American Economic Association

Structural Change in a Multisector Model of Growth

Author(s): L. Rachel Ngai and Christopher A. Pissarides
Source: The American Economic Review, Vol. 97, No. 1 (Mar., 2007), pp. 429-443
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/30034402
Accessed: 06-04-2017 21:49 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
American Economic Review

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
 Structural Change in a Multisector Model of Growth

By L. RACHEL NGAI AND CHRISTOPHER A. PISSARIDES*

Economic growth takes place at uneven employment share used to produce consump-
rates across different sectors of the economy. tion goods vanishes from all sectors except
This paper has two objectives related to this for the one with the smallest TFP growth rate,
fact: (a) to derive the implications of different but the employment shares used to produce
sectoral total factor productivity (TFP) capital goods and intermediate goods con-
growth rates for structural change, the name verge to nontrivial stationary values. If the
given to the shifts in industrial employment utility function in addition has unit intertemporal
shares that take place over long periods of elasticity of substitution, during structural change
time; and (b) to show that even with ongoing the aggregate capital-output ratio is constant and
structural change, the economy's aggregate the aggregate economy is on a balanced growth
ratios can be constant. We refer to the latter as
path.
aggregate balanced growth. The restrictions Our results contrast with the results of Cris-
needed to yield structural change consistent tina Echevarria (1997), John Laitner (2000),
with the facts and aggregate balanced growth Francesco Caselli and Wilbur Coleman II
are weak restrictions on functional forms that
(2001), and Douglas Gollin, Stephen Parente,
are frequently imposed by macroeconomists and Richard Rogerson (2002), who derived
in related contexts. structural change in a two- or three-sector
We obtain our results in a baseline model economy with nonhomothetic preferences.
Our results also contrast with the results of
of many consumption goods and a single cap-
ital good, supplied by a sector that we label
Piyabha Kongsamut, Sergio Rebelo, and Dan-
yang Xie (2001) and Reto Foellmi and Josef
manufacturing. Our baseline results are con-
sistent with the existence of intermediate Zweimuller (2005), who derived simulta-
goods and many capital goods under some neous constant aggregate growth and struc-
reasonable restrictions. Production functions tural change. Kongsamut, Rebelo, and Xie
in our model are identical in all sectors except(2001) obtain their results by imposing a re-
for their rates of TFP growth, and each sector striction that maps some of the parameters of
produces a differentiated good that enters their
a Stone-Geary utility function onto the
constant elasticity of substitution (CES) util-
parameters of the production functions, aban-
ity function. We show that a low (below one) doning one of the most useful conventions of
elasticity of substitution across final goodsmodern macroeconomics, the complete inde-
leads to shifts of employment shares to sec- pendence of preferences and technologies.
tors with low TFP growth. In the limit, the Foellmi and Zweimuller (2005) obtain their
results by assuming endogenous growth
driven by the introduction of new goods into
a hierarchic utility function. Our restrictions
are quantitative restrictions on a conventional
* Ngai: Centre for Economic Performance, London School
CES utility function that maintains the inde-
of Economics and CEPR (e-mail: 1.ngai@lse.ac.uk); Pissar-
ides: Centre for Economic Performance, London Schoolpendence
of of the parameters of preferences
and
Economics, CEPR, and IZA (e-mail: c.pissarides@lse.ac.uk). technologies.
We have benefited from comments received at several presen- Our results confirm William J. Baumol's
tations (the CEPR ESSIM 2004 meetings, the SED 2004
(1967) claims about structural change. Baumol
annual conference, the NBER 2004 Summer Institute, the
divided the economy into two sectors, a "pro-
2004 Canadian Macroeconomic Study Group, and at several
gressive"
universities), and from Fernando Alvarez, Francesco Caselli, one that uses new technology and a
Antonio Ciccone, Nobu Kiyotaki, Robert Lucas, Nick Oulton,
"stagnant" one that uses labor as the only input.
Danny Quah, Sergio Rebelo, Robert Shimer, Nancy Stokey,
He then claimed that the production costs and
Richard Rogerson, Jaume Ventura, and two anonymous refer-
prices of the stagnant sector should rise indefi-
ees. Funding from the CEP, a designated ESRC Research
Centre, is acknowledged. nitely, a process known as "Baumol's cost dis-
429

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
 430 THE AMERICAN ECONOMIC REVIEW MARCH 2007

