The Evolution of an Industrial Cluster in China October 28,

The Evolution of an Industrial Cluster in ChinaOctober 28, 2008Belton FleisherDepartment of EconomicsOhio State UniversityColumbus, OH 43210&China Center for Human Capital and Labor Market ResearchCentral University of Finance and EconomicsBeijing, China&IZAEmail: fleisherecon.ohio-state.eduDinghuan HuChina Academy of Agricultural SciencesChinaEmail: dinghuanhuWilliam McGuireDepartment of Agricultural, Environmental, and Development EconomicsOhio State UniversityColumbus, OH 43210Email: william.h.mcguireXiaobo ZhangInternational Food Policy Research Institute (IFPRI)Washington, DC&Zhejiang UniversityEmail: x.zhangcgiar.orgThe Evolution of an Industrial Cluster in ChinaAbstractWe use two rounds of surveys, in 2000 and 2008, in the Zhili Township childrens garment cluster in Zhejiang Province to examine in depth the contract arrangements between vertically separated firms. Output per firm tripled in real terms since the year 2000 and is immensely larger than in 1990. The variation in firm size is very large both in terms of output and employment, although firms are small on average. The increasing divergence in firm sizes is associated with a significant increase in specialization and outsourcing among firms in the cluster. We track changes in financing, firm size, evidence of quality control as evidenced by trademarking and ISO certification. Although initial investments have more than tripled, they were still so low that formal bank loans remain an insignificant source of finance. Accompanying lower entry barriers, there have been an increasing number of firms in the cluster, which have driven down profit and bid up wages. Profit relative to invested capital has declined sharply while the ratio of wage to value added has increased steadily, except for firms whose founders have schooling of twelve years or more. Facing severe competition, more firms have begun to upgrade their product quality. By the year 2007, nearly half of the sampled had established registered trademarks and nearly 20 percent had become ISO certified. JEL Codes: L22, O14, P231. IntroductionChinas rapid rural industrialization seems to defy the conventional wisdom that high transactions costs imply vertical integration. Its rapid industrialization has been accompanied by the emergence of numerous “specialty cities” of a particular kind. Thousands of firms, large and small, many specialized in a finely defined production step, are agglomerated in a densely populated region, where some particular manufactured consumer good is churned out in millions (if not billions) annually. Many formerly rural towns in the coastal areas have become so specialized, boasting themselves as the worlds Socks City, Sweater City, Kids Clothing City, Footwear Capital, and so on. Despite numerous popular media reports on this phenomenon, few studies have rigorously investigated the mechanisms behind the emergence of these clusters. In this paper, we make use of two rounds of surveys, in 2000 and 2008, in the Zhili Township childrens garment cluster in Zhejiang Province to examine in depth the contract arrangements between vertically separated firms. Three decades ago, most people in this rural town were paddy farmers. Nowadays Zhili township embodies one of the largest children garment production centers in China and the world. In the year 2006, a trade report stated that there were over 5000 childrens garment manufacturers in Zhili Township. (http:/www.esupplychain.eu/en/info/viewart/73,Supply_Clusters_A_Key_to_China_s_Cost_Advantage). In this cluster, the production of garments has been divided into very fine phases. Many of the production tasks are undertaken by family workshops. Recently, because of the high labor and land costs, some enterprises have begun to subcontract production to firms in neighboring Anhui and Jiangxi Provinces. An in-depth cast study on the evolution of this rural industrial cluster can help shed light on the path of Chinas rapid industrialization over the past several decades. Quality control has been argued to be one of the key problems associated with increasing division of labor among different producers (North, 1981). We pay particular attention to the interrelationship between quality concerns and the development of trademarks and patent registration in a cluster. Between the time the two surveys were conducted in Zhili, 2000 and 2008, two major deadly fires resulted from overemphasis on cost cutting. In response, the township government developed a regulation that requires “three-in-one” workshops where workers live, dine and work in the same place to install fire exits. This greatly increased the cost of some workshops and induced a series of production and marketing innovations, such as outsourcing and branding. To study whether this virtuous cycle has occurred in clusters in Zhili (and by extrapolation can be