Professional Documents
Culture Documents
Endogenous Growth Model
Endogenous Growth Model
The new growth models emerging after 1986 depart from the Solow model in three
main ways. One group of models generates continuous growth by abandoning the
assumption of diminishing returns to capital accumulation. To achieve this, Paul Romer
(1986) introduced positive externalities from capital accumulation so that the creation of
economy-wide knowledge emerges as a by-product of the investment activity of individual
firms, a case of ‗learning by investing‘ (Barro and Sala-i-Martin, 2003). A second approach
models the accumulation of knowledge as the outcome of purposeful acts by entrepreneurs
seeking to maximize private profits; that is, technological progress is endogenized (P. Romer,
1990). A third class of model claims that the role of capital is much more important than is
suggested by the α term in the conventional Cobb–Douglas production function. In their
‗augmented‘ Solow model, Mankiw et al. (1992) broaden the concept of capital to include
‗human capital‘. The first two classes of model constitute the core of endogenous growth
theory whereas the Mankiw, Romer and Weil (MRW) model constitutes what Klenow and
Rodriguez-Clare (1997a, 1997b) call a ‗neoclassical revival‘. The central proposition of
endogenous growth theory is that broad capital accumulation (physical and human capital)
does not experience diminishing returns. The growth process is driven by the accumulation of
broad capital together with the production of new knowledge created through research and
development.
The new growth theory provides a theoretical framework for analyzing persistent
growth of output that is determined within the system governing the production process. In
the new growth theories, there exist technological spillovers, externalities and increasing
returns to scale. They do not expect convergence, rather they accept the fact that
disparities among countries can persist or even enlarge. They emphasize on the
importance of investments in human capital and potential gains from technology
improvements – inventions and innovations. The theories of endogenous growth can be
broadly divided into main parts
1. Where growth is driven by Research and Development (R&D) and
2. Where growth is driven by Human Capital Accumulation.
The models so far do not provide satisfying answers to the central questions about
economic growth. The models‘ principal result is a negative one: if capital‘s earnings reflect
its contribution to output, then capital accumulation does not account for a large part of either
long-run growth or cross-country income differences. And the only determinant of income in
the models other than capital is a mystery variable, the ―effectiveness of labour‖ (A), whose
exact meaning is not specified and whose behaviour is taken as exogenous. During the mid-
1980s several economists, most notably Paul Romer (1986, 1987b) and Robert Lucas (1988),
sought to construct alternative models of growth where the long-run growth of income per
capita depends on ‗investment‘ decisions rather than unexplained technological progress. The
key to endogenous steady state growth is that there should be constant returns to broad capital
accumulation‘. Hence in order to construct a simple theory of endogenous growth, the long-
run tendency for capital to run into diminishing returns needs to be modified to account for
the extraordinary and continuous increases in observed per capita incomes across the world‘s
economies.
Research & Development (R&D) based models originate from the work of Romer
(1990), and Grossman and Helpman (1991). In all these models economic growth is the result
of technological change that comes from purposive R&D activities by firms. Patents and
blueprints are non-rival goods that can be accumulated without bounds and so the
diminishing returns to capital accumulation can be avoided and growth continues.
𝑌 𝑡 = [ 1 − 𝑎 𝑘 𝐾 𝑡 ]𝛼 𝐴 𝑡 1 − 𝑎 𝐿 𝐿 𝑡 1− 𝛼
, 0<α<1
This equation implies constant returns to capital and labour: with a given technology,
doubling the inputs doubles the amount that can be produced. The production of new ideas
depends on the quantities of capital and labour engaged in research and on the level of
technology. Given the assumption of generalized Cobb–Douglas production, we therefore
write
𝐾(𝑡) = 𝑠 𝑌(𝑡)
Finally, the initial levels of A, K, and L are given and strictly positive. This completes
the description of the model. Because the model has two state variables whose behaviour is
endogenous, K and A, it is more complicated to analyze than the Solow model. We therefore
begin by considering the model without capital; that is, we set α and β to zero. This case
shows most of the model‘s central messages. We then turn to the general case.