ease," and labor should move in

change andthe direction
balanced aggo
the stagnant sector.' tions IV and V, we study
In the more recent empirical literature,
baseline model, two
one whe
competing explanations (which
can also can
be used ascoexis
inter
have been put forward where there are change
for structural many
our explanation, whichpendixis sometimes
discusses the term
imp
"technological" because it attributes
extension structur
and differenc
change to different ratessectors,
across of sectoral TFP
and conta
results. explanation, whic
growth; and a utility-based
requires different income elasticities for diffe
ent goods and can yield I. An Economy
structural changewith M
eve
with equal TFP growth in all sectors. Baumol,
Sue Anne Batey Blackman, and Edward
The baseline N
economy
Wolff (1985) provide trary number
empirical of m
evidence at sec
th
two-digit industry level,
m - 1consistent with
produce only o
co
model. Irving B. Kravis, Alan W.
last sector, Heston,
which an
is den
Robert Summers (1983) manufacturing, produ
also present eviden
sumptionexplanation,
that favors the technological good and t
stock.
least when the comparison isWe derive
between the eq
manufa
tionfeatures
turing and services. Two to a social plannin
of their da
tive function is
satisfied by the technological explanation pro
posed in this paper are: (a) relative prices refle
differences in TFP growth rates; and (b) re
consumption shares vary a lot less over tim
(1) U = e-tLv(c1, ..., cm) dt,
than nominal consumption shares.2 Our mode
0
is also consistent with the observed positi
correlation between employment growth a
where p across
relative price inflation > 0, ci ax 0 are per capita consump-
two-digit sector
and with historical tionOECD evidence
levels, and the instantaneous utility func- present
by Simon Kuznets (1966) and Angus Maddison
tion v(.) is concave and satisfies the Inada
(1980) for one-digit sectors.4
conditions. The constraints of the problem are
as follows.
Section I describes our model of growth with
many sectors and Sections IIis and
The labor force III,
exogenous respe
and growing
tively, derive the conditions for structur
at rate v, and the aggregate capital stock is
endogenous and defines the state of the econ-
omy. Sectoral allocations are controls that
1 Baumol controversially also claimed that as more
satisfy
weight is shifted to the stagnant sector, the economy's
growth rate will be on a declining trend and eventually
m m
converge to zero. This claim contrasts with our finding that
the economy is on a balanced-growth path. We get our
result because we include capital in our analysis, ironically
(2) ni = 1; niki = k,
i=1 i=1
left out of the analysis by Baumol (1967, 417) "primarily for
ease of exposition ... that is [in]essential to the argument."
where ni ax 0
2 See Rodney E. Falvey and Norman Gemmell (1996) for an
update of some of their results. Falvey and Gemmell find a unit
0 is the capita
income elasticity and a small (negative) price elasticity for ser- is the aggrega
vices in a cross section of countries, consistent with our results. mobility for b
3 These correlations are shown in the working paper All productio
version of this paper, Ngai and Pissarides (2004).
consumed, bu
4 Kuznets (1966) documented structural change for 13
OECD countries and the USSR between 1800 and 1960, and either consumed or invested. Therefore:
Maddison (1980) documented the same pattern for 16 OECD
countries from 1870 to 1987. They both found a pattern with
the same general features as the predictions that we obtain
(3) ci = F'(niki, ni) Vi Cm m;
when the ranking of the average historical TFP growth rates is
agriculture followed by manufacturing followed by services. (4) k = Fm(nmkm, nm) - cm - (5 + v)k,

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
VOL. 97 NO. 1 NGAI AND PISSARIDES: STRUCTURAL CHANGE IN A MULTISECTOR MODEL OF GROWTH 431

where 8 > 0 is the depreciation rate. Production mand functions have constant price elasticity
function Fi( , ) has constant return to scale, -e and unit income elasticity. With this utili
positive and diminishing returns to inputs, and function, (8) yields
satisfies the Inada conditions.
The social planner chooses the allocation of
factors ni and ki across m sectors through a set
w-=- vi.
pm Cm 0,iM( Ai)
of static efficiency conditions:
The new variable xi is the
(5) ViIv,m = F/F = F/F' V i. expenditure on good i to
ture on the manufacturin
The allocation of output to consumption and useful in the subsequent
capital is chosen through a dynamic efficiency behind this formula is in
condition: ities, given that all goo
elasticity. The ratio of con
(6) - im/Vm = FK - (6 + p + V), is a weighted average of th
each good in the utility func
where F' and FP are the marginal products of prices. A higher price ratio
labor and capital in sector i.5 By (5), the rates expenditure on good i to goo
of return to capital and labor are equal across common price elasticity.
sectors. We also define aggrega
In order to focus on the implications of penditure
dif- and output p
ferent rates of TFP growth across sectors,manufacturing:
we
assume production functions are identical in all
sectors except for their rates of TFP growth: m m

Pi Pi
(7) Fi= AiF(niki, ni); Ai/Ai = yi; V i. (11) c =IPm
- ci; y
PmFi. i=1

With these production functions, we show in the Us

Appendix that static efficiency and the resource
constraints (2) imply (1

(8) ki = k; pi/pm = vilVm = AmIAi; V i, w

where pi is the price of good i in the decentral- II.

ized economy.
The utility function has constant elasticities W
both across goods and over time: which at least some of the labor shares are
changing over time, i.e., ri, f 0 for at least some
4()1-0 - 1 i. We derive in the Appendix (Lemma A2) the
(9) (C, ... , Cm) 1 -0 employment shares

( m ) E - 1) (13) ni = - V i m,
xi

i=1

(14) ifnm
where 0, e, oi > 0 and Ic i = 1. Of course,
0 = 1, v(-) = In (-.), and if e = 1, In 0(-) =
The first
I1' oiln ci. In the decentralized economy, de- term on the right side of (14) parall
5 The corresponding transversality condition
exp(-f'8 (FI - 8 - v) dr) = 0.
+ 1 - -.
the term in (13) and so represents the emplo
ment needed to satisfy the consumption dema
for the manufacturing good. The second brac
is lim,_ k
eted term is equal to the savings rate and re

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
432 THE AMERICAN ECONOMIC REVIEW MARCH 2007

