Chapter Three: Endogenous Economic growth

3.1 Introduction

• Solow model does not provide satisfying answers to our central questions about
economic growth.

• The models’ principal result is a capital accumulation does not account for a large
part of either long-run growth or cross-country income differences.

• The only determinant of income in the models other than capital is a mystery
variable, the ―effectiveness of labor‖ (A), whose exact meaning is not specified
and whose behavior is taken as exogenous.

• This model treat capital accumulation and its role in production in ways that are
similar to Solow model but they differ from the earlier models in explicitly
interpreting the effectiveness of labor as knowledge and in modeling the
determinants of its evolution over time.
• Incorporating endogenous technology into growth theory forces as to

work on increasing returns and the associate market structure - imperfect

competition.

• The conclusions derived from endogenous growth model are mixed.

• Models of knowledge accumulation provide a plausible and appealing

explanation of worldwide growth but, they provide little help in
understanding cross-country income differences.
 3.2. Frame work and Assumptions
 The model is a simplified version of the models of R&D and growth developed by P.
Romer (1990), Grossman and Helpman (1991a), and Aghion and Howitt (1993).

 There are two sectors:

o Goods producing sector - where output is produced.

o R &D/Ideas sector - where additions to the stock of knowledge are made.

 Therefore, the economy produces two things: consumption goods and ideas.

 We make two major simplifications:

o First, both the R&D and goods production functions are assumed to be generalized Cobb–
Douglas functions..

o Second, the model takes the fraction of output saved and the fractions of the labor force
and the capital stock used in the R&D sector as exogenous and constant.
Specific

 The model involves four variables: labor (L), capital (K ), technology

(A), and output (Y ).

 Labor and capital are split between the two sectors .Fraction 𝛼𝐿 of the
labor force is used in the R&D sector and 1- 𝛼𝐿 in the goods-producing
sector. Similarly, fraction 𝛼𝑘 of the capital stock is used in R&D and the
rest in goods production.

 Both 𝛼𝐿 and 𝛼𝑘 are exogenous and constant. Because the use of an idea
or a piece of knowledge in one place does not prevent it from being used
elsewhere, both sectors use the full stock of knowledge, A.

 The model is set in continuous time.

 New technology is produced every period, but it never depreciates, so it

accumulates like capital.
 Production Function Goods Producing Sector
 Production function of the goods-producing sector is a CRTS Cobb-Douglas,
 The quantity of output produced at time t is thus

Y(t ) = [(1 − 𝛼𝑘 )K(t )]α [A(t )(1 − 𝛼𝐿 )L(t )]1−α , 0< α < 1……………(3.1)
 Note that equation (3.1) implies constant returns to capital and labor: with a
given technology.
Production Function for R and D sector.
 Production function of the R&D sector is a generalized Cobb-Douglas PF but the
sum of their exponents on the inputs is not necessarily restricted to 1.
 The production of new ideas depends on the quantities of capital and labor
engaged in research and on the level of technology.
 Given our assumption of generalized Cobb–Douglas production, we therefore
write
A(t ) = B[𝛼𝑘 K(t )]𝛽 [𝛼𝐿 L(t )]𝛾 A(t )𝜃 , B > 0, 𝛽 ≥ 0, γ ≥ 0……….……(3.2)
where B is a shift parameter.
 Production function is not necessarily restricted to CRS because
o Production of new ideas (old ideas do not have to be replaced). Thus, this
function allows for decreasing, constant or increasing returns to scale .
o The value can be both positive or negative as it reflects the effect of existing
knowledge stock on the success of R&D.
 The parameter θ reflects the effect of the existing stock of knowledge on the
success of R&D.
o This effect can operate in either direction. On the one hand, past discoveries
may provide ideas and tools that make future discoveries easier. In this case, θ
is positive.
o On the other hand, the easiest discoveries may be made first. In this case, it is
harder to make new discoveries when the stock of knowledge is greater, and so
θ is negative.
o Because of these conflicting effects, no restriction is placed on θ in (3.2).
 As in the Solow model, the saving rate is exogenous and constant.
 In addition, depreciation is set to zero for simplicity. Thus,
𝐾 (t)=sY(t)…………………………………………..(3.3)
 The labour force is assumed to grow at rate n .
𝐿(t) = nL(t), n> 0…………………………………(3.4)
 Because the model has two state variables whose behavior 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.
 Finally, as in our earlier models, the initial levels of A, K, and L are given
and strictly positive. This completes the description of the model.
 3.3. The model without capital
The Dynamics of Knowledge Accumulation
 When there is no capital (α and β to zero) in the model, the production
function for output (equation [3.1]) becomes
Y(t ) = A(t)(1 − 𝛼𝐿 )L(t)…………………………………….(3.5)
 Similarly, the production function for new knowledge (equation [3.3]) is now
A(t) = B[𝛼𝐿 L(t )]𝛾 )(A(t)𝜃 ………………………………….(3.6)
 Given constant population growth (Equation (3.4), Equation (3.5)) implies
that output per worker is proportional to A, and thus that the growth rate of
output per worker equals the growth rate of A.
𝐴(𝑡)
𝑔𝐴 (t ) = ……………………………………………….(3.7)
𝐴(𝑡)

