Skip to content
United States Department of Transportation - Federal Highway Administration FHWA HomeFeedback

FHWA > Policy > Economic Returns From Transportation Investment

Appendix C

REVIEW OF INTERNATIONAL STUDIES OF THE PRODUCTIVITY EFFECTS OF HIGHWAY INFRASTRUCTURE

By
Dr. T.R. Lakshmanan, Director, Bureau of Transportation Statistics
U.S. Department of Transportation
July 1996

Table C1: Summary of International Studies of the Productivity Effects of Highway Infrastructure or Other Capital
Study Country/Sample Type of Estimation Sector or Output Measure Public Infrastructure
Type
Public Infrastructure
Elasticity
Private Inputs
Type
Private Inputs
Elasticity
Notes
Mera 1973, '75Japan
8 regions
1954-1963 (time series/cross section)
production function, Cobb-Douglasprimary
secondary
tertiary
transportation and communication; 3 other categories0.35** for transport and communications in secondary; 0.39-0.42 in tertiary; transp. and comm. Not used in primarycapital
labor
K: 0.15-0.33
L: 0.55-0.85, according to sector
 
Elhance and Lakshmanan 1988India
national 1950-79 (time series)
and 6 states 1970-71 (cross section)
cost function, translog, with endogenous private adjustment to publicly supplied infrastructuremanufacturing
value added
economic infrastructure (roads, rail, electric capacity); social infrastructure (hospital beds, etc.)economic infrastr. cost elasticities; at state level, -0.04 to -0.47; at national level -0.01 to -0.03capital
labor
energy
materials
 adjustment period of about 5 years
Johansson 1992 Johansson and Karlsson 1994Sweden
280 and 284 munic.
1980-88 (time series/cross section)
production function, Cobb-Douglasaggregate income,
industry-level income
highway capital public transit capital road accessibilityhighway 0.12-0.18; public transp. 0.18-0.20; road accessibility 0.20-0.27capital
labor
K: 0.47-0.50adjustment periods of 14-26 years, depending on industry
Seitz 1993Germany
1970-89
(time series/cross section)
cost function, translog, industry fixed effectime seriesreal output, 31 industriesroad mileage
road capital
negative and significantcapital
labor
 roads complementary to private capital, substitute for labor; these effects small
Seitz 1994Germany
1970-89
(time series/cross section)
cost function, translog, industry fixed effectime seriesreal output, 31 industriespublic capital
"core" public capital
on demand for K: 0.36
on demand for L: -0.13 to -0.15
capital
labor
rental elasticity of K demand, -0.04; wage elasticity of L demand, -0.09 
Conrad and Seitz 1994Germany,
1961-89 (time series/cross section)
cost function, transloggross output, 3 sectors: manufacturing, construction., trade and transportationpublic capitalnegative and significantcapital
labor
material
 estimated shadow price of public infrastructure approx. half of user cost of private capital; decrease in pub infrastr. Partially responsible for observed productivity decline
Lynde and Richmond 1992United Kingdom
aggregate, 1958-89 (time series)
cost function, translogmanufacturingpublic capitalsignificant cost reduction effectcapital labor public capital complemntary to private capital; aggregate constant returns to scale with public capital
Lynde and Richmond 1993United Kingdom
aggregate 1966: 1-1990; 2 (time series)
cost function translog (cointegrated)manufacturing value addedpublic capitalsignificant cost reduction effectcapital
labor
materials
 public capital contributes 17% of manufacturing productivity growth
Anderson, Anderstig and Harsman 1990Sweden
70 commuting regions, 1970 and 1980 (cross section)
production function, variable returns to scaleaggregate incomemain roads
railroads
airport
capacity accessibility
roads significant in 1980; rail significant in 1970; airports significant only when interacted with R&Dcapital
labor
K: 0.25 L: 0.28; 0.57- 0.75 combined with knowledge capital 
Prud'hommeFrance,
22 regions, 1981-88 growth (cross section), 1988 (cross section)
production function, Cobb-Douglasaggregate income/Lpublic capitalsignificant only when measured per unit of L times land areacapital/laborK/L:0.20-0.26 

Table C2: Summary of U.S. Studies of the Productivity Effects of Highway Infrastructure or Other Capital
Study Country/Sample Type of Estimation Sector or Output Measure Public Infrastructure
Type
Public Infrastructure
Elasticity
Private Inputs
Type
Private Inputs
Elasticity
Notes
Ratner 1983U.S.
aggregate, 1949-73 (time series)
production function, Cobb-Douglasprivate business sector outputpublic capital: non-military, gov't owned equipment and structures0.05 - 0.06labor/capital, capital adjusted for capacity utilizationL: 0.71-0.72
K: 0.22
 
Costa, Ellson and Martin 1987U.S.
48 states, 1972
(cross section)
production function, translogall sectors
manufacturing
non-agriculture
state and local
public capital
0.20 for all sectors, .19 for manufacturing, 0.21 - 0.26 for non-agrcapital
labor
K: -0.06 - 0.11
L: 0.77 - 1.02
public capital complementary to labor
Keeler and Ying 1988U.S.
9 regions, 1950-73 (time series/cross section)
cost function,
translog
truckingindustryfederal-aidhighway capitalstockcost elasticity -0.07capital, labor
fuel, other materials and services
  
