Long-Run Growth in Open Economies:

Export-Led Cumulative Causation or

a Balance-of-Payments Constraint?


Robert A. Blecker*


Revised draft
December 2010









To appear in: Handbook of Post-Keynesian Economics, edited by Geoff Harcourt and Peter
Krielser (Oxford University Press, 2011, forthcoming).


*The author is Professor and Chair, Department of Economics, American University,
Washington, DC 20016 USA, blecker@american.edu. An earlier version of this chapter was
presented as a lecture at the 2
nd
Summer School on Keynesian Macroeconomics and European
Economic Policies, sponsored by the Research Network Macroeconomics and Macroeconomic
Policies, 2-9 August 2009, Berlin, Germany. The author would like to thank Geoff Harcourt,
Mark Setterfield, Tony Thirlwall, and participants at the Summer School for their comments and
suggestions on the earlier draft.

1. Introduction

Post-Keynesian economists concerned with long-run growth in open economies have developed
two related but fundamentally different theoretical approaches. On the one hand, models of
export-led cumulative causation (ELCC) stress the possibility that some countries can achieve
ever-widening virtuous circles of faster technological progress, improving competitiveness,
rising exports, and rapid output growth (although, in this view, other countries may be doomed to
suffer vicious circles of slower technological progress, worsening competitiveness, stagnant
exports, and sluggish output growth). Providing that labor supply constraints can be overcome,
exports are seen as the key limiting constraint on demand-driven growth in open economies
(Cornwall, 1977, p. 163). On the other hand, models of balance-of-payments-constrained growth
(BPCG) emphasize the limitations placed upon a nations growth by the need to finance
necessary imports through either export growth or financial inflows (McCombie and Thirlwall,
1994, 2004). In this latter view, virtuous circles may be impossible to achieve because rapid
output growth is likely to make imports rise too fast to be compatible with equilibrium in the
balance of payments (Thirlwall and Dixon, 1979).
These two views do coincide in certain respects. Both maintain the Keynesian belief that
aggregate demand constraints are paramount in determining a nations output, even in the long
run, and see those constraints as lying primarily in the international domain rather than the
domestic economy.
1
Both theories agree that increasing the growth rate of exports is key to
raising a countrys long-run growth rate of output. Beyond that, however, the two views disagree
strongly on core theoretical assumptions and policy implications. In regard to theory, the BPCG

1
In this respect, both views are at odds with other post-Keynesian perspectives that put more emphasis on domestic
investment demand as constraining profits and growth. See Robinson (1962) and later expositions by Marglin
(1984) and Harcourt (2006), among others.
2

model puts primary emphasis on import demand and the balance of payments, while the ELCC
model implicitly assumes that these are not limiting factors in the growth process. The ELCC
model focuses on changes in relative cost competitiveness driven by endogenous technological
progress as driving export success (or failure), while the BPCG model assumes that such changes
either dissipate in the long run (due to relative purchasing power parity holding) or else have
small effects on trade flows (so-called elasticity pessimism).
In terms of policy, perhaps the most radical implication of some ELCC models is that a
stimulus to domestic demand can potentially spark a virtuous circle of export-led growth,
because of the positive response of technology and productivity to faster domestic expansion. In
contrast, the BPCG model implies that a stimulus to domestic demand is unlikely to bring
sustained long-run benefits in an open economy because it is likely to raise import demand
without boosting exports. The two models also have different implications for what kinds of
policies can be effective for promoting exports in the long run. The ELCC view implies that cost
reductions or currency depreciations could provoke self-sustaining increases in both exports and
output, while the BPCG view implies that such policies are likely to be ineffective in the long
run and that nothing can raise export growth except either faster growth of foreign economies or
an increase in the income elasticity of export demand. Thus, qualitative competitiveness matters
in both theoretical approaches, but cost competitiveness and real exchange rates matter only in
the ELCC version.
2

This chapter will identify the key theoretical differences between the ELCC and BPCG
approaches and evaluate how and to what extent they can be reconciled by representing both in a
common analytical framework. In spite of its motivation as a theory of cumulative causation, the

2
Some extensions of the BPCG approach allow for relative price or real exchange rate effects (e.g., Blecker, 2002;
Dutt, 2002; Perraton, 2003), but I am referring here to the original version of Thirlwall (1979).
3

export-led approach can be represented by a model that has an equilibrium solution. It will be
shown that this solution is not a sustainable long-run equilibrium precisely because it lacks a
plausible balance-of-payments constraint. However, in some interpretations the ELCC model
was not intended to represent a long-run equilibrium in a conventional sense, but rather a kind of
provisional equilibrium that is only a weak attractor for the economy in the medium run, and
which is subject to path-dependent displacements due to endogenous negative feedbacks
(Setterfield, 2002). Also, we will show that the incompatibility of the ELCC model with balance-
of-payments equilibrium can be demonstrated without making all the strong assumptions
typically made in the simplest BPCG models, particularly the assumptions that rule out relative
price (real exchange rate) effects.
Although early versions of the BPCG model assumed that the trade balance (current
account) must be zero in the long run, subsequent developments of this approach have
incorporated financial (capital) flows and identified the conditions for these to be sustainable in
the long run. Similar results were obtained by McCombie and Thirlwall (1997), who analyzed
the conditions for the ratio of a countrys external debt to gross domestic product (GDP) to
stabilize at a constant level, and Moreno-Brid (1998, 1998-99), who analyzed the conditions for
the ratio of the current account balance to GDP to stabilize also at a constant level. By allowing
for sustainable financial flows in this sense and also incorporating the relative price effects that
are assumed away in Thirlwalls (1979) original approach, we find that a broader solution of the
BPCG modelone that allows for cumulative causation to have some impact at least in the
medium runreconciles the core contributions of both approaches.

4

2. Literature Survey

The ELCC concept harks back to Adam Smiths (1776 [2003], p. 27) famous dictum that the
division of labour is limited by the extent of the market. Extrapolating from this principle,
Smith deduced that one of the distinct benefits of international trade was that:
By means of [foreign trade], the narrowness of the home market does not hinder
the division of labour in any particular branch of art or manufacture from being
carried to the highest perfection. By opening a more extensive market for
whatever part of the produce of their labour may exceed the home consumption, it
encourages them to improve its productive powers, and to augment its annual
produce to the utmost, and thereby to increase the real wealth and revenue of the
society. (Smith, 1776 [2003], pp. 561-62)
Smiths emphasis on dynamic feedbacks from exports to productivity growth was largely
forgotten after Ricardo (1821 [1951]) shifted the focus of international trade theory to static
efficiency gains based on comparative advantage. However, the idea of dynamic feedbacks was
revived by some of the early post-World War II development economists, especially Myrdal
(1957) in his principle of circular and cumulative causation. Nicholas Kaldor borrowed this
concept in his work on explaining differences in growth rates among the industrialized nations
(Kaldor, 1966, 1970), and also drew upon the work of Young (1928) in arguing that the
pervasiveness of increasing returns invalidates the general equilibrium approach to economics
(Kaldor, 1972). More specifically, Kaldor came to believe that the analysis of economic growth
should be founded on a series of empirical generalizations or stylized facts, which have come
to be known as Kaldors Growth Laws. Four of these laws, as summarized by Thirlwall (1983,
pp. 345-47, italics in original) are most relevant to the present discussion:
i) The faster the rate of growth of the manufacturing sector, the faster will be the
rate of growth of Gross Domestic Product (GDP)....
ii) The faster the rate of growth of manufacturing output, the faster will be the rate
of growth of labor productivity in manufacturing owing to static and dynamic
5