= B[𝛼𝐿 L(t )]𝛾 (A(t )𝜃−1

 Taking logs of both sides and differentiate w.r.t. to time gives us the growth rate of
A(𝑔𝐴 (t))= (the growth in technological progress). eq3.8.jpg
𝑔𝐴 (𝑡)
=𝛾n + (𝜃− 1) 𝑔𝐴 (𝑡)…………………………………………(3.8)
𝑔𝐴 (𝑡)

 Multiplying both sides of this expression by 𝑔𝐴 (𝑡) yields

𝑔𝐴 (𝑡)=𝛾n𝑔𝐴 (𝑡) + (𝜃− 1) (𝑔𝐴 (𝑡))2 ……………………………………(3.9)
 The initial values of A and L determines the initial value of 𝑔𝐴 (t) , which determines
the value 𝑔𝐴 (t) but all depends on θ.
 The behavior of A (how does the growth rate of A behaves?)
 There are three cases :
Case 1:𝜽 < 𝟏
 If 𝜽 < 1, then 𝜽 < 0 and each increase in A(t ) lowers the growth rate of A(t), other factor inputs
held constant.
 In this case, technology exhibits a kind of diminishing returns ( technology is not self-sustaining‖)
with respect to its own production that is similar to that of capital in the Solow model.
 Past discoveries make future discoveries more costly in terms of resources Thus, 𝜽 < 𝟏, 𝑔𝐴 , is
positive for small values of 𝑔𝐴 and negative for large values. See Figure 3.1.
  We will use 𝑔𝐴 ∗ to denote the unique
positive value of 𝑔𝐴 that implies that
𝑔𝐴 is zero.
 From (3.9), 𝑔𝐴 ∗ is defined by 𝛾n + (𝜃−
1) 𝑔𝐴 ∗ =0, solving for 𝑔𝐴 ∗ yields
𝛾
𝑔𝐴 ∗ = 1−𝜃n…………………(3.10)

 Regardless of the economy's initial

conditions 𝑔𝐴 converges to 𝑔𝐴 ∗ . .

 For example, if the parameter values and the initial value of L and A imply
𝑔𝐴 (0)<𝑔𝐴 ∗ , 𝑔𝐴 is positive; i.e. 𝑔𝐴 is rising].
 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
 Note the following properties:
o The steady-state growth rate(𝑔𝐴 ∗ ) of output per worker is proportional to
population growth(n). (see Eq. 3.10): technological progress requires sustained
population growth and countries with faster population growth grow faster
(weakness: this property does not seem to be confirmed by actual data);this
seems true at larger scales, but not at smaller ones (i.e. larger population larger
chance of innovation).
o Technological growth at the steady-state growth rate does not depend on the
fraction of labor(𝛼𝐿 ) employed in the production of knowledge (R&D). A
change in 𝛼𝐿 has a temporary level effect (see Equ.3.7 and 3.6) but does not
affect the steady-state rate of growth (see Equ.3.9 & 3.10).
 Therefore, an increase in the share of the population working in R&D can
produce a level effect on technology and output but, no growth effect.
 This analysis implies that, paralleling the impact of a rise in the saving rate on
the path of output in the Solow model.
 In contrast to Solow, Diamond models, this model displays endogenous
growth: the steady-state growth of output per worked 𝑔𝐴 is determined
within the model rather than by an exogenous rate of technological
progress.
 Finally, capital-generated growth in the Solow model was limited by the
fact that capital faced diminishing returns in reproducing itself.
Eventually, the economy settled into a steady state in which capital
could no longer grow relative to other factors.
 The same thing can happen to technology in the R&D model when θ <
1: Eventually, technology-induced growth is limited and, without growth
in the non-produced factor (labor), the economy becomes stationary
(zero growth)
Case 2: 𝜽 > 𝟏
 If θ > 1, the rate of technological progress may grow explosively.
 Each discovery opens up multiplying new opportunities so that future discoveries
become less costly to find. Production of knowledge is self-reinforcing: each new
discovery encourages further technological progress. If n = 0, there is no point to the
right of the origin at which the curve intersects the horizontal axis, so there is no
nonzero steady-state rate of technical progress.
 Thus, when 𝜽 > 𝟏 , 𝑔𝐴 is positive for all values of 𝑔𝐴 , implying that the economy is
characterized by ever-increasing growth rather than convergence.