resents the manufacturing employment needed

to satisfy investment demand. rii (cy)
(18) -.drive
Conditions (13) and (14) . -c/y
ni + (1 - our
e)(Y - 1i); Vi 4 m;
structural
change results. To see the intuition behind them,
note that by aggregation over all i, we obtain
that in our economy the employment share used
to produce consumption goods is equal to c/y,
(19) nm = +(1 - e)(- )
and the employment share used to produce cap-
ital goods is 1 - c/y. Conditions (13) and (14)
state that the same holds for each sector i. From S(cly(xmX) ( 1y
nm 1 - c/y\ nm /
(10) and (12), the consumption expenditure
share of each sector is PiCilPmC = xilX. So the
where ' = I', (xilX) Yi is a weighted aver-
employment share of age consumption
of TFP growth rates, with thegoodweight i is
the consumption sharegiven
of bygood
each good'si consumption
multiplied share. by
the employment share of total
Equation consumption.
(18) gives the growth rate in the
Equivalently, the employment
employment share of eachshare ofgood
consumption con-
as a linear
sumption good i is the function of
average its own TFP growth to
propensity
consume good i : ni rate. The intercept and slope of this function
= PiCipmy.
Condition (13) has the important
are common across sectors, butimplica-
although the
slope isof
tion that the growth rate a constant,
two thesectors'
intercept is in general
relative
employment depends only
a function of time because both c/y and T are be-
on the difference
tween the sectors' TFP growth
in general functions of rates and the
time. Manufacturing,
however,
elasticity of substitution does not conform
between to this rule, be-
goods:
cause its employment share is a weighted
n. n.
average of two components, one for the pro-
(15) ni = nj
(1 -e)(y - y,) V i,j m. duction of the consumption good, which con-
forms to the rule, and one for the production
of capital goods, which behaves differently.
But (8) implies that the growth rate of relative
prices is: The properties of structural change follow
immediately from (18) and (19). Consider, first,
the case of equality in sectoral TFP growth
(16) = "i V
Pi Pj
pi pj- -y rates, i.e., let y = y,, Vi. In this case, our
economy is one of balanced TFP growth, with
and so, relative prices remaining constant but with
many differentiated goods. Because of the con-
n stancy of relative prices, all consumption goods
, p.
(17) - =(1 -e) -IVij 0m.
ni nji jp
can be aggregated into one, so we effectively
have a two-sector economy, one sector produc-
ing consumption goods and one producing cap-
PROPOSITION 1: The
ital goods. rate
Structural change of
can still take place ch
relative price of good
in this economy, but i onlyto
betweengoodj
the aggregate is
difference between
of the consumption the
sectors andTFP
the capital sec-gro
sector j and sector i.if c/y
tor, and only In sectors
changes over time. If c/y is pr
consumption goods, relative
increasing over employ
time, the investment rate is fall-
grow in proportion to
ing and labor is movingrelative
out of the manufactur- pri
ing sector and into the consumption
factor of proportionality given sectors.by
elasticity of substitution between
Conversely, if c/y is falling over time, labor is g
moving out of the consumption sectors and into
The dynamics of the individual em
manufacturing. In both cases, however, the rel-
shares satisfy
ative employment shares in consumption sec-
tors are constant.

If c/y
6 All derivations and proofs, unless trivial, are collected is constant over time, structural
in the Appendix. change requires e 0 1 and different rates o

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
VOL. 97 NO. 1 NGAI AND PISSARIDES: STRUCTURAL CHANGE IN A MULTISECTOR MODEL OF GROWTH 433

a small e can reconcile the small changes in

sectoral TFP growth rates. It follows imme-
diately from (16), (18), and (19) that if
the relative real consumption shares with the
large changes in relative nominal consump-
(W/y) = 0, e = 1 implies constant employment
tion shares found by Kravis, Heston, and
shares but changing prices. With constant em-
Summers (1983). The authors concluded that
ployment shares, faster-growing sectors pro-
duce relatively more output over time. Price
their finding is evidence in favor of the tech-
changes in this case are such that consump- nological explanation of structural change.
tion demands exactly match all the output More recently, Daniel E. Sichel (1997) found
the same pattern for relative output shares,
changes due to the different TFP growth rates.
and Falvey and Gemmell (1996) found that
But if e v 1, prices still change as before but
consumption demands are either too inelasticthe real consumption share of services (a sec-
(in the case e < 1) to match all the output tor with low TFP growth rate) falls very grad-
ually with income, both of which are
change, or are too elastic (e > 1) to be sat-
isfied merely by the change in output due toconsistent with our model when e < 1.
TFP growth. So if e < 1, employment has to
move into the slow-growing sectors, and if e
> 1, it has to move into the fast-growing III. Aggregate Growth
sectors.

We now study the aggregate growth path of

PROPOSITION 2: If yTi y m Vi # m, economy, this a with the objective of finding a
necessary and sufficient condition for structural
sufficient set of conditions that satisfy struc-
tural
change is 6/c $ 0~/y. The structural change in change as derived in the preceding sec-
this case is between the aggregate of consump-
tion, and in addition satisfy Kaldor's stylized
tion sectors and the manufacturing sector.facts of aggregate growth. Recall that, for the
If 6/c = fly, necessary and sufficient condi-
analysis of structural change, we imposed a
Hicks-neutral
tions for structural change are e 4 1 and 3i E technology. It is well-known
1, ..., m - 1) s.t. yji y. The structural
that with this type of technology, the econ-
omy can be on a steady state only if the
change in this case is between all sector pairs
with different TFP growth rates. If e < 1,production
em- function is Cobb-Douglas. We
ployment moves from the sector with the higher
therefore let F(niki, ni) = k ni, a E (0, 1).7
TFP growth rate to the sector with the lower
With TFP in each sector growing at some rate
TFP growth rate; the converse is true if eTi, the aggregate economy will also grow at
> 1.
some rate related to the yis. The following
Proposition 2 for e < 1 confirms the struc-
proposition derives the evolution of the ag-
gregate economy.
tural change facts identified by Baumol, Black-
man, and Wolff (1985). When demand is price
PROPOSITION 3: Given any initial k(O), the
inelastic, the sectors with the low productivity
growth rate attract a bigger share of labor, de-
equilibrium of the aggregate economy is a path
spite the rise in their price. From the for
static
the pair [c, k] that satisfies the following two
efficiency results in (8) and (12), we find differential
that equations:
the nominal output shares (defined as PiFi/py)
k c
are equal to the employment shares in all sec-
(21)
tors, and by (10) the nominal consumption Amk'-
k k (6 + v),
shares are given by xilX, so the results obtained
for employment shares also hold for nominal
consumption and output shares. But real con-
sumption growth satisfies
7 Daron Acemo
ined the implic
(20) C,/ci - cj/cj = e(Yi - yj); Vi, j, economic grow
capital deepeni
an expression also satisfied by real output shares
unbalanced grow
Vi, j 4: m. plications of dif
A comparison of (15) and (20) reveals our
that
model.