1
𝑌 𝑡 = 𝐴 𝑡 1 − 𝑎𝐿 𝐿 𝑡
Y(t ) = A(t )(1 − aL)L(t )
Output per worker is proportional to A, and thus that the growth rate of output per
worker equals the growth rate of A. We therefore focus on the dynamics of A, which are
given by the above equation. This equation implies that the growth rate of A, denoted gA, is
gA(t ) ≡ 𝐴(𝑡)
𝐴(𝑡 )
𝐵[𝑎𝐿 𝐿(𝑡)]𝛾 𝐴(𝑡)𝜃
=
𝐴(𝑡)
𝛾
= B𝑎𝐿 𝐿(𝑡)𝛾 𝐴(𝑡)𝜃−1
𝑔𝐴 (𝑡)
= 𝛾𝑛 + 𝜃 − 1 𝑔𝐴 (𝑡)
𝑔𝐴 𝑡
Multiplying both sides of this expression by gA(t ) yields
𝑔 𝐴 (𝑡)
. 𝑔𝐴 (t) = 𝛾𝑛𝑔𝐴 (𝑡) + (𝜗 − 1)𝑔𝐴 𝑡 . 𝑔𝐴 (𝑡)
𝑔𝐴 𝑡
The initial values of L and A and the parameters of the model determine the initial
value of gA. The above equation determines the subsequent behaviour of gA. To describe
further how the growth rate of A behaves (and thus to characterize the behaviour of output per
worker), let us distinguish among the cases θ <1, θ >1, and θ = 1.
Case 1: θ <1
The following figure shows the phase diagram for gA when θ is less than 1. That is, it
plots 𝑔𝐴 as a function of A. The production function for knowledge implies that gA is always
positive, the diagram considers only positive values of gA. The diagram shows that for the
case of θ less than 1, 𝑔𝐴 is positive for small positive values of gA and negative for large
values. Let 𝑔𝐴 denote the unique positive value of gA that implies that 𝑔𝐴 is zero. Now, 𝑔𝐴∗ is
defined by γ n + (θ− 1)𝑔𝐴∗ = 0. Solving this for 𝑔𝐴∗ yields
𝛾
𝑔𝐴∗ = 1− 𝜃 n.
𝑔𝐴
0 𝑔𝐴∗ 𝑔𝐴
This analysis implies that regardless of the economy‘s initial conditions, gA converges
to 𝑔𝐴∗ . If the parameter values and the initial values of L and A imply gA (0) <𝑔𝐴∗ , for example,
𝑔𝐴∗ is positive; that is, gA is rising. It continues to rise until it reaches 𝑔𝐴∗ . Similarly, if gA(0)
>𝑔𝐴∗ , then gA falls until it reaches 𝑔𝐴∗ . Once gA reaches 𝑔𝐴∗ , both A and Y/L grow steadily at
rate 𝑔𝐴∗ .Thus the economy is on a balanced growth path.
This model is the first example of a model of endogenous growth. In this model, in
contrast to the Solow, the long-run growth rate of output per worker is determined within the
model rather than by an exogenous rate of technological progress.
The model implies that the long-run growth rate of output per worker,𝑔𝐴∗ , is an
increasing function of the rate of population growth, n. Indeed, positive population growth is
necessary for sustained growth of output per worker.
The fraction of the labour force engaged in R&D does not affect long-run growth.
This too may seem surprising: since growth is driven by technological progress and
technological progress is endogenous, it is natural to expect an increase in the fraction of the
economy‘s resources devoted to technological progress to increase long-run growth. To see
why it does not, suppose there is a permanent increase in aL starting from a situation where A
is growing at rate 𝑔𝐴∗ . aL does not enter expression for 𝑔𝐴 : 𝑔𝐴 (𝑡)= 𝛾𝑛𝑔𝐴 𝑡 +(θ − 1)[𝑔𝐴 (𝑡)]2 .