economies of scale, or increasing returns in the widest sense....
vi) The growth of manufacturing output is not constrained by labor supply but is
fundamentally determined by demand from agriculture in the early stage of
development and exports in the later stages....
vii) A fast rate of growth of exports and output will tend to set up a cumulative
process, or virtuous circle of growth, though the link between output growth and
productivity growth.
Of these propositions, number ii) is based on Verdoorns (1949) empirical finding of a positive
correlation between the growth rates of labor productivity and output in manufacturing across
countries, and thus is often referred to as Verdoorns Law.
Several economists developed explicit models of export-led growth that sought to
incorporate these ideas of cumulative causation. Early contributions included Beckerman (1962)
and Lamfalussy (1963), but the most widely known and accepted version is due to Dixon and
Thirlwall (1975) and was further developed by Cornwall (1977), Setterfield and Cornwall
(2002), and Setterfield (2002), among many others. Figure 1 shows the basic logic of a
simplified version of the ELCC model, showing the circular and cumulative causation between
export growth, output growth, productivity growth, and international competitiveness (measured
by the rate of increase in the real exchange rate or relative price of foreign goods). The diagram
is drawn to represent a virtuous circle of increases in all these factors; a vicious circle of
decreases in all these factors could be represented simply by reversing the upward direction of
the arrows shown next to the variables (inside the text boxes). Unlike some of the earlier models
in the literature, the graphical representation in Figure 1 ignores the special role of the
manufacturing sector and focuses only on aggregate output, productivity, and exports.
[Figure 1 about here]
In terms of empirical support, it is worth remembering that the ELCC model was
developed (e.g., by Kaldor, 1966) as a generalization of empirical regularities found in early
6

cross-country regression analysis. More recently, Len-Ledesma (2002) estimated an extended
version of an ELCC model, in which a fifth endogenous variable (a measure of R&D
expenditures) is added, and a number of exogenous variables are included to help identify the
structural equations in a simultaneous equations framework.
3
Len-Ledesma finds that most of
the coefficients representing the key causal relationships in his extended ELCC model have the
theoretically expected signs and are statistically significant. However, Len-Ledesma does not
test the validity of the ELCC model versus any other particular model (such as BPCG) as a
predictor of long-run average growth rates.
Almost as soon as it was developed, the ELCC growth model received an important
challenge from Thirlwall (1979) and others who, although sympathetic to the Kaldorian
approach, believed that the ELCC models erred in ignoring the role of import demand and
neglecting to incorporate a balance-of-payments (BP) equilibrium condition. Thirlwall and
Dixon (1979, p. 173) criticized these models (including their own earlier version) because No
consideration is given to the possibility that the rate of growth of income determined by the
model may generate a rate of growth of imports in excess of the rate of growth of exports,
thereby imposing a constraint on the export-led growth rate if balance of payments equilibrium
must be preserved. If import demand is incorporated into the model and a BP constraint is
imposed, exports continue to play a key role in determining long-run growth because faster
growth of exports allows faster growth of imports without risking a chronic BP (current account)
deficit.
4
Nevertheless, Thirlwall and Dixon (1979) showed that, under certain assumptions, the

3
Len-Ledesma uses data for 17 countries averaged over four time periods between 1965 and 1994 and employs
two- and three-stage least squares to solve identification problems.
4
Although this literature generally refers to balance of payments equilibrium, it is clear from the context that what
is really meant is balance on current account. Furthermore, the models usually ignore all other components of the
current account besides trade in goods and services.
7

cumulative causation mechanism is thwarted and the growth rate consistent with BP equilibrium
is determined solely by the ratio of the growth rate of exports to the income elasticity of import
demand, regardless of whether Verdoorns Law (Kaldors second law, which incorporates
dynamic increasing returns) holds.
This solution for the BP-constrained growth rate, which is also found in Thirlwall (1979),
is sometimes referred to (following Davidson, 1990-91) as Thirlwalls Law. This simple
version of the Law depends on certain strong assumptions, however. One of these assumptions
is that the current account must be balanced in the long run. The BPCG model can easily be
modified to allow for a long-run current account imbalance matched by a sustainable level of
capital (net financial) inflows or outflows.
5
One way of incorporating financial flows was
introduced by Thirlwall and Hussain (1982), who assumed a given growth rate of net financial
inflows. However, Thirlwall and Hussains approach potentially allows for perpetually rising
ratios of financial inflows (net borrowing) or external debt to GDP, which cannot be sustainable
in the long run. In contrast, later models assumed that either a countrys net external debt
(McCombie and Thirlwall, 1997) or its current account balance (Moreno-Brid, 1998, 1998-99)
must be a constant share of GDP in the long run. These two approaches lead to equivalent
solutions for the BP-constrained growth rate, which will be presented below.
A second key assumption in Thirlwalls Law is that there are no relative price effects in
the long run. This assumption is crucial for ruling out cumulative causation when these effects
are modeled as lowering the relative price of a countrys exports by raising its productivity
growth rate. Thirlwall (1979) argued that it is realistic to assume that relative purchasing power
(PPP) holds in the long run, so that relative prices of home and foreign goods (measured in a

5
This chapter will adopt the terminology of modern balance-of-payments accounting (e.g., in the IMF and US
government presentations) in referring to financial flows and the financial account, rather than capital flows
and the capital account, regardless of the terminology used in earlier studies.
8

common currency) do not permanently change. In this case, any competitive gains from more
rapid productivity growth must be offset either by exchange rate appreciation or a rise in
domestic prices (perhaps, although Thirlwall did not say so, because of increasing nominal
wages or raw materials costs; he simply cited the famous law of one price).
6
Alternatively,
Thirlwall and Dixon (1979) point out that the same result is achieved if the price elasticities of
export and import demand (in absolute value) sum to approximately unity, so that the Marshall-
Lerner condition
7
is not satisfied and changes in international relative prices have no effect on
the trade balance (this assumption will be referred to below as elasticity pessimism).
8
Thirlwall
and Hussain (1982) accept that changes in export prices (which, in their model, are the same as
domestic prices) may be significant for developing countries, but only insofar as they affect the
real value of net financial inflows measured in home currencynot for cumulative causation
reasons (which are implicitly ruled out by the assumption of PPP).
9

It is not so clear, however, that relative price or real exchange rate effects can be
completely neglected on either of these grounds. Alonso and Garcimartn (1998-99) find
econometric evidence for the elasticity pessimism view in most of the industrialized countries
covered in their study (Canada and J apan are two notable exceptions, where they find that
Marshall-Lerner holds), but they have an unusual way of modeling lagged price effects, and
other studies have found price elasticities of exports and imports that sum to more than unity for

6
For an analysis of the stability of the equilibrium in a Thirlwall-type BPCG model focusing mainly on labor
markets and wage dynamics, see Pugno (1998).
7
The Marshall-Lerner condition states that a real devaluation of a countrys currency will improve its balance of
trade if the absolute value of the sum of the price elasticities of export and import demand exceeds unity, under
certain assumptions including initially balanced trade. The extended Marshall-Lerner condition with imbalanced
trade is stated below.
8
There is a subtle difference in the Thirlwalls Law solutions under these two assumptions, i.e., PPP and elasticity
pessimism, as discussed below.
9
Perraton (2003) finds that including terms of trade (relative price) effects, i.e., not assuming PPP, improves the fit
of BPCG models applied to data for developing countries.
9

most countries (e.g., Cline, 1989).
10
In studies of various individual countries, ranging from the
US to India, Marshall-Lerner is often found to hold (e.g., Lawrence, 1990; Blecker, 1992;
Razmi, 2005), although one recent study finds that it holds only barely in the US case (Chinn,
2004). At best, the evidence on elasticity pessimism is mixed, and elasticity estimates vary
widely across different countries, time periods, and econometric methodologies. Moreover,
according to standard J -curve logic, we would expect price elasticities to be relatively low and
Marshall-Lerner to be violated in the short run (i.e., up to a year or two following a devaluation),
but elasticities to increase (in absolute value) and satisfy Marshall-Lerner over longer time
periods when trade flows are (for well-known reasons) easier to adjust.
The empirical evidence on long-run, relative PPP is also mixed, and highly sensitive to
the currencies, price indexes, time periods, and econometric methods used (see Rogoff, 1996).
11