Figure 3.2 The dynamics of the growth rate of knowledge when θ >1
 The impact of an increase in the share of the labor force employed in the
R&D sector is substantial through its impact on 𝑔𝐴 which in turn affect 𝑔𝐴 as
shown in the figure .
 In this case, there is no steady-state growth rate of technology because except
for 𝑔𝐴 = 0, 𝑔𝐴 is always positive .
 Growth accelerates over time and the economy never converges to a balanced
growth path and then overall economy never reaches a steady-state.
 Since production of knowledge is self-reinforcing( each new discovery
encourages further technological progress), any increase in 𝛼𝐿 increases 𝑔𝐴
permanently (𝑔𝐴 (t ) = B[𝛼𝐿 L(t )]𝛾 (A(t )𝜃−1 and further accelerates 𝑔𝐴 .
 Thus changes in 𝛼𝐿 affect the long-run growth rate of the economy.
 This is implausible, but shouldn't be completely dismissed.
 This is one type of fully endogenous model of growth.
 Case 3: 𝜽 = 1
 Production of knowledge is proportional to its stock i.e. existing knowledge is just
productive enough in generating new knowledge i.e. technology is neither enhanced nor
retarded by the pre-existing level of technology.
 In this case, expressions (3.7) and (3.9) for 𝑔𝐴 and , 𝑔𝐴 simplify to
𝑔𝐴 (t) == B[𝛼𝐿 L(t )]𝛾 ) …………………………………………………(3.11)
 Growth of ideas depends on the proportion of the population working in research and
development.
𝑔𝐴 (𝑡) =𝛾𝑛(𝑡)𝑔𝐴 (t) …………………………………………………(3.12)
 If population growth is positive (n> 0), 𝑔𝐴 is growing over time; in this case the
dynamics of the model are similar to those when 𝜽 >1.
 If population growth zero (n=0) and (𝛾=0) , 𝑔𝐴 is constant regardless of the initial
situation, in this case the economy is always on a balanced growth path: 𝑔𝐴 = 0.
 Therefore, growth of ideas is accelerating when population growth is positive, and growth
of ideas stops when population growth stops~steady state.
• These models are simple and plausible .
The Importance of Returns to Scale to Produced Factors:
 The reason that these three cases have such different implication is that
whether 𝜃 is less than, greater than or equal to 1, it determines whether
there are decreasing ,increasing, or constant returns to scale to produced
factors of production.
 The growth of labor is exogenous, and we have eliminated capital from
the model; thus knowledge is the only produced factor.
 Therefore whether there are on the whole increasing, decreasing, or
constant returns to knowledge in this economy is determined by the
returns to scale to Knowledge in knowledge production-that is , by θ.
 The Importance of Population Growth
 To understand importance of population, consider equation (3.7) for knowledge
accumulation:
𝑔𝐴 (t )= B[𝛼𝐿 L(t )]𝛾 (A(t ))𝜃−1 .
 Built into this expression: when there are more people to make discoveries,
more discoveries are made.
 And when more discoveries are made, the stock of knowledge grows faster, and
so (all else equal) output per person grows faster.
o In the particular case of θ = 1 and n = 0, this effect operates in a special way: long-run
growth is increasing in the level of population. Therefore, long run growth an increasing
function of level of population.
o When θ > 1, the effect is even more powerful, as increases in the level or growth rate of
population lead to ever-rising increases in growth. Therefore, long run growth is an
increasing function of level of population or population growth(n).
 when θ > 1 (or θ = 1 and n > 0), one can show that an increase in population
growth causes income per person to be higher than it otherwise would have
been by an ever increasing amount.
 When θ < 1, there are decreasing returns to scale to produced factors. In this
case, although knowledge may be helpful in generating new knowledge, the
generation of new knowledge rises less than proportionally with the existing
stock. This model has the surprising implication that positive population
growth is necessary for long-run growth in income per person, and that the
economy’s long-run growth rate is increasing in population growth.
 Note: when there are more people to make discoveries, more discoveries are
made and stock of knowledge grows faster, and so (all else equal) output per
person grows faster.
 A natural interpretation of the model, A represents knowledge that can be
used anywhere in the world. With this interpretation, the model does not
imply that countries with larger populations, or countries with greater
population growth, enjoy greater income growth; it only implies that higher
worldwide population.
 This implication is plausible: because people are an essential input into
producing knowledge, it makes sense that, at least up to the point where
resource limitations (which are omitted from the model) become important,
higher population growth is beneficial to the growth of worldwide
knowledge.
3.4. The General Case: The model with capital
The dynamics of knowledge (A) and capital (K)
A: Capital
 Now there are 2 endogenous state variables, A &K in the model.
 Substituting the production function [Eqn 3.1] into the expression for capital growth
[Eq 3.3] yields:
 Ch3 E3.14 &3.15.jpg
𝐾 (t ) = s (1 − 𝛼𝑘 )α (1 − 𝛼𝐿 )1−α K(t )α A(t )1−α L(t )1−α . ………(3.13)
 Dividing both sides by K(t ) and defining ck =s(1 −𝛼 )α (1 −𝛼 )1−α gives us
𝑘 𝐿
𝐾 (𝑡)
𝑔𝐾 (t ) = 𝐾(𝑡)
………………….(3.14)
1−α
(A(t )𝐿(𝑡)
= ck
𝐾(𝑡)

