GEOGRAPHICAL ECONOMICS

(1)

GEOGRAPHICAL ECONOMICS

(2)

(3)

(4)

ELTE Faculty of Social Sciences, Department of Economics

Geographical Economics

week 4

THE BACKGROUND OF GEOGRAPHICAL ECONOMICS: GROWTH AND CONVERGENCE

Author: Gábor Békés, Sarolta Rózsás Supervised by Gábor Békés

June 2011

(5)

week 4 Gábor Békés

Economic Growth and International Convergence

The Mankiw- Romer-Weil analysis Interpretation of convergence

Outline

1 Economic Growth and International Convergence The Mankiw-Romer-Weil analysis

Interpretation of convergence

(6)

Growth

Economic Growth and the neoclassical model Solow-Swan model

Mankiw, G., Romer, D. and Weil, D. (1992), A contribution to the empirics of economic growth, QJE

Once a country has a growth rate below the steady-state level, they grow at a faster rate.

= convergence to the steady-state level, which is the same for each country.

. . . if technologies and preferences are identical.

(7)

MRW: basic Solow model

Solow model, Cobb-Douglas, CRS

Y(_t) =_K(_t)^α[_A(_t)_L(_t)]¹^−α ₍₁₎

A(t) total factor productivity

K· =sY−δK (2)

L_t =_L₀_e^nt _{és A}_t =_A₀_e^gt

s is investment rate,δis depreciation rate, g is the rate of technological progress and n is the exogenous rate of population growth

k =K/L, y =Y/L

k· =_sy(_t)−(_n+_g+δ)_k(_t) =_sk(_t)^α−(_n+_g+δ)_k(_t) ₍₃₎

(8)

MRW: basic equilibrium

Steady-state capital per eective labor: k^∗

k^∗= [s/(n+g+δ)]¹^/¹⁻^α (4) Steady state output - log (GDP per capita): y^∗

ln y^∗=ln A₀+gt+ ^α

1−αln s− ^α

1−αln(n+g+δ) (5) As in realityα=_{1/3, thus}_α/1−α=₀.5, that is the two important elasticities (s, n+_g+δ) are around 0.5-0.5 When A(_t) =_a+e_i, i.e. a constant plus a shock

ln y^∗=_a+_gt+ ^α

1−αln s− ^α

1−αln(_n+_g+δ) +e (6)

(9)

MRW: empirical assumptions

Assumptions

n,s are independent frome. Why?

The eect of A(_t) what does it mean? how could we introduce it in a dierent way (Hicks)?

g+δ=0.05 OLS estimation

With or without the assumption of elasticities (parameter restrictions)

(10)

MRW: First results

(11)

Results

They conrm some important theoretical results The coecients of saving and population growth have opposite signs

We cannot reject that the two eects are equal in magnitude.

We can explain a lot of things. . . . . . but: αis too high

. . . but: developed countries are dierent

(12)

MRW: Augmented Solow model

Lucas (1988) - human capital

Y(t) =K(t)^αH(t)^β[A(t)L(t)]¹^−α−β (7) Human capital accumulation has a similar fashion to physical capital

k· =s_ky(t)−(n+g+δ)k(t),h^· =s_hy(t)−(n+g+δ)h(t) (8) Steady state output - log (GDP per capita): y^∗

ln y^∗ =_{ln A}₀+_gt+^α

z ln s_k−^α+β

z ln(_n+_g+δ) + ^β

z ln s_h (9) where z =₁−α−β

(13)

MRW: Augmented Solow model

Some remarks β'1/3−1/5

If h^∗ is the steady state level of human capital, then ln y^∗=ln A₀+gt+ ^α

1−αln s_k

− ^α

1−αln(_n+_g+δ) + ^β

1−αln h^∗ (10) up to this point h was a part of the error term

h positively correlates with s, and negatively correlates with n omitting it can cause biased estimates

Ability to choose from the two model specications what type of data can we use the level or the rate of accumulation of human capital? NOW we are using `the fraction of eligible population enrolled in secondary school' (s_h)

(14)

Augmented Solow model

(15)

MNW: Augmented model results

Human capital measure has a strong, signicant eect The model explains almost 80 percent of variation in income per capita (increased)

The sum of the coecients is equal to zero.

Restriction can not be rejected β'₀.3−₀.4

Consistency with the data, e.g. Y =L¹^/³K¹^/³H¹^/³

(16)

MNW: Conslusion

Consistency with the data, e.g. Y =L¹^/³K¹^/³H¹^/³ On the whole, these models indicate dierent steady state levels that are aected by exogenous determinants.

Steady-state: accumulation of human and physical capital, growing population

Convergence only if we control for the determinants of the steady-state

= conditional convergence The speed of convergence:

λ= (_n+_g+δ)(₁−α−β)'0.02

= GDP growth rate is the function of steady state, exogenous parameters (s_k, s_h,n+g+δ) and the initial level of income (y_o) (Deduct this result - homework!)

(17)

Convergence estimation

(18)

Convergence estimation

(19)

Convergence estimation

(20)

Empirical results

Basic setup: no convergence, poor countries do not grow faster (Table 3)

When we take into account the steady state conditions, the eect of the initial level of income is signicant and negative;

conditional convergence (Table 4)

Human capital matters, stronger convergence (Table 5) However, the convergence occurs as a quite slow process The Solow model is not so bad...

(21)

Two concepts of convergence: remarks

Barro and Sala-i-Martin (1995): Economic Growth, Chapter 11

Up until now we've dealt with β-convergence (poor countries catching up with rich nations)

According to another concept, convergence occurs when the dispersion of log per capita income within a group of countries declines over time. We call this process σ-convergence.

(22)

Two concepts of convergence: remarks

In the case ofβ-convergence there are no overtaking or big jumps

β-convergence tends to generate σ-convergence however, this is not a rule

To put it in a more accurate way, β-convergence is a necessary but not a sucient condition forσ-convergence.

(Derive this statement! - homework!)

The dispersion of income can be determined by exogenous shocks (e.g. oil-price shock), that aect only a limited number of countries. It can inuence our convergence estimation.

(23)

Two concepts of convergence: EU8

(24)

Two concepts of convergence: EU8

Convergence relative to the country mean

(25)

Two concepts of convergence: EU8

(26)

Two concepts of convergence: EU8

β σ interesting

Germany 2,4% 0,310,19

UK 2,8% 0,170,12 *

Italy 1,5% 0,420,27 * France 1,2% 0,210,14 Spain 1,8% 0,350,22

How could we explain the dierences? - homework!

(27)

A dierent model: Quah

Danny Quah: Convergence Clubs

What if there appear plenty of connections within a group of countries, wherefore individual growth cannot explain precisely the convergence path?

Empirical results: instead of unconditional convergence, evolving two groups and countries converge to one of these two `peaks'

=σ-convergence within the two groups

=`twin peaks'

e.g. Asian convergence and EU convergence to a dierent level

(28)