Globalization of Work and Innovation: Evidence from Doing Business in China Jan Bena and Elena Simintzi May 2017 Abstract We study how access to “cheap” offshore labor due to the 1999 U.S.-China bilateral agreement affects U.S. firms’ innovation. To this end, we decompose innovations into new goods (product innovations) and new production methods (process innovations). We find that U.S. firms operating in China decrease their share of process innovations by 12% and that this adjustment is purely driven by a lower quantity of process innovations. We obtain the same result using the inter-temporal variation in ownership restrictions on foreign investment in China across industries. Our findings suggest that using cheap and abundant offshore labor substitutes for labor-saving innovation.

JEL classification: O33, J31, L23 Keywords: Globalization of work, technological change, product and process innovation, labor-saving innovation, China. Affiliations: Sauder School of Business, The University of British Columbia. E-mails: [email protected]; [email protected]. Acknowledgments: We would like to thank Ramin Baghai, Nicholas Bloom, Philip Bond, Will Cong, Nancy Gallini, Ron Giammarino, Yi Huang, Adrien Matray, Jilian Popadak, John Van Reenen, F. M. Scherer, and Martin Schmalz, as well as seminar participants at Boston College, Center for Economic Research and Graduate Education - Economics Institute (CERGE-EI), McMaster University, Simon Fraser University, University of Alberta, UBC Finance, UBC Economics, UBC Strategy and Business Economics, UCLA, UNC, University of Washington, West Virginia University, and conference participants at IFN Stockholm Conference, Economics of Entrepreneurship and Innovation Conference at Queen’s Business School, Stockholm School of Economics Finance Symposium, NBER Productivity, Innovation, and Entrepreneurship, Yale SOM Conference, and LFG Conference for helpful discussions and comments.

Manufacturers who had been automating U.S. and European factories to shave labor costs stopped once they set up in China. (WSJ, 11/23/2015)

I

Introduction

Globalization and technological change are the main forces that drive increasing polarization of job opportunities and rising wage inequality in developed economies.1 The “end of labor”—as the unintended consequences of these forces are often called in popular terms— is a topic of heated public debate, in which a lower integration of labor markets is often suggested to be an obvious remedy. In this paper, we argue that public policy aimed at addressing these challenges needs to be more nuanced, because globalization and technological change are connected through corporate investment. Specifically, restricting access to “cheap” offshore labor (“less globalization”) can create incentives for firms to alter their investment policy by directing R&D toward devising production methods that save costs (“more technological change”) instead of employing more local workers.2 In this paper, we examine the relationship between two main ways U.S. firms can lower their production costs: substituting U.S. for cheaper offshore labor and investing in the development of labor-reducing technologies. Our hypothesis is that, if substituting U.S. for offshore labor becomes more attractive, firms decrease their R&D investments in developing cost-reducing technologies. To measure firms’ investment into cost-reducing 1

See Acemoglu and Autor (2011); Autor, Katz, and Kearney (2006, 2008); Autor, Levy, and Murnane (2003); Blinder (2006); Blinder and Krueger (2013); Goos, Manning, and Salomons (2014). 2

A longstanding literature in economics points to the relation from the abundance and price of the production factors to technology adoption. John Hicks (1932) in his Theory of Wages notes: “...a change in the relative prices of the factors of production is itself a spur to invention, and to invention of a particular kind—directed to economizing the use of a factor which has become relatively expensive...”. This reasoning agrees with recent anecdotal evidence. For example, the CEO of United Technologies comments on Trump-Carrier jobs deal to keep a plant in Indiana open instead of shipping production to Mexico: “GREG HAYES: So what’s good about Mexico? Wages are obviously significantly lower. About 80% lower on average. ... The result of keeping the plant in Indiana open is a $16 million investment to drive down the cost of production, so as to reduce the cost gap with operating in Mexico. ... What does that mean? Automation. What does that mean? Fewer jobs, Hayes acknowledged.” (Business Insider, Dec 6, 2016.)

–1–

technologies, we decompose innovations into two types, new products and processes, which we differentiate by analyzing texts of firms’ patent claims. Process innovations refer to inventions of new methods that lower production cost (Scherer, 1982, 1984; Link, 1982; Eswaran and Galini, 1996), while product innovations result in new goods. Our main dependent variables are the share of process to total innovations and the quantities of process and product innovations. An important offshoring destination for U.S. firms is China mainly because the size of labor force and low wages in China create a prime opportunity for U.S. firms to save costs.3 U.S. firms operating in China, however, cannot capture all the benefit of low wages because Chinese partners (for example, joint venture counterparts, suppliers, and distributors) obtain a share of the profits of U.S. firms’ subsidiaries in China.4 As a result, the effective labor costs of U.S. firms from their Chinese operations do not only depend on the actual wage paid to Chinese workers, but also on the share of profits of Chinese subsidiaries captured by the Chinese partners. We use the 1999 U.S.-China bilateral agreement to identify the effect of firms’ improved ability to access cheap and abundant offshore labor on their investment in cost-reducing technologies. The agreement, which was largely unanticipated due to the turbulent political landscape, made using Chinese labor easier for U.S. firms by lifting restrictions on doing business in China, such as: the removal of local content and export performance requirements, the withdrawal of the requirement to use domestic suppliers, or the liberalization of distribution services. While a large share of the profits of U.S. firms’ Chinese subsidiaries accrued to Chinese partners before 1999, the agreement increased the share 3

“China’s average manufacturing wages, at about $0.25 per hour, are about one-fifth as great as Mexico’s, and about one-fiftieth as much as total compensation for manufacturing workers in the United States. China’s labor force is 18 times that of Mexico and five times that of the United States” (CSR Report for Congress, 2000). 4

China is a prominent example of hold-up problems due to the fact that foreign companies have to deal with local partners. In Poorly Made in China, Midler (2009) describes how Chinese suppliers extract surplus from Western companies by manipulating prices and quality, and argues that solutions like relationship contracting were not effective in the case of China. See also Antràs (2005, 2014).