One recent survey summarizes the descriptive evidence from US-UK exchange rates as follows:
Neither absolute nor relative PPP appear to hold closely in the short run, although
both appear to hold reasonably well as a long-run average and when there are
large movements in relative prices, and both appear to hold better between
producer price indices than between consumer price indices. (Taylor and Taylor,
2004, p. 139).
An empirical literature from the 1990s that found that real exchange rates were not mean-
reverting even in the long run has now been overturned by more powerful econometric methods,
but more recent studies also find that the speed of adjustment to PPP can sometimes be slow and

10
Alonso and Garcimartn (1998-99) estimate desired levels of exports and imports and assume that actual trade
flows adjust gradually to these levels, while most other studies use distributed lags of the independent variables to
explain actual exports and imports. Alonso and Garcimartn also find evidence supporting the BPCG view that
output growth rates, rather than international relative prices (real exchange rates), are the variable that adjusts to
restore BP equilibrium when countries have current account imbalances. However, Len-Ledesma (2002) finds that
relative prices do change as predicted in the ELCC model, i.e., in response to productivity growth differentials, and
have a significant impact on export growth.
11
In fairness to Thirlwall (1979), the evidence for PPP was stronger at the time he conceived the BPCG model,
based mainly on data from the Bretton Woods period of fixed but adjustable pegs for exchange rates. The behavior
of real exchange rates since the collapse of Bretton Woods and the advent of flexible rates has been less kind to the
PPP hypothesis.
10

mean-reversion may be non-linear (i.e., faster for larger deviations from PPP) (Taylor and
Taylor, 2004). Even if relative PPP holds over very long periods of time such as a half century or
longer for certain major currencies, it does not generally hold for shorter time periods (or all
currencies).
12

Furthermore, there is no consensus that long-run equilibrium real exchange rates should
be constant over time (and some of the newer studies finding faster rates of adjustment also
allow for time-varying equilibrium rates). Many economists have argued that real exchange rates
need to change in the long run in response to factors such as international differences in income
elasticities of export and import demand (Houthakker and Magee, 1969; Chinn, 2004),
international differences in relative rates of productivity growth between tradable and
nontradable goods (Balassa, 1964; Samuelson, 1964; Chinn, 2000), and changes in international
asset or debt positions (Lane and Milesi-Ferretti, 2002).
Although it is difficult to summarize the subtle and ever-changing literature on PPP, it
does seem safe to conclude that the longer the time period one considers, the more likely it is that
relative PPP will hold. It also seems that PPP is more likely to hold between countries that are
more structurally similar and hence do not have reasons for their equilibrium real exchange rates
to change over long periods of time. Recent empirical studies in the BPCG framework using
modern time-series methods have found that the model without relative price effects applies
better over very long time periods than over medium-run periods. For example, Razmi (2005)
finds that the BPCG model explains Indian growth well for his full sample period of 1950-1999,
as well as two subperiods of about 30 years, and concludes that income was by far the dominant
influence in determining the balance of payments constraint in the long run (p. 682). But for

12
For example, Bahmani-Oskoee (1995) found that PPP held for only 8 out of 22 developing countries studied.
11

shorter periods (such as decades), he found substantial variations of actual growth rates from
BPCG predictions and concluded that relative price effects could not be neglected. This
conclusion opens the door to incorporating cumulative causation and relative price effects in
BPCG models applied to medium-run time periods, even if those effects are not plausible over
very long-run periods.

3. The Export-Led Growth Model

The following model, which is adapted with some modifications from Setterfield and
Cornwall (2002), represents the core ideas of the ELCC approach.
13
Although the model is
simplified in certain respects compared to earlier ELCC models discussed in the previous section
(especially in not treating manufacturing and other sectors separately), the model has been
specified in aggregate terms to facilitate comparisons with the BPCG model in later sections.
This aggregative approach can also be justified by appeal to Kaldors first law, which, as stated
above, says that faster growth of manufacturing output results in faster growth of aggregate
output (i.e., the manufacturing sector is the engine of growth for the whole economy).
Starting on the export side, export demand is specified as a conventional, constant-
elasticity function of the real exchange rate and foreign income. With all variables measured in
instantaneous rate-of-change form (differences in natural logarithms) for convenience, the export
function is:
(1) x =c
x
(e +p* p) +q
x
y* ,
where x is the growth rate of exports, e is the rate of nominal currency depreciation (with the

13
The first three equations of the model correspond closely to the earlier model of Dixon and Thirlwall (1975); the
fourth equation is Setterfield and Cornwalls innovation.
12

exchange rate measured in home currency per unit of foreign currency), p and p* are the rates of
change in the home and foreign price indexes, respectively, y* is the growth rate of foreign (rest-
of-world) income, and c
x
and q
x
are the price and income elasticities of export demand,
respectively (defined so that c
x
, q
x
>0). Note that (e +p* p) represents the rate of real
depreciation of the home currency, or rate of increase in the relative price of foreign goods.
Foreign price inflation p* is taken as exogenously given on the small country assumption,
but domestic price inflation p is determined by changes in unit labor costs and the gross profit
markup:
(2) p =t +w q ,
where (again with all variables in growth rate form) t is the rate of change in (one plus) the
markup over unit labor cost, w is the rate of wage inflation, and q is the rate of labor productivity
growth. Although markups can change due to changes in non-labor costs or competitive
conditions in markets, we abstract from changes in markups here for simplicity and assume t =
0.
14
Productivity growth is assumed to be endogenous according to an aggregative version of
Verdoorns Law:
(3) q =q
0
+oy ,
where q
0
is a shift factor representing autonomous technological dynamism (including catch-up
possibilities) and technology policies (e.g., R&D subsidies, intellectual property rights, etc.), o
represents the Verdoorn effect (dynamic increasing returns or positive feedbacks), and q
0
, o >0.
Finally, aggregate demand (national income) increases at a rate determined by the
weighted average of the growth rates of export demand and domestic expenditures multiplied by

14
See Blecker (1998) for a BPCG model that incorporates flexible markups with partial pass-through of exchange
rate changes.
13

the Keynesian multiplier :
15

(4) y =(e
a
a +e
x
x) ,
where a is the growth rate of autonomous domestic expenditures and e
a
and e
x
are the shares of
domestic expenditures and exports (respectively) in total demand. Equations (1), (2), and (4) can
then be combined (assuming t =0) to yield what Setterfield and Cornwall (2002) call the
demand regime (DR) equation,
16

(5) q y
x x
c e + O =
where ]} ) ( [ {
* *
y w p e a
x x x a
q c e e + + + = O and the variables a, e, p*, w, and y* are all treated
as exogenously given constants. This means that domestic demand, the nominal exchange rate,
foreign prices, domestic wages, and foreign income are all assumed to grow at constant rates.
While treating foreign variables as exogenously given can be justified on the small country
assumption, assuming that the other variables are fixed independently of home country output
and productivity growth is more dubious, and the implications of this strong assumption will be
discussed below.
Then, defining the Verdoorn equation (3) as the productivity regime (PR), Setterfield
and Cornwall note that (3) and (5) together constitute a system of two linear equations in two

