Convergence Hypothesis: A Cross Country Analysis

• Author: KALSOOM ZULFIQAR, MUHAMMAD ASLAM CHAUDHARY and ARIBAH ASLAM

Abstract. Convergence debate has been an important topic of economic growth literature. This article aims to investigate convergence at assorted level of disaggregation among a sample of almost 60 countries. It has tested absolute and conditional convergence hypotheses for a set of developed and developing countries by applying pooled least square methodology. The results suggest absolute convergence for countries having similar characteristics and conditional convergence for countries having heterogeneous structures. Disparity level for each country is also calculated with reference to average steady state income. The study has also scrutinized the role of investment, openness and population growth in accelerating the convergence process.

Keywords: Absolute convergence, Conditional convergence, Solow swan growth model, Pooled ordinary least square

1. INTRODUCTION

   There has been substantial inquisition into the nature and sources of differences in growth rates across countries and regions over time. This was essentially necessitated owing to the considerable potential impact caused by even marginal differences in growth rates, over a long period of time, on the standards of living of people. Convergence, as a phenomenon of diminution in growth rate disparity among different regions, denotes the course by which comparatively poorer regions or countries grow quicker than the rich countries. It can be described as a process of catching up or narrowing down the gap between per capita incomes of less developed and developed countries. It is absolute if all countries accomplish the same level of long-term income growth. It also suggests that the less developed countries grow more rapidly than developed countries resulting in catching up by poorest countries.

   Conditional convergence, on the other hand, suggests that a country or a region will converge to its own steady state as every country or region has its own distinguished set of endowments.

   Convergence hypothesis was initially advocated by Solow (1956) and further refined and developed by Baumol (1986) and Barro and Xavier-Sala-i-Martin (1991). Barro (2000) concluded that absolute convergence can occur only if all countries have same identical inherent features. Conditional convergence, on the other hand, implies that economies with homogeneous features are more likely to experience income convergence irrespective of their preliminary situation. These findings were further proven by Barro and Sala-I-Martin (1992) and Barro (2000). Murphy and Ukpolo (1999) conducted a detailed analysis of conditional convergence hypothesis for African region. Empirical results verified the occurrence of conditional convergence in the region for the period 1960 to 1985. Romer (1986) however raised questions about the validity of convergence hypothesis while presenting his endogenous growth theory.

   The absolute and conditional convergence hypotheses have been tested by several researchers using different methodologies and data sets. The outcome appears to have attracted a mixed response from unmitigated rejection by some to ardent acceptance by others. It is in this background that current study is conducted for a set of developed and developing counties to furnish evidence regarding the convergence hypothesis. The analysis is based on latest data sets and is expected to improve understanding of convergence process in various developed and developing countries. An important contribution of this research work is the calculation of disparity intensity for each country which helps to find out how far away a country is from the average steady state. Besides, the study has also investigated the role of investment, population growth and openness in convergence process. The empirical findings are expected to help policy makers in devising relevant policies in this regard.

2. LITERATURE REVIEW

   Abramovitz (1986) and Baumol (1986) conducted maiden empirical analysis of the convergence premise using Maddison’s (1982) dataset. Abramovitz (1986) authenticated the convergence hypothesis by employing relative variance and rank correlation coefficient. Baumol (1986) estimated a simple regression equation to show the strong inverse association among the growth rate of Gross Domestic Product and its preliminary value. Delong (1988) in his analysis, nonetheless, concluded no income convergence rather divergence by using the same data set. Baumol and Wolff (1988) verified convergence hypothesis for a set of developed countries by applying piecewise linear and quadratic regression. Dowrick and Nguyen (1989) established the existence of income convergence for developed countries by using parameter stability test. The study commended increase in total factor productivity (TFP) as the basis of income convergence.

   Barro and Sala-i-Martin (1990) studied absolute convergence for the United States of America. Barro (1991) tested the convergence hypothesis for a comprehensive data set consisting of 98 countries and rejected the absolute convergence hypothesis. The analysis also recommended that main factor causing income diversion was disparity regarding human capital stocks possessed by various countries.

   Paci and Pigliaru (1997) rejected the convergence hypothesis for European region. The study was conducted for 109 regions of 12 European countries for the decade of 1980s. It also analyzed the trend of labor productivity convergence in sample countries. The results suggested that labor productivity in these countries was converging at the rate of 1.2%. Blomstrom and Wolf (1994) found that in most of the world economies labor productivity rates were experiencing convergence. The study also concluded no convergence for manufacturing sectors in these countries.

   Johnson (2000) analyzed income convergence across the United States for the period 1929-1993. The analysis was performed using the per capita personal income in each state relative to the United States average by applying the nonparametric methodology proposed by Quah (1996). The study however, found no empirical evidence of divergence in the cross-state income distribution. Esteban (2000) studied Regional convergence in Europe using shift-share analysis and concluded that regional specialization had a very trivial role in regional convergence. The analysis suggested policy measures to bridge the gap between developed and underdeveloped countries based on improvement of infrastructures and human capital.

   Cole and Neumayer (2003) analyzed the absolute convergence hypothesis for 110 countries for the time period 1960-96 based on population weighted per capita GDP. Knack (1996) analyzed the factors influencing the speed and convergence ability of a country and concluded quality of institutions as an imperative factor in building up the convergence potential of a country. Mankiw et al. (1992), conducted comprehensive empirical analysis of conditional convergence based on cross-section data. The study was primarily based on the empirical test of various versions of Solow growth model.

   Nonneman and Vanhoudt (1996) tested convergence hypothesis for a set of OECD countries. The analysis established strong empirical support for convergence among the homogeneous countries. Cho and Graham (1996) resolved that most of the poor economies typically exceed their steady state levels and consequently approach their steady state from above. Murthy and Ukpolo (1999) conducted empirical analysis of convergence phenomenon for African region. The study assessed conditional convergence by utilizing Solow model. The study concluded that African economies were converging at an overall rate of 1.7% and this sluggish convergence was ascribed to the structural problems in the region. Dobson and Ramlogan (2002) rejected conditional convergence hypothesis for Latin American region. The study was based on cross-section data from various Latin-American countries.

   Karras (2010) inspected convergence hypothesis for various regions. Levine and Renelt (1992) utilized extreme-bounds analysis (EBA) to empirically analyze the conditional convergence hypothesis. The results suggested occurrence of conditional convergence for the period 1960-89.

   Andres et al. (1996) analyzed convergence hypothesis for OECD region by including inflation rate, exports growth and public-sector expenditure in their model. However, the analysis did not recommend any notable...