Fear of Relocation? Assessing the Impact of Italy’s

FDI on Local Employment

Stefano Federico (Banca d’Italia)Gaetano Alfredo Minerva (Università del Piemonte Orientale)

Vienna, June 3, 2005

Background

• Widespread fears that jobs are being lost as a consequence of increased FDI (i.e. increased capital mobility by multinational firms)

• Debate on policy measures against relocation to protect domestic employment

• The issue must be addressed at an empirical level

Empirical evidence

• Labor substitution in multinational companies: how do changes in foreign affiliates’ wages affect employment in parent company? (Brainard and Riker 1997, Braconier and Ekholm 2000, Konings and Murphy 2001, Becker et al. 2005)

Limitation: Results are conditional on the multinational firm already having invested

• Employment performance of firms investing abroad for the first time compared with national firms (Barba Navaretti and Castellani 2004)

Limitation: Results do not take into account the destination of FDI

Our contribution• Extend the analysis to the whole local area to capture the

presence of domestic external effects

• Estimate a “modified” employment growth regression, which includes a measure of FDI outflows by industry and local area (derived from BOP data)

• Apply method to data on 103 Italian provinces and 12 manufacturing industries

• Disentangle the effects of FDI according to the country of destination of FDI

Our data

• Local employment, by industry and local area Data on the universe of local plants and firms Years: 1996, 2001 Allows to compute local employment growth and industrial structure Source: Istat (National Institute of Statistics)

• BOP data on FDI outflows, by industry, destination country and source local area

Include greenfield investments and foreign takeovers Years: from 1997 to 2001 Distribution similar to employment in foreign affiliates (Reprint database) Source: Ufficio Italiano dei Cambi

FDI by destination area

FDI manufacturing outflows (1997-2000)

% total FDI

FDI Advanced 76.8

FDI Developing 13.8

FDI “Small” 9.3

FDI World 100.0

Employment and FDI: IndustryEmpl. Change

(thousands)FDI

(% total manuf.) Biggest losers Textiles, apparel, leather -108.5 5.9Transport vehicles -5.4 13.8Chemical products -3.2 10.2

Biggest gainers Office equipment 6.9 17.0Industrial machinery 44.8 14.2Metal products 82.5 3.9

Employment and FDI: Local areaEmpl. Change

(thousands)FDI

(% total manuf.) Biggest losers Lombardia -53.2 43.3Piemonte -16.7 31.3Lazio -13.7 3.9

Biggest gainers Veneto 16.6 5.8Puglia 16.7 0.7Emilia Romagna 25.9 7.2

Econometric approach

• OLS estimation of modified employment growth regression:

where

• Local industrial structure (at 1996): Specialization, Diversity, Plant size (all normalized to the industry average)

• Spatial controls: 103 dummies (Problem: High number; remedy: stratification, to be added)

)ln()ln(1996

2001

1996,

2001,,

i

i

ip

ipip

LL

LLy

p,ipp,ip,ip,ip,ip,i uXα)(sizeα)(divα)(specαααy 654321 lnlnlnfdipro

The FDI variable: fdiprop,i • Sum of FDI outflows over the years 1997-2000 (excluding

therefore 2001) • We divide FDI by employment in order to control for the size

of the economic activity in each {p,i}• We normalize it to the industry average• This is equivalent to the share of FDI on the industry total

divided by the corresponding share in terms of employment:

i

ip

p ttip

ttip

i

p ttip

ip

ttip

ip

LL

FDI

FDI

L

FDIL

FDI

fdipro,

,,

,,

,,

,

,,

,

Interpretation of fdiprop,i

• If larger than one, FDI per employee in a given {p,i} is larger than the national average in the same industry

OR

• If larger than one, the share of FDI over the industry total is larger than the corresponding share of workers

• Both interpretations point to a high propensity to invest abroad

Results (OLS)[1] [2] [3] [4] [5]

specialization -0.022 -0.022 -0.023 -0.016 -0.024

diversity 0.119*** 0.117*** 0.126*** 0.091* 0.083

size -0.127*** -0.125*** -0.125*** -0.180*** -0.106**

fdipro (world) 0.016** - - - -

fdipro (advanced) - 0.012*** - 0.009** -

fdipro (developing)

- - 0.001 - -0.006

spatial controls 103 103 103 99 81

adj R-squared 0.23 0.23 0.22 0.31 0.22

no. observations 1160 1160 1160 737 435

Tails of dependent variable excluded. In (4) and (5): i) only {p,i} with positive FDI are considered; ii) fdiprop,i is taken in logarithm.

Robustness analysis

• FDI only towards new EU Member States from Central and Eastern Europe

• Sector specificities (Interaction between fdiprop,i and sector dummies)

• Different construction of the FDI variable (share on industry total)

• Employment in plants located beyond the local area

Discussion• No evidence that local employment growth, relatively to the industry average, is

negatively affected by FDI propensity (relationship is even positive for FDI to advanced countries)

• Problem (endogeneity): Unobserved variables (local firms’ average productivity, other competitive advantages), correlated with FDI.

• Other caveats: 1) short time period; 2) “outsourcing” (non-equity FDI) is missing