 Taking logs of both sides and differentiating with respect to time yields:
𝑔𝑘 (𝑡)
= (1 − a) 𝑔𝐴 𝑡 + n −𝑔𝑘 (𝑡) …………………………………(3.15)
𝑔𝑘(𝑡)
 From Eq.(3.13) the growth rate of capital, 𝑔𝑘, is always positive and thus,
• Rising if 𝑔𝐴 𝑡 +n-𝑔𝑘 (𝑡) > 0
• Falling if 𝑔𝐴 𝑡 +n-𝑔𝑘 (𝑡) < 0
• Zero if 𝑔𝑘 (𝑡) = 𝑛 + 𝑔𝐴 𝑡

 In (𝑔𝐴 , 𝑔𝑘 ) space, the locus of

points where 𝑔𝑘 is constant
with intercept n and slope 1.
 Above the locus, 𝑔𝑘 is falling;
below the locus, it is rising.

Figure 3.3. The dynamics of the growth rate of capital in general version of the model
B. Knowledge
 Similarly, dividing both sides of equation (3.2),
𝛽 𝛾
𝐴 t = B 𝛼𝑘 𝐾(𝑡) 𝛼𝐿 𝐿 𝑡 A(t)𝜃 𝑏𝑦 A yields an expression for the growth rate of A (
knowledge) and it is expressed by
𝐴(𝑡)
𝑔𝐴 (𝑡)= 𝐴(𝑡) = B 𝛼𝑘 𝐾(𝑡) 𝛽
𝛼𝐿 𝐿 𝑡 𝛾
A(t) 𝜃−1 …….(3.16)

𝑔𝐴 (𝑡)= CA 𝐾(𝑡) 𝛽 𝐿 𝑡 𝛾 A(t) 𝜃−1

Where ,CA ≡ B𝛼𝑘 β 𝛼𝐿 γ
 Taking logs and differentiating with respect to time gives, growth rate of the knowledge
accumulation as
𝑔𝐴 (𝑡)
= β𝑔𝑘 𝑡 + 𝛾𝑛 + (𝜃 − 1) 𝑔𝐴 𝑡 … (3.17)
𝑔𝐴 (𝑡)

 Implying that the growth rate of capital when knowledge grows at constant rate is
1−𝜃 𝛾𝑛
𝑔𝐾 ∗ =[ 𝛽
]𝑔𝐴 ∗ - 𝛽
 𝑔 (𝑡)
From 𝑔𝐴(𝑡) (𝑡) = β𝑔𝑘 𝑡 + 𝛾𝑛 + (𝜃 − 1) 𝑔𝐴 𝑡 ( Eq 17)
𝐴

 𝑔𝐴 is rising if β𝑔𝑘 𝑡 + 𝛾𝑛 + (𝜃 − 1) 𝑔𝐴 is positive,

 falling if β𝑔𝑘 𝑡 + 𝛾𝑛 + (𝜃 − 1) 𝑔𝐴 is negative
 and constant if β𝑔𝑘 𝑡 + 𝛾𝑛 + (𝜃 − 1) 𝑔𝐴 is zero
 This is shown in the next figure drown for case 𝜃 < 1
 The set of points where 𝑔𝐴 is constant has an intercept of −γn/β and a slope of
(1− 𝜃)/β. Above this locus, 𝑔𝐴 is rising; and below the locus, it is falling.