𝛾
aL does enter expression for gA: gA(t ) =B𝑎𝐿 L(t )γA(t )θ−1. The increase in aL therefore causes
an immediate increase in gA but no change in 𝑔𝐴 as a function of gA.
The increase in the growth rate of knowledge is not sustained. When gA is above 𝑔𝐴∗ ,
𝑔𝐴∗ is negative. gA therefore returns gradually to 𝑔𝐴∗ and then remains there. In other words, θ
less than 1 means that the contribution of additional knowledge to the production of new
knowledge is not strong enough to be self-sustaining.
Case 2: θ >1
The second case to consider is θ greater than 1. This corresponds to the case where
the production of new knowledge rises more than proportionally with the existing stock. We
know that 𝑔𝐴 = 𝛾𝑛𝑔𝐴 + (𝜃 − 1)𝑔𝐴2 .When θ exceeds 1, this equation implies that 𝑔𝐴 is
positive for all possible values of gA. Further, it implies that 𝑔𝐴 is increasing in gA (since gA
must be positive). The phase diagram is shown below.
𝑔𝐴
0 gA
The implications of this case for long-run growth are very different from those of the
previous case. As the phase diagram shows, the economy exhibits ever-increasing growth
rather than convergence to a balanced growth path. In this case, knowledge is so useful in the
production of new knowledge that each marginal increase in its level results in so much more
new knowledge that the growth rate of knowledge rises rather than falls. Thus once the
accumulation of knowledge begins—which it necessarily does in the model—the economy
embarks on a path of ever-increasing growth.
The impact of an increase in the fraction of the labour force engaged in R&D is now
dramatic. An increase in aL causes an immediate increase in gA. But 𝑔𝐴 is an increasing
function of gA; thus 𝑔𝐴 rises as well. And the more rapidly gA rises, the more rapidly its
growth rate rises. Thus the increase in aL causes the growth rate of A to exceed what it would
have been otherwise by an ever-increasing amount.
Case 3: θ = 1
When θ is exactly equal to 1, existing knowledge is just productive enough in
generating new knowledge that the production of new knowledge is proportional to the stock.
In this case, expressions for gA and 𝑔𝐴 are
𝛾
𝑔𝐴 𝑡 = 𝐵𝑎𝐿 𝐿(𝑡)𝛾
𝑔𝐴 (𝑡) = 𝛾𝑛𝑔𝐴 (𝑡)
If population growth is positive, gA is growing over time. In this case the dynamics of
the model are similar to those when θ >1. If population growth is zero, on the other hand, gA
is constant regardless of the initial situation. Thus there is no adjustment toward a balanced
growth path: no matter where it begins, the economy immediately exhibits steady growth.
The growth rates of knowledge, output, and output per worker are all equal to
𝛾
𝐵𝑎𝐿 𝐿𝛾 in this case. Thus changes in aL affect the long-run growth rate of the economy.
Since the output good in this economy has no use other than in consumption, it is
natural to think of it as being entirely consumed. Thus 1 − aL is the fraction of society‘s
resources devoted to producing goods for current consumption, and aL is the fraction devoted
to producing a good (namely, knowledge) that is useful for producing output in the future.
Thus one can think of aL as a measure of the saving rate in this economy.
With this interpretation, the case of θ =1 and n = 0 provides a simple example of a
model where the saving rate affects long-run growth.