–2–

of the profits the U.S. firms capture post-1999, effectively reducing their labor cost. In our main analyses, we use the difference-in-differences estimator to compare the effect of the 1999 U.S.-China bilateral agreement on U.S. high-patenting firms with a subsidiary in China two years prior to the agreement (treated) relative to U.S. high-patenting firms with a subsidiary in a low-wage Asian country but not China (control). We find that, after 1999, the treated firms have a lower share of process to total innovations relative to the control firms by 4 percentage points compared to pre-treatment years, which is a 12% reduction relative to the median ratio. This change in the process-product innovation mix is driven by a lower quantity of process innovations, which is 25% lower for the treated firms. In contrast, the agreement has no differential effect on the quantity of product innovations of the treated relative to control firms. Our findings suggest that a better ability to tap cheap offshore labor is an economically important determinant of the process-product innovation mix, inducing technological change, as it decreases return on investing in labor-saving innovation, namely innovation substituting for more “expensive” U.S. workers. In our regressions, we control for time-invariant firm characteristics by including firm fixed effects, for time-varying firm characteristics by including firm-level controls, and for time-varying industry characteristics by including interacted industry and year fixed effects. Our key identifying assumption is that, conditional on these controls, the assignment of firms into the treated versus control group is “as good as random”. In that regard, we define our control group as U.S. firms with subsidiaries in low-wage Asian countries (but not in China) pre-treatment, since firms that select into having a subsidiary in China or other lowwage Asian country have arguably similar production and cost structures and thus similar incentives to economize on labor costs. Furthermore, such firms may be affected by the same shocks experienced by countries in similar geographies, which are thereby differenced out in our specifications. Although the identifying assumption is practically untestable, we support it in a number of ways. First, we show that there are no significant differences in firm characteristics between the treated and control groups in 1998, that is, before the onset of the treatment. Second, we find no significant effect of the agreement in pre-treatment

–3–

years, while the effect persists after 1999. Third, when we control for potential differential trends between the treated and control firms by interacting the value of the dependent variable in 1997 with a full set of year dummies, our results continue to hold. Fourth, we conduct a placebo test by generating “pseudo” treated groups and re-estimating our baseline specifications: we conclude that the regression coefficients we estimate in our main tests are results that cannot occur mechanically in our data. We discuss and rule out three alternative explanations. First, it may be that our estimates capture the effect of Chinese import competition or the opening of the Chinese market to U.S. firms on technological change (Bloom, Draka, and Van Reenen, 2015). To address this concern, we show that Chinese import competition and U.S. exports to China have no differential effect on the process-product innovation mix and the levels of process and product innovations of treated firms after 1999. Second, it may be that our estimates capture changes in U.S. firms’ patenting practices around 1999, such as U.S. firms transferring their R&D centers to China, trade secrets substituting for patenting of process innovations, or, relatedly, secrecy incentives affecting patent quality. We find no support for these explanations. Third, we do not find support for the possibility that the burst of the “Dot-com bubble” around 2000 affected U.S. firms’ innovation in a way that is driving our results. More generally, we show that the differences in intensities with which treated and control firms are involved in different technological fields of innovation are not biasing our results. We provide evidence that the change in the firms’ innovation mix is due to the labor channel. To this end, we first examine subgroups of treated firms where we expect to observe differential effects. We exploit cross-sectional variation in expected wage bills of U.S. firms’ Chinese subsidiaries and find that the treatment effect is attenuated for firms whose Chinese subsidiaries expect to pay relatively higher wage bills. We further exploit cross-sectional variation in the equity shares of U.S. firms vis-à-vis their Chinese counterparts in the Chinese subsidiaries, which proxies for the reduction in effective labor costs due to U.S. firms’ ability to capture a higher share of the subsidiaries’ profits. We find that the treated firms with higher U.S. equity relative to Chinese equity respond more to the agreement.

–4–

Our process innovation measure does not capture firms’ overall investments in production costs saving technologies as it does not include improvements in production processes that occur due to purchases of innovations embedded in the new intermediate capital goods. Our estimates should therefore be interpreted as a lower bound on the overall effect of the agreement on the adoption of labor-saving technologies. To examine the importance of such externally sourced improvements in production processes, we show that our results are not driven by the substitution between internally produced process innovations (captured by our measure) and purchases of product innovations in capital goods from suppliers. We also examine the effect of the agreement on treated firms’ capital investments. We find that U.S. firms with operations in China relatively decrease their capital investments following the agreement. This suggests that the reduction in effective labor costs leads the firms to rely relatively less on physical capital, which is consistent with the firms substituting investment in labor-saving production methods for utilizing cheap Chinese labor. Finally, we examine whether our results are robust to using an alternative setting. Specifically, we use the variation across industries and over time in ownership restrictions imposed by the Chinese government on foreign investments in China. We extract this information from the Foreign Investment Industry Catalogues issued six times in the 19952012 period. The staggered loosening of ownership restrictions implied by the catalogues changes the split of the profits of Chinese subsidiaries in favor of U.S. firms, effectively reducing their labor cost. We find that the loosening of ownership restrictions decreases the ratio of process to total innovations and the quantity of process innovations for highpatenting firms with subsidiaries in China as compared to those with subsidiaries in other low-wage Asian countries, while there is no differential effect on the quantity of product innovations. This evidence further supports a causal interpretation of our findings. Our paper relates to several literatures. The labor and finance literature examines how labor interacts with corporate investment (Atanassov and Kim, 2009; Besley and Burgess, 2004; Ouimet and Zarutskie, 2015; Tate and Yang, 2016) and financing (Benmelech, Bergman, and Seru, 2015; Giroud and Mueller, 2015, 2016; Benmelech, Frydman, and Papanikolaou, 2017) decisions. Our paper contributes to this literature by studying

