Regional Efficiency Dispersion, Convergence, and Efficiency Clusters: Evidence from the Provinces of Indonesia 1990-2010

Improving production efficiency at the regional level is often considered a means to reduce regional inequality. This presentation is about regional efficiency convergence across provinces in Indonesia over the 1990–2010 period. Through the lens of both classical and distributional convergence frameworks, the dispersion dynamics of the following three indicators are contrasted: overall efficiency, pure efficiency, and scale efficiency. Results from the classical convergence approach suggest that—on average—there is regional convergence in all these three efficiency measures. However, results from the distributional convergence approach indicate the existence of two local convergence clusters within the overall and pure efficiency distributions. Moreover, since scale efficiency is characterized by only one convergence cluster, the two clusters of pure efficiency appear to be driving the overall regional efficiency dynamics in Indonesia. The presentation concludes highlighting the importance of monitoring and evaluating heterogeneous (beyond average) behavior, multiple convergence clusters, and geographic proximity when formulating regional policies that aim to reduce regional inequality.

Transcript

1. Regional Efficiency Dispersion, Convergence, and Efficiency Clusters: Evidence from the Provinces of Indonesia 1990-2010 Carlos Mendez https://carlos-mendez.rbind.io Associate Professor Graduate School of International Development Nagoya University, JAPAN Prepared for the 56th Annual Meeting of the Japan Section of the RSAI [ Slides available at: http://bit.ly/rsai2019japan ]
2. Motivation: Large per-capita income differences across provinces in Indonesia Persistent income differences despite several policy efforts Differences in efficiency explain a larger fraction of the differences in income (Caselli 2005; Hall and Jones 1999; Hsieh and Klenow 2010) Research Objective: Study efficiency convergence/divergence across provinces in Indonesia over the 1990-2010 period. Methods: Classical convergence framework (Barro and Sala-i-Martin 1992) Distributional convergence framework (Quah 1996; Hyndman et. al 1996; Menardi and Azzalini 2014) Data: Overall efficiency = Pure technical efficiency x Scale efficiency (DEA framework) 26 provinces over the 1990-2010 period (Kataoka 2018)
3. Main Results: 1. Convergence on average in the three measures of efficiency 2. Regional heterogeneity matters: Local convergence clusters 3. Clustering dynamics Overall efficiency: Two convergence clusters Pure technical efficiency: Two convergence clusters Scale efficiency: One convergence cluster Policy Implication: Policy should be focalized at the cluster level
4. Outline of this presentation 1. Global convergence "on average": Using classical summary measures Beta convergence Sigma convergence 2. Let's go beyond the average: Regional heterogeneity still matters Distribution dynamics framework Distributional convergence 3. Local convergence clusters: Overall efficiency: Two convergence clusters Pure technical efficiency: Two convergence clusters Scale efficiency: One convergence cluster
5. (1) Global convergence "on average" Using classical summary measures Beta convergence Sigma convergence
6. Beta convergence
7. Sigma convergence
8. (2) Let's go beyond the average Regional heterogeneity still matters Distribution dynamics framework Distributional convergence
9. Regional heterogeneity matters Let's GO beyond the average! Study the dynamics of the entire regional distribution Let's move from conditional mean to conditional density estimation. Recent advances in nonparametric econometrics: Distribution dynamics
10. The distribution dynamics framework
11. (3) Local convergence clusters Overall efficiency = Pure technical efficiency x Scale efficiency Overall efficiency: Two convergence clusters Pure technical efficiency: Two convergence clusters Scale efficiency: One convergence cluster
12. Overall e ciency: Two convergence clusters
13. Pure technical e ciency: Two convergence clusters
14. Scale e ciency: One convergence cluster
15. Spatial distribution of overall e ciency clusters
16. Spatial distribution of pure e ciency clusters
17. Concluding Remarks A happy ending "on average" : Differences in overall efficiency and its two determinants (pure technical efficiency and scale efficiency) have decreased over the 1990-2010 period. Global convergence on average Focus beyond the average : Regional differences are still important Multiple local convergence clubs: Overall efficiency: Two convergence clusters Pure technical efficiency: Two convergence clusters Scale efficiency: One convergence cluster Implications and further research Convergence clusters help us identify regions facing similar challenges Call for better coordination of regional policies at the cluster level What is the role of geographical neighbors in accelerating convergence? What alternative clustering frameworks could be implemented?
18. Thank you very much for your attention https://carlos-mendez.rbind.io Slides available at: http://bit.ly/rsai2019japan This research project was supported by JSPS KAKENHI Grant Number 19K13669