15
See Setterfield and Cornwall (2002, pp. 81-82, Appendix 5.A) for derivation. Some earlier ELCC models assume
that y is an increasing function of x, but do not explicitly include a. The model of Beckerman (1962), as presented in
Cornwall (1977, pp. 165-167), assumes that y =x but only for the manufacturing sector.
16
This specification differs in certain respects from Setterfield and Cornwalls DR relation. They ignore exchange
rates and wages, effectively assuming that e =w =0. They also assume that p* is determined by foreign equations
for markup pricing and Verdoorns Law parallel to equations (2) and (3) here; in our notation, (2') p* =t* +w* q*
and (3') q* =q
0
* +o*y*, where the *s indicate foreign variables or parameters (the authors omit the * on q
0
*, but I
presume this was a typographical error). Since Setterfield and Cornwall also assume w* =t* =0, p* ultimately
depends only on y*, which is taken as exogenously given. Thus, in their model (but in our notation, and after
correcting the typographical error in their published solution), the constant term in their DR relation is
] ) ( [
* * *
0
y q a
x x x x x a
c o q e c e e + = O . We prefer to treat p* as exogenously given and to allow explicitly for e
and w to assume nonzero values, since this is not only more general but also facilitates comparisons with the BPCG
model below.
14

endogenous variables (output growth y and productivity growth q), taking all other variables as
exogenously given. Provided that there is not too much cumulative causation, this system then
solves for a unique and stable equilibrium. Graphically (see Figure 2), this requires that the PR
line be steeper than the DR line in qy space, which is equivalent to the slope condition 1/o >
e
x
c
x
or oe
x
c
x
<1. Thus, the mere existence of cumulative causation is not sufficient to create
a disequilibrium situation; only if the forces of cumulative causation are very strong (i.e.,
1 >
x x
c oe , or PR is flatter than DR) would an equilibrium not exist. Assuming that the
equilibrium exists, the equilibrium ELCC growth rate y
E
is determined by the simultaneous
solution of equations (3) and (5):
(6)
x x
x x
E
q
y
c oe
c e

+ O
=
1
0

.

[Figure 2 about here]
Although it may seem contrary to the spirit of Kaldor (1972) to represent his ideas using
a model that has an equilibrium solution, it should be noted that this demand-determined growth
equilibrium is quite different from a conventional model of a long-run equilibrium growth path
uniquely determined by exogenous increases in factor supplies and factor productivity.
Disequilibrium in the ELCC model presented here would imply ever-rising or ever-falling
growth rates, which do not seem plausible in the long run (China may have grown at a 10% clip
in recent years, but this rate has not continued to increase). Setterfield (2002) has suggested that
the ELCC approach does not define a stable long-run equilibrium, but rather defines a sort of
temporally punctuated equilibrium in which a country settles for some period of time on a
growth path defined by a certain set of demand and productivity conditions (i.e., DR and PR
regimes), but then moves along a traverse toward a new equilibrium as the underlying
15

parameters of the system endogenously adjust.
17
In this view of the ELCC model, the
equilibrium solution of the model is at best a weak attractor for the medium run (Setterfield,
2002, p. 227), and negative as well as positive feedbacks are admitted into the long-run evolution
of an economy (which is seen as a path-dependent process).
Returning to the logic of the ELCC model as specified above, its comparative dynamic
properties are easily analyzed. Any policy that would exogenously stimulate productivity growth
(for example, an R&D subsidy or improved technical education) would increase q
0
and shift the
PR line down and to the right, thereby having a positive effect on the equilibrium growth rate y
E
.
Similarly, any event that would stimulate exports to grow faster (such as a faster rate of currency
depreciation e, faster foreign income growth y*, or an opening of foreign markets that raised the
income-elasticity of export demand q
x
) would shift the DR line upward, also increasing the
ELCC growth rate y
E
. What is most surprising in this model, however, is that a stimulus to
domestic demand (increase in a) would have the same effect as a stimulus to export demand in
shifting the DR line up, thereby permanently raising the equilibrium growth rate y
E
. In fact,
given the logic of this model, the domestic demand stimulus would actually increase export
growth by causing productivity to rise faster, thereby making exports more competitive.
This may seem like a strong, if not unbelievable, conclusion. Of course, this strong result
stems in part from the oversimplified, aggregative nature of the present model. In a more
complete ELCC framework, in which the Verdoorn relationship (endogenous productivity
growth) was limited to the manufacturing sector, only the part of increased domestic demand that
went toward the purchase of domestically produced manufactures would be able to kick-start the

17
Mark Setterfield also comments (in e-mail correspondence, 16 November 2010) that the ELCC model as specified
here has an innately non-long-run set-up because equation (4) allows exports, domestic expenditures, and output
to grow at different rates, which would not be possible in a truly long-run steady state (and also the weights e
a
and
e
x
would not be expected to remain constant in the long run).
16

process of cumulative causation; demand increases that were spent on services or imports would
not have the same effect. Furthermore, it is possible that exposure to the discipline of
international competition induces more innovative effort and quality control than sales in the
domestic market, as argued by Amsden (1989) for South Korea.
Even leaving these issues of disaggregation aside, the ELCC model laid out above
ignores other economic forces that could limit the cumulative growth gains from any type of
demand stimulus, especially a domestic one but also an export-led one. Perhaps most obviously,
the assumption that the nominal rates of currency depreciation e and wage inflation w would
remain constant irrespective of an increase in domestic growth of output and productivity could
be called into question. Much theory and intuition suggest that a country experiencing an export-
led boom might be expected to confront pressures toward currency appreciation (lower e) or
faster wage increases (higher w). The former can be avoided through currency market
intervention (as in China in the past decade), provided that the central bank can sterilize the
resulting reserve accumulation. The latter, however, is harder to avoid.
Advocates of the ELCC approach (e.g., Cornwall, 1977) emphasize that the labor supply
is not an inelastic constraint on long-run growth, as it appears in conventional models of the
Solow (1956) variety. Labor supplies can vary elastically in the growth process due to factors
such as international migration (including temporary guest workers as well as more permanent
forms of migration) and changes in social norms regarding age and gender in the workplace. In
multisectoral models, labor can potentially be drawn out of less-productive agricultural or
service sectors into manufacturing. In developing countries with dual economies, migration of
surplus labor from rural or pre-modern areas may augment modern sector labor supplies. But
unless the labor supply is perfectly elastic at the current wage level, it would seem that some
17

upward pressure on wages is unavoidable in a rapidly growing economy (and China, where
wages are increasing rapidly, is no exception). Thus, some of the optimistic aspects of the ELCC
model stem from the implicit assumption that improvements in cost competitiveness due to faster
productivity growth are not counterbalanced by offsetting currency appreciation or wage
increases. While one cannot necessarily assume that these adjustments will fully offset all
competitive gains in a process of cumulative causationexchange rates do not always behave as
predicted, and wage increases may lag behind productivity growthneither should one ignore
these types of adjustments altogether.

4. The Balance-of-Payments-Constrained Growth Model

A related but distinct critique of the ELCC approach concerns the fact that the ELCC model
lacks an import function and a balance-of-payments equilibrium condition, and hence the
equilibrium growth rate specified may be inconsistent with the long run requirement of payments
balance (Thirlwall and Dixon, 1979, p. 173). This section presents the model of Thirlwall and
Dixon (1979), who sought to correct this problem, modified to incorporate financial flows.
To enhance comparability with the ELCC model in the preceding section, equations (1)
through (3) are all retained, including the Verdoorn relation to incorporate cumulative causation.
The model is then augmented by adding a conventional import demand function with constant
elasticities
(7) m =c
m
(e +p* p) +q
m
y ,
where m is the growth rate of imports and c
m
and q
m
are the price and income elasticities of
import demand, respectively (defined so that c
m
, q
m
>0).
18

Next, we define BP equilibrium as implying that the current account balance (surplus or
deficit) must equal a constant, sustainable ratio to national income. Following the approach of
Moreno-Brid (1998),
18
but converted into our notation, a constant ratio of the current account
balance to GDP implies that
(8) u(x y) =e +p* p +m y ,
where u is the ratio of the value of exports to the value of imports, both measured in domestic
currency.
19
Note that (8) now replaces (4) as the equation that determines the output growth
rate.
20
If we then substitute equations (1), (2), (3), and (7) into (8) and solve for y, again
assuming t =0, we obtain a very general expression for the BP-constrained growth rate, y
B
:
21