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
434 THE AMERICAN ECONOMIC REVIEW MARCH 2007

0=1, e4:1;
(22) 0 - k (0 - + - )
C

and 3iE{1,..., m - 1}s.t. yi fym.

+ otAmka-1 -(6 + p + ,).
Recalling the definition of Y following equation
We define an aggregate balanced growth path (19), Proposition 3 implies that the contribution of
such that aggregate output, consumption, and each consumption sector i to aggregate equilib-
capital grow at the same rate. It follows from rium is through its weight xi in Y. Because each xi
Proposition 3 that a necessary condition for the depends on the sector's relative TFP level, the
existence of an aggregate balanced growth path weights here are functions of time. So y cannot be
is that the expression (0 - 1)(ym - ) be a constant during structural change and the only
constant. To show this, let way that 4 can be constant is through 0 = 1,
which yields 4 = 0. In this case, our aggregate
economy in c and k becomes formally identical to
(23) (0 - 1 )(y, - y) =, constant.
the one-sector Ramsey economy with growth rate
,,,. There are two other conditions that give a
Define aggregate consumption and the capital- constant 4 and so yield balanced aggregate
labor ratio in terms of efficiency units, ce growth: yi = Ym Vi or e = 1. But as Proposition
cA~l/(-a) and ke kAm1/(1-), and let gm 2 demonstrates, neither condition permits struc-
7m/(1 - a), the rate of labor-augmenting tech- tural change on the balanced growth path, where
nological growth in the capital-producing sector. c/y is constant.
The dynamic equations (21) and (22) become Proposition 4 requires the utility function to be
logarithmic in the consumption composite 0,
which implies an intertemporal elasticity of sub-
(24) 6e/Ce stitution equal to one, but also to be nonlogarith-
mic across goods, which is needed to yield
= [ak-' + - ( + + p)]0 - ; nonunit price elasticities. A noteworthy implica-
tion of Proposition 4 is that balanced aggregate
(25) ke = k~- ce - (gm + 6 + v)ke. growth does not require constant rates of growth
of TFP in any sector other than manufacturing.
Equations (24) and (25) parallel the two dif- Because both capital and labor are perfectly mo-
ferential equations in the control and state ofbile across sectors, changes in the TFP growth
the one-sector Ramsey economy, making the rates of consumption-producing sectors are re-
aggregate equilibrium of our many-sectorflected in immediate price changes and realloca-
economy identical to the equilibrium of the tions of capital and labor across sectors, without
one-sector Ramsey economy when 4, = 0, and effect on the aggregate growth path.
trivially different from it otherwise. Both To give intuition for the logarithmic inter-
models have a saddlepath equilibrium and temporal utility function, we recall that bal-
stationary solutions (Ce, ,e) that imply bal-anced aggregate growth requires that aggregate
anced growth in the three aggregates. Theconsumption be a constant fraction of aggregate
capital-labor ratio is growing at the rate ofwealth. With our homothetic utility function, this
growth of labor-augmenting technologicalcan be satisfied either when the interest rate is
progress in the sector that produces capitalconstant or when consumption is independent of
goods, gm,. Aggregate consumption and out- the interest rate. The relevant interest rate here is
put deflated by the price of manufacturing the rate of return to capital in consumption units,
goods are also growing at the same rate. which is given by the net marginal product of
Proposition 2 and the requirement that capital, ay/k - 6, minus the change in the relative
be constant yield the following importantprice of the consumption composite, ym - Y. The
proposition. latter is not constant during structural change. In
the case e < 1, - is falling over time (see Lemma
PROPOSITION 4: Necessary and sufficient A3 in the Appendix for proof), and so the real
conditions for the existence of an aggregate interest rate is also falling, and converging to
ay/k - 6. With a nonconstant interest rate, the
balanced growth path with structural change are:

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
VOL. 97 NO. 1 NGAI AND PISSARIDES: STRUCTURAL CHANGE IN A MULTISECTOR MODEL OF GROWTH 435

hump-shaped or declines monotonically. As-

consumption-wealth ratio is constant only if con-
ymptotically,
sumption is independent of the interest rate, which the economy converges to an
requires a logarithmic utility function.8 economy with
Under the conditions of Proposition 4,
there is a steady state characterized by aggre-
nm 01 +v+p+g
gate balanced growth, in the sense that in this
steady state the aggregate ratios are constant.
In order to achieve this balance, the aggre- n= 1 -
gates c and y are divided by manufacturing
price, to conform to the aggregate k. If some
where 6" is the savings rate along the
other price index is used as deflator, the rate
balanced growth path.
of growth of the aggregates is constant only if
the rate of growth of the price index is con- Proposition 5 follows immediate
(18)-(19) and Lemma A3. Consider
stant, but of course the aggregate ratios are
e < 1, the one for e > 1 followi
still constant. The published aggregate series
studied by macroeconomists usually use an corresponding argument. For e < 1
average price as deflator which does not have
expands if and only if its TFP growt
fixed weights. If the price index used to de-
smaller than y-, and contracts if and

flate national statistics is some /f, the pub-

growth rate exceeds it. But if e
lished real aggregate income is y/ft. If theweighted average y is decreasing over
weights used to construct fi are the sector Lemma A3 in the Appendix). Therefor
shares, ft changes during structural change.of expanding sectors is shrinking ov
But because sector shares do not change rap-
more sectors' TFP growth rates excee
feature of the model implies that sec
idly over time, visually there is virtually noth-
ing to distinguish the "stylized fact" of TFP growth rates below the initial ~
constant growth in reported per capita GDP hump-shaped employment share, an im
with another "stylized fact" of constant
that we believe is unique to our mo
growth in our per capita output measure.9 employment shares first rise, but onc
Next, we summarize the dynamics of em-
down to their own '/i, they fall.1o
ployment shares along the aggregate balanced In contrast to each sector's employment
growth path. share, once the economy is on the aggregate
balanced growth path, output and consump-
PROPOSITION 5: Let sector 1 denote the sec- tion in each consumption sector grow accord-
tor with the smallest TFP growth rate when e ing
< to
1, or the sector with the biggest TFP growth
rate when e > 1. On the aggregate balanced F' Ai ki fri
growth path,innlthe
Employment increases monotonically.
other sectors is either (26)PF=Ai +k ni
-k
=EYi + agm + (1 - E).

