21/04/23
Wages and accessibility: the impact of transport infrastructure
Anna MatasJosep LLuis Raymond
Josep LLuis Roig
Universitat Autònoma de Barcelona
ERSA Congress, Barcelona 2011
21/04/23
Contents
1. Introduction
2. Productivity and market accessibility
3. Data
4. Estimated models
5. Results
6. Conclusions
Introduction
• Objective: to assess the impact that investment in road infrastructure can have on firm productivity.
• Moreover, we also look at the impact of agglomeration effects on productivity.
• Productivity is approached by individual wages. • Workplace is allocated at NUTS III level.• Repeated cross-section data for 1995, 2002 and 2006.• Empirical evidence using micro-data: Graham (2007),
Combes et al. (2008), Mion and Naticchioni (2009), Gibbons et al (2010),and for a review Melo et al (2009).
• Empirical evidence for Spain (on firm location): Alañón and Arauzo (2008), Albarrán, Carrasco and Holl (2008), Arauzo (2005), Holl (2004, 2006), Matas and Roig (2004)
21/04/23
2. Productivity and market accessibility
- Improved accessibility allows for larger economies of scale due to wider potential markets
- Improved accessibility increases exposition to competition and generates incentives to higher efficiency. (Sorting of higher productivity firms)
21/04/23
- Improved accessibility can affect economies of agglomeration in two ways:
1. Better accessibility can attract new and relocating firms to positively affected areas, increasing the size of agglomeration
2. Better accessibility increases the spatial scope of economies of agglomeration
21/04/23
3. Data
• Source: Spanish Earnings Structure Survey (EES): 1995, 2002 and 2006.
• Micro data on individual wages• Spatial unit: provinces (NUTS III)• Data on workers and firm characteristics• Sectors: manufacturing + services
21/04/23
Summary statistics
1995 2002 2006 Individual continuous variables Mean Mean Mean Gross monthly wage (euros, 2006) 1620 1616 1561 Age (years) 38.5 38.0 38.5 Tenure (years) 10.8 8.0 7.5 Individual discrete variables (shares) Gender Male 74.7% 66.1% 61.1% Female 25.3% 33.9% 38.9% Working time Full time 96.0% 90.1% 85.0% Part time 4.0% 9.9% 15.0% Observations 109596 114368 86456
21/04/23
Summary statistics
Spatial variables
1995 2002 2006 2002/1995 2006/2002 2006/1995 Mean Mean Mean Market potential 3195 4205 4766 31.6% 13.4% 49.2% Employment density 91.3 121.8 147.5 33.4% 21.1% 61.6% Specialization 1.3 1.4 1.3 1.1% -4.6% -3.5% Human capital 16.0 21.0 23.4 31.2% 11.4% 46.2% Observations 109596 114368 86456
21/04/23
Market potential 2006
21/04/23
Change (%) in travel time 1995-2006
21/04/23
Change (%) in market potential: 1995-2006
21/04/23
CHANGE IN TIME (%) VS CHANGE IN POTENTIAL (%)
40
45
50
55
60
65
70
-20 -16 -12 -8 -4 0
Change in travel time 1995-2006 (percent)
Ch
ang
e in
inte
rna
l pot
en
tia
l 19
95
-20
06
(p
erc
en
t)
Correlation = -0.82
21/04/23
Estimated equation
Dependent variable: ln (monthly wage) Explanatory variables ln (market potential) ln (employment density) (ln (employment density))^2 ln (specialization) ln(share of working population with tertiary degree) Distance to French border Control variables ln (age) (ln (age))^2 Gender Education in level (9 levels) Tenure (years) Working time (full/partial) Type of contract (indefinite/fixed term) Industrial sector (18 sectors) Occupation (12 categories) Year (dummy)
21/04/23
Estimated coefficients OLS
All sectors Manufacturing Services ln (market potential) 0.104726
(24.77) 0.093585 (19.08)
0.133138 (19.41)
ln (employment density) 0.121009 (27.52)
0.079249 (15.05)
0.171342 (23.43)
(ln (employment density))^2 -0.01221 (20.86)
-0.00637 (9.08)
-0.01929 (19.65)
ln (specialization) 0.041588 (35.75)