Figure 3.3. The dynamics of the growth rate of knowledge in general version of the model
 As before, the model’s implications depend on the returns to scale in the produced
factors of production (capital and knowledge) in the production of knowledge.
 The production function has CRTS; the production of knowledge displays IRTS if β + 𝜃
> 1, DRTS if β + 𝜃 < 1, and CRTS if β + 𝜃 = 1.
 The key determinants of the economy's behavior is now not how θ compares with 1 but
how 𝛽 +θ compares with 1.
Case 1: 𝜷 + 𝜽 < 𝟏

(1− 𝜃)
 If β + 𝜃 < 1, β
is greater than 1 (rearranging case1) -the locus of points where

𝑔𝐴 =0 [the previous figure] is steeper than the locus where 𝑔𝐾 =0 [the one before the last
one]: see in the diagram below.
 The equilibrium is given by the intersection of the two lines: Thus the values of 𝑔𝐴 and
𝑔𝐾 at Point E, which we denote 𝑔𝐴 ∗ and 𝑔𝐾 ∗ , must satisfy
𝑔𝐴 ∗ +n- 𝑔𝐾 ∗ = 0→ 𝑔𝐾 ∗ = 𝑔𝐴 ∗ +n…………………(3.18)
𝛽+𝛾
β𝑔𝑘 ∗ + 𝛾𝑛 + (𝜃 − 1) 𝑔𝐴 ∗ =0→ 𝑔𝐴 ∗ = n…(3.19)
1−(𝜃+𝛽)
 Figure 3.4 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.

Figure 3.4. The dynamics of the growth rates of capital and knowledge when β + 𝜃 < 1
 The economy converges to a balanced-growth path, regardless of the initial
 conditions. In the steady state, both K and A grow at constant rates.
 The model again displays endogenous growth; the steady-state growth rate of
output per worker is proportional to population growth i.e. long run growth is
an increasing function of population growth [from 3.19] .
 The fractions of the labor force (𝛼𝐿 ) and capital stock devoted to R&D (𝛼𝑘 or
s ) hove no long-run effect on growth rates, but a level effect [are not in 3.19
but are there on 3.16].
 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 reason that these parameters do not affect long-run growth is essentially the
same as the reason that 𝛼𝐿 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 <1

are 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, as the name implies, growth seems only somewhat endogenous.

Case 2: The dynamics of the growth rates of knowledge and capital when
𝜷 + 𝜽 > 𝟏 and n= 𝟎
 There is no steady state because the 𝑔𝐴 = 0 and 𝑔𝐾 = 0 do not intersect:

1− 𝜃
𝑔𝐴 = 0 𝑙𝑖𝑛𝑒 𝑖𝑠 𝑓𝑙𝑎𝑡𝑡𝑒𝑟 𝑡𝑕𝑎𝑛 < 1 t𝑕𝑎𝑛 𝑔𝐾 =0
β
Case 3: The dynamics of the growth rates of knowledge and capital when 𝜷 + 𝜽 = 𝟏
 In this case (1-θ)/β equals 1 and hence the loci 𝑔𝐴 and 𝑔𝐾 have the same slope
(see next figure)
 In n is positive the 𝑔𝐾 =0 line lies above 𝑔𝐴 =0 line (shown in panel a figure)
 When n is 0 on the other hand the two loci lie on to of each other (shown in panel
b of the figure next)
 The figure (panel b) shows regardless of where the economy begins it converges
to the balanced growth path.
 As in the model with θ =1 and n=0 in the model without capital, panel b doesn't
tell us what balanced growth path the economy converges to.
 Irrespective of initial conditions, the economy always converges to the steady
state. Once the steady state, 𝑔𝑘 (𝑡) = 𝑔𝐴 (𝑡) and the economy grows at a constant
rate.
 Case 3 (a): 𝜷 + 𝜽 = 𝟏; n> 𝟎
 There is convergence to steady state growth; however, the growth rate is not unique.
3.5. The Nature of Knowledge and the Allocation of Resources to R&D [Romer 1990
Model]
A. The Nature of Knowledge
 So far we have simply described the ―A‖ variable produced by R&D as knowledge
but knowledge comes in many forms, from highly abstract scientific results to highly
applied every-day solutions found by each and every individual.
 Any of these different types of knowledge play important roles in economic growth.
 There is no reason to expect the determinants of the accumulation of these different
types of knowledge to be the same. There is thus no reason to expect a unified theory
of the growth of knowledge.
 Rather, we should expect to find various factors underlying the accumulation of
knowledge.
Non-rivalry
 At the same time, all types of knowledge share one essential feature: they are no rival:
it can be used by others without diminishing the amount of knowledge available to the
original user or inventor.
 Because it is non-rival, knowledge cannot be efficiently produced under
competitive markets: the marginal cost of replicating past inventions is close to
zero.
 Patents and protection of intellectual property rights make knowledge excludable,
i.e. others may be prevented from using it.
 Excludability creates private incentives for R&D.
o If knowledge is non-excludable, private incentives for its production are
essentially non-existent.
o Because of the externalities inherent in the production of knowledge, there may
be strong arguments for public funding of R&D.
o When knowledge is excludable, the producers of new knowledge can license the
right to use the knowledge at positive prices, and hence hope to earn positive
returns on their R&D efforts.
 The major forces governing the allocation of resources to the development

of knowledge are four forces:

