Industrial Development Policy and Innovation in Southern China: Government Targets and Firms' Behaviour

Abstract

The paper investigates the relation between firms' innovation behaviour and the industrial innovation policy promoted by the Guangdong Province Government in the framework of its “Specialized Towns Program”. In this context there are very few academic studies, and non-Chinese scholars are not involved in this debate; moreover, the attempt of self-evaluation of government institutions appears weak (or at least not accessible). In other words, little evidence is offered to check the real response of firms to the government policy apart from that diffused by the government itself. With the support of specific town and firm-level data, we investigate firms' responses to the local governments' innovation policy. In doing so, we suggest a set of relevant variables that should be considered as well as possible ways to measure them. We then run an empirical econometric analysis. The main findings suggest that among the most relevant determinants of a positive attitude towards government policies are the ownership structure, the background of the entrepreneurs in terms of their engagement in government activities and, to a lesser extent, the strength of the policy. We believe that, although these issues find in Guangdong a unique institutional setting, they are relevant not only for Guangdong, but can shed light on more general dynamics of contemporary industry.

Notes

There is a wider body of literature (especially in Chinese language) that reviews the different types of industrial agglomerations in Guangdong, their birth and growth dynamics (Wang, 2000a; Li, 2002; Wang, 2004; Chen & Cheng, 2005; Tang & Wang, 2005; Tang et al., 2005; Huang et al., 2006; Li, 2006; Huang & Hu, 2007; Zhao, 2007 to name some). However, in this paper we concentrate on the experience of STs only, since it is the only case where a specific government policy was designed and a clear definition of “ST” provided.

We refer in particular to the “co-operation agreement on improving self-innovation capability and accelerating the economic and social development in Guangdong” (Lin, 2007).

These include collaborations between: Tangkeng town in Meizhou and South China University of Technology; Huangpu town in Zhongshan, South China University of Technology and South China Agriculture University; Leliu town in Shunde and China Geology University; Jun'an town in Shunde and Donghua University; Shiling town in Guangzhou and Sichuan University; and Dongcheng in Zhaoqing and South China University of Technology and Guangdong Industrial University (for further information see DST, 2007).

Including the: water heating facilities cluster in Shuikou in Jiangmen 1999; motorbike cluster in Pengjiang in Jiangmen 2003; rosewood furniture cluster in Dachong in Zhongshan 2003; ceramic cluster in Nanzhuang in Nanhai, Foshan 2003; electronics cluster in Qingxi in Dongguan in 2003; Section Aluminum cluster in Dali in Nanhai, Foshan in 2003; textile cluster in Xichao in Nanhai, Foshan 2003; hardware cluster in Jinsha in Nanhai; Electronics cluster in Songgang in Nanhai; shoemaking cluster in Pingzhou in Nanhai; underwear cluster in Yanbu in Nanhai 2003; toy cluster in Guanyao in Nanhai 2003; agriculture cluster in Heshun in Nanhai 2003.

Wang (2002a) gave evidence from the case of Xiaolan town to support this conclusion.

A few theoretical papers have used game theory to analyse possible firm behaviour (Liu & Yang, 2002; Luo & Liu, 2004; Tang & Wang, 2005; Gao, 2006; Chen & Li, 2007; Zhu, 2007), but they have not been applied to any case study.

The criteria are in order of relevance: i.e. if no firm in the town satisfies the first criteria, we look for a firm that satisfies the second one and so forth. Should we find two firms in one town that satisfy the first criteria, we choose the one with the highest market share.

By the global (or national) market we mean that if a firm is among the top 10 in market share in the global market, then it can surely be considered as the leading firm of the town. If this condition is not satisfied, we see if it is instead verified on the national market.

The selection herewith specified provides us with a sample of relatively large firms; however, the data for the exact number of employees are missing for nearly half of the firms. Therefore, we could not explicitly include such a control in our regression model. We do not think this invalidates the results. The selection criteria use sales as a first proxy of size. Moreover, we are studying the relation between one firm and the innovation centre within each town, so it is not size per se that matters, but rather the relative size and the fact that the firm is the leading firm within its own specific context. On the other hand, we control for the sector of specialization, which seems to be more influencing per se on the firms' attitude towards innovation.

There is no foreign firm whose entrepreneur has a background or present position related to the government.

The classification of resource-based, low, medium and high-tech sectors is the same as that adopted by United Nations Industrial Development Organization (UNIDO). For further information, see UNIDO, International Yearbook of Industrial Statistics (Cheltenham: Edward Elgar, 2002).

Here, the different level of governments includes town level, county level and prefecture city level governments.

The company dataset Alibaba is available on http://www.alibaba.com/ and http://china.alibaba.com/.

There is no equivalent of the variance inflation factor (VIF) test for multi-collinearity for ologit models, however, some authors suggest to try the VIF test anyway (Tardanico, 2005). We tried the test and it does not evidence any multi-collinearity.

We refer in particular to the test for the parallel regression assumption (or proportional odds assumptions). See Long and Freese (2006) for details.