–5–

firms’ investment in labor-saving technological innovation. Our paper is also related to the literature on the effect of trade with low-wage countries on corporate innovation activities (Bernard, Jensen, and Schott, 2006; Bloom, Draka, and Van Reenen, 2015; Hombert and Matray, 2016). While these papers study how firms in developed countries react to a surge in Chinese imports, we show that an improved ability to tap cheap offshore labor alters U.S. firms’ incentives to innovate in new production methods. Our paper is the first to differentiate empirically innovations into new products and processes and examine drivers of the composition of corporate innovation activities. There is also prior work examining the impact of regulatory frictions on international trade and investment (Moran, 2001; Antràs, 2005). More broadly, seminal papers in the corporate finance literature describe how hold-up problems due to contract incompleteness distort investments (Williamson, 1979; Grossman and Hart, 1986; Hart and Moore, 1990). We add to this literature by showing that regulatory changes that allow U.S. firms to benefit more from offshore workers have implications for their R&D investment decisions. The paper is organized as follows. Section II describes how we decompose innovations into new products and processes and defines key variables. Section III gives details on the 1999 U.S.-China bilateral agreement and section IV describes the sample. In sections V-VII, we present the main results. Section VIII presents results obtained using an alternative empirical setting based on the Foreign Investment Industry Catalogues, and section IX concludes. The Internet Appendix provides details on data and construction of variables, and contains validation and robustness checks.

II

Process-Product Innovation Mix Measure

We categorize firms’ innovative projects into two main types, products and processes, and create novel measures of firms’ process-product innovation mix. To proxy for firms’ process and product innovations, we examine the output of corporate R&D activities as measured by patents, the exclusive rights over an invention of a product or a process (Griliches, 1990). We collect the complete set of patent grant publications issued weekly by the United States

–6–

Patent and Trademark Office (USPTO) from January 1976 to June 2013, which contain full texts of the universe of patents awarded by USPTO to U.S. and international companies, individuals, and other institutions. We parse the structured-texts of utility patent grants to, first, identify the section that contains patent claims, and, next, to classify each claim within this section as process or product. We are also able to classify claims into independent, that is, those that stand alone and do not reference any other claim in the patent, or dependent.5 Claims define—in technical terms—the scope of protection conferred by a patent, and thus define what subject matter the patent protects. Claims are critical defining elements of a patent and are the primary subject of examination in patent prosecution. Claims are also crucial in patent litigation cases. To measure a firm’s process-product innovation mix, we define Share of Process Innovationsit as the ratio of the number of process claims to the total number of claims included in patents applied for by firm i in year t.6 To measure the level of process (product) innovation output, we define Process Innovationsit (Product Innovationsit ) as the natural logarithm of one plus the number of process (product) claims included in patents applied for by firm i in year t. In robustness checks, we use count-data models, and we also compute these measures (i) using independent claims only, (ii) by weighting claim counts by the number of citations received, and (iii) by using claim counts adjusted by the average number of claims in each technological class and time period. We assume that the claim count is zero for firm-years with missing USPTO information. We also measure a firm’s process-product innovation mix at the patent level. To this end, we classify each patent as: i) a process patent, if all patent’s claims are process claims; ii) a product patent, if all patent’s claims are product claims; iii) a process-apparatus patent, 5

We download the publications from the ‘United States Patent and Trademark Office Bulk Downloads’ page hosted by Google Inc. at http://www.google.com/googlebooks/uspto.html in July 2013. A detailed description of how we distinguish claim types is provided in the Internet Appendix. 6 To assign patents to firms in Compustat we use the name matching procedure from Bena, Ferreira, Matos, and Pires (2016). Note that the NBER patent database provides GVKEY-patent number links for patents awarded by USPTO until 2006, while our matching is based on patents awarded until June 2013. See Bena, Ferreira, Matos, and Pires (2016) for a detailed description of the matching procedure and a comparison of the matches to those in the NBER patent database.

–7–

if the first patent’s claim is a process claim and there exists at least one independent claim that is a product claim; iv) a product-method patent, if the first patent’s claim is a product claim and there exists at least one independent claim that is a process claim. Our patentlevel measure of a firm’s process-product innovation mix is the sum of the number of process and process-apparatus patents (or, alternatively, the sum of the number of process, processapparatus, and product-method patents) divided by the total number of patents applied for by firm i in year t. We also construct a patent-level measure of a firm’s process-product innovation mix adjusted by the number of citations received. All variable definitions are provided in the Appendix. A process innovation, by definition, describes a new way to produce an existing good, while a product innovation describes a new good that did not exist before. Prior literature argues that a process innovation is aimed at improving a firm’s own production methods in order to lower its production cost, while a product innovation is an improvement sold to others–either to consumers or to other firms (Scherer 1982, 1984; Link 1982; Cohen and Klepper, 1996; Eswaran and Gallini, 1996). Using our measure of the process-product innovation mix and the level of process innovation output, we provide four pieces of evidence consistent with process innovation acting like a labor-saving technology. First, we show that, consistent with the predictions of the labor economics inequality literature, industries with a high share and quantity of process innovations are associated with subsequently lower intensity of routine tasks, namely tasks substitutable by technology (Table IA-B2 in the Internet Appendix). Second, we show that industries involving easily offshorable tasks are associated with a lower share and quantity of process innovations (Table IA-B3 in the Internet Appendix). Third, we show that industries with a high share and quantity of process innovations become subsequently more capital intensive as proxied by investment in equipment over industry employment (Table IA-B4 in the Internet Appendix). Fourth, we search abstracts and background description sections of patent documents for texts directly mentioning labor costs reductions, and we document a positive correlation between the share and quantity of process innovations and the share of a firm’s patents that contain such texts (Table IA-B5 in the Internet Appendix). Overall,

–8–

these analyses provide evidence that our measures of process-product innovation mix and level of process innovation output are meaningful proxies for labor-saving innovation.