1. support for basic scientific research,

2. private incentives for R&D and innovation,

3. alternative opportunities for talented individuals, and

4. learning-by-doing.
1. Support for Basic Scientific Research
 Basic scientific knowledge are available relatively freely; this research is
not motivated by the desire to earn private returns in the market.
 The economics of this type of knowledge are relatively straightforward.
Since it is useful in production and is given away at zero cost, it has a
positive externality. Thus its production should be subsidized.
2. Private Incentives for R&D and Innovation

 Many innovations receive little or no external support and are motivated almost

entirely by the desire for private gain.

 As described above, for R&D to result from economic incentives, the

knowledge that is created must be at least somewhat excludable. Thus the

developer of a new idea has some degree of market power.

 Typically, the developer is modeled as having exclusive control over the use of

the idea and as licensing its use to the producers of final goods.

 There are in fact three distinct externalities from R&D: the consumer-surplus

effect, the business-stealing effect, and the R&D effect.

 The consumer-surplus effect is that the individuals or firms licensing ideas
from innovators obtain some surplus, since innovators cannot engage in
perfect price discrimination. Thus this is a positive externality from R&D.
 The business-stealing effect is that the introduction of a superior technology
typically makes existing technologies less attractive, and therefore harms the
owners of those technologies. This externality is negative.
 The R&D effect . The development of new knowledge has a positive
externality on others engaged in R&D.
 The net effect of these three externalities is ambiguous. It is generally
believed, however, that the normal situation is for the overall externality from
R&D to be positive. In this case the equilibrium level of R&D is inefficiently
low, and R&D subsidies can increase welfare
3. Alternative Opportunities for Talented Individuals
 Major innovations and advances in knowledge are often the result of the work of
extremely talented individuals.
 The ability to enjoy private returns from innovative activity is crucial for deter
mining whether talented individuals choose to engage in entrepreneurial activity
and innovation or in rent seeking
 Hence economic incentives and social forces influencing the activities of highly
talented individuals may be important to the accumulation of knowledge.
 Three factors that influence talented individuals’ decisions whether to pursue
activities.
1. The size of the relevant market: the larger is the market from which a talented
individual can reap returns, the greater are the incentives to enter a given activity.
2. The degree of diminishing returns.
3.The ability to keep the returns from one’s activities- Thus, clear property rights
tend to encourage entrepreneurship.
4. Learning-by-Doing
• When learning-by-doing is the source of technological progress the rate of
knowledge accumulation depends not on the fraction of the economy’s
resources devoted to R&D but on how much new knowledge is generated by
the conventional economic activity.
• Knowledge is modeled as by-product of the production process rather than an
outcome of a separate production process (Romer, 1990). Hence, all inputs are
engaged in the goods sector:
Y(t ) =K(t)𝛼 A(t)L(t 1−𝛼
…………………………………….(3.34)
 knowledge is created as a side effect of product of accumulation of new
physical capital (learning by doing):the stock of knowledge is a function of the
stock of capital and knowledge accumulation is a function of the increase in
capital or capital/Labor ratio:
• A(t)=BK(t)∅
• Y(t)=K(t)𝛼 𝐵1−𝛼 𝐾(𝑡)∅(1−𝛼) L(t)(1−𝛼) ………………….(3.35)
 Since capital grows according to 𝐾 (t) =sY(t , the dynamics of the capital stock is given by
𝛼
𝐾 (t) =s𝐵1−𝛼 K(t) 𝐾(𝑡)∅(1−𝛼) L(t)(1−𝛼) ……………………..….(3.36)
 The returns to scale depend on how 𝛼+ ∅ 1 − 𝛼 compares with 1 or on how ∅ compares
with 1.
 A steady state exists if either ∅ < 1 or ∅ = 1 and n = 0.
 In all other cases, growth is explosive.
 If ∅ <1, long-run growth is a function of n,
o If ∅ > 1, long-run growth is explosive,
o If ∅ =1 long-run growth is explosive if n is positive and constant (steady state) if n = 0.
o In the case with ∅ = 1 and n = 0: production function from 3.35 becomes
o Y(t)=(BL)1−𝛼 K(t), if b=(BL)1−𝛼 then Y(t)=bK(t).
o Capital accumulation is therefore given by
𝐾 (t) =s(BL)1−𝛼 K(t)=sbK(t)
o Thus, capital grows at the constant rate 𝐾 (t) =sbK(t)
 Output grows at the same rate. The long-term growth rate of the economy depends on the
savings rate. This is because capital makes two-fold contribution to economic growth:
savings raise output through the higher stock of capital and through the higher stock of
3.6. Empirical Application: Time-Series Tests of Endogenous Growth Models
 Jones (1995b) raises a critical issue about these models: Does growth in fact
vary with the factors identified by the models in the way the models predict?
Are Growth Rates Stationary?
 Jones considers two approaches to testing the predictions of fully endogenous
growth models about changes in growth.
 The first starts with the observation that the models predict that changes in the
models’ parameters permanently affect growth.
 For example, in the model with β + 𝜃 =1 and n = 0, changes in s, a𝐿 , and a𝐾
change the economy’s long run growth rate.
 He therefore asks whether the actual growth rate of income per person is
stationary or non-stationary and considers several tests of stationarity versus
non-stationarity.
 Jones argues that because models of fully endogenous growth imply that long-run
growth is easily changed, they predict that growth rates are no stationary.
 He therefore considers several tests of stationary versus non stationarity.
 First test: A simple one is to regress the growth rate of income per person, on a
constant and a trend, and then test the null hypothesis that b = 0.
𝑔𝑡 =𝛼 + 𝑏𝑡 +𝑒𝑡 ……………………………………………….(3.37)
 A second test is an augmented Dickey-Fuller test. Consider a regression of the form
∆𝑔𝑡 =μ + 𝜌𝑔𝑡−1 + 𝛼1∆𝑔𝑡−1 + 𝛼2∆𝑔𝑡−2 +…+ 𝛼𝑛∆𝑔𝑡−𝑛 + 𝜀𝑡 ..(3.38)
 With null hypothesis, ρ = 0.
 Jones examines data on U.S. income per person over the period 1880–1987.
 His statistical results seem to provide powerful evidence that growth is stationary.
 The augmented Dickey-Fuller test overwhelmingly rejects the null hypothesis that ρ
= 0, thus appearing to indicate stationarity.
b. The Magnitudes and Correlates of Changes in Long-Run Growth
 Jones’s second approach is to examine the relationships between the
determinants of growth identified by endogenous growth models and
actual growth rates.
 He begins by considering learning-by-doing models with φ = 1. Recall
that that model yields a relationship of the form
 This implies that the growth rate of income per person is
𝑌(𝑡) 𝐾(𝑡)
=b …………………………………………....…….(3.38)
𝐿(𝑡) 𝐿(𝑡)