(9)
) 1 ( 1
) )( 1 (
*
0
*
+ +
+ + + +
=
m x m
x m x
B
y q w p e
y
c uc o u q
uq c uc
.
It is at this point that BPCG theorists typically introduce certain strong assumptions to
rule out relative price effects. First, suppose that the price elasticities of import and export
demand are too low to satisfy the extended Marshall-Lerner condition with financial flows; more
specifically, suppose uc
x
+c
m
~ 1.
22
Under this assumption, (9) reduces to
23


18
As observed earlier, equivalent results would be obtained using the approach of McCombie and Thirlwall (1997).
19
See Moreno-Brid (1998) for the derivation. The more traditional assumption of balanced trade (i.e., a current
account balance equal to zero) in the long run is the special case in which u =1 and (8) becomes p +x =e +p* +m.
20
Since equation (4) still has to holdit is, after all, nothing more than a dynamic version of the standard Keynesian
multiplier for national income determinationthe domestic expenditure growth rate a in equation (4) must now
become endogenous in the long run so that national income will grow at the rate implied by (9). This conforms with
Kaldors mature view that exports are the only truly exogenous constraint in the long run (Palumbo, 2009).
21
This equation combines the solution of Thirlwall and Dixon (1979, p. 183, equation 19) for a BPCG model with
cumulative causation with the solution of Moreno-Brid (1998, p. 418, equation 8) for a model including financial
flows, both translated into the present notation. To obtain intuitively plausible results, the denominator of this
solution must be assumed to be positive.
22
From now on, the term elasticity pessimism will be used to represent this assumption in models that allow for
financial flows. Note that the extended Marshall-Lerner condition uc
x
+c
m
>1 is less likely to hold, and the
alternative that uc
x
+c
m
~ 1 is more likely to hold, the lower is u (i.e., the larger is the countrys trade deficit), for
any given values of c
x
and c
m
. Thus, elasticity pessimism is more warranted in countries with large initial deficits.
23
Again, for intuitively sensible results (a positive growth rate), the denominator must be positive. This is very
19

(10)
u q
uq
+
=
1
*
m
x
B
y
y
and o (the Verdoorn coefficient representing endogenous feedbacks from income growth to
productivity growth) has no effect on the BP-constrained equilibrium growth rate. If we further
add the assumption that trade must be balanced in the long run, so that u =1, then (10) becomes:
(11)
m
x
B
y
y
q
q
*
= .
Second, consider what happens if relative PPP is assumed, so that e +p* p =0. To see
the implications, note that when e +p* p =0 is assumed, these relative price change (real
depreciation) terms drop out of both the export and import demand functions (equations 1 and 7)
and also from the equilibrium condition (8). Since Verdoorns Law (3) and markup pricing (2)
enter the model only through the relative price effects, these two relationships also disappear
from the solution. In this case, x =q
x
y* by (1) and m =q
m
y by (7), and substituting these into the
equilibrium condition
(8') u(x y) =m y ,
the model again solves for equation (10)or (11) if we also assume u =1. Furthermore, in this
case (10) and (11) respectively simplify to
24

(10')
u q
u
+
=
1
m
B
x
y
and (assuming u =1)

likely in equation (10), since the only way it could be negative would be for a country to have a very low income
elasticity of import demand q
m
and a very large trade deficit (u <<1), which seems like an unlikely combination. If
q
m
>1, as seems to be empirically true in most countries, then q
m
1 +u >0 regardless of the size of u (since u >0).
24
Perraton (2003) refers to equation (11) as the strong form of Thirlwalls BPCG hypothesis and (11') as the
weak form.
20

(11')
m
B
x
y
q
= ,
Now, several important points emerge immediately. First, there are a variety of different
solutions for the BP-constrained growth rate, depending on what we assume about three factors:
price elasticities, real exchange rates, and financial flows. Equation (11') states the simplest and
perhaps best known version of Thirlwalls Law, but it is a special case that requires both PPP and
the absence of financial flows. Second, which of these versions of the BPCG model applies is
likely to depend on the time frame considered. As noted earlier, elasticity pessimism is more
likely to hold in the short run than in the long run, while PPP (i.e., the absence of real exchange
rate changes) is more likely to hold in the long run than in the short or medium run (and the
validity of both assumptions can vary across countries). Third, none of these solutions (i.e.,
equations 9, 10, 10', 11, or 11') is equivalent to the ELCC growth rate shown in equation (6).
Thus, regardless of what we assume about these factors, the economy is not likely to grow at the
ELCC growth rate (6) if the BP constraint is effective, even if cumulative causation effects are
present and relative price effects are significant. Fourth, merely assuming that the BP must be in
equilibrium (in the sense that the current account balance is either zero or a sustainable, constant
fraction of GDP) does not by itself rule out any impact of cumulative causation effects on the
economys long-run growth rate. As long as PPP does not hold and the extended Marshall-
Lerner condition does hold, cumulative causation can affect the most general BP-constrained
growth rate (9) even though this is not the same as the ELCC growth rate (6).
25
For the reasons
noted earlier, this is more likely to occur in medium-run time periods (e.g., decades) than over
longer ones (e.g., generations or half centuries).

25
Note that, assuming uc
x
+c
m
>1 and e +p* p = 0, and also assuming that the denominator of (9) is positive as
discussed above, then cy
B
/co >0 in (9).
21


5. Reconciling the Two Growth Rates

To fix ideas, it is helpful to compare the ELCC and BPCG solutions graphically. Here,
we use the most general solution for the BP-constrained growth rate, i.e., equation (9), which
allows for cumulative causation effects and financial flows, does not assume PPP, and does
assume that the extended Marshall-Lerner condition holds. To represent this solution on the same
type of diagram as was used for the ELCC model in Figure 2, it is convenient to substitute
equations (1), (2), and (7) into (8)again assuming t =0 and taking e and w as exogenously
givenand obtain the following equation for y as a function of q:
(12)
u q
uq c uc
+
+ + + +
=
1
) )( 1 (
* *
m
x m x
B
y q w p e
y .
This relationship, which is upward sloping as long as the extended Marshall-Lerner condition
holds and PPP does not hold (and horizontal if either Marshall-Lerner is violated or PPP holds),
26

is represented by the dashed BP line in Figures 3 and 4.
[Figures 3 and 4 about here]
In Figures 3 and 4, the BPCG solution y
B
is represented by the point where the BP
constraint (12) intersects the PR relation (3), while the ELCC solution y
E
is represented by the
point where DR (5) intersects PR (3). In general, the BP relation may lie either above or below
the DR relation, as there is no reason (from inspection of equations 5 and 12) that one is
necessarily higher than the other.
27
Figure 3 shows the case where y
E
>y
B
and Figure 4 shows the

26
Note that, if the extended Marshall-Lerner condition does not hold, (12) becomes (10), and if PPP does hold, then
(12) becomes (10'). Either way, the BP relationship then becomes horizontal in Figures 3 and 4, i.e., y
B
is
independent of q.
27
Since this statement applies to the slopes as well as the intercepts in equations (5) and (12), the DR and BP
22

opposite situation. We shall focus, with no loss of generality, on the case shown in Figure 3,
where the ELCC growth rate exceeds the BP-constrained growth rate.
Now, it might be thought that allowing for financial flows to be somewhat elastic might
relax the BP constraint, effectively permitting the BP constraint to shift upward until y
B
would
coincide with y
E
. After all, financial flows do not have to be fixed at a given percentage of GDP;
some range of current account imbalances can be sustainable as long as the interest rate on the
foreign debt is not too high relative to the countrys growth rate. Since the BP equation (12)
assumes a given ratio of the current account balance to GDP, cant we just assume that this ratio
might fall somewhat (i.e., become more negative, assuming the country has a current account
deficit and a financial account surplus), thereby enabling the economy to borrow more and
thereby to reach the ELCC growth rate without risking a chronic payments disequilibrium?
The answer, somewhat surprisingly, is no, at least not in the present framework. To see
this point, recall that the BP equilibrium condition (8) is specified in growth rate form, and
assumes a constant ratio of the current account balance to GDP. A decrease in this ratio does not
change the form of the equilibrium condition (8). Of course, greater net financial inflows would
imply a lower ratio of exports to imports u, but this has ambiguous effects on the height of the
BP relation (12) and may very well shift it down (away from DR) instead of up (toward DR).
28