III

1999 U.S.-China Bilateral Agreement

The bilateral agreement signed between the U.S. and China in November 1999 was a landmark in the economic relations of the two countries, and it paved the way to China’s entry into the World Trade Organization (WTO). The agreement involved significant concessions from China, including the elimination of a number of restrictions on investment by U.S. firms, tariff reductions, and trade barriers removals. The agreement was unexpected due to turbulent political relations between the two countries. In mid-1997, the U.S. puts aside multilateral negotiations with China and starts bilateral talks instead—a decision driven mainly by political reasons. In 1998, little progress is being made. A milestone in the talks is the visit of Premier Zhu Rongji in the U.S. in April 1999, when he made—for the first time—significant concessions. No agreement was signed however, and the negotiations were seriously threatened a few weeks later when the U.S. mistakenly bombed the Chinese embassy in Belgrade. The agreement was finally signed on November 15, 1999 when the U.S. Trade Representative (USTR) Charlene Barshefsky visited China. To emphasize the uncertainty surrounding the negotiations, it is worth mentioning that USTR threatened to leave China three times and the negotiations were completed only after she decided to stop at the Chinese trade ministry on her way to the airport (Devereaux and Lawrence, 2004). Historically, U.S. firms operating in China faced numerous restrictions and regulatory interventions. They had to work with Chinese partners, mainly, joint-venture counterparts, suppliers, distributors, and local governments, which led to hold-up problems, disrupting firms’ operations and lowering profits. Hold-up problems stem from contact incompleteness, which is predominant in international contracts (Rodrik, 2000). Antràs (2014) highlights the nature of incomplete contracts in China by citing a Chinese old saying: “signing a contract is simply a first step in negotiations”.

–9–

The hold-up problems were substantially alleviated by the agreement. China lifted ownership restrictions on foreign investment and agreed to comply with the WTO Trade Related Investment Measures agreement. The agreement thereby expanded the space of applicable contracts. In particular, it allowed U.S. firms to side step working with Chinese partners. Specifically, China ceased to impose trade and foreign exchange balancing requirements, local content requirements, which require foreign firms to use domestic materials and parts for production, and export performance requirements. China committed that U.S firms are not required to work with domestic suppliers or distributors, and withdrew requirements of any kind such as offsets, transfer of technology, production processes, or the conduct of R&D in China. Furthermore, China committed to ensure fair competition between private and state-invested enterprises and liberalize distribution services. Overall, the agreement secured that China is moving toward “rule of law” and will be held accountable for the contracts that it makes (Charlene Barshefsky, 18 November 1999). The agreement therefore increased the share of the profits from Chinese operations accruing to U.S. firms, effectively reducing U.S. firms’ labor cost.

IV

Sample Construction and Summary Statistics

From 10-K filings, we hand collect information on Compustat firms with subsidiaries in low-wage Asian countries as of 1997, the first year of comprehensive reporting of this information. The treated group consists of firms with a subsidiary in China, and the control group consists of firms with subsidiaries in low-wage Asian countries other than China. We pick November 15, 1999, that is, the date when the U.S.-China bilateral agreement is signed, as the date of the event. Our sample period starts in 1995 and ends in 2004 thereby using 10 years of data around the event. Our main dependent variable Share of Process Innovationsit is defined for firm-years with at least one patent and it provides a meaningful measure of the changes in the processproduct innovation mix over time only for firms with a nontrivial number of patents. Our main results are therefore based on a sample of high-patenting firms, which we define as firms that applied for at least 100 patents with the USPTO during our sample period. This

– 10 –

threshold corresponds to the 15th percentile of the firms’ patent distribution. We drop this sample selection restriction and use different thresholds in our robustness checks. The majority of our sample firms are manufacturing firms (SIC 20-39, 87% of firms) followed by services (SIC 70-89, 9.5% of firms), while the remaining 3.5% of firms are evenly distributed across the remaining industries. Table 1 provides summary statistics of 282,337 patents in our sample. On average, a patent has 20.4 claims, of which 7.9 are process, 12.5 are product, 3.5 are independent, and 16.9 are dependent. These statistics are similar for patents of treated and control firms, respectively.7 Table 2 provides summary statistics of our sample firms’ characteristics. On average, our sample firms’ share of process to total innovations is about 33%, based on both the claim and the patent based measures, as well as when we use the citationweighted measures. On average, a firm in our sample has assets $10.9 billion, sales $8.4 billion, capital expenditures $17 thousands per employee, and profits $47.4 thousands per employee. It also has sales growth of 8.8% and the market to book equity ratio of 4.6. All firm-level variables are winsorized at the 1% level before all analyses. Table 2 further provides summary statistics separately for the treated and control firms computed in 1998, that is, the year before the onset of treatment. We find no significant differences between the treated and control firms suggesting that the two groups are similar in terms of observable characteristics pre-treatment.

V V.1

Main Results Effect of the 1999 U.S.-China bilateral agreement on the process-product innovation mix

To identify the effect of an improved access to offshore labor on firms’ process-product innovation mix, we estimate the following difference-in-differences regression

yi,t = αi + βt + δ · Agreement(t>1999) · Chinai + γ · Xi,t−1 + i,t ,

(1)

7 Table IA-A1 in the Internet Appendix shows that a typical patent in our sample closely resembles a typical patent issued by the USPTO.