8 After reexamining the evidence, Robert Barro and
Xavier Sala-i-Martin (2004, 13) concluded, consistent with
our model, "it seems likely that Kaldor's hypothesis of a
roughly stable real rate of return should be replaced by a
tendency for returns to fall over some range as an economy
develops." In our model, it is converging from above to a
positive value.
9 Nicholas Kaldor (1961, 178) spoke of a "steady trend
rate" of growth in the "aggregate volume of production." In
Ngai and Pissarides (2004, fig. 4) we plot our series of per
capita real incomes and the published chain-weighted series
for the United States since 1929, and show that they are
virtually indistinguishable from each other.
If E - 1,
output in the
eachrate of isgrowth
sector positiveof(provided
consumption
yi ax and
0), and so sectors never vanish, even though
their employment shares in the limit may van-
ish. If e > 1, the rate of growth of output may
1O Maddison (1980, 48), in his study of historical OECD
data, found a "shallow bell shape" for manufacturing em-
ployment for each of the 16 OECD countries, which can be
reproduced by our model if the manufacturing TFP growth
rate takes values between the TFP growth rates of agricul-
ture and services.

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
436 THE AMERICAN ECONOMIC REVIEW MARCH 2007

be negative in some low-growth sectors,

fraction of output in virtually and
all sectors of the
since by Lemma A3 Y is economy
rising is sold to businesses."
over time in this
case, their rate of growth As in the baseline model,
remains sectors are of two
indefinitely
types.
negative until they vanish. The first type produces perishable goods
that are either
Finally, we examine briefly the consumed by households or
implications of used
as intermediate inputs
0 4 1. When 08 : 1, balanced aggregate growth by firms. We continue
referring to change,
cannot coexist with structural these sectors as consumptionthe
because sec-
tors. The second type of sector produces goods
term 4 = (0 - 1)(%m - ) in the Euler condition
that canBut
(24) is a function of time. be used asas capital.
shown For generality's
in the
Appendix Lemma A3, ' sake, we assume that the output
is monotonic. As oftthe -->capital-
o,
l converges to the constant
producing sector (0 - be
can also 1)(y,
processed-into
yi),
both
consumption
where Yi is the TFP growth goods and
rate inintermediate
the limiting inputs.
The output growing
sector (the slowest or fastest of consumption sector i is now
consump-
tion sector depending ci +on
hi, where hi is the output
whether e that
< oris used as
>1).
Therefore, the economy an intermediate
with 0good. Manufacturing output
1 converges to
can be with
an asymptotic steady state consumed, cm, used
the sameas an growth
intermediate
input,0hm,
rate as the economy with = or
1. used as new capital, k. We
What characterizes theassume that all intermediate
dynamic path goods,
of the hi, are
aggregate economy when used as0 an 0
input
1?intoBy an aggregate
differen- CES pro-
tiation and using Lemma A3, we obtain
duction function F(ha, ..., hm) = [= 1
cpih~ly-1)/T'I]"/(-1) which produces a single
m(27)
intermediate good (@, with qr > 0, (pi > 0, and
i 9p = 1. The production functions are mod-
(27) if = (0 - 1)(1 - e) > (xi/X)(yi- -)2,
i=1
ified to Fi = Ainikfqi, Vi, where qi is the
ratio of the intermediate good to employment
which is of second order comparedinwith sectorthe
i and p is its input share, with a, 3 >
growth in employment shares in (15), 0given
and athat
+ p < 1. When P = 0, we return to
the ys are usually small numbers our centered
baseline model. We show in the Appendix
that a
around 0.02. Therefore, the rate of growth ofnecessary
the and sufficient condition for
economy during the adjustment to the an aggregate
asymp- balanced growth path with struc-
totic steady state with 0 0 1 is very close to the
tural change is rl
Cobb-Douglas.'2 = 1, D(-)
When i.e., is
4(.)Cobb-Doug-
should be
constant growth rate of the economy with 0 =
las, in
1, despite ongoing structural change our both
central results from the baseline
economies. model carry through, with some modifica-
tions.
The aggregate equilibrium is similar to the
IV. Intermediate Goods one in the baseline model:

Our baseline model has no intermediate

(28)
inputs and has only one sector producing cap- Ak(ax* - )-) - (6 + p + v),
ital goods. We now generalize it by introduc-
ing intermediate inputs and (in the next
section) by allowing an arbitrary number of
" According to input-output tables for the United States,
sectors to produce capital goods. The in
key
1990 the percentage distribution of the output of two-
difference between intermediate goodsdigit andsectors across three types of usage, final consumption
demand, intermediate goods, and capital goods, was 43, 48,
capital goods is that capital goods are reus-and 9, respectively. In virtually all sectors, however, a large
able, while intermediate goods depreciate fraction of the intermediate goods produced are consumed
fully after one usage. The motivation for theby the same sector.
introduction of intermediate inputs is that 12 Nicholas Oulton (2001) claims that if there are inter-

many of the sectors that may be classifiedmediate

as goods, and if the elasticity of substitution between
the intermediate goods and labor is bigger than one, Bau-
consumption sectors in fact produce for busi-mol's "stagnationist" results could be overturned (in the
nesses. Business services is one obvious ex- absence of capital). No such possibility arises with Cobb-
ample. Input-output tables show that a large Douglas production functions.

