Weitere Kapitel dieses Buchs durch Wischen aufrufen
While we have examined productivity and innovation in the previous two chapters, this chapter discusses capital-labor ratio and composition of workers in outsourcing firms. Capital intensity is as important as innovation for analyzing the determinants of productivity, particularly labor productivity. In this sense, this chapter is arrayed just after the previous two chapters on productivity and innovation. However, the investigation of capital-labor ratio has been critical for analyzing the outsourcing decision since the seminal work by Grossman and Hart (1986). As some firms are more capital-intensive than other firms even within narrowly defined industries, we exploit firm-level data to consider within-industry variations across firms in capital intensities. We inspect how capital-labor ratio is associated with the firm’s outsourcing decision and how outsourcing affects the boundary of firms, especially in the lens of employment. Labor is further divided into production versus non-production workers, as well as skilled versus unskilled workers. Although it is not directly about outsourcing, this chapter also discusses the firm’s decision of multiplant operation, as it links critically to the ownership and firm boundary.
Bitte loggen Sie sich ein, um Zugang zu diesem Inhalt zu erhalten
Sie möchten Zugang zu diesem Inhalt erhalten? Dann informieren Sie sich jetzt über unsere Produkte:
Imports from branch offices/factories are not included as they are not independent legal entities.
In incomplete contract models, transaction-cost theory since Williamson (1975) and property-right theory since Grossman and Hart (1986) differ in several dimensions. For instance, the latter allows inefficiency even within an integrated firm. However, we do not discuss them separately in our empirical analyses, as our data do not allow us to identify sources of transaction costs necessary to distinguish these theories. Antràs (2016a, b) are excellent expositions of incomplete contract theory in international economics.
He also reported that the share of intra-firm trade tends to be high in capita-abundant countries. This observation is consistent with the theory under the premise that capital-abundant countries tend to export capital-intensive goods, as predicted by the orthodox factor proportions trade theory.
While the regression of import share is useful to directly compare our results with previous literature, no data on the value of offshoring is available in the RIETI survey.
R&D expenditures for firms without R&D data are set at zero. Negligible 10 −8 is added to R&D-sales ratio before taking logarithm.
Firms without K data are excluded. Less than 2% of the surveyed firms are without K data. Even if we treat K data of these firms as zero, the estimates remain virtually unchanged. Although the capital intensity of offshored tasks may vary depending on the destinations, the K/L data at the firm level cannot distinguish dispersions across destinations.
The invariance in the share of outsourcers across regions contradicts with the argument that offshoring to China is distorted by FDI ownership regulation, although our results might pick up the effects of Special Economic Zones (e.g., majority-owned subsidiaries not allowed outside of SEZ).
We use the capital-labor ratio in 2000 as the instrumental variable, since the same firm’s capital-labor ratios even in different years are likely to be correlated each other but the capital-labor ratio is unlikely to directly affect the firm’s sourcing mode choice six years later. Own variable in 2000 is similarly assigned to other right-hand variables as well.
From the MITI survey sample, Tomiura (2007a) also found that firms outsourcing production offshore are on average more labor-intensive than FDI firms, but the MITI survey contains no data on outsourcing destinations.
We need not distinguish one-sided and two-sided tests in our case as two curves do not cross in all relevant ranges.
Girma et al. (2005) found, in the case of the UK, that the productivity distribution of multinational plants stochastically dominates that of exporters, which, in turn, dominates that of non-exporters.
The width of each K/L interval is set equal.
This also suggests that—even if labor-intensive firms are sorted to China based on factor abundance differences—these firms do not necessarily choose outsourcing because the make-or-buy decision is relatively insensitive to K/L in China. Therefore, the high share of outsourcers in China is not explained by the North–South difference in factor abundance.
According to 2006 White Paper on Small and Medium Enterprises by METI, the most impressive China-ASEAN gap in the perceived business problems for Japanese firms is the bill collection difficulty.
Feenstra and Hanson (2005) found that more than 80% of Chinese processing exports choose only one mode in their four ownership/control regimes (foreign versus Chinese in the factory ownership and in the control of input purchases).
As explained in Chap. 5, Japanese manufacturing census no longer collects data on production workers separated from non-production workers in total employment of plants.