– 11 –

where i and t index firms and years, respectively; yi,t stands for the Share of Process Innovationsit , Process Innovationsit , or Product Innovationsit ; Agreement(t>1999) is an indicator variable that takes a value of one in the post-1999 period; Chinai is an indicator variable that takes a value of one for firms in the treated group (49% of firms in our sample); Xi,t−1 are time-varying firm-level control variables lagged by one year; αi and βt denote firm and year fixed effects, respectively; and i,t is the error term. Coefficient δ captures the change in the dependent variable at firms with a subsidiary in China as of 1997 following the 1999 U.S.-China bilateral agreement as compared to years before the agreement, relative to firms with subsidiaries in low-wage Asian countries other than China.8 In Columns 1-3 of Table 3, we present estimates of regression (1) with Share of Process Innovationsit as the dependent variable. The specification in Column 1, which does not include any firm-level control variables, shows that the treated firms decrease the share of process innovations relative to control firms post-1999 by 4 percentage points, which is a 12% reduction relative to the median ratio in the sample. The estimate of coefficient δ is significant at the 1% level. In Column 2, we additionally control for time-varying firm-level variables: the natural logarithm of sales and the market-to-book equity ratio. In Column 3, we add interacted year and two-digit SIC industry fixed effects to account for any timevarying industry-level omitted variables. We show that including these additional controls has little impact on the magnitude and significance of our δ estimate. These results suggest that our findings are not driven by differences in size, investment opportunities, or industry trends between the two groups of firms. The reduction in the ratio of process to total innovations we document may be due to less process innovations, more product innovations, or process and product innovations changing at different rates. The agreement alleviated hold-up problems of U.S. firms operating in China, allowing them to benefit more from the abundant and cheap Chinese labor and thereby reducing U.S. firms’ effective labor costs. To the extent that firms have fewer incentives to invest in R&D to reduce production cost, we should find that a lower share 8

Variables Agreement(t>1999) and Chinai are absorbed by the fixed effects and their coefficients are thus not estimated.

– 12 –

of process innovations is due to less process innovations. To this end, Columns 4-9 of Table 3 examine the effect of the agreement on the quantities of process and product innovations separately. The dependent variable is Process Innovationsit in Columns 4-6 and Product Innovationsit in Columns 7-9. All columns include firm and year fixed effects and control for the overall intensity of firms’ innovation using the logarithm of one plus the number of patents in each firm-year. Columns 5-6 and 8-9 additionally control for firm size and the market to book ratio, while Columns 6 and 9 also add interacted year and two-digit SIC industry fixed effects. We find that the quantity of process innovations decreases after the agreement. The estimate of δ, significant at the 1% level across all specifications, shows a 25% reduction in the quantity of process innovations (Column 6). On the contrary, the quantity of product innovations does not change as δ estimates are close to zero and not statistically significant. The effects of the 1999 U.S.-China bilateral agreement on corporate innovation we estimate are consistent with the argument that U.S. firms’ improved ability to utilize cheap offshore labor decreases the return on investing in process innovation, namely, innovation that is aimed at reducing production cost. U.S. firms invest in China to take advantage of lower labor cost. The hourly average factory-worker wage in China was $0.5 in 2000 versus $16.6 in the U.S. (a ratio of 0.03), while the same ratio is 0.04 in 2005, the final year in our sample.9 Our results can also be interpreted in light of the real options literature that uncertainty creates an opportunity cost of investing today in the form of a positive option value of waiting (Dixit and Pindyck, 1994). Lower uncertainty over input costs, following the agreement, eliminates U.S. firms’ option value from delaying changes in their 9

See Exhibit 1 in a Boston Consulting Group report: “Why manufacturing will return to the U.S.”. Multinational Monitor comments on the agreement: “U.S. businesses want the right to exploit its [China’s] cheap labor, or at least to import goods made in China with cheap labor.” Porter and Rivkin (2012) asked 10,000 Harvard alumni running businesses what are the main reasons for moving production out of the U.S. 70% of the respondents mention lower wage rates as the main reason for moving existing activities out of the U.S. When the same respondents were asked which are the countries they consider transferring their production to, China was the most common response (42% of the answers).

– 13 –

innovation mix (Pindyck, 1993).10 In the Internet Appendix, we show that our main results are robust to changes in the sample construction, refinements of the process-product innovation mix measures, and alternative estimation techniques. Specifically, our main results are robust to: i) using different thresholds to define high-patenting firms and to dropping the sample restriction on high-patenting firms altogether (Table IA-C1); ii) using measures of product-process innovation mix that are based on independent claims only, are adjusted for citation-weighting, or are defined at the patent level (Table IA-C2); iii) normalizing the quantities of process and product innovations by R&D expenditure or employment (Table IA-C3); iv) estimating the effect on the quantities of process and product innovations by the Negative binomial count data model and when dropping observations with zero process or product claim counts (Table IA-C4); v) estimating the effect on the share of process innovations using the Fractional response model (Table IA-C5).

V.2

Selection into treatment

U.S. firms may not randomly establish subsidiaries in China, but rather choose in which country to move their production to. This raises an identification concern as ex-ante differences in observable and unobservable firm characteristics between the “treatment” and “control” groups may lead to differential intention-to-treat. In light of this concern, we condition our control group on firms with subsidiaries in low-wage Asian countries, except China. Firms that operate in these countries have arguably similar incentives to move their production outside of the U.S. and are also subject to similar shocks (e.g., technology, productivity, demand) that are common among countries in similar geographies and, specifically, Asia. We perform multiple analyses to further mitigate the selection concern. 10

“U.S. companies expect to benefit from billions of dollars in new business and an end to years of uncertainty in which they had put off major decisions about investing in China. The business relationship has grown rapidly but remains lopsided, partly because of Chinese market restrictions and partly because of the vast discrepancy in wealth between the countries.” (The New York Times, September 2000).