Or, to put the point another way, note that at the ELCC equilibrium (y =y
E
) in Figure 3, the

relations could also intersect, thereby creating additional cases not shown in Figures 3 and 4. These other cases are
not considered for reasons of space, and because the key theoretical points can be illustrated with full generality by
reference to the simpler cases shown in these two figures.
28
Consider, for example, the simple case of a horizontal BP constraint given by equation (10), which would apply if
either elasticity pessimism or PPP holds. In this equation, the shift in BP due to a change in u is given by
2
*
) 1 (
) 1 (
u q
q q
u +

=
c
c
m
x m B
y y

which has the same sign as q
m
1. If q
m
>1, as is often found empirically, the BP relation would shift down when u
falls. If elasticity pessimism or PPP is violated so that the more general BP equation (12) applies, this derivative
becomes more complex but does not cease to be ambiguous in sign.
23

economy is not merely experiencing a current account deficit, but rather a current account
balance that is continuously falling relative to GDP. In other words, it is not possible to grow at
the rate y
E
with any constant ratio of the current account deficit (or external debt) to GDP, and
therefore y
E
is simply unsustainable in the long run. Similarly, in Figure 4, the economy could
not grow at y
E
unless it could have a perpetually rising ratio of the current account deficit (or
external debt) to GDP, which is also not sustainable.
The impossibility of growing at y
E
derives simply from the assumptions that BP
equilibrium must be maintained in the long run and financial flows cannot increase or decrease
indefinitely as a percentage of GDP (i.e., as long as the ratio of the current account deficit or
external debt to GDP cannot rise or fall continuously). The strong assumptions of either PPP or
elasticity pessimism are not required for this result. If neither of these assumptions holds, the
correctly specified BP-constrained growth rate incorporates cumulative causation effects, as in
equation (9). It is this growth rate, not the ELCC growth rate (6), that should be considered the
long-run equilibrium growth rate in a properly specified post-Keynesian open economy model
though, as noted earlier, many ELCC advocates dont necessarily view (6) as a truly long-run
equilibrium solution. Thirlwall and Dixon (1979) were correct in saying that either PPP or
elasticity pessimism rules out cumulative causation effects on the BP-constrained growth rate,
but they were incorrect in implying that these assumptions are necessary to show that the ELCC
growth rate is not a sustainable long-run equilibrium.
This then raises the question of how an economy that is experiencing a virtuous circle of
export-led growth (e.g., at the point where y =y
E
in Figure 3) in the short run can adjust to grow
instead at the lower BP-constrained growth rate y
B
in the long run. Since the BP constraint
cannot be expected to rise to intersect DR at y
E
, for the reasons discussed above, the DR relation
24

must instead fall to intersect BP at y
B
, and the question becomes what adjustment mechanism(s)
can be expected to make DR fall toward BP in the long run. The first and most obvious candidate
is a decrease in the rate of growth of domestic expenditures a, which lowers the intercept term O
in equation (5) and thereby shifts DR downward. This could be accomplished either through
deliberate government policy (e.g., contractionary fiscal or monetary policies),
29
or through
private sector spending restraint (perhaps induced by rising debt burdens during the period of
booming growth). Thus, an expenditure reducing policy (or a private sector expenditure
reduction) is a plausible way of making the ELCC equilibrium shift to coincide with the BPCG
equilibrium.
30

A second possibility is a relative price adjustment. As noted earlier, an export-led boom
(such as where y
E
>y
B
in Figure 3) could be expected to lead to either faster nominal wage
growth (a rise in w) or currency appreciation (a fall in e), either of which would also lower O and
shift DR downward. However, whether this is a stable adjustment process or not depends on
whether relative price effects are allowed in the BPCG solution. Consider first the cases assumed
by Thirlwall and Dixon (1979), where either PPP or elasticity pessimism holds, so that equation
(12) reduces to (10) or (10'). In this situation, the BP curve is horizontal and is not affected by
changes in w or e, so the DR relation can shift down toward a fixed BP relation and there can be
a stable adjustment to the BPCG equilibrium where y =y
B
.
Thus, under Thirlwall and Dixons (1979) assumptions, and even allowing for financial

29
This statement applies to reductions in the growth rate of government spending or increases in interest rates
designed to slow the growth of private consumption and investment spending. Tax increases that raise marginal tax
rates would instead lower the Keynesian multiplier , which would reduce the slope as well as the intercept of the
DR relation (5).
30
Several empirical studies have researched to what extent fiscal and monetary policies as well as private sector
spending respond endogenously to limit current account imbalances. See, for example, Summers (1988), Bayoumi
(1990), Artis and Bayoumi (1990), and Epstein and Gintis (1992).
25

flows, their BP-constrained growth rate is a stable attractor for the long run equilibrium. Indeed,
the rise in w or fall in e in this situation can be seen as a mechanism that brings about the PPP
condition assumed by Thirlwall and Dixon for the long run. However, if the adjustments in w
and e are not sufficient to establish PPP and Marshall-Lerner holds, then the relative price
adjustment process may be unstable. Assuming uc
x
+c
m
>1 and e +p* p = 0, then cy
B
/ce >0
and cy
B
/cw <0 in (12), so a fall in e or rise in w would also shift the BP relation downward in
Figure 3, and, depending on the other parameters, the downward shift in DR might never catch
up with the downward shift in BP so that the BPCG equilibrium would never be reached. The
longer the time period considered, however, the more likely it is that PPP would eventually rule,
and hence the more likely that the adjustment process will eventually prove to be stable.
Similar considerations would apply in the opposite case depicted in Figure 4. If an
economy was growing at the rate y
E
<y
B
in Figure 4, the country would be experiencing a
continuously increasing ratio of current account surplus (or net foreign assets) to GDP, which is
not plausible in the long run. Consequently, the country would have to adjust through either
expansion of domestic demand (a rise in a, perhaps through expenditure increasing policies) or
else a real depreciation (rise in e or fall in w relative to p*), assuming that the relative price
changes would bring about a stable adjustment process (which is more likely, if the changes in w
or e bring the country toward PPP in the very long run).
There is, however, another way of modeling the BP constraint with financial flows that
could make it flexible so that the economy could possibly adjust to grow at the rate y
E
(in the
situation shown in either Figure 3 or 4). That is, if we model financial flows following the
approach of Thirlwall and Hussain (1982), who did not assume constancy of either the current
account balance or external debt to GDP ratio, the BP equilibrium condition can be written as:
26

(8) (1|)(x +p) +|f =e +p* +m ,
where (using lower case letters to represent growth rates and upper case Roman letters to
represent levels of variables) f is the growth rate of financial inflows (F, measured in domestic
currency) and | =F/(PX +F) is financial inflows as a percentage of total receipts in the
balance of payments (if there are net financial outflows then F <0, and also | <0 assuming F <
PX). If we then substitute equations (1), (2), (3), and (7) into (8), we obtain the following
(modified) version of Thirlwall and Hussains most general solution for the BP-constrained
growth rate as a function of a given rate of increase in financial inflows f:
31