Based on (3.26) Y(t ) = bK(t ), b ≡ 𝐵1−𝛼 𝐿1−𝛼 .

 This implies that the growth rate of income per person is
𝑔𝑌/𝐿(𝑡)= 𝑔𝑘(𝑡) -𝑔𝐿(𝑡) ………………………..…………….(3.39)
where gx denotes the growth rate of x. Based on K(t ) = sbK(t). 𝑔𝑘 is given by
𝐾(𝑡) 𝑆𝑌(𝑡)
= 𝐾(𝑡) -𝜎…………………………………………..…….(3.40)
𝐾(𝑡)
 where s is the fraction of output that is invested and δ is the depreciation rate.
 Jones observes that Y/K, δ, and 𝑔𝐿𝑡 all appear to be fairly steady, while investment
rates have been trending up. Thus the model predicts an upward trend in growth.
 Jones reports that in most major industrialized countries, Y/K is about 0.4 and the ratio
of investment to GDP has been rising by about 1% per decade.
3.7. Empirical Application: Population Growth and Technological Change
 Kremer (1993): model of endogenous accumulation of knowledge and
population
growth since 1 Million B.C.
 Kremer (1993), argues that endogenous growth provide insights into the
dynamics of population, technology, and income over the broad sweep of human
history.
 Kremer begins his analysis by noting that essentially all models of the
endogenous growth of knowledge predict that technological progress is an
increasing function of population size.
 If knowledge is non-rival, technological advances can be shared costless and a
larger population size translates into more technological advances.
 He then argues that over almost all of human history, technological progress has
led mainly to increases in population rather than increases in output per person.
A Simple Model
 Kremer’s formal model is a straightforward.
o First, output depends on technology, labor, and land:
Y(t)=𝑇 𝛼 [A(t)L(t)] 𝛼−1 , ………………………….….(3.41)
 where T denotes the fixed stock of land. For simplicity there is no capital.