– 14 –

Pre-treatment trends: Figure 1 plots, for treated and control firms separately, the within-firm average share and quantity of process innovations around the time of the agreement. Both variables follow an upward trend that is common for treated and control firms prior to 1999. That trend breaks for treated firms after 1999, the year of the agreement, while it keeps following the same trend for control firms. To test for the presence of pre-trends using regression analysis, we include the interaction term of Chinai with an indicator variable that takes a value of one in year 1999 into the specification in Column 3 of Table 3. The result, reported in Column 1 of Table 4, shows that δ estimate remains largely unchanged and the estimated coefficient on the new term is positive and statistically insignificant. We obtain similar results in Columns 3 and 5 of Table 4, where we examine the quantities of process and product innovations, respectively. This evidence shows that the treated and control firms do not have different shares and levels of process and product innovations pre-treatment. In Columns 2, 4, and 6 of Table 4, we estimate a further augmented version of equation (1) where we interact Chinai with an indicator variable for each year t.11 In Column 2 for the share of process innovations, we find that no interaction term is significant pretreatment; the estimate is in fact positive in 1999, the year the agreement is signed. The estimate becomes negative in 2000 and is close to significance with a p-value of 0.103, and the estimates remain negative and significant both economically (large fairly stable magnitudes between 3.4% and 4.7%) and statistically (significant at 10% in 2001, significant at 5% in 2002, p-value of .11 in 2003, and significant at 5% in 2004). We find similar results in Column 4 for the quantity of process innovations where coefficients are significant at the 5% or 1% level post 1999, while no coefficient estimate is statistically significant when we look at the quantity of product innovations in Column 6. These results show that our results are not due to differential pre-treatment trends. 11 We omit the 1996 interaction term and thus set 1996 as the baseline year. Note that year 1995 is dropped because we lag control variables by one year.

– 15 –

Placebo test: To alleviate the selection concern further, we conduct a placebo test by randomly assigning firms from our sample into the ‘placebo treatment’ group in 1999 (matching the 49% treated firms) and repeating our baseline analyses in Table 3. We repeat this procedure 1,000 times, each time estimating the ‘placebo treatment’ effect. We report the distribution of the obtained estimates in Figure 2, and summarize the results in Table 5. Table 5, Column 1, shows that the average of the placebo treatment coefficients is 0.0026 and the standard deviation of these coefficients is 0.0119. This result suggests that the true coefficient estimate of -0.0402 reported in Table 3 is a very unlikely event. Specifically, we find that 100% of the coefficients obtained from the placebo treatment are above the true coefficient estimate in Column 1, Table 3. The results in other columns of Table 5 are similar. Overall, we show that our estimated true treatment coefficients are always in the very left tail of the generated distributions, suggesting that non-random location of subsidiaries across countries is unlikely to explain our findings.

Mean-reversion in firms’ innovation activities: A related concern might be that treated firms’ shares of process innovations mean-revert to some firm-specific equilibrium levels post 1999, which is captured by our interaction term. We address this concern in Table IA-C6 in the Internet Appendix. We interact the values of the dependent variables in 1997 (Columns 1, 3, 5) and the value of the number of patents (log-transformed) in 1997 (Columns 2, 4, 6) with the full set of year indicator variables, and add these interaction terms to the specification in Column 3, Table 3. The estimates of δ are almost identical to our baseline results.12 We conclude that mean reversion in firms’ pre-treatment innovation activities cannot explain our findings.

Allowing for entry into China: Since entry into China is possibly endogenous to the agreement, we define Chinai in 1997—two years before the agreement is signed— throughout our main analyses. To the extent that all U.S. firms with a presence in China, including those that enter China after 1999, benefit from the agreement, we repeat the 12

We get very similar results when we interact instead values of the dependent variables in 1998 instead of 1997.

– 16 –

analyses from Table 3 using a time-varying measure of having presence in China. Specifically, we construct an indicator variable Chinait that takes a value of one if a firm has a subsidiary in China in year t according to its 10-K filings (30% of our control firms enter China in 1999 or later), and use it in the interaction with Agreement(t>1999) . The results, reported in Table IA-C7 in the Internet Appendix, are analogous to our baseline results. The negative and significant estimates of δ for the ratio and quantity of process innovations are slightly bigger in magnitude compared to those reported in Table 3.

V.3

Supply of product innovations and capital investment

A process innovation is aimed at improving a firm’s own production methods, while a product innovation is an improvement sold to end consumers in the form of final goods or to downstream firms in the form of intermediate capital goods used in their own production processes (Scherer, 1984). Our measure of (internally produced) process innovations does not capture firms’ overall investments in process innovation, because it does not include improvements in production processes that occur due to purchases of innovations embedded in the new intermediate capital goods from suppliers. According to this broader view of process innovation, the estimates in Table 3 give a lower bound on the effect of the agreement on firms’ overall, that is, internally produced and externally sourced, investment in process innovation. In this section, we perform additional analyses to explore the importance of externally sourced process innovations. To this end, we first examine whether our baseline results capture the substitution between internally produced process innovations and purchases of product innovations embedded in intermediate capital goods from suppliers, and, second, we ask how the agreement affects firms’ total investment in purchases of capital goods. To examine the degree of substitution, we define an industry-level variable Supply of Product Innovationsjt to capture the flow of product innovations from originating industries to using industries. We use the 2002 U.S. Benchmark Input-Output Accounts to construct weights based on the share of outputs of using industries that are due to the supply of inputs from originating industries (Acemoglu et al., 2012). Assuming that the

– 17 –

same weights apply to the supply of product innovations embedded in intermediate goods (Delgado and Mills, 2017), we multiply the weights with the quantity of product innovations (lagged by one year) of firms in the originating industries and compute the weighted sum of product innovations for each using industry in each year. In Table 6, we estimate our baseline regressions additionally controlling for the industrylevel supply of product innovations to our sample firms. In all specifications, the estimates of δ are very similar to those reported in Table 3, both in terms of economic magnitude and statistical significance. This evidence shows that changes in the technological content of the intermediate capital goods is not affecting our results, which suggests that our results cannot be explained by the substitution between internally produced and externally sourced process innovations. Next, we examine the effect of the agreement on U.S firms’ total investments in purchases of capital goods. We measure capital investment as the logarithm of capital expenditure over employment in Table 7, Columns 1-2, and as the logarithm of capital expenditure over total assets in Table 7, Columns 3-4. We find a negative and statistically significant effect of the agreement on capital investment: treated firms’ capital investment is relatively lower by about 10% following the agreement, irrespective of the measure and specification we use. These results support the view that the reduction in the quantity of process innovations obtains due to the labor cost channel: treated firms have relatively lower incentives to invest in production cost saving technologies, leading to their lower capital intensity.

