Premium
WHY DO RATES OF β‐CONVERGENCE DIFFER? A META‐REGRESSION ANALYSIS
Author(s) -
Dobson Stephen,
Ramlogan Carlyn,
Strobl Eric
Publication year - 2006
Publication title -
scottish journal of political economy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.4
H-Index - 46
eISSN - 1467-9485
pISSN - 0036-9292
DOI - 10.1111/j.1467-9485.2006.00374.x
Subject(s) - estimation , econometrics , convergence (economics) , economics , meta regression , meta analysis , regression , regression analysis , investment (military) , rate of convergence , panel data , statistics , mathematics , macroeconomics , computer science , medicine , computer network , channel (broadcasting) , management , politics , political science , law
The roles played by differences in study design and methodology in influencing the estimates of β ‐convergence have been hinted at in narrative reviews of the empirical convergence literature. While such reviews are useful, they only provide informal evidence as to the reasons for the study‐to‐study variation in reported convergence rates. In contrast, meta‐regression analysis is a way of formally measuring the roles played by study design and methodology in influencing β ‐convergence. In cross‐national studies, convergence rates are found to be higher when panel estimation methods are used, and when human capital development, investment rates and spatial factors are controlled for. The longer the time span covered by the estimation, the lower the rates. In intra‐national studies, β ‐convergence is higher when the investment rate is included as a conditioning variable and when GMM estimation methods are used. Rates are found to be lower in studies of developing countries.