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
VOL 97 NO. 1 NGAI AND PISSARIDES: STRUCTURAL CHANGE IN A MULTISECTOR MODEL OF GROWTH 437
(29) k = (1 - P)Ak(ax + -/(1-
c
- (8 + 9),
k
where A [Am(/axlm)13]l l(-) and (m is the
marginal product of the manufacturing good in
(D. The growth rate of A is constant and equal to
Y = ym + (/3 >I=1 - P), where axi is the
input share of sector i in (. Therefore, we can
define aggregate consumption and the aggregate
capital-labor ratio in terms of efficiency units
and obtain an aggregate balanced growth path
with growth rate (ym -+ 3 I1 pIrij)/(1 - a -
p), which is the sum of labor-augmenting tech-
nological growth in the capital-producing sector
and a p fraction of labor-augmenting techno-
logical growth in all sectors that produce inter-
mediate goods. Recall the aggregate growth rate
in the baseline model depended only on the TFP
growth rate in manufacturing. In the extended
model with intermediate goods, the TFP growth
rates in all sectors contribute to aggregate growth.
The employment shares (13) and (14) are
now modified to
(30) n, = Xxi+ c)
axPij; V i 0 m;
[X (Y)/ \Y)
(31) nm = x c + P + 1- - P c
For the consumption sectors, the extra term in
(30) captures the employment required for pro-
ducing intermediate goods. pi is the share of
sector i's output used for intermediate purposes
and p is the share of the aggregate intermediate
input in aggregate output. For the manufactur-
ing sector, the terms in the first bracket parallel
those of the consumption sectors. The second
term captures the employment share for invest-
ment purposes.
Our results on structural change now hold for
the component of employment used to produce
consumption goods, (xi/X)(cly). The definition
of xi and X is the same as in the absence of
intermediate goods. The contribution of inter-
mediate goods to sectoral employment dynam-
ics is the addition of the constant employment
share pip/, with no impact on the other two
This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
components of employment. Following on from
this, the asymptotic results in Proposition 5 are
also modified. Asymptotically, the employment
share used for the production of consumption
goods still vanishes in all sectors except for the
slowest growing one (when e < 1), but the
employment share used to produce intermediate
goods, ji/3, survives in all sectors.
V. Many Capital Goods
In our second extension we allow an arbitrary
number of sectors to produce capital goods. We
study this extension with the baseline model
without intermediate inputs.
We suppose that there are K different capital-
producing sectors, each supplying the inputs
into a production function G, which produces a
capital aggregate that can be either consumed or
used as an input in all production functions F'.
Thus, the model is the same as before, except
that now the capital input ki is not the output of
a single sector but that of the production func-
tion G. The Appendix derives the equilibrium
for the case of a CES function with elasticity jL,
i.e., when G = [I=1 m (FmJ)( ) - ll)/I AQLl- ),
where
each capitalggood's
> 0,sector
4, - mj.
0 and
G nowF'm is the
replaces the output of
output of the "manufacturing" sector in our
baseline model, F'.
It follows immediately that the structural
change results derived for the m - 1 consump-
tion sectors remain intact, as we have made no
changes to that part of the model. But there are
new results to derive concerning structural
change within the capital-producing sectors.
The relative employment shares across the
capital-producing sectors satisfy
(32) nmj/nm = (axmj4/m,)"(Am,/Am)1-L;
V i, j=l,...., K.
(33) =nmi
(1 -nmi
( )(7mi- 'm);
V i, j = 1, ... , .
These equations parallel (13) and (15) of the
baseline model and the intuition behind them is
the same.
When there are many capital goods, the Am of
438 THE AMERICAN ECONOMIC REVIEW MARCH 2007

the baseline model is replaced

capital-producing by
sectors, GmAm
their f
relative employ-
each sector m., wherementGm sharesdenotes
remaining constant
the independently
sector
marginal product in the production
of their TFP growth rates. of aggrega
capital, and Am is theThe model with e < 1TFP
sector's and k = 1level.
has clear This
contrasting
term measures the rate predictions about
of return tothe relation be- in t
capital
jth capital-producing
tween sector, which
the dynamics of sectoral is equ
employment
across all K sectors because
shares and TFPof
growththe(or relative
free prices).
mobilit
Sec-
of capital. In the Appendix we derive
tors that produce primarily consumption goods the a
gregate growth rate should exhibit a well-defined linear relation be-
tween their employment share growth and their
K
TFP growth rate; sectors that produce many
(34) = > I jYm'; intermediate goods should still have a positive
j=1 linear relation, but less well-defined because of
the constant term due to the production of in-
termediate goods. But sectors that produce pri-
M iA'-'M m marily capital goods should exhibit no linear
S A ( t 1 mAmi,4 relation at all between their employment share
growth and their relative TFP growth rate.
which is a weighted average of TFP growth
rates in all capital-producing sectors. The dy- VI. Conclusion
namic equations for c and k are the same as in
the baseline model, given the new definition We have shown that different TFP growth
of Ym.
rates across industrial sectors predict sectoral
If TFP growth rates are equal across all employment changes that are consistent with
capital-producing sectors, c and k grow at a the facts if the substitutability between the final
common rate in the steady state. But then all goods produced by each sector is low. Balanced
capital producing sectors can be aggregated into aggregate growth requires, in addition, a loga-
one, and the model reduces to one with a single rithmic intertemporal utility function. Underly-
capital-producing sector. If TFP growth rates ing the balanced aggregate growth there is a
are different across the capital-producing sec- shift of employment away from sectors with a
tors and 4:f 1, there is structural change within high rate of technological progress toward sec-
the capital-producing sectors along the transi- tors with low growth, and eventually, in the
tion to the asymptotic state. Asymptotically, limit, all employment converges to only two
only one capital-producing sector remains. In sectors, the sector producing capital goods and
the asymptotic state, c and k again grow at a the sector with the lowest rate of productivity
common rate, so there exists an asymptotic ag- growth. If the economy also produces interme-
gregate balanced growth path with only one diate goods, the sectors that produce these
capital-producing sector. goods also retain some employment in the limit,
A necessary and sufficient condition for the which is used to produce the intermediate
coexistence of an aggregate balanced growth goods.
path and multiple capital-producing sectors Our results are consistent with the observa-
tion of simultaneous growth in the relative
with different TFP growth rates is /t = 1. The
reason for this result is that a balanced aggre- prices and employment shares of stagnant sec-
tors such as community services, with the near-
gate path requires a constant ym, which is un-
attainable if the relative TFP levels in the constancy of real consumption shares when
compared with nominal shares. It is also con-
capital-producing sectors are allowed to influ-
sistent with the long-run evidence of Kuznets
ence it. From (34), the influence of the produc-
(1966) and Maddison (1980) concerning the
tivity levels disappears only when / = 1. The
aggregate growth rate in this case is ym/(1decline
-- a), of agriculture's employment share, the
rise and then fall of the manufacturing share,
where y,, = I,axm.m. Using (32), the relative
employment shares across capital-producing and the rise in the service share. The key re-
quirement
sectors are equal to their relative input shares in for these results is again a low sub-
G. There is no structural change within the
stitutability between final goods. Of course, at a