VI

Alternative Explanations

In this section, we discuss and subsequently refute alternative explanations of our findings. First, we show that our results are not driven by the response of U.S. firms to increasing Chinese import competition or due to U.S. exports to China. Second, we show that changes in U.S. firms’ patenting practices cannot explain our findings. Third, we show that our results cannot be explained by the burst of the “Dot-com bubble” in 2000/2001, or by U.S. firms refocusing R&D on different technology fields or functional areas following the

– 18 –

agreement.

VI.1

U.S. trade with China

Increases in import competition from low-wage countries may impact firms’ innovation. Specifically, a reduction in the profitability of making low-tech products due to cheaper imports may give U.S. firms stronger incentives to innovate new goods and climb the quality ladder in order to escape import competition. Bernard, Redding, and Schott (2011) show that a reduction of trade costs with a low-wage country leads to a change in the product mix offered by Northern firms toward more high-tech products. Bloom, Draka, and Van Reenen (2015) examine the effect of import competition on innovation and find a positive effect for firms affected by Chinese imports. Hombert and Matray (2016) show that R&D intensive firms are able to differentiate their products to escape from Chinese competition and are more resilient to Chinese imports. Therefore, a possible concern is that our results capture U.S. firms’ differential responses to accelerating Chinese imports triggered by the agreement. Inconsistent with this alternative explanation, we show in Table 3 that there is no differential effect of the agreement on the quantity of product innovations of treated firms. To further rule out this alternative explanation, in Table 8 we add in our baseline specification variable Importjt and interact this variable with Agreement(t>1999) . Following Bernard, Jensen, and Schott (2006), Importjt is measured, for each manufacturing 4-digit SIC industry, as the level of lagged Chinese import penetration in the U.S. in Column 1 and as the logarithmic growth rate of Chinese import penetration in the U.S. in Columns 2 and 4-6. We show that the estimated coefficient on the new interaction term is positive and not statistically significant in all columns, while the estimates of δ remain negative and statistically significant for the share and quantity of process innovations. The magnitude of the effect of the agreement on the process-product innovation mix is in fact larger compared to the baseline estimates reported in Table 3. This evidence shows that the differential response of U.S. firms to increasing Chinese import competition is not driving our results. A related argument is that removal of trade barriers increases market size and induces

– 19 –

firms to innovate by reducing the fixed cost of innovation (Krugman, 1980; Grossman and Helpman, 1991, 1992; Lileeva and Trefler, 2010). Specifically, firms may pursue more product innovation in order to adapt to local product markets and cater to new consumers (Utterback and Abernathy, 1975; Klepper, 1996; Mitchell and Skrzypacz, 2014). Therefore, a possible concern is that our results capture U.S. firms’ differential responses to an improved access to the Chinese large and rapidly developing market due to the agreement. Again, we note that our results on the quantity of product innovations are inconsistent with this alternative explanation. To further rule out this possibility, in Table 8 we add in our baseline specification variable Exportjt and interact this variable with Agreement(t>1999) . Following Schott (2008), Exportjt is measured, for each manufacturing 4-digit SIC industry, as the logarithmic growth rate of exports from the U.S. to China. In all specifications we consider, the estimated coefficient on this interaction term is not statistically significant, while the estimates of δ remain negative and statistically significant for the share and quantity of process innovations. We conclude that our findings cannot be explained by the response of U.S. firms to increasing Chinese import competition or due to their improved market access to China following the agreement.

VI.2

Patenting practices

We examine three possible alternative explanations of our results that are related to changes in U.S. firms’ patenting practices. First, U.S. firms may transfer their R&D centers to China following the agreement. This transfer may be more likely for firms with subsidiaries in China, for example, due to positive knowledge spillovers between the production facilities in China and nearby R&D centers. Under this scenario, we may observe an increase in (process) patenting activity by treated firms’ Chinese subsidiaries following the agreement, which compensates for a decrease in (process) patenting activity by the treated firms. To address this possibility, we hand-collect information on the number of patents applied for by the treated firms’ Chinese subsidiaries over the 1999-2012 period. When we look for such patents applied for at the USPTO, we are unable to find any. Next, we look for patents applied for at the Chinese State Intellectual Property Office (CSIPO). We find that

– 20 –

55% of the subsidiaries in our sample do not have any CSIPO patents. For the remainder 45% of the subsidiaries, we compute the ratio of the total count of patents filed by the subsidiary over the total count of patents filed by its U.S. parent firm. The median ratio is 0.003. The small magnitude of the ratio shows that this alternative explanation cannot be driving our results. Second, it may be that trade secrets substitute for process innovations. This is possible, for example, if treated firms transfer more of their production to China following the agreement, which may elevate concerns regarding China’s weak intellectual property rights protection. Such concerns may be particularly relevant for process innovations since these innovations are easier to steal or less enforceable (Levin et al., 1987). To address this possibility, following Ang, Cheng, and Wu (2014), we exploit the crosssectional variation in the degree of enforcement of intellectual property rights across Chinese provinces. We collect information on subsidiaries’ locations from the 2001 Survey of Foreign Invested Enterprises (FIEs) conducted by the National Bureau of Statistics in China.13 We interact our treatment variable with an ordinal variable decreasing in the intellectual property rights protection at the subsidiaries’ locations. In Table 9, we find no statistically or economically significant differential effect of the agreement on the share and quantity of process innovations for firms whose subsidiaries are located in provinces with different intellectual property rights enforcement.14 This evidence suggests that substitution between process patents and trade secrets is unlikely to explain our results. Third, it may be that the change in the mix of patented innovations we document reflects changes in patent quality. For example, due to aggravated secrecy considerations, treated firms may patent only process innovations that absolutely need to be patented, 13 The survey covers FIEs set up by U.S. investors in China that account for 75% of the total number of U.S. FIEs operating in China in 2001 as reported by China Statistical Yearbook 2002 (Du, Lu, and Tao, 2008). The survey is available only in Chinese. We translate it into English and hand-match to our Compustat sample. 14

The differences in enforcement across provinces affect (Chinese) firms’ financing and investment (Ang, Cheng, and Wu, 2014) and R&D and innovation (Fang, Lerner, and Wu, 2016), suggesting that they are meaningful.

