Serge Coulombe, ECO 6120 –Convergence and growth regressions
Conditional convergence
The rate of convergence to the steady state
The speed of convergence β is the speed at which economies approach their steady state A positive β implies that the growth rate of an economy decreases as the economy approaches its steady state With α = 1/3, n = 2 %, δ = 5 %, x = 2 %, β is around 6 %
In practice, log-linear estimation on the growth rate of y:
Investment per unit of effective worker (n+x+δ)k sf(k) k k*k(1)k(0) Time series considerations Cross-section considerations: absolute convergence
n+x+δ sf(k)/k k k* k(rich) k(poor) γ(poor) γ(rich) Absolute convergence
n(p)+x+δ sf(k)/k k k(p)* k(r) k(p) γ(poor) γ(rich) conditional convergence n(r)+x+δ k(r)*
Testing the convergence hypothesis and growth regressions Absolute convergence hypothesis: Barro and Sala-i-Martin (1992) U.S. states (B&siM2004, section 11.2.) Point estimate of 0.02 for β Coulombe and Lee (CJE, 1995), panel data approach to Canadian provinces,
Conditional convergence The underlying framework is the relationship between the growth rate and the gap between the actual level of Y versus the steady-state level (here assume y = log(y): Proxy variables are used for Y*: The long-run equilibrium is:
A wave of growth empirics study Absolute convergence hypothesis not rejected for developed countries But was rejected for a broad set of countries when less- developed countries were included using new Heston and Summers data bank But conditional convergence not rejected (Mankiw, Romer and Veil (1992) and Barro and Sala-i-Martin (1995)