Is the finance led growth hypothesis robust to alternative measures of financial development?

Abstract

In this article we employ tests of noncausation to measure the impact of financial development on 34 developing countries experiencing vastly different stages of economic development and covering annual observations over the period 1960–2009 for most countries. Focusing on the dual role of financial development and economic growth as proposed in much of the endogenous growth literature, we draw upon individual country evidence from 34 developing countries at alternative stages of economic development. Our contribution to the literature lies in our use of several tests based on multiple measures selected to represent financial development. Testing each one separately, we come up with varying patterns of correlation between finance and growth for individual countries. The variation or consistency of each measure as an influence on economic growth is both telling for policy and the determination of future patterns of growth for emerging market economies.

Acknowledgements

This article has greatly benefited from helpful comments and criticisms provided by Hashem Pesaran, Ron Smith, Sean Holly and seminar participants at Harvard, Cornell, Cambridge and the Australian National University. The views expressed in this article are of the authors and not necessarily shared by JP Morgan. Only the authors are responsible for any remaining errors and omissions.

Notes

¹ The PWT may be accessed freely from the following web address: http://pwt.econ.upenn.edu/download/download.html. The version used in this analysis was 5.6.

²The exact line references of variables taken from the CD-ROM are provided in the Appendix.

³ A broad ranging overview of some of the following recently advocated procedures is pro-vided in Toda and Phillips (1994) who not only survey the methodological issues that may arise from application of alternative techniques, but also make suggestions based upon a simulation exercise. In particular, Toda and Phillips (1994) provide a very thorough theoretical overview along with a simulation study of the sampling properties of sequential Granger causality tests in vector ECMs compared with standard causality tests in levels and first-difference VARs.

⁴ See, for example, King et al. (1991), Mosconi and Giannini (1992) and Toda and Phillips (1993a, b).

⁵ See Toda (1994a, b) for further elaboration of these problems, as well as the associated asymptotic theory.

⁶ See also Saikkonen and Lutkepohl (1996) for a related approach to estimation and inference in infinite order cointegrated VAR processes.

⁷ Incidentally, these were also the three criteria used to select countries in seperate country Granger non-cuasality tests conducted by Demetriades and Hussein (1996).

⁸ Although the European countries of Greece and Portugal, along with a number of Asian countries, are now formally recognized as members of the Organization for Economic Co-operation and Development (OECD), they were not classified as being member countries in 1960. Also, our hypothesis of interest is to what extent financial development contributes to the process of economic development. This process should ideally include those periods (which may comprise several years) through which the country made a gradual transition from a less to more developed economy.