0.040170 (31.67)
0.054697 (11.18)
ln(share tertiary degree) 0.067284 (17.53)
0.081762 (17.04)
0.046082 (7.19)
Distance to French border -0.0000666 (17.16)
-0.000117 (23.11)
-0.0000193 (2.86)
R square 0.5928 0.5476 0.6206
Robustness: coefficients of interest are highly stable to the selection of control variables
21/04/23
Possible endogeneity problems
• Troublesome explanatory variables: market potential, employment density, specialization and human capital
• Instruments: market potential in 1980, population density in 1860, specialization in 1980 and demographic structure in 1991 (share of 19-15 years minus share of 9-5 years)
21/04/23
Estimated coefficients IV
All sectors Manufacturing Services ln (market potential) 0.079132
(12.58) 0.085238 (12.15)
0.085729 (7.91)
ln (employment density) 0.085643 (11.72)
0.061729 (7.14)
0.111801 (9.5)
(ln (employment density))^2 -0.00693 (7.08)
-0.00374 (3.24)
-0.01043 (6.50)
ln (specialization) 0.045020 (32.52)
0.040690 (26.66)
0.039886 (5.53)
ln (share tertiary degree) 0.045728 (7.51)
0.058211 (7.94)
0.033145 (3.14)
Distance to French border -0.000076 (19.24)
-0.000123 (23.60)
-0.000025 (3.14)
R square 0.5926 0.5475 0.6203
21/04/23
Test for instrument’s validity
Shea's Adjusted Partial R- square All sectors Manufacturing Services ln (market potential) 0.4198 0.4858 0.4066 ln (employment density) 0.3548 0.3780 0.3631 (ln (employment density))^2 0.3517 0.3764 0.3556 ln (specialization) 0.7205 0.7077 0.4558 Human capital 0.3825 0.3832 0.3647
21/04/23
Hausman exogeneity test
Ho: variables are exogenousAll sectors: Robust score chi2(5) = 500.269 (p=0.0000)Robust regression F(5,310309) = 101.226 (p = 0.0000)
Manufacturing:Robust score chi2(5) = 136.947 (p=0.0000)Robust regression F(5,164840) = 27.7033 (p = 0.0000)
Services:Robust score chi2(5) = 287.792 (p = 0.0000)Robust regression F(5,145431) = 58.1808 (p = 0.0000)
21/04/23
Results
• Positive and significant effect of road accessibility on wages with a similar impact on manufacturing and services industries.
• Non linear effect of employment density with a different pattern for manufacturing and services.
• Evidence of positive human capital externalities• Positive specialization effect more significant for
manufacturing than for services.• Distance to French border is significant for
manufacturing but only marginally for services.
21/04/23
Elasticity of wages (productivity)
All sectors Manufacturing Services Market potential 0.0791 0.0852 0.0857 Specialization 0.0450 0.0407 0.0399 Human capital 0.0547 0.0582 0.0331 Density (minimum) 0.0700 0.0533 0.0882 Density (mean) 0.0193 0.0260 0.0122 Density (maximum) 0.0036 0.0174 -0.0116
21/04/23
Effects of employment density on wages
0.96
1.00
1.04
1.08
1.12
1.16
1.20
1.24
1.28
1.32
1.36
25 50 75 100 125 150 175 200 225 250 275 300
All sectors
Services
Manufacturing
Density
Wa
ge
s
21/04/23
Elasticity wages-employment density
-.02
.00
.02
.04
.06
.08
.10
.12
25 50 75 100 125 150 175 200 225 250 275 300
All sectors
Services: 212
Manufacturing
21/04/23
Conclusions
• Accessibility measured through market potential positively affects wages (Productivity)
• Evidence of agglomeration economies measured by employment density
• Evidence of human capital externalities: After controlling for individual educational level, the human capital of the province is positive and statistically significant
21/04/23
Pending task
•Infrastructure investment reduces travel time, increases potential, increases productivity and this leads to a further increase in market potential• Take these feedback effects into account must be the target of further research• Include measures of accessibility for other transport modes