Deno 1988U.S.
36 metropolitan areas, 1970-78 (time series/cross section)
profit function,
translog
manufacturinghighways and bridges;
water;
sewer
output supply elasticity of highway / bridge
0.31-0.57; 0.06 (not sig.) in declining
regions
capital
labor
 highway capital complementary to private capital in growing regions; complementary
to labor in full sample
Duffy-Deno and Eberts 1991U.S.
28 metropolitan areas, 1980-84 (time series/cross section)
simultaneous equations: personal income and public investmentper capita
personal income
public capitalpersonal income elasticity of public capital 0.08capital
labor
energy
  
Aschauer 1989U.S.
aggregate, 1949-85
(time series)
production function, Cobb-DouglasGNP/private capitalnon-military public capital and
core public capital
non-military 0.39
core 0.24
labor/ private capital0.38 
Munnell 1990U.S.
aggregate, 1948-87 (time series)
production function, Cobb-DouglasGNPcore public capital0.31 - 0.39capital
labor
K: 0.56
L: 0.11
 
Aschauer 1990U.S.
48 states (cross section)
determinants of
average annual
growth rate
state per capita incomeaverage total highway mileage, 1960-85;
disaggregated into rural and urban mileage
total: 0.22-0.37
rural: 0.24-0.40
urban: 0.10-0.17
  higher pavement quality contributes to growth;
vehicles/highway
mile as congestion measure retards
growth
Eisner 1991U.S.,
48 states, 1970-86 (time series, cross section, and time series/cross section)
production function Cobb-Douglas and translogGSPpublic capital; highway capital, water and sewer, otherpublic capital insignificant, highway 0.05-0.07 in time
series; public capital
0.16, highway 0.06 in
cross section
capital
labor
K: 0.29
L: 0.77
uses data from Munnell
1990; estimates of first differences sensitive to scale constraints
Tatom 1991U.S.
aggregate 1950-88 (time series)
production function, translog, 1st differencesprivate busines
ssector output
public capital0.03, not significantcapital labor
energy
L/K: 0.69 E(price): -0.06energy price variable derived from 1st order condition on energy
use
Garcia-Mila and McguireU.S.
48 states, 1969-83 (time series/cross section)
production function, Cobb-DouglasGSPhighway capital0.04capital structures, capital equipment labortotal K: 0.47
L: 0.36-0.45
 
McGuire 1992U.S.
48 states, 1969-83 (time series/cross section)
production function, Cobb-Douglas (state fixed effects);
translog
GSPhighway capitalC-D (fixed eff.): not significant
translog: 0.24
capitallaborK: 0.23-0.46
L: 0.70-75
return to scale of 1.2-
1.3 in state fixed
effects model; CRS in
random effects model
Pinnoi 1992U.S.
48 states, 1970-86 (time series/cross section)
production function, Cobb-DouglasGSPhighway capital0.06 - 0.08capital laborK: 0.30
L: 0.70
first-differencing to
correct measurement
errors in public capital
yielded insignificant
highway output
elasticities
Jones, Miaow,
Lee and Rickert
1993
U.S.
48 states, 1983-89 (time series/cross section)
production function, Cobb-Douglas, 5-
eqn system
GSP/Llane mileage,
vehicle miles traveled, separate urban and rural, different groups of highway classes
VMT: 0.16-0.23
LM, total: 0.09-0.14 urban VMT, LM: positive
rural VMT,LM:
negative
capital/laborK/L: 0.36-0.38endogenous
congestion and VMT
Moomaw and
Williams 1991
U.S.
48 states, 1954-76 (time series/cross section)
multifactor
productivity
manufacturinghighway densitysignificant positive
effect
  use regional dummies
with state data
Hulten and
Schwab 1991
U.S.
9 regions, 1951-76
(time series)
Multifactor
productivity
manufacturingnone capital
labor
 no observed role for
public infrastructure in
different interregional
growth rates
Hulten and
Schwab 1991
U.S.
9 regions, 1951-86
(time series)
multifactor
productivity
manufacturingpublic capitalnot significantcapital
labor
materials
 model predicts
significant coef.ficient
on K if public capital
operates through
production function.;
but K coef.ficient is
significant and public
capital coefficient is
insignificant; interpretations unclear
Carlino and Mills 1987U.S.
approx. 3000 counties,
1970-80 growth (cross
section)
determinants of county
employment and
population growth,
simultaneous
equation estimation
total employment,
manufacturing
employment, population
interstate highway
density
over decade: 0.06 total
employment; 0.06
manufacturing
employment.; 0.03
population
   
Mills and Carlino 1989U.S.
approx. 3000 counties,
1970-80 growth (cross
section)
determinants of county
employment and
population growth, simultaneous
equation estimation
total employment,
manufacturing employment,
population
interstate highway
density
over decade: 0.54 total
employment.:
0.17 population
   

Note- all reported coefficients are statistically significant unless explicitly noted otherwise
Source: Dr. T.R. Lakshmanan, Director, Bureau of Transportation Statistics, USDOT


< Previous | Table of Contents


FHWA Home | Feedback
FHWA