(13)
| |
| | ) 1 ) 1 (
) ( ) 1 ( ) ( ) 1 ) 1 (
0
*
0
*
| c c | o q
| q | c c |
+ +
+ + + + +
=
m x m
x m x
B
q w f y q w p e
y
Assuming that the denominator of (13) is positive, as seems necessary for intuitively
plausible results, then cy
B
/cf >0. Now the outcome depends on whether f is exogenously fixed or
can adjust elastically in a situation where y
B
= y
E
. Suppose, for example, y
B
<y
E
as in Figure
3.
32
It is plausible that, in a country experiencing an export-led boom, financial inflows would be
attracted into the countrys growing economy, thereby increasing f and raising the BP-
constrained growth rate y
B
toward the higher rate y
E
. Similarly, if y
B
>y
E
as in Figure 4, it is
plausible that in a country stuck in a vicious circle of slow export growth and slow
productivity growth, financial inflows would be reduced or outflows would increase, thereby

31
Thirlwall and Hussain (1982) assumed PPP, in which case (13) simplifies to
m
x
B
p f y
y
q
| q | ) ( ) 1 (
*
+
=
.

32
Note that, if (8) replaces (8) as the BP equilibrium condition, then the BP constraint (12) shown in Figures 3 and
4 (i.e., y
B
written as a function of q) would have to be replaced by the following expression:
| | | |
m
m x x m x
B
q w f y w p e
y
q
| c c | | q | c c | ) 1 ) 1 ( ) ( ) 1 ( ) ( ) 1 ) 1 (
* *
+ + + + + + +
=
,

which would be upward sloping if (1 |)c
x
+c
m
1 +| >0. Also note that if this BP constraint replaces (12), then
the meaning of being above or below the BP line also changes: points above it represent higher rates of growth of
financial inflows (f) and points below it represent lower rates f.
27

depressing f and pushing y
B
downward toward y
E
. On the other hand, if financial markets are
relatively closed and financial inflows or outflows are inelastic, then y
B
cannot adjust toward y
E

and the BP constraint will remain in force at a fixed level of f.
However, even if f is flexible, it cannot change too drastically relative to the rate of
nominal income growth (y +p) or the country may experience a potentially unsustainable
explosion of its foreign debt or asset position. In the case shown in Figure 3, if f <y +p initially
and f does not have to increase too much to make y
B
=y
E
, the BP constraint could be relaxed and
the country could enjoy its export-led growth boom as long as the financial inflows kept growing
at the requisite rate and the current account deficit and external debt did not become
unsustainably large. On the other hand, if f would have to rise so much that f >y +p would
result, then the current account to GDP ratio would begin to fall (and the external debt to GDP
ratio would begin to rise) continuously, which (as noted previously) is implausible in the long
run.
Even so, an export-led boom financed by financial inflows could easily persist for a
while, perhaps sustained by bubble behavior in financial markets (since it is typical in a bubble
that investors suspend disbelief and ignore warning signs of unsustainable financial positions
see Shiller, 2008), leading eventually to some kind of financial crash or debt crisis marked by a
shift from a bubble to a panic mentality in the financial markets. In this case, the financial crisis
would be the enforcement mechanism that would impose a stricter BP constraint in the long
run, but at tremendous cost to the country involved.

28

6. Conclusions

In spite of their differences, the BPCG and ELCC models share common roots in
Kaldors ideas about the centrality of export markets for facilitating demand-led growth in open
economies. Both models have found empirical support in the literature, although upon closer
examination it appears that support for the strongest versions of the BPCG model (i.e., the
versions that exclude relative price effects) is found mainly in very long-run data while support
for ELCC models is to be found mostly in studies that adopt a more medium-run perspective. If
relative price effects are allowed in BPCG-type modelsand the evidence suggests that these
may be significant for some countries over periods of a decade or longerthen it is possible to
incorporate the cumulative causation effects emphasized in the ELCC approach in models that
embed a BP constraint. However, the solution of this synthetic model is different from the one
that emerges from a pure ELCC model without a BP constraint.
Allowing for financial flows, which is important for considering open economies in an
epoch of financial liberalization, adds further insight to the comparison of the two approaches. If
we assume, following Thirlwall and Hussain (1982), that countries obtain financial inflows that
grow at steady rates, then it is conceivable that a country undergoing rapid export-led growth la
the ELCC model could relax its BP constraint and continue to grow rapidly by increasing the
growth rate of its financial inflows within certain limits. However, as pointed out by McCombie
and Thirlwall (1997) and Moreno-Brid (1998, 1998-99), financial inflows cannot grow at any
arbitrary rate in the long run without potentially causing the current account deficit and external
debt to rise without limit as percentages of GDP, which would be unsustainable. If we assume
instead that either the current account balance or external debt must be a stable fraction of GDP
29

in the long run, then it becomes clear that the BP constraint really is binding and the ELCC
equilibrium cannot generally be reached. Or, to put it another way, increasing financial inflows
can at most be a temporary way of relaxing the BP constraint, but they do not allow a country to
grow at the ELCC growth rate in the long run.
The conclusion that the strict BPCG model holds only in the very long run when relative
price effects can be ignored (e.g., due to PPP holding) should not, of course, be surprising. Since
its earliest formulations (e.g., Thirlwall, 1979), the BPCG model without relative price effects
has always been intended as a long-run model. But, this should not be taken to imply that
medium-run models such as ELCC are irrelevant. Even if Thirlwalls Law holds in growth rate
form in the very long run, we may conjecture that it is possible that the level of income at which
a countrys BP constraint is satisfied could be permanently affected by its virtuous or vicious
circles of cumulative causation over the intervening medium-run periods. Future theoretical
research in this framework may wish to address this possibility, while more empirical work is
required to identify the conditions and time frames under which different versions of these
models operate.

30

References

Alonso, J ., and C. Garcimartn. 1998-99. A new approach to the balance-of-payments constraint:
some empirical evidence. Journal of Post Keynesian Economics 21(2): 259-282.
Amsden, A. H. 1989. Asias Next Giant: South Korea and Late Industrialization. New York:
Oxford University Press.
Artis, M., and T. Bayoumi. 1990. Saving, investment, financial integration, and the balance of
payments. In International Monetary Fund, Staff Studies for the World Economic Outlook
1990. Washington, DC: International Monetary Fund.
Bahmani-Oskoee, M. 1995. Real and nominal effective exchange rates for 22 LDCs: 1971:1 -
1990:4. Applied Economics 27: 591604.
Balassa, B. 1964. The purchasing-power parity doctrine: a reappraisal. Journal of Political
Economy 72(6): 584-596.
Bayoumi, T. 1990. Saving-investment correlations: immobile capital, government policy, or
endogenous behavior? International Monetary Fund Staff Papers 37 (J une): 360-387.
Beckerman, W. 1962. Projecting Europes growth. Economic Journal 72 (December): 912-925.
Blecker, R. A. 1992. Structural roots of U.S. trade problems: income elasticities, secular trends,
and hysteresis. Journal of Post Keynesian Economics 14(3): 321-346.
Blecker, R. A. 1998. International competition, relative wages, and the balance-of-payments
constraint. Journal of Post Keynesian Economics 20(4): 495-526.
Blecker, R. A. 2002. The balance-of-payments-constrained growth model and the limits to
export-led growth. In P. Davidson, ed., A Post Keynesian Perspective on Twenty-First
Century Economic Problems. Cheltenham, UK: Edward Elgar.
Chinn, M. D. 2000. The usual suspects? productivity and demand shocks and Asia-Pacific real
exchange rates. Review of International Economics 8(1): 20-43.
Chinn, M. D. 2004. Incomes, exchange rates, and the U.S. trade deficit, once again. International
Finance 7(3): 451-469.
Cline, W. R. 1989. United States External Adjustment and the World Economy. Washington,
DC: Institute for International Economics.
Cornwall, J . 1977. Modern Capitalism: Its Growth and Transformation. New York: St. Martins.
Davidson, P. 1990-91. A Post Keynesian positive contribution to theory. Journal of Post
Keynesian Economics 13(2): 298-303.
31