expected to occur elsewhere under similar circumstances), we apply historical comparative methods to understand the major development paths, institutional innovations, and the interaction between entrepreneurs and local governments. This paper represents a preliminary and descriptive analysis of the Zhili survey data. In the next section we describe basic features of the Zhili childrens garment cluster; in section 3 we outline major aspects of the evolution of the Zhili cluster since 1990; section 4 concludes.2. Basic Features of the IndustryTable 1 reports basic sample statistics for these firms in the Zhili childrens garment cluster. The top half of the table reports summary statistics for firms surveyed in 2008 and the bottom half for firms surveyed in 2000. One of the most striking statistics in both surveys is the rapid growth of the number of firms. Of the 121 firms sampled in 1999, only 31, approximately 25%, existed in 1990. If the 135 firms sampled in 2008, only 35, again approximately 25% existed in 2000. The number of firms expanded by approximately 16% per year in the 1990s and at about 21% per year in the first seven years of the new century. Mean output value grew rapidly as well, especially in the firms covered in the 2008 survey. By the year 2007, the average firm produced nearly three times as much as in 2000; in the 1990s, growth was significant, but less rapid, output rising by 71% between 1990 and 1997, but declining between 1997 and 1999.The data indicate that these are small enterprises, but they also exhibit large variation in size and performance. The median number of employees was as few as 8 in 1990 and is only 26 in the last year of the 2000 sample. In the last year covered in the 2000 survey, the ratio of the mean number of employees in the top quartile of firms to the mean in the lowest quartile was 5.24; in the last year covered in the 2008 survey, the comparable ratio is nearly 11. The dispersion of output value is even large, equal to over 30 in the last year of the 2000 survey and nearly 30 in 1999, the last full year covered in the 2008 survey.A major feature of these clustered enterprises is low entry barriers in terms of initial investment. Firms starting up in the 1980s report median initial investment of approximately 7,700 yuan in the 2000 survey and 36,200 yuan in the 2000 survey (in real terms with 1990 as the base year). Firms reported in 2000 survey starting in the 1990s reported a median startup investment of 16,000 yuan, while those in the 2008 survey reported 75,000 yuan. Among the firms in the 2000 survey, virtually all initial financing was from the initial principal or his family. Among the firms surveyed in 2008, the proportion of initial investments obtained from formal bank loans has been small. However, the mean proportion of initial investment obtained by means of formal bank loans grew from negligible amount for firms that started operations prior to 1990 almost 13% in 2007. The findings are consistent with early studies (Huang, Zhang and Zhu, 2008; Ruan and Zhang, 2009) on two other rural industrial clusters in Zhejiang Province that lower entry barriers due to clustering enable more firms to participate in the nonfarm production process. However, the quantity expansion in the early stage of industrial clustering often squeezes profit and drives up wages. Table 3 summarizes features of wage payments and profit. Firms in both samples show a sharply increase trend of the proportion of wages in value added, suggesting. increasing wage pressure between labor and capital less than unity. In the 2000 sample there was a particularly sharp increase in the period 1997 to 1999. Firms in the 2008 sample exhibit a rather steady increase in the wage share over the 2000-2007 period. Real monthly wages for women (who make up just over half of employment) increased by approximately 4 percent annually between 2000 and 2006 but jumped by 17.5 percent between 2006 and 2007, rising by nearly 50 percent from 2000 to 2007. Real pay for men also jumped, but in a somewhat different pattern, rising about 38 percent from 2000 to 2007. Although the profit ratio to value added is larger than the wage ratio at the beginning of the 2000 sample, it falls to less than the wage ratio by 1997 and is less than the ratio in every year of the 2008 sample except the first year, 2000. The wage and profit ratios do not sum to 1 for three reasons: (i) profit is based on sales, while production includes goods produced but not sold during the calendar year; (ii) profit is not the total return to capital, because it excludes interest payments; (iii) the statistics reported in table 3 represent the median firm for each item listed, and the median firm for the profit ratio may not be the median firm for wage share. With lower profit and higher wages, it is hard for a firm to survive by just relying on the strategy of expanding production. 