THE CONSTRUCTION SECTOR IN ECONOMIC DEVELOPMENT*
| Published date | 01 November 1970 |
| Date | 01 November 1970 |
| Author | W. Paul Strassmann |
| DOI | http://doi.org/10.1111/j.1467-9485.1970.tb00715.x |
THE CONSTRUCTION SECTOR
IN
ECONOMIC DEVELOPMENT*
W.
PAUL
STRASSMANN
I
Does the construction sector, like agriculture
or
manufacturing, follow a
pattern of change that reflects a country’s level
of
development? This article
will suggest that
it
does-although the pattern
is
often obscured by short-run
fluctuations. After lagging in early development, construction accelerates
in
middle income countries
($350-$900
per capita) and then falls
off.
An
explanation of this pattern and some of its consequences will be given.
A
more detailed comparison is then made between patterns of change in
construction and manufacturing. Simple regression analysis
is
applied
to
a sample of all countries with a population of over one million for which
construction and manufacturing output, employment, and labour earnings data
were available during
1955-64.
The sharply contrasting, yet statistically
signi-
ficant, relations for sub-sets of developed and underdeveloped countries are
consistent with the hypothesis that the construction sector has distinctive
characteristics at various levels of development.
I1
Other writers have not observed this construction growth pattern for a
variety of reasons.
For
a sample of fifty-one countries, Chenery
(1963)
com-
puted linear logarithmic regressions
of
value added in major sectors on
gross
national product per capita. Construction rose from
4
per cent
of
national
product at the
$100
per capita income level to almost
6
per cent. at the
$1,000
level.’ Thirty-four countries were in the construction sub-set, but for
*Thanks for a variety
of
useful suggestions and criticisms are due to
Dr.
E.
J.
Howenstine and
to
Professors Subbiah Kannappan, Jan Kmenta, Simon Kuznets, Carl
Liedholm, and Thomas Mayer. Remaining errors are beyond their control. Mr. Gary
Feuerbacher showed great ingenuity in reconciling the computer to the data. Apprecia-
tion
is
also due for an MSU-International Programs-Ford Foundation Grant.
1
See Chenery
(I
963,
p.
634).
Several economic historians claim
to
have explained
18-20
year building cycles in Western countries before World War
I.
Brinley Thomas
(1
954)
was particularly interested in the inverse relationship between British and over-
seas (including American) building cycles, since the late
1840s.
The basic mechanism
was a cycle
of
birth rates, leading to occasional peaks in migration. Slack demand
for
housing in Britain released capital
for
overseas railway construction and, after a slight
lag, building [pp.
102-3, 174-791.
Paul Cootner
(1963)
explained these same cycles on the basis
of
production
functions. In developed countries manufacturing
grows
with more
or
less constant
returns to scale. Food and raw materials are produced at increasing costs. When the
discrepancy becomes great, that
is,
with a lag, investment
is
shifted overseas to develop
new sources of food and materials. Due
to
poor foresight, long gestation periods, and
indivisibilities, the building
of
social overhead capital abroad is overdone, and manu-
facturing investment at home resumes being more profitable. Employment expands,
emigration slows, and housing is in short supply.
392
W.
PAUL
STRASSMANN
a fluctuating industry the one or two Korean war years used were not pre-
sented as yielding conclusive results. Anyway, manufacturing trends were
Chenery’s primary target.
A
comparison of manufacturing employment trends with tertiary employ-
ment including construction was the aim of a cross-country regression analysis
by Galenson
(1963).
Manufacturing employment in less developed countries
grew at higher percentage rates, but the absolute increases in tertiary employ-
ment were larger, hence the most striking fact seemed
to
be the stability
of
the manufacturing to tertiary employment ratio. Construction could not be
singled out for separate analysis of statistical significance because, as in
Chenery’s study, the use
of
single first and terminal years, a decade or less
apart, must obscure secular trends in a fluctuating industry. Galenson tenta-
tively concluded that commerce and services had
to
absorb the bulk of new
employment,
A
United Nations study
(1967)
also concluded that
‘
the wide scatter
of
data revealed by the table only confirms the apparent lack of correlation
already pointed out above between the share of construction in
GDP
and
either the per capita GDP or the rate
of
growth
’.
Kuznets
(1960)
reported that
in
his sample
of
thirty-four countries the
cross-sectional rise of construction was not striking. In post-war years through
1957,
construction averaged
8-5
per cent.
of
gross domestic product in
countries in the
$100
and
$200
annual per capita income classes, as well as
in
the
$270
and
$400
classes. For the
$650
and
$1,000
classes. it was
11.0
per
cent. and for the
$1,700
per capita income class, it was
11.9
per cent. of gross
domestic product. Since construction remained between
50.7
and
55.5
per
cent. of gross domestic capital formation, and without any clear trend in this
percentage at that, the sector naturally seemed like
a
variable dependent on
national income and investment.
Kuznets noticed, however, that
within
construction, the share of dwellings
rose from
30.3
per cent. in the poorest two classes of countries to
42.5
per
cent. in the
$650
and
$1,000
classes, and then fell again to
37.3
per cent. in
the
$1,700
class. He found it tempting to attribute the generally rising share
of
dwellings to a rise in cost compared with other construction: a supply
side explanation.2
A year later, Kuznets
(1961)
reported
on
the long-term trends
of
construc-
tion in eleven developed countries. He found that neither of the earlier cross-
sectional patterns held. Construction since the late
1800s
has been not a
constant, but a
declining
share of fixed capital formation, except for Canada
and South Africa. Nor did the share of residential construction rise, but
instead remained roughly constant (United Kingdom, Italy) or declined
(United States, Canada, Australia, Japan).
A
demand side explanation was suggested as an answer
:
falling popula-
tion growth for less housing; the shift from extensive social overhead capital
2
Kuznets
(1960,
p.
34)
referred
to
the calculations of Gilbert and Kravis
(1954,
p.
119).
US
price
weights would raise the volume of Italian residential construction
by
50
per
cent, and lower
the
value
of
other construction by
5
per cent. in
1950.
To continue reading
Request your trial