Maurin et al. (2002) is a rare example that empirically distinguished skilled and unskilled non-production workers at the firm level, though for exporters, not for outsourcers. They defined skilled non-production workers by the sum of engineers and technicians.
The same problem remains when we discuss the white-collar/blue-collar dichotomy. Even if we classify workers based on their educational attainments, only academic achievement is considered.
While it differs from U.S. practice, we define non-regular workers as those employed for prefixed periods with termed contracts. Some part-time workers are employed without specifying employment period and thus could be included in regular workers.
It is generally unclear whether the supply disruption risk is higher in foreign outsourcing compared with domestic outsourcing. While supply chains stretched over long distance may be vulnerable, suppliers concentrated in a narrow region has its own risk. The impact of serious earthquakes in Japan gives us an opportunity to consider this issue. Barrot and Sauvagnat (2016) found that firm-level idiosyncratic shocks propagate through production networks especially when firms produce specific inputs, based on the data on major natural disasters in the U.S. over the past three decades.
OECD (2007) summarized descriptive statistics related to this hypothesis. Empirical analyses of this issue will be referred to later.
Buch and Lipponer (2010) and Navaretti et al. (2003), for example, analyzed how labor demand differs between multinational versus domestic firms.
In Japan, the wage gap between production and non-production workers is slim, and the supply of production workers relative to non-production workers decreased due to declining population and rising college enrollment.
As a pertinent theoretical prediction, Cuñat and Melitz (2007) showed that countries with more flexible labor markets specialize in sectors with higher volatility.
Hashimoto (1993) reported that non-regular employment is twice as variable as regular employment in Japan by estimating the output elasticity.
For example, Kawaguchi and Ueno (2013) documented the long-run decline of age-specific mean job tenure by birth cohort uniformly across firm sizes and industries in Japan after the end of the high-growth period.
In Japan, production workers and non-production workers are not separated in manufacturing census and in BSJBSA. Production workers occupy less than one-fifth of total employment in Japan, according to Labor Force Survey.
We cannot estimate the cost-share equation because wage data are not disaggregated. The employment share equation shows the labor demand impact net of wage effects. Berman et al. (1994) and Hijzen et al. (2005) confirmed that the results from employment shares are very similar to those from cost shares in their analysis of skilled/unskilled workers.
This measure of capital is superior to the often used tangible fixed assets, since the latter is inevitably affected by volatile fluctuations in values of land and plant constructions.
From the denominator, we exclude temporary workers and workers dispatched from personnel service companies.
As the standard deviation is strongly affected by firm size, we normalize it by the mean.
In our standard specifications, χ 2 test statistics take negative values, which we can regard as the non-rejection of the null hypothesis. Besides, as limited number of firms (less than 6%) switched their offshoring mode between the two years, it is difficult to precisely estimate the coefficient on offshoring dummy by the fixed-effect model.
Before taking logarithm, we add negligible 10 −8 after replacing unavailable data with zero.
We define wage as total wage payment divided by the total number of long-term workers. As short-term workers are mobile across firms, their wages should be determined by the market and are likely to be largely common across firms.
Slaughter (2001) also detected rising U.S. elasticity, but attributed it to the time trend effect. Bergin et al. (2009) found large employment volatility in Mexican maquiladoras.
Based on OECD input–output tables, Hijzen and Swaim (2010) reported that the effect of imported intermediates on labor demand elasticity is stronger in countries with weaker employment protection legislation. Based on Indian sector-region data, Hasan et al. (2007) found that labor demand elasticity weakly increases with the region’s flexibility of labor markets.
As related evidence, Munch (2005) found that offshoring increases the unemployment risk based on Danish data on individual workers.
Senses (2010) argued that the impact of increased offshoring is offset by declining share of unskilled workers in the U.S.
Thesmar and Thoenig (2007) reported that the share of workers employed under interim contracts rises with the intermediate consumption-sales ratio in French firms, but offshore and domestic sourcing are not distinguished in their analysis.
As a pertinent finding, Chen and Kamal (2016) reported that Internet-based computer connections are positively associated with in-house production ratio in the U.S. sample.
Kremer and Maskin (1996) formalized the shift “from firms such as General Motors, which use both high- and low-skill workers, to firms such as Microsoft and McDonald’s, whose workforce are much more homogeneous” (p. 1). Our results suggest that offshoring appears to segregate a part of regular workers into non-regular workers outside of strong employment protection.
On the other hand, the domestic sourcing of the same type of tasks has no significant relation with regular employment, indicating that tasks outsourced within the same country are likely to be limited to tasks previously performed by non-regular workers (e.g., document filing, data entry).
Crinò (2010) reported that service offshoring increases high skill workers based on detailed U.S. occupation data, but the tasks types within service offshoring are not distinguished.
We have confirmed that virtually all other interaction terms are statistically insignificant even if we add them to ( 8.5).
Although we regard offshoring and regulation of dispatched workers as separate issues, these might be interrelated. For example, Boulhol (2009) argued that trade liberalization leads to labor market deregulation.
Japanese traditional employment practice was once characterized by, though not exactly permanent employment, but lifetime employment commitment, especially in large-sized firms in manufacturing industries. However, as discussed in the beginning of this subsection, the average job tenure declined securely across any firm sizes and industries since the 1980s in Japan.
As a result, we have no information how many plants each firm operates or which plants are operated by the same firm. Transactions between plants operated by the same firm are not distinguished either.
This includes expenditures not only on materials, but also fuels and electricity.
The value-added is defined by the value of shipment minus expenditures on materials.
The estimation results reported below are from Okubo and Tomiura (2011).
We control for regional effects at the level of city-town-village, the most geographically detailed geographic unit in Japan.
As a related finding from the U.S., Aarland et al. (2007) reported that the probability of multiplant operation increases as the firm is industrially diversified.
High labor productivity could be due to high capital-labor ratio, but we find that these plants are labor-intensive. High labor productivity may also reflect rich human skills, but no employment data disaggregated by educational attainment or occupations/skills are available within our plant-level data and we have controlled for plant size, which tends to correlate with human skills.
Forman and McElheran (2013) reported that the share of within-firm transfers declined more among plants that adopted IT compared with plants that did not in the sample of about 2500 U.S. manufacturing plants.
Her measure of plant size is relative to the parent firm; that is, the share of the parent firm’s total manufacturing revenue that is shipped by the plant. Due to the data constraint, we cannot link plant-level data with their parent firms.
They use the number of software programmers as a proxy for resource in discussing the impact on investment in within-establishment Internet in over 300,000 U.S. establishments.
Offices inactive in production are not captured by the manufacturing census. Aarland et al. (2007) and Henderson and Ono (2008) used the U.S. data on “auxiliary establishments,” which are non-production offices providing services to other plants/offices of the firm.
We must note the difference in sample coverage—all plants included in the analysis of multiplant decision in the previous table, but only single-plant firms covered in this table. This difference is due to the design of Japanese manufacturing census. The census asks the question on the HQ separation only to single-plant firms probably due to the possibility that the HQ could be collocated with another plant of the same firm even if it is separated from the plant itself. As shown by the number of observations in these two tables, overwhelming majority of firms captured by manufacturing census are single-plant firms if we count the number of plants.
The HQ services for plants have also been analyzed theoretically in models such as Ekholm and Forslid (2001), and Fujita and Thisse (2006), though they focus on the effect of communication costs. Our result is in line with their prediction if we suppose that plant-specific communication costs are relatively low in productive plants.
Aarland et al. (2007) reported that firms with separated HQ are substantially bigger than those without.
We can construct longitudinal data of plants for these later years. The year 2008 was the most recent year with available plant-level data at the time of starting that research.
The decline of middle-skilled labor, rather than low-skilled labor, characterizes the changes in the U.S. and Germany (7.4 percentage point in Germany, 5.9 in the U.S., but 2.1 in Japan), while the rise of high-skilled labor is similar, including Japan (3.1 percentage point increase in Japan, 3.4 in Germany, 4.0 in the U.S.). The former could be consistent with the labor market polarization, but we should be cautious in this interpretation at least from the latter observation.
- Capital, Labor, and Boundaries of Offshoring Firms
- Springer Singapore
- Chapter 8
Neuer Inhalt/© Stellmach, Neuer Inhalt/© Maturus, Pluta Logo/© Pluta, Rombach Rechtsanwälte/© Rombach Rechtsanwälte