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
VOL. 97 NO. 1 NGAI AND PISSARIDES: STRUCTURAL CHANGE IN A MULTISECTOR MODEL OF GROWTH 439

approach that we suggested for intermediate and

finer sector decomposition, the elasticity of sub-
many capital goods, namely the existence of
stitution between two goods may reasonably
exceed unity, as for example between the output
subsectors that produce an aggregate that enters
the utility or production function, is an obvious
of the sector producing typewriters and the out-
approach to the analysis of these cases. Within
put of the sector producing electronic word pro-
cessors. Our model in this case predicts that the subsectors, there is structural change toward
labor would move from the sector with low TFPthe high TFP goods, but between the aggregates
growth to the one with high TFP growth. The the flow is from high to low TFP sectors.

APPENDIX: PROOFS

LEMMA Al: Equations (2), (5), and (7) imply equation (8).

PROOF:

Defining f(k) = F(k, 1), and omitting subscript i, (7) implies FK = Af(k) and FN = A[f(k) -
kf(k)]. So FN/FK = f(k)/f (k) - k, which is strictly increasing in k. Hence, (5) implies ki = km Vi
m, and together with (2), results follow.

LEMMA A2: Vi : m, ni satisfy (13) and (18), and nm satisfies (14) and (19).

PROOF:

ni follows from substituting F' into (10), and nm is derived from (2). Given 4i/xi = (1 - e)(ym
'i) and X/X = (1 - e)(ym - Y), the result follows for ri, i : m. Using (2),

-
iom
=
(cly)n c/y -
i~m
),

so result fo

PROPOSITION 3

PROOF:
Use (2) and (8) to rewrite (4) as

k/k =Amk-a- I > ni) -cm/k --( + v).

i:Am

But pilPm = Am/Ai and by the definition of c,

k/k = Amk ' I- clk - (8 + v).

Next, 4 is homogenous of degree one:

m m

S=
i=1i= 1
AiCi = >

But 0 Om(O/Cm)1/" and

(to/(- )X 1))1 - 0)c-0-ax, so (6)

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
440 THE AMERICAN ECONOMIC REVIEW MARCH 2007

LEMMA A3: d/ldt 5 0 e 1.

PROOF:
Totally differentiating Y as defined in equations (18) and (19),

m m

d//dt
i=1 j=1 =

m m

= (1 -
i= 1Lj=l

m m

(1
i= 1 i=1
-

Since the sum

Capital Share
and (13)-(17))
production in
share and zi =
1, i.e., the fix

ni/ni = (P

so the result
independent
modified as f

m 3 jam'
ki= Xikm
Different capital shares add the term (am - ai)km/km in (16). In a growth equilibrium with km
growing, lower ai is another reason for higher relative price in sector i. Combining the relative price
and relative employment equations, different capital shares add the term (1 - e)(aj - ai)km/km in
(15). The existence of a fixed factor modifies (15) to

[1-(1n,-)(1 )= (1
n, - e)( - yax)
nj + (1 -km
)(ai - a) Vj m.
If n1 is falling, then the p
and (14) are modified to

Xi C pi xm C pi c Pi
ni = y n, Vi m; nm= X y n +1
J J J

whereimplies
system c = Xcm, y =canAmkn
n1,..., nm Ii simultaneously.
be solved (1mnilaxi), and xi= (ol(/m)[cik[m-i)k aZa+Pi-l(Am/Ai)]1-". The new

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
VOL. 97 NO. 1 NGAI AND PISSARIDES: STRUCTURAL CHANGE IN A MULTISECTOR MODEL OF GROWTH 441

Intermediate Goods.-Vi, Fi Ainikq, a, 3 E (0, 1), a + 1 < 1. We have

(Al) Fm = Cm + hm + ( + v)k + k, F = ci + hi, Vi m.

The planner's problem is similar to the baseline with (Al) replacing (3) and (4), {hi, ci, qii=
m as additional controls, and I~m niqi = ((h1, ... , hm) as an additional constraint, where
homogenous of degree one, (i > 0, and ii < O0. The static efficiency conditions are

(A2) vilv, = F1/F / = F /F/ = Fm/F' = m V i,

which implies ki = k, qi = (, and pi = Am/Ai Vi, so

m m m

y
i=1 i=1 i=1= A

Optimal conditi

k= i:Am
Amk(P(

The dynamic e

(A3) 6/c = ah

Constant 6/c req

To derive c
Pihi/hm = ((Pil(p

h = Zhm, ~

where Z = zi. Hence,

h = (D/(D! = (PAm ka) /( - P)(ax(n- 1)Z1(n- 1))p3/( -3),

and so

(1 - P)//h = (ym + ak/k) +i=11 ( (z,/Z)yi - 79 ,

constant if I= 1 ziy, is constant. Given y, differs across all i, constancy requi