𝑌 𝑡 = [ 1 − 𝑎 𝑘 𝐾 𝑡 ]𝛼 𝐴 𝑡 1 − 𝑎 𝐿 𝐿 𝑡 1− 𝛼
, 0<α<1
Now, let us substitute the production function equation into the equation
{𝐾(𝑡) = sY(t)}, we get
𝛼
Dividing both sides by K(t ) and defining cK= 𝑠 1 − 𝑎𝐾 ( 1 − 𝑎𝐿 )1− 𝛼 gives us
𝐾(𝑡)
𝑔𝐾 𝑡 =
𝐾(𝑡)
1− 𝛼 𝛼 1− 𝛼 1− 𝛼
𝑠 1− 𝑎𝐾 𝛼 ( 1− 𝑎𝐿 ) 𝐾(𝑡) 𝐴(𝑡) 𝐿(𝑡)
=
𝐾(𝑡)
𝐾 𝑡 𝛼 𝐴 𝑡 1−𝛼 𝐿 𝑡 1−𝛼
= 𝐶𝑘 [ ]
𝐾 𝑡
𝐴 𝑡 1−𝛼 𝐿 𝑡 1−𝛼
= 𝐶𝐾 [ ]
𝐾(𝑡)1 𝐾(𝑡)−𝛼
𝐴 𝑡 1−𝛼 𝐿 𝑡 1−𝛼
= 𝐶𝐾 [ ]
𝐾(𝑡)1−𝛼
𝐴 𝑡 𝐿 𝑡 1− 𝛼
= 𝐶𝐾 [ ]
𝐾 𝑡
𝑔𝐾 (𝑡)
= 1 − 𝛼 [𝑔𝐴 𝑡 + 𝑛 − 𝑔𝐾 𝑡 ]
𝑔𝐾 𝑡
gK 𝑔𝐾 = 0
(𝑔𝐾 < 0)
𝑔𝐾 > 0
n
0
gA
Similarly, dividing both sides of equation 𝐴= B(𝑎𝐾 𝐾)𝛽 (𝑎𝐿 𝐿)𝛾 𝐴𝜃 , by A yields an
expression for the growth rate of A:
𝐴(𝑡) 𝐵[𝑎𝐾 𝐾 𝑡 ]𝛽 [𝑎𝐿 𝐿(𝑡)]𝛾 𝐴(𝑡)𝜃
=
𝐴 𝑡 𝐴(𝑡)
𝛽 𝛾
= 𝐵𝑎𝐾 𝐾(𝑡)𝛽 𝑎𝐿 𝐿(𝑡)𝛾 𝐴(𝑡)𝜃−1
𝛽 𝛾
= 𝐵𝑎𝐾 𝑎𝐿 𝐾(𝑡)𝛽 𝐿(𝑡)𝛾 𝐴(𝑡)𝜃−1
𝑔𝐴 𝑡 = 𝐶𝐴 𝐾(𝑡)𝛽 𝐿(𝑡)𝛾 𝐴(𝑡)𝜃−1
𝛽 𝛾
Where cA≡ B𝑎𝐾 𝑎𝐿 . Taking logs and differentiating with respect to time gives
𝛽 𝛾 𝜃−1
In 𝑔𝐴 𝑡 = 𝐼𝑛[𝐶𝐴 𝐾 𝑡 𝐿 𝑡 𝐴 𝑡 ]
𝑔𝐴 𝑡
= 𝐼𝑛 𝐶𝐴 + 𝛽𝐼𝑛 𝐾 𝑡 + 𝛾𝐼𝑛𝐿 𝑡 + 𝜃 − 1 𝐼𝑛 𝐴 𝑡
𝑔𝐴 𝑡
𝑔𝐴 𝑡
= 0 + 𝛽𝑔𝐾 𝑡 + 𝛾𝑛 + (𝜃 − 1)𝑔𝐴 (t)
𝑔𝐴 𝑡
𝑔𝐴 (𝑡)
= 𝛽𝑔𝐾 𝑡 + 𝛾𝑛 + 𝜃 − 1 𝑔𝐴 (𝑡)
𝑔𝐴 𝑡
gK 𝑔𝐴 = 0
0
gA
𝛾𝑛
−
𝛽
The production function for output exhibits constant returns to scale in the two
produced factors of production, capital and knowledge. Thus whether there are on net
increasing, decreasing, or constant returns to scale to the produced factors depends on their
returns to scale in the production function for knowledge. As that equation shows, the degree
of returns to scale to K and Ain knowledge production is β + θ: increasing both K and A by a
factor of X increases A by a factor of Xβ+θ. Thus the key determinant of the economy‘s
behaviour is now not how θ compares with 1, but how β + θ compares with 1. We will limit
our attention to the cases of β + θ <1 and of β + θ =1 with n = 0.
