Spatial dependence in the growth process and implications for convergence rate: evidence on Vietnamese provinces

ABSTRACT

Existent studies on Vietnamese provinces tend to assume that province-specific growth is independent of that in its neighbours. However, many studies analysing regional economic growth in China, Brazil and Mexico report the existence of spatial spill-over effects. This paper investigates whether this is the case for 60 Vietnamese provinces for the time-period 1999–2010, using a system-GMM estimator and a Solow growth model augmented with human and physical capital and spatial-lag covariates. We report that spatial dependence is a significant determinant of growth and conditional convergence in Vietnamese provinces. We also demonstrate that the rate of convergence decreases as the distance between neighbouring provinces increases. Given these findings, we recommend testing for spatial dependence in growth models for Vietnam and beyond to avoid omitted variable bias and inform evidence-based regional policies that take account of spatial externalities.

KEYWORDS:

• Economic growth
• spatial dependence
• regional convergence, GMM

JEL CLASSIFICATION:

• C21
• O11
• R11

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. List of provinces: An Giang, BRVT, Bac Giang, Bac Kan, Bac Lieu, Bac Ninh, Ben Tre, Binh Dinh, Binh Duong, Binh Phuoc, Binh Thuan, Ca Mau, Can Tho, Cao Bang, Da Nang, Dak Lak, Dong Nai, Dong Thap, Gia Lai, HCMC, Ha Giang, Ha Nam, Ha Noi, Ha Tinh, Hai Duong, Hai Phong, Hoa Binh, Hung Yen, Khanh Hoa, Kien Giang, Kon Tum, Lai Chau, Lam Dong, Lang Son, Lao Cai, Long An, Nam Dinh, Nghe An, Ninh Binh, Ninh Thuan, Phu Tho, Phu Yen, Quang Binh, Quang Nam, Quang Ngai, Quang Ninh, Quang Tri, Soc Trang, Son La, TT-Hue, Tay Ninh, Thai Binh, Thai Nguyen, Thanh Hoa, Tien Giang, Tra Vinh, Tuyen Quang, Vinh Long, Vinh Phuc, Yen Bai.

2. In order to double-check, we use Cook's distance (1982) to detect outliers. Cook's distance measures the effect of excluding an observation. It has been suggested that Cook's distance values greater than one should be investigated. None of Cook's distance values obtained from the regressions in Table 3 are greater than one, but those that stand out relatively to the rest of the values suggested the same outcome as winsorizing, e.g. excluding three outliers.

Additional information

Notes on contributors

Bulent Esiyok

Bulent Esiyok is an Assistant Professor in the Department of Economics, Baskent University, Turkey. His research interests include international trade, foreign direct investment, economic growth and competition.

Mehmet Ugur

Mehmet Ugur is Professor of Economics and Institutions at the University of Greenwich and member of the Greenwich Political Economy Research Group (GPERC). His research focuses on microeconometrics of innovation and productivity and evidence synthesis methods, particularly meta-analysis. He teaches Applied Econometrics and supervises PhD students working on FDI, innovation, early warning systems for banking crises and firm performance.