ABSTRACT
This study examines the productivity growth and its four components for Japanese regional economies by analyzing 47 prefectures for the period 1995–2012. The productivity changes are measured by the Hicks–Moorsteen–Bjurek productivity growth, and the four components consist of technical change, efficiency change, scale change, and input and output mix effects. Data envelopment analysis (DEA) is applied to the measurement. From the results, this study provides two policy recommendations for Japan’s new economic growth strategy, which are associated with (1) government support for the diffusion of advanced technology over regions and (2) the creation and development of unique innovation by regional industries.
ACKNOWLEDGEMENTS
The authors thank Dr Dieter F. Kogler and the reviewers, whose comments have improved the quality of this paper.
DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.
SUPPLEMENTAL DATA
Supplemental data for this article can be accessed http://dx.doi.org/10.1080/00343404.2017.1413238.
ORCID
Mika Goto http://orcid.org/0000-0001-7214-2233
Akihiro Otsuka http://orcid.org/0000-0001-7852-4590
Notes
1. The data are available from the Cabinet Office (see http://www.esri.cao.go.jp/jp/sna/data/data_list/kakuhou/files/h26/h26_kaku_top.html). The growth rate is the arithmetic average of each year’s growth rate.
2. For a continuing discussion of ‘industrial competitiveness’ at a committee created at the Cabinet Office, Japan Productivity Center, see http://www.kantei.go.jp/jp/singi/keizaisaisei/skkkaigi/kaisai.html/.
3. For example, Fukao and Kwon (Citation2006) discuss the problems of zombie corporations in Japan.
4. This non-parametric feature of DEA is associated with its methodological weakness because the traditional use of DEA does not cope with statistical tests at the SFA level. However, in empirical studies, both DEA and SFA are popular for productivity measurement. Examples of them applied to regional productivity analysis include the studies by Otsuka and Goto (Citation2015), Badunenko and Romero-Ávila (Citation2014), and Enflo and Hjertstrand (Citation2009). The last example uses DEA, but tried to overcome the weakness arising from the non-parametric feature by applying a bootstrap method.
5. The fundamental structure of HMB-PI is almost identical to the ‘encompassing index’ defined by Balk (2001). That study defined ‘new encompassing index numbers of productivity change’ from four components – technological change, technical efficiency change, scale efficiency change, and output or input mix change – and examined differences between the Malmquist index and the proposed index. The former two components comprise the Malmquist index; the latter two components are added to create the new productivity index. The fundamental structure of HMB-PI is similar to the encompassing index in Balk. The former, in this study, is defined by integration of input- and output-based distance functions with the application of DEA, while the latter is defined by either an input- or an output-based parametric distance function with translog form.
6. See Appendix A in the supplemental data online for a formal presentation of computational process and mathematical descriptions of DEA models, the HMB-PI and each component of the index. The decomposition originates from Nemoto and Goto (Citation2005). Löthgren and Tambour (Citation1996) originally proposed the scale change component in a context of DEA-based returns to scale and scale elasticity measurements.
7. Otsuka and Goto (Citation2015) examined agglomeration economies in the Japanese regional economy. Agglomeration economies include not only scale economy but also technology advancement and production efficiency.
8. This study employs a gross value added-KLE (capital, labour and energy)-type production model. Intermediate energy is not exactly consistent with the output in this study because the final energy consumption includes the energy used for producing not only final products but also intermediates. This is because of data availability, that is, regional energy consumption exclusively used for producing gross value added is not publicly available. Therefore, we use final energy consumption data as an alternative approximation variable. Such treatment may cause larger bias in the calculation with industry-level productivity. It is our future challenge to conduct an industry-level productivity examination using more consistent energy data. In addition, this study does not use gross output data because a corresponding input of intermediates in the real term cannot be available in the long-term period covered in this study.