– 21 –

resulting in a lower number of process innovations with higher average quality. To address this possibility, in Table 10, we estimate our baseline regressions using citation-weighted measures. Columns 1-6 show that the estimates of δ remain negative and statistically significant, and are similar in magnitude to those reported in Table 3. These results suggests that there is no change in patent quality. We reach the same conclusion when we instead estimate the effect of the agreement on the number of citations per process or product innovation in Table 10, Columns 7-8. Overall, we conclude that our findings cannot be explained by changes in U.S. firms’ patenting practice following the agreement.

VI.3

Technology focus

An important event occurring shortly after the date of the agreement is the burst of the so called “Dot-com bubble” in 2000/2001. Around that time, there was a considerable legal uncertainty concerning patents granted for business methods implemented in software. These events may shifted U.S. firms’ innovation efforts away from such types of patents, affecting patenting of process innovations differentially across firms. To address this possibility, we identify patent technology classes broadly related to software and define software-related patents to be those with at least a single reference to these technology classes.15 We then compute the share of software-related patents in each firm-year. The median share of such patents in our sample is 2.46%. In Table 11 Panel A, we repeat our baseline analysis when dropping firms with at least one software-related patent in 2000, the peak of the bubble. The results remain strong in statistical significance with somewhat bigger magnitudes.16 Relatedly, it may be that U.S. firms refocus their R&D more broadly on different technology fields or functional areas, which may affect patenting of process innovations of treated firms for reasons unrelated to the agreement. To address this possibility, we test 15

Cooperative Patent Classification (CPC) classes “G06: Computing; Calculating; Counting”, “G11: Information Storage”, and “H04: Electric Communication Technique”. 16

The sample is considerably reduced given the stringent criterion to drop any firm with any one patent citing even a single software-related technology class.

– 22 –

the effect of the agreement on a normalized version of our dependent variables where we scale process/product claim counts by the average number of process/product claims taken across all firms that applied for at least one patent in the same technology class in the same year (Bena and Li, 2014). Table 11, Panel B, shows that our results are analogous to those reported in Table 3. We conclude that our findings cannot be explained by the burst of the “Dot-com bubble” in 2000/2001, or by changes in U.S. firms’ focus on different technology fields or functional areas following the agreement.

VII

Heterogeneous Treatment Effects

In this section, we exploit variation within treated firms to highlight the underlying mechanism explaining our findings. First, we find a weaker response to the agreement for the firms that expect to pay higher wage bills in China. Second, we show that the negative effect on the treated firms is more pronounced when the U.S. firm’s equity share in its Chinese subsidiary is higher.

VII.1

Wage bill of Chinese subsidiaries

We argue that the effect we are identifying is driven by U.S. firms utilizing low cost labor in China. This argument suggests that the effect of the agreement should be attenuated for treated firms whose subsidiaries in China expect to pay relatively higher wages. To proxy for the expectation of future wage bills of Chinese subsidiaries, we create a time-invariant indicator Wage Billi that takes a value of one if the number of employees at the U.S. firm’s subsidiary in China at the time of its registration is higher than the sample median and also the minimum wage growth rate in the county where the subsidiary is located in the year prior to the agreement is higher than the sample median, and is zero otherwise. Wage Billi varies across treated firms and is zero for control firms. Information on minimum wages is from Huang, Loungani, and Wang (2015).17 Information on employment and location of U.S. firms’ subsidiaries in China is from the 2001 Survey of Foreign Invested Enterprises 17

The data are originally collected by the Ministry of Human Resources and Social Security in China from official reports of local governments.

– 23 –

(FIEs) conducted by the National Bureau of Statistics in China. In Table 12, we augment our baseline specifications with the interaction term of our treatment variable with Wage Billi . We find that the differential effect on the share of process innovations is positive and statistically significant at the 5% or 10% level (Columns 1-2). The differential effect on the quantity of process innovations is also positive and it is statistically significant at the 5% level (Columns 3-4). These results show that U.S. firms that expect wages of their subsidiaries to increase by more cut their process innovation activities by less, which is consistent with the effect being driven by the labor cost channel.

VII.2

Equity share in Chinese subsidiaries

We argue that the effect we are identifying operates through the ability of U.S. firms to capture a higher portion of the profits from their Chinese operations vis-à-vis the Chinese partners due to the agreement, increasing U.S. firms’ benefit from a lower labor cost in China. This argument suggests that the effect of the agreement should be more pronounced for treated firms that have a higher equity stake in their Chinese subsidiaries relative to the Chinese partners. We collect information on equity stakes of U.S. and Chinese parties in Chinese subsidiaries from FIEs. Using this data, we define a time-invariant variable Equity Ratioi as the U.S. firm’s capital invested in the Chinese subsidiary divided by the capital provided by the Chinese party at the subsidiary’s registration. Equity Ratioi varies across treated firms and is zero for control firms, and ranges from 0.05 (1st percentile) to 382 (99th percentile) with median equal to four. In Table 13, we augment our baseline specifications with the interaction term of our treatment variable with Equity Ratioi . Columns 1-2 show that the differential effect on the share of process innovations is negative and significant at the 1% or 5% level. The effect is also economically large. If the ratio of invested capital at registration increases from 1 (10th percentile) to 100 (90th percentile), the share of process innovations decreases by three percentage points (Column 2). The differential