S= hi", Z = 1/Pm, zi = ji/(axm V i.

i=l

(A2) implies ( = hm I H1 (zAj/Am)Ii' and so 4)= p rD/hm

[PAmkFm]1t1-), so h = = (PAmk")l/(1- - P/ (A3) be

6/c + 8 + p + v = aAkaxl('I-))-1"; + c + (6 + v)k = (1 -

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
442 THE AMERICAN ECONOMIC REVIEW MARCH 2007

where A [Am(3(Fpm)ax]11) 1 - ). Define

Ce cA-(1-P)/(1-a-P); ke - kA
We have y = [y,n + Y~ ( 1iyi- -

,e/Ce = ak oa+l- )/(1 - (6 + p + v +

which imply the existence and uniqueness of an

g (1- )/(1) -a- 3) -ax(m-- i=1

We obtain ni using F' = ci + hi, Vi * m, i.e.,

Ainik"aDPpi pi(ci + hi) = XiCm + Zihm = cxi/X + axih.

Substitute pi and h to obtain ny = cxilX + cpi3y, so (30) and (31) follow.

Many Capital-Producing Sectors.-Vj, Fj Amnmk., which together produce good

K K

j=1
G
j=1

The planner

k = G

and (kmj, n

F~F = Fm/F"I, Vi m, Vj,

so ki = km = k. Also
my

Gmj = FK1/FI - AAm;, i, j,

which implies nm/nm. = (m/axm)"(Am!/Am.) and grows at rate (1 - I)('mi Ym). Let nm C K
nm; Mi
we have nm = nml
I (m/'m)(Am/Am
I I )1-C. Next,
Mi

Pi = Vi/Vm

where Am GmA,m
To derive the aggre
same as the baseline
we now derive. Given

Gin, = ml (G/Fm')l/i; G/Fm'

j=l
= mj (mnAmn/(Am, nmi))l)/

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms
VOL. 97 NO. 1 NGAI AND PISSARIDES: STRUCTURAL CHANGE IN A MULTISECTOR MODEL OF GROWTH 443

using the result on nmnm we have GIFm Am thus Am

GmAm, mj
K K

j=l j=lYm
constant if (p
the model redu

producing sectors and an ABGP requires (2), i.e., G = Ij> (Fmj)C and ym = = Im'Ym,-

REFERENCES and Economic Growth." In The Theory of

Capital, ed. F. A. Lutz and Douglas C.
Hague, 177-222. New York: St. Martin's
Acemoglu, Daron, and Veronica Guerrieri. 2005.
Press.
"Capital Deepening and Non-Balanced Eco-
nomic Growth." Unpublished. Kongsamut, Piyabha, Sergio Rebelo, and Dan-
Barro, Robert J., and Xavier Sala-i-Martin. 2004.
yang Xie. 2001. "Beyond Balanced Growth."
Economic Growth: Second Edition. Cam- Review of Economic Studies, 68(4): 869-82.
bridge, MA and London: MIT Press. Kravis, Irving B., Alan W. Heston, and Robert
Baumol, William J. 1967. "Macroeconomics Summers.
of 1983. "The Share of Services in
Economic Growth." In Global Econometrics:
Unbalanced Growth: The Anatomy of Urban
Crisis." American Economic Review, 57(3):
Essays in Honor of Lawrence R. Klein, ed. F.
415-26. Gerard Adams and Bert G. Hickman, 188-
Baumol, William J., Sue Anne Batey Blackman, 218. Cambridge, MA, and London: MIT
and Edward N. Wolff. 1985. "Unbalanced Press.
Growth Revisited: Asymptotic Stagnancy Kuznets, Simon. 1966. Modern Economic
and New Evidence." American Economic Re- Growth: Rate, Structure, and Spread. New
view, 75(4): 806-17. Haven: Yale University Press.
Caselli, Francesco, and Wilbur John Coleman II. Laitner, John. 2000. "Structural Change and
2001. "The U.S. Structural Transformation Economic Growth." Review of Economic
and Regional Convergence: A Reinterpreta- Studies, 67(3): 545-61.
tion." Journal of Political Economy, 109(3): Maddison, Angus. 1980. "Economic Growth and
584-616. Structural Change in the Advanced Coun-
tries." In Western Economies in Transition:
Echevarria, Cristina. 1997. "Changes in Sectoral
Composition Associated with Economic Structural Change and Adjustment Policies
Growth." International Economic Review,in Industrial Countries, ed. Irving Leveson
38(2): 431-52. and Jimmy W. Wheeler, 41-65. London:
Falvey, Rodney E., and Norman Gemmell. 1996.Croom Helm.
"Are Services Income-Elastic? Some New Ngai, L. Rachel, and Christopher Pissarides.
Evidence." Review of Income and Wealth, 2004. "Structural Change in a Multi-Sector
42(3): 257-69. Model of Growth." Center for Economic Pol-
Foellmi, Reto, and Josef Zweimuller. 2005. icy Research Discussion Paper 4763.
"Structural Change, Engel's Consumption Oulton, Nicholas. 2001. "Must the Growth Rate
Cycles and the Kaldor's Facts of Economic Decline? Baumol's Unbalanced Growth Re-
Growth." Unpublished. visited." Oxford Economic Papers, 53(4):
Gollin, Douglas, Stephen Parente, and Richard 605-27.
Rogerson. 2002. "The Role of Agriculture in Sichel, Daniel E. 1997. "The Productivity Slow-
Development." American Economic Review down: Is a Growing Unmeasurable Sector the
(Papers and Proceedings), 92(2): 160-64. Culprit?" Review of Economics and Statis-
Kaldor, Nicholas. 1961. "Capital Accumulation tics, 79(3): 367-70.

This content downloaded from 190.102.151.4 on Thu, 06 Apr 2017 21:49:46 UTC
All use subject to http://about.jstor.org/terms