Dixon, R. J ., and A. P. Thirlwall. 1975. A model of regional growth-rate differences on
Kaldorian lines. Oxford Economic Papers 27(2): 201-214.
Dutt, A.K. 2002. Thirlwalls law and uneven development. Journal of Post Keynesian
Economics 24(3): 367-390.
Epstein, G., and H. Gintis. 1992. International capital markets and the limits of national
economic policy. In T. Banuri and J . B. Schor, eds., Financial Openness and National
Autonomy: Opportunities and Constraints. Oxford: Oxford University Press.
Harcourt, G. C. 2006. The Structure of Post-Keynesian Economics: The Core Contributions of
the Pioneers. Cambridge: Cambridge University Press.
Houthakker, H. S., and S. P. Magee. 1969. Income and price elasticities in world trade. Review of
Economics and Statistics 51(2): 111-125.
Kaldor, N. 1966. The Causes of the Slow Rate of Economic Growth of the United Kingdom.
Cambridge: Cambridge University Press.
Kaldor, N. 1970. The case for regional policies. Scottish Journal of Political Economy 17(3):
337-348.
Kaldor, N. 1972. The irrelevance of equilibrium economics. Economic Journal 82 (December):
1237-1255.
Lamfalussy, A. 1963. The United Kingdom and the Six: An Essay on Economic Growth in
Western Europe. Homewood, IL: Irwin.
Lane, P., and G. M. Mileri-Ferretti. 2002. External wealth, the trade balance and the real
exchange rate. European Economic Review 46(6), 1049-1071.
Lawrence, R. Z. 1990. U.S. current account adjustment: an appraisal. Brookings Papers on
Economic Activity 1990(2): 343-389.
Len-Ledesma, M. A. 2002. Accumulation, innovation and catching-up: an extended cumulative
growth model. Cambridge Journal of Economics 26(2): 201-216.
Marglin, S. A. 1984. Growth, Distribution, and Prices. Cambridge, MA: Harvard University
Press.
McCombie, J . S. L., and A. P. Thirlwall. 1994. Economic Growth and the Balance-of-Payments
Constraint. New York: St. Martins.
McCombie, J . S. L., and A. P. Thirlwall. 1997. Economic growth and the balance-of-payments
constraint revisited. In P. Arestis, G. Palma, and M. Sawyer, eds., Markets,
Unemployment and Economic Policy: Essays in Honour of Geoff Harcourt, vol. 2.
London: Routledge.
32

McCombie, J . S. L., and A. P. Thirlwall, eds. 2004. Essays on Balance of Payments Constrained
Growth: Theory and Evidence. London: Routledge.
Moreno-Brid, J . C. 1998. Balance-of-payments constrained economic growth: the case of
Mexico, Banca Nazionale del Lavoro Quarterly Review, no. 207 (December): 413-433.
Moreno-Brid, J . C. 1998-99. On capital flows and the balance-of-payments-constrained growth
model. Journal of Post Keynesian Economics 21(2): 283-298.
Myrdal, G. 1957. Economic Theory and Underdeveloped Regions. London: Duckworth.
Palumbo, A. 2009. Adjusting theory to reality: the role of aggregate demand in Kaldors late
contributions on economic growth. Review of Political Economy 21(3): 341-368.
Perraton, J . 2003. Balance of payments constrained growth and developing countries: an
examination of Thirlwalls hypothesis. International Review of Applied Economics 17(1):
1-22.
Pugno, M. 1998. The stability of Thirlwalls model of economic growth and the balance-of-
payments constraint. Journal of Post Keynesian Economics 20(4): 559-581.
Razmi, A. 2005. Balance-of-payments-constrained growth model: the case of India. Journal of
Post Keynesian Economics 27(4): 655-87.
Ricardo, D. 1821. Principles of Political Economy and Taxation, 3
rd
ed. Reprinted Cambridge:
Cambridge University Press, 1951.
Robinson, J . 1962. Essays in the Theory of Economic Growth. London: Macmillan.
Rogoff, K. 1996. The purchasing power parity puzzle. Journal of Economic Literature 34(2):
647-668.
Samuelson, P. A. 1964. Theoretical notes on trade problems. Review of Economics and Statistics
46(2): 145-154.
Setterfield, M. 2002. A model of Kaldorian traverse: cumulative causation, structural change and
evolutionary hysteresis. In M. Setterfield, ed., The Economics of Demand-led Growth:
Challenging the Supply-side Vision of the Long Run. Cheltenham, UK: Edward Elgar.
Setterfield, M., and J . Cornwall. 2002. A neo-Kaldorian perspective on the rise and decline of the
golden age. In M. Setterfield, ed., The Economics of Demand-led Growth: Challenging
the Supply-side Vision of the Long Run. Cheltenham, UK: Edward Elgar.
Shiller, R. J . 2008. The Subprime Solution: How Todays Global Financial Crisis Happened, and
What to Do about It. Princeton: Princeton University Press.
Smith, A. 1776. The Wealth of Nations. Reprinted New York: Bantam/Dell, 2003.
33

Solow, R. M. 1956. A contribution to the theory of economic growth. Quarterly Journal of
Economics 70 (February): 65-94.
Summers, L. H. 1988. Tax policy and international competitiveness. In J . A. Frenkel, ed.,
International Aspects of Fiscal Policies. Chicago: University of Chicago Press.
Taylor, A. M., and M. P. Taylor. 2004. The purchasing power parity debate. Journal of
Economic Perspectives 18(4): 135-158.
Thirlwall, A. P. 1979. The balance of payments constraint as an explanation of international
growth rate differences. Banca Nazionale del Lavoro Quarterly Review, no. 128 (March):
45-53.
Thirlwall, A. P. 1983. A plain mans guide to Kaldors growth laws. Journal of Post Keynesian
Economics 5(3): 345-358.
Thirlwall, A. P., and R. J . Dixon. 1979. A model of export-led growth with a balance of
payments constraint. In J . K. Bowers, ed., Inflation, Development and Integration: Essays
in Honour of A. J. Brown. Leeds: Leeds University Press.
Thirlwall, A. P., and M. N. Hussain. 1982. The balance of payments constraint, capital flows and
growth rate differences between developing countries. Oxford Economic Papers 34(3):
498-510.
Verdoorn, P. J . 1949. Fattori che regolano lo sviluppo della produttivit del lavoro. LIndustria 1
(March): 3-10.
Young, A. A. 1928. Increasing returns and economic progress. Economic Journal 38
(December): 527-542.
34

More rapid export growth , |x
Faster output growth, |y
Faster labor productivity growth, |q
Increased competitiveness or
faster real currency depreciation,
| (e +p* p)
Keynesian multiplier
effects, increased
capacity utilization,
stimulus to
investment
Increasing returns to
scale, induced
technological
innovation
(Verdoorns Law)
Mark-up pricing over unit
labor costs, taking nominal
wage increases as given
Export demand
function with a high
relative price elasticity

Figure 1 Basic export-led growth model, schematic form (italics explain causal mechanisms
indicated by large arrows; see text for definitions)



Labor productivity growth rate (q)
Output
growth
rate (y)
O
(5) Demand Regime (DR)
y q q o + =
0
(3) Productivity
Regime (PR)
q
0
q
0
/o
q y
x x
c e + O =
y
E

Figure 2 Solution of export-led growth model with cumulative causation (see text for
definitions)
35

Labor productivity growth rate (q)
Output
growth
rate (y)
(5) Demand regime (DR)
(3) Productivity regime (PR)
q
0
q
0
/o
y
E
(12) Balance-of-payments
constraint (BP)
y
B

Figure 3 Comparing the BPCG and ELCC solutions: the case of y
E
>y
B




Labor productivity growth rate (q)
Output
growth
rate (y)
(5) Demand regime (DR)
(3) Productivity regime (PR)
q
0
q
0
/o
y
E
(12) Balance-of-payments
constraint (BP)
y
B

Figure 4 Comparing the BPCG and ELCC solutions: the case of y
E
<y
B