 Second, additions to knowledge are proportional to population, and also depend

on the stock of knowledge:
𝐴(𝑡)=BL(𝑡)A(t)𝜃 ……………………………………………...…….(3.42)
 Third, Population growth is limited by the stock of land (Malthusian
assumption).
𝑌(𝑡)
=𝑦………………………………………………………....…….(3.43)
𝐿(𝑡)

 Where 𝑦 is subsistence level income.

 We solve the model in two steps.
 Solving this condition for L(t ) gives us

1 1/𝛼 1−𝛼
− 𝛼
𝐿(𝑡)= A(t) T……………..…………….…………..….….(3.45)
𝑦

 This equation states that the population that can be supported is decreasing in the subsistence
level of output, increasing in technology, and proportional to the amount of land.
 The second step is to find the dynamics of technology and population. Since both 𝑦 and T are
constant, (3.45) implies that the growth rate of L is (1 − α)/α times the growth rate of A:
𝐿(𝑡) 1−𝛼 𝐴(𝑡)
= ……………………………………………………….(3.46)
𝐿(𝑡) 𝛼 𝐴(𝑡)

 Thus in this case, (3.46) implies that the growth rate of population is proportional to the level
of population.
 Based on 3.42 and 3.46, thus population growth is increasing in the size of the population unless α is
large or θ is much less than 1 (or a combination of the two). Intuitively, Kremer’s model implies
increasing growth even with diminishing returns to knowledge in the production of new knowledge (that
is, even with 𝜃 <1) because labor is now a produced factor: improvements in technology lead to higher
population, which in turn leads to further improvements in technology.
 Kremer tests the model’s predictions using population estimates extending back to 1
million B.C. that have been constructed by archaeologists and anthropologists.
 Figure 3.9 shows the resulting scatter plot of population growth against population.
 Each observation shows the level of population at the beginning of some period and the
average annual growth rate of population over that period.
 The figure shows a strongly positive, and approximately linear, relationship between
population growth and the level of population. A regression of growth on a constant and
population (in billions) yields.
 A regression of growth on a constant and population (in billions) yields

𝑛𝑡 = −0.0023 0.524𝐿𝑡 2
+ , 𝑅 =0.92, D.W.=1.10 …………………………….3.47
(0.0355) (0.026)
 where n is population growth and L is population, and where the numbers in
parentheses are standard errors.
 Thus there is an overwhelmingly statistically significant association between
the level of population and its growth rate.
3.7. Models of Knowledge Accumulation and the Central Questions of Growth Theory
 Our analysis of economic growth is motivated by two issues: the growth over time in
standards of living, and their disparities across different parts of the world.
 What the models of R&D and knowledge accumulation have to say about these issues?
 Explaining cross-country income differences on the basis of differences in knowledge
accumulation faces a fundamental problem, however: the nonrivalry of knowledge.
 Thus there is no inherent reason that producers in poor countries cannot use the same
knowledge as producers in rich countries. If the relevant knowledge is publicly
available, poor countries can become rich by having their workers or managers read the
appropriate literature.
 And if the relevant knowledge is proprietary knowledge produced by private R&D, poor
countries can become rich by instituting a credible program for respecting foreign firms’
property rights.
 With such a program, the firms in developed countries with proprietary knowledge would open
factories in poor countries, hire their inexpensive labor, and produce output using the proprietary
technology. The result would be that the marginal product of labor in poor countries, and hence
wages, would rapidly rise to the level of developed countries.
 There are numerous examples of poor regions or countries, ranging from European
colonies over the past few centuries to many countries today, where foreign investors can
establish plants and use their know-how with a high degree of confidence that the political
environment will be relatively stable, their plants will not be nationalized, and their profits
will not be taxed at exorbitant rates.

 Yet we do not see incomes in those areas jumping to the levels of industrialized countries.
 One might object to this argument on the grounds that in practice the flow of knowledge
is not instantaneous. In fact, however, this does not resolve the difficulties with
attributing cross-country income differences to differences in knowledge.

 If economies but do not all have access to the same technology, the lags in the diffusion
of knowledge from rich to poor countries that are needed to account for observed
differences in incomes are extremely long—on the order of a century or more.

 It is hard to believe that the reason that some countries are so poor is that they do not
have access to the improvements in technology that have occurred over the past century.

 One may also object on the grounds that the difficulty countries face is not lack of access
to advanced technology, but lack of ability to use the technology.

 But this objection implies that the main source of differences in standards of living is not
different levels of knowledge or technology, but differences in whatever factors allow
richer countries to take better advantage of technology. Understanding differences in
incomes therefore requires understanding the reasons for the differences in these factors.