0000000001133720
AUTHOR
G Di Vaio
CONVERGENCE AND DIVERGENCE: A NEW APPROACH, NEW DATA, AND NEW RESULTS
Recently, Penn World Tables include new data that enable calculation of total factor productivity in addition to output for a large set of countries. We use these new data to examine convergence and divergence across countries by applying a new approach, which differentiates between the dynamics of output and of productivity. Our empirical results lead to two main new contributions to the literature. The first is on the interpretation of “β-convergence” in “growth regressions.” It means that output per worker in each country converges to productivity but does not imply convergence across countries, since productivity tends to diverge from the global frontier. The second contribution is to t…
A new look at global growth convergence and divergence
A key question in economics is whether poor countries will automatically close the income gap with rich countries. However, different empirical methods yield different answers – growth regressions suggest convergence, whereas tests of distribution dynamics suggest divergence. This column discusses recent research that reconciles these two strands of the literature. It extends the benchmark growth regression model to include a parameter that determines the share of new technologies a country can adopt each year. The result is that, although each country converges to a growth path, the growth paths themselves may diverge
Global Divergence in Growth Regressions
This paper extends the standard growth regression model by adding an assumption that a country follows the global technology frontier either fully or partially. This additional assumption changes significantly the growth regression model and its results in three main ways. First, it shows that although a country converges to its long-run growth path, this path can diverge from the countries at the global frontier. We measure the degree of divergence for each country and find that most indeed diverge from the frontier. Second, we estimate growth dynamics without controlling for additional variables. Third, our new method enables us to disentangle the effects of the explanatory variables on t…