Motivation Data description Model and Estimation Issues Preliminary results
18 Production Function Results Size-based ln (Employment)0.479*** ln(Capital)0.058*** ln(Materials)0.461*** … … Observations151,775 Number of clusters41898 R-squared overall0.85
19 Basic specification (XT-FE) SizeDiversity ln(Local employment)0.040 (0.022) ln(# firms) 0.059* (0.028) ln(average employment) (0.027) “Anti-diversity index” (0.020) Multi-plant0.115*** 0.112*** (0.029) (0.030) Firms41,898 41,240 obs151, ,857 Overall R
20 Olley-Pakes Two Stage (TFP) SizeDiversity ln(Local empl)0.043* (0.019) ln(# firms) 0.044* (0.022) ln(Average empl) (0.025) “Anti-diversity index” ** (0.018) Multi-plant firm (0.028) Firms25,906 25,541 obs110, ,737 Overall R
21 “Urban depreciation” Ratio: Current value of all fixed assets in the city to Historical value of all fixed assets in the city Predicted effect: the higher, the better
22 “Urban depreciation” SizeOP (1 stage)Diversity Local employment0.104*** 0.418*** (0.038) (0.550) # firms 0.003*** (0.001) “Anti-diversity index” (0.023) Urban depreciation0.080***0.085*** *** (0.038)(0.037)(0.060)(0.038) Industry* Year f.e.yes City f.e.yes firms42,02641,89826,90941,240 obs150,311151,77573,870147,857 Overal
23 Ownership variables PO: privately owned DO: private, majority domestic FO: private, majority state PO = FO + DO
24 Ownership variables StateDomesticForeignTotal 20011,07219, , , , , , , , , ,855 Total4,751105,1113,845113,707
25 Ownership results SizeDiversity ln(Local employment)0.032 (0.023) Primarily domestic0.169*** 0.174*** (0.021) Primarily foreign0.285*** 0.292*** (0.028) ln(# firms)0.026 (0.030) ln(average employment)0.038 (0.028) "Anti-diversity index" ** (0.021) Multi-plant firm0.068**0.069**0.067** (0.025)
26 Innovations and ownership Ownership+ Innovations ln(Urban employment) ln(Urban # of firms)0.363***0.388*** Employment - based HHI Firm-based HHI-0.166***-0.165*** ln(Number of innovating firms) ln( innovation expenditures)0.007* Majority private, domestic0.285*** Majority private, foreign0.593*** 0.059
27 Conclusions and implications Urbanization economies seem to be present More pronounced for firm counts based measures, than labor based Urban capital depreciation matters Ownership effect: foreign – private domestic – state.
28 Innovations seem important, but total stock would be a better measure Past values of innovation expenditures insignificant Ownership is clearly important, but cross- effects and lags may better show channels Estimates may vary by sub-industries Past capital stock (lock-in effect)? Conclusions and implications
29 Volodymyr Vakhitov Kyiv School of Economics :