3. Evolution of the IndustryWe discuss several measures of changes in industry structure and the complexity of manufacturing. Integration of production and outsourcing.Table 4 shows the median ratio of raw materials to value added for firms in the 2000 and 2008 surveys. The most notable feature of this table is that the ratio of raw material to value added is much lower among firms in the 2008 survey than among firms in the 2000 survey. This suggests that a typical firm in the 2008 survey was “adding more” in the way of factor inputs to the raw material inputs in the manufacture of childrens garments than a typical firm in the 2000 survey. This observation is consistent there being more production processes within individual firms in the later sample, and it is consistent with mean output per firm that is much larger in the 2008 sample. In the nearly overlapping years of 1999 and 2000, mean output of firms in the 2000 survey was less than half that of firms in the 2008 survey, respectively. One is tempted to conclude that an industry trend toward more integrated production processes underlies the difference observed between the two surveys, but the upward trend in the variable within both surveys suggests otherwise. It is noteworthy that the raw materials-value added ratio increased sharply in the 2000 survey between 1997 and 1999, and the mean firm output declined by approximately 20 percent. Among firms in the 2008 sample, there a modest increase in the ratio between 2006 and 2007, but firms entering the data (beginning operations) in 2007, have a ratio of 1.29, lower than the sample median of 1.51 for 2007 and even lower than the median of 1.61 firms that were in continuous operation from the year 2000.The evidence of a slowdown in any trend toward integrated production within both the 2000 and 2008 samples suggests a relationship to another trend, that toward the outsourcing of components of production. Indeed, studying the relationship between outsourcing, subcontracting, and interfirm contractual relationships is one of the principal long-term goals of this project. There is no information on outsourcing in the 2000 survey. Figure 1 shows the proportion of firms outsourcing in the 2008 survey, and the trend is sharply upward. Between 2000 and 2006, the proportion of firms in the full sample who report outsourcing some production rises from about 2.5 percent to about 12.5%, and the proportion of firms outsourcing among the 16 firms entering the sample in 2007 is about 18%. We notice also that the growth of mean firm output slowed down considerably between 2006 and 2007, while the ratio of mean output value of firms in the top quartile of mean production to those in the bottom quartile increased from approximately 24 to 30. This suggests that firms entering the industry tended to be smaller than average, acting as subcontractors to larger firms that increasingly outsourced some tasks to other production units.Branding and qualityOne of the most important issues facing this industry has been increasing competition creating increased pressure on profits and a tendency for firms to cut costs resulting in a self-defeating “race to the bottom” in terms of product quality and with negative implications for product and employee safety. In the 2008 sample, inflation-corrected profits were only 19% higher in 2008 than in 2000 despite a near tripling of real output. Real profits were actually lower in 2008 than in 2005. Industry leaders have understood the negative implications of this trend and have promoted quality certification and development of brand names in order to create credible indicators of product quality and integrity. Branding and quality certification require investments of entrepreneurs time and financial costs. They are a signal to the market that firm stands to lose its investment if its products fall short of promised quality. Moreover, they provide a barrier to effective competition from firms cutting costs but lowering quality. Figures 2 through 5 show the proportion of firms in the 2008 sample that have obtained and that have applied for trademarks and certification from the International Office of Standardization (ISO). The proportion of firms in the survey with trademarks grew from 10% in 2000 to about 38% in 2005 and to 40% by 2007. There were no firms with ISO certification in 2000, although 10% of firms had applied. Nearly 45% of firms had achieved ISO certified status by 2006. It is noteworthy that among firms appearing in the sample in 2007, less than 10% had registered trademarks and a similar number had applied; a similar gap appears for firms with and applying for ISO certification. We take this as further evidence of increasing specialization in production, with firms that exist mainly as suppliers of semifinished products to firms farther down the production chain not depending on trademarks or ISO certification to establish their bona