Is the secularization hypothesis valid? A panel data assessment for Taiwan

Abstract

The secularization thesis is based on the idea that a country becomes more secular as it becomes richer and more industrialized. In order to investigate whether this hypothesis is valid, we examine the relationship between religion in Taiwan and economic development across 23 counties by adopting the panel data approach for the period 1998 to 2003. We select six religious activity proxy variables and five economic development variables in a cross-county panel data framework. The evidence indicates that the ratio of females to males has a significantly positive effect and that the literacy rate and population density each have a significantly negative effect on the religious variables. Finally, except for the Christian culture, the unemployment rate has a positive effect on the religious variables. Our findings thus support the view that the religious secularization hypothesis is valid in Taiwan.

Acknowledgements

The authors are appreciative of the highly constructive comments provided by the Editor of this journal and the anonymous reviewers. This work is also supported in part by the Ministry of Education, Taiwan, ROC under the ATU plan. Chun-Ping Chang is grateful to the National Science Council of Taiwan for financial support through grant NSC 97-2410-H-158-002, as well as NSC 97-2410-H- 158-003. We are responsible for any errors that remain in this article.

Notes

¹ To investigate the effect of religious beliefs on economic development, traditional studies were usually based on qualitative research, until Barro and McCleary (2003) used international survey data on religiosity for a broad panel of countries.

² Economists figure that religion, by teaching honesty, fairness, trust and integrity, can promote economic development (Grier, 1997; Blum and Dudley, 2001). However, our interest here is not to assess the determinants of religiosity, but to facilitate the analysis of the effects of religious activity on economic development.

³ It is a part of what is often referred to as ‘Modernization Theory’ (Barro and Mitchell, 2004).

⁴ We are grateful to an anonymous referee for this suggestion.

⁵ Due to the changes in the statistical basis of the official data, with regard to the definitions of the religious variables, this is the longest period that we can currently select.

⁶ Levin et al. (2002) proposed a panel-based ADF test that restricts parameters, keeping them identical across cross-sectional regions. Im et al. (2003) relaxed the assumption of the identical first-order autoregressive coefficients of the LLC test and allowed parameters to vary across regions under the alternative hypothesis.

⁷ Based on Monte Carlo experiment results, IPS demonstrate that their test has more favourable finite sample properties than the LLC test. Moreover, Chen and Lee (2007) recently applied a new panel unit root testing procedure with structural breaks and cross-sectional correlations introduced into the model to re-investigate the stationarity of energy consumption per capita for seven regional panel sets covering the 1971 to 2002 period. The structural breakpoints identify the likely causes of major changes in energy consumption in the past.

⁸ All the empirical results in our study are estimated using EViews software, and if the reader is interested in replication, the dataset can be provided if requested. We are grateful to our referees for this suggestion.

⁹ McCleary and Barro (2006) stressed that people often raise their monetary contributions as a substitute for participating personally.

¹⁰ It is possible to allow the slopes to vary across counties, but this gives rise to some new methodological issues and considerable complexity in the calculation (see Greene, 1997, Chap. 14).