CHAPTER 3 Classic Theories of Economic Growth and Development 155

Appendix 3.2
The Solow Neoclassical Growth Model
The Solow neoclassical growth model, for which Robert Solow of the Massachu-
setts Institute of Technology received the Nobel Prize, is probably the best-known
model of economic growth.1 Although in some respects Solow’s model describes
a developed economy better than a developing one, it remains a basic reference
point for the literature on growth and development. It implies that economies will
conditionally converge to the same level of income if they have the same rates
of savings, depreciation, labor force growth, and productivity growth. Thus, the
Solow model is the basic framework for the study of convergence across countries
(see Chapter 2). In this appendix, we consider this model in further detail.
The key modification from the Harrod-Domar (or AK) growth model, con-
sidered in this chapter, is that the Solow model allows for substitution between
capital and labor. In the process, it assumes that there are diminishing returns
to the use of these inputs.
The aggregate production function, Y = F(K, L) is assumed characterized
by constant returns to scale. For example, in the special case known as the
Cobb-Douglas production function, at any time t we have
Y1t2 = K1t2 α 1A 1t2L 1t22 1 - α (A3.2.1)
where Y is gross domestic product, K is the stock of capital (which may include
human capital as well as physical capital), L is labor, and A(t) represents the
productivity of labor, which grows over time at an exogenous rate.
Because of constant returns to scale, if all inputs are increased by the same
amount, say 10%, then output will increase by the same amount (10% in this
case). More generally,
γY = F 1γK, γL)
where γ is some positive amount (1.1 in the case of a 10% increase).
Because γ can be any positive real number, a mathematical trick useful in
analyzing the implications of the model is to set γ = 1 >L so that
Y>L = f1K>L, 12 or y = f1k2 (A3.2.2)
Lowercase variables are expressed in per-worker terms in these equations. The
concave shape of ƒ(k)—that is, increasing at a decreasing rate—reflects dimin-
ishing returns to capital per worker, as can be seen in Figure A3.2.1.2 In the
Harrod-Domar model, this would instead be a straight, upward-sloping line.
This simplification allows us to deal with just one argument in the production
function. For example, in the Cobb-Douglas case introduced in Equation A3.2.1,
y = Akα (A3.2.3)
This represents an alternative way to think about a production function, in
which everything is measured in quantities per worker. Equation A3.2.3 states
that output per worker is a function that depends on the amount of capital
per worker. The more capital with which each worker has to work, the more
output that worker can produce. The labor force grows at rate n per year, say,
156 PART ONE Principles and Concepts

FIGURE A3.2.1 Equilibrium in the Solow Growth Model

y = f(k)

f(k)

(n + δ)k

sf (k)

k
k*

and labor productivity growth, the rate at which the value of A in the pro-
duction function increases, occurs at rate λ. The total capital stock grows when
savings are greater than depreciation, but capital per worker grows when
savings are also greater than what is needed to equip new workers with the
same amount of capital as existing workers have.
The Solow equation (Equation A3.2.4) gives the growth of the capital-labor
ratio, k (known as capital deepening), and shows that the growth of k depends on
savings sf(k), after allowing for the amount of capital required to service depre-
ciation, δk, and after capital widening, that is, providing the existing amount of
capital per worker to net new workers joining the labor force, nk. That is,
∆k = sf1k2 - 1δ + n 2k (A3.2.4)
Versions of the Solow equation are also valid for other growth models, such as
the Harrod-Domar model.
For simplicity, we are assuming for now that A remains constant. In this
case, there will be a state in which output and capital per worker are no longer
changing, known as the steady state. (If A is increasing, the corresponding state
will be one in which capital per effective worker is no longer changing. In that
case, the number of effective workers rises as A rises; this is because when
workers have higher productivity, it is as if there were extra workers on the
job.) To find this steady state, set ∆k = 0:
sf1 k* 2 = 1 δ + n 2k* (A3.2.5)
The notation k* means the level of capital per worker when the economy is in
its steady state. That this equilibrium is stable can be seen from Figure A3.2.1.3
The capital per worker k* represents the steady state. If k is higher or lower
than k*, the economy will return to it; thus k* is a stable equilibrium. This sta-
bility is seen in the diagram by noting that to the left of k*, k < k*. Looking at
the diagram, we see that in this case, 1n + δ2 k < sf(k). But now looking at the
Solow equation (Equation A3.2.4), we see that when (n + δ)k < sf(k), ∆k > 0.
As a result, k in the economy is growing toward the equilibrium point k*. By
similar reasoning to the right of k*, (n + δ)k > sf(k), and as a result, ∆k < 0
 CHAPTER 3 Classic Theories of Economic Growth and Development 157