fides to their customers. Figure 6 shows mean investment in design over the sample period. The average investment (including firms that reported spending nothing on product design) jumped about 5-fold in real terms between 2000 and 2005 and another 40 percent between 2005 and 2007.PerformanceWe discuss two measures of firm performance, total factor productivity (TFP) and profit per unit of invested capital. TFP is obtained by subtracting the weighted sum of the logarithms of labor and total invested capital inputs from the logarithm of value added, where the weights are the share of wages in value added and its complement. Trends in TFP are depicted in figures 7 and 8. TFP rose throughout the period covered in the 2000 sample but remained roughly constant throughout the 2008 sample, rising somewhat between 2006 and 2007 for firms that were in the sample from 2000 through 2008. The different trends of TFP in the two samples suggests that in the early stage of the market, innovative entrepreneurs led an expansion of overall productivity, while in the later stage covered by the 2008 survey, the low-lying fruit of available manufacturing and marketing technology had been harvested and that the industrys productivity growth was converging to a longer term, slower growth trend.Figure 9 shows the trend of profit as a proportion of invested capital in the 2008 sample. Profit is defined as reported value of sales deflated by the GDP deflator and invested capital is deflated as described in the discussion of table 2. The profit ratios are indexed to the year 2000 = 100. The calculation of profit is problematic for two reasons in addition to normal recall and record keeping problems associated with these small firms: (i) Although we have data for initial investment in all firms, for firms established prior to 2000, we lack data on investment in years between the date of founding and the 2000; (ii) “profit” includes the remuneration of owner/managers, and we have not yet attempted to subtract this item from the residual remaining after other costs are deducted from revenue. For the entire sample, profit/invested capital rises between 2000 and 2005, despite the rise in the median share of wages in value added between these years. The profit ratio declines abruptly between 2005 and 2006, but rebounds somewhat in 2007. It is striking, though, that for firms entering the sample in 2007, the profit ratio is much lower than for any other observation in the chart. We note three possible causes of the low profit rate for these firms: (i) we have better estimates of their investment because of the shorter period of time covered by the data; (ii) rising wages have cut into the profit rate of new firms more severely than into the profit rate for older firms, perhaps due to sluggish wage adjustments among the currently employed; (iii) as noted in the introduction, in recent years, safety requirements have raised costs, and these requirements for safe workplaces may impinge more severely on new firms.Figures 10 and 11 relate TFP and the profit ratio, respectively, to the human capital of the firms founders. In almost all cases, the firms founder was also the general manager of the firm at the time of sampling. We divide schooling level into three categories, less than 8 years, 8-11 years, and 12 years and above. In figure 10, we find little evidence of a systematic difference in TFP by schooling level of founder either within years or over time. In figure 11, we observe two very interesting patterns: (i) The profit rate declines throughout the sample period on average; (ii) the relative standing of firms whose founders have less than 8 years of schooling has dropped substantially. Enterprises whose founder has 12 years of schooling or more actually show a much higher profit rate in 2007 than in 2000, in sharp contrast to firms whose founders fall into the two lower schooling categories. A simple regression of the of the profit ratio for firms with founders in the highest schooling group divided by the profit ratio for firms in the lowest schooling group yields a highly significant slope coefficient of 0.11 per year over the period 2000 through 2007. Moreover, while the founders of a third of the firms in 2000 were in the lowest schooling category, only 11 percent of the founders in the year 2000 were in the highest schooling category, they constituted 19 percent of the founders in the year 2007. This pattern suggests the possibility that schooling was becoming more important to firm profitability as the industry expanded and competitive pressures increased. Finally in figure 12 we show a scatter relating investment in design among firms that reported information for the year 2000 to their profit ratios in 2007. That there is no system relationship between these variables is suggestive of the intense competition in the industry.4. SummaryThe average firm in Zhili in 2007, the last