The remaining cases (β + θ >1 and β + θ =1 with n >0) have implications similar to
those of θ >1 in the simple model;
Case 1: β +θ <1
If β + θ is less than 1, (1 − θ)/β is greater than 1. Thus the locus of points where 𝑔𝐴 = 0
is steeper than the locus where 𝑔𝐾 = 0. The initial values of gA and gK are determined by the
parameters of the model and by the initial values of A, K, and L. Their dynamics are then as
shown in the figure.
𝑔𝐴 =0
gK 𝑔𝐾 =0
E
𝑔𝐾∗
0
𝑔𝐴∗ gA
𝛾𝑛
-𝛽
The figure shows that regardless of where gA and gK begin, they converge to Point E in
the diagram. Both 𝑔𝐴 and 𝑔𝐾 are zero at this point. Thus the values of gA and gK at Point E,
which we denote g* A and g*K , must satisfy
𝑔𝐴∗ + 𝑛 − 𝑔𝐾∗ = 0
and
𝛽𝑔𝐾∗ + 𝛾𝑛 = 𝜃 − 1 𝑔𝐴∗ = 0
Or, 𝛽𝑔𝐴∗ + 𝛽 + 𝛾 𝑛 + 𝜃 = 1 𝑔𝐴∗ = 0
𝛽 +𝛾
Or, 𝑔𝐴∗ = 𝑛
1− 𝜃+𝛽
From above, 𝑔𝐾∗ is simply 𝑔𝐴∗ + n, and when A and K are growing at these rates, output
is growing at rate 𝑔𝐾∗ . Output perworker is therefore growing at rate 𝑔𝐴∗ . This case is similar
to the case when θ is less than 1 in the version of the model without capital. Here, as in that
case, the long-run growth rate of the economy is endogenous, and again long-run growth is
an increasing function of population growth and is zero if population growth is zero. The
fractions of the labour force and the capital stock engaged in R&D, aL and aK, do not affect
long-run growth; nor does the saving rate, s. The reason that these parameters do not affect
long-run growth is essentially the same as the reason that aL does not affect long-run growth
in the simple version of the model.
Models like this one and like the model without capital in the case of θ <1are often
referred to as semi-endogenous growth models. On the one hand, long-run growth arises
endogenously in the model. On the other, it depends only on population growth and
parameters of the knowledge production function, and is unaffected by any other parameters
of the model. Thus, growth seems only somewhat endogenous.
Case 2: β +θ = 1 and n = 0
The locus of points where 𝑔𝐾 = 0 is given by gK= gA+n, and that the locus of points
where 𝑔𝐴 = 0 is given by gK= −(γn/β) +[(1−θ)/β]gA. When β+θ is 1 and n is 0, both
expressions simplify to gK= gA. That is, in this case the two loci lie directly on top of each
other: both are given by the 45-degree line. The dynamics of the economy in this case.
gK
𝑔𝐾 = 𝑔𝐴 = 0
450
gA
As the figure shows, regardless of where the economy begins, the dynamics of gA and
gK carry them to the 45-degree line. Once that happens, gA and gK are constant, and the
economy is on a balanced growth path. As in the case of θ =1 and n = 0 in the model without
capital, the phase diagram does not tell us what balanced growth path the economy converges
to. However, one can show that the economy has a unique balanced growth path for a given
set of parameter values, and that the economy‘s growth rate on that path is a complicated
function of the parameters. Increases in the saving rate and in the size of the population
increase this long-run growth rate Because long-run growth depends on a wide range of
parameters, models like this one, as well as the model of the previous section when θ ≥1 and
the model of this section when β + θ >1 or β + θ =1 and n >0, are known as fully endogenous
growth models.