(again refer to Equation A3.2.4), and capital per worker is actually shrinking
toward the equilibrium k*.4 Note that in the Harrod-Domar model, sf(k) would
be a straight line, and provided that it was above the (n + δ)k line, growth in
capital per worker—and output per worker—would continue indefinitely.
Equation (A3.2.5) has an interpretatation that the savings per worker,
sf(k*), is just equal to δk*, the amount of capital (per worker) needed to replace
depreciating capital, plus nk*, the amount of capital (per worker) that needs to
be added due to population (labor force) growth.
The Solow model has a (single) equilibrium income per worker, again
given by Equation (A3.2.5) above. In contrast, the Harrod Domar equilibrium
is (constant, balanced) growth—there is no equilibrium income per worker.
Essentially, this is because f(k)—and hence sf(k)—does not exhibit diminishing
returns; rather, it is a straight line. That is, growth continues as long as the line
sf(k) stays above the line (δ + n)k.
It is instructive to consider what happens in the Solow neoclassical growth
model if we increase the rate of savings, s. A temporary increase in the rate of
output growth is realized as we increase k by raising the rate of savings. We
return to the original steady-state growth rate later, though at a higher level
of output per worker in each later year. The key implication is that unlike in
the Harrod-Domar (AK) analysis, in the Solow model an increase in s will not
increase growth in the long run; it will only increase the equilibrium k*. That
is, after the economy has time to adjust, the capital-labor ratio increases, and
so does the output-labor ratio, but not the rate of growth. The effect is shown
in Figure A3.2.2, in which savings is raised to s′. In contrast, in the Harrod-
Domar model, an increase in s raises the growth rate. (This is because in the
Harrod-Domar model, sf(k) becomes a straight line from the origin that does
not cross (n + δ)k; and so, as we assume that sf(k) lies above (n + δ)k, growth
continues at the now higher Harrod-Domar rate—a result that was repre-
sented, for example, in the comparison of Equations 3.8 and 3.9.)
Note that the neoclassical growth model (Equation A3.2.5 and Figure
A3.2.1) implies that while economies will (conditionally) converge to the same
level of income per worker other things equal, it does not imply unconditional
convergence. This can be seen clearly in Figure A3.2.2: We can interpret the
alternative savings rates (s and s’) in the figure as corresponding to those pre-
vailing in two different countries; the country with the higher savings rate
converges to a higher equilibrium income per worker.
Note carefully that in the Solow model, an increase in s does raise equi-
librium output per person—which is certainly a valuable contribution to
development—just not the equilibrium rate of growth. And the growth rate
does increase temporarily as the economy kicks up toward the higher equi-
librium capital per worker. Moreover, simulations based on cross-national
data suggest that if s is increased, the economy may not return even halfway
to its steady state for decades.5 That is, for practical purposes of policymaking
in developing countries, even if the Solow model is an accurate depiction of
the economy, an increase in savings may substantially increase the growth
rate for many decades to come. (Both theoretically and empirically, the link
between the rate of savings and the rate of growth remains controversial.)
Finally, it is possible that the rate of savings (and hence investment) is posi-
tively related to the rate of technological progress itself so that the growth
of A depends on s. This could be the case if investment uses newer-vintage
158 PART ONE Principles and Concepts

FIGURE A3.2.2 The Long-Run Effect of Changing the Savings

Rate in the Solow Model

y
f(k)

(n + δ)k

s'f(k)
sf(k)

k
k* k* 

capital and hence is more productive, if investment represents innovation

in that it solves problems faced by the firm, and if other firms see what the
investing firm has done and imitate it (“learning by watching”), generating
externalities. This leads to a model between the standard Solow model and the
endogenous growth models such as the one examined in Appendix 3.3.

Notes

1. Robert M. Solow, “A contribution to the theory of 5. See N. Gregory Mankiw, David Romer, and David
economic growth,” Quarterly Journal of Economics N. Weil, “A contribution to the empirics of eco-
70 (1956): 65–94. nomic growth,” Quarterly Journal of Economics
2. Note that the symbol k is used for K > L and not for 107 (1992): 407–437. This article shows that
K > Y, as it is used in many expositions (including when human capital is accounted for, as well as
previous editions of this text) of the AK or Harrod- physical capital, the Solow model does a rather
Domar model. good job of explaining incomes and growth
across countries. For a critical view, see William
3. Readers with more advanced mathematical training
Easterly and Ross Levine, “It’s not factor accu-
may note that Figure A3.2.1 is a phase diagram,
mulation: Stylized facts and growth models,”
which applies given that the Inada conditions hold:
World Bank Economic Review 15 (2001): 177–219,
that the marginal product of k goes to infinity as
with the reply by Robert M. Solow, “Applying
k goes to zero and goes to zero as k goes to infinity
growth theory across countries,” World Bank
(this follows from Inada conditions assumed sepa-
Economic Review 15 (2001): 283–288. For time-
rately for capital and labor inputs). This diminishing-
series evidence that the Solow model does a
returns feature drives results of the Solow model.
good job of explaining even the case of South
4. Note that in the Solow model with technological Korean growth, see Edward Feasel, Yongbeom
progress, that is, growth of A, the capital-labor Kim, and Stephen C. Smith, “Investment,
ratio grows to keep pace with the effective labor exports, and output in South Korea: A VAR
force, which is labor power that is augmented by approach to growth empirics,” Review of Devel-
its increasing productivity over time. opment Economics 5 (2001): 421–432.
 CHAPTER 3 Classic Theories of Economic Growth and Development 159

Appendix 3.3
Endogenous Growth Theory
Motivation for Endogenous Growth Theory
The mixed performance of neoclassical theories in illuminating the sources
of long-term economic growth has led to dissatisfaction with traditional
growth theory. In fact, according to traditional theory, there is no intrinsic
characteristic of economies that causes them to grow over extended periods
of time. The literature is instead concerned with the dynamic process through
which capital-labor ratios approach long-run equilibrium levels. In the
absence of external “shocks” or technological change, which is not explained
in the neoclassical model, all economies will converge to zero growth. Hence,
rising per capita GNI is considered a temporary phenomenon resulting from
a change in technology or a short-term equilibrating process in which an
economy approaches its long-run equilibrium.
Any increases in GNI that cannot be attributed to short-term adjustments
in stocks of either labor or capital are ascribed to a third category, commonly
referred to as the Solow residual. This residual is responsible for roughly 50% Solow residual The pro-
of historical growth in the industrialized nations.1 In a rather ad hoc manner, portion of long-term economic
growth not explained by
neoclassical theory credits the bulk of economic growth to an exogenous or growth in labor or capital and
completely independent process of technological progress. Though intuitively therefore assigned primarily
plausible, this approach has at least two insurmountable drawbacks. First, to exogenous technological
using the neoclassical framework, it is impossible to analyze the determinants change.
of technological advance because it is completely independent of the decisions
of economic agents. And second, the theory fails to explain large differences in
residuals across countries with similar technologies.
According to neoclassical theory, the low capital-labor ratios of devel-
oping countries promise exceptionally high rates of return on investment. The
free-market reforms impressed on highly indebted countries by the World
Bank and the International Monetary Fund should therefore have prompted
higher investment, rising productivity, and improved standards of living. Yet
even after the prescribed liberalization of trade and domestic markets, many
developing countries experienced little or no growth and failed to attract new
foreign investment or to halt the flight of domestic capital. The frequently
anomalous behavior of developing-world capital flows (from poor to rich
nations) helped provide the impetus for the development of the concept of
endogenous growth theory or, more simply, the new growth theory. Endogenous growth theory
The new growth theory provides a theoretical framework for analyzing (new growth theory)
Economic growth gen-
endogenous growth, persistent GNI growth that is determined by the system erated by factors within the
governing the production process rather than by forces outside that system. In production process (e.g.,
contrast to traditional neoclassical theory, these models hold GNI growth to increasing returns or induced
be a natural consequence of long-run equilibrium. The principal motivations technological change) that are
studied as part of a growth
of the new growth theory are to explain both growth rate differentials across model.
countries and a greater proportion of the growth observed. More succinctly,
endogenous growth theorists seek to explain the factors that determine the size
of λ, the rate of growth of GDP that is left unexplained and exogenously deter-
mined in the Solow neoclassical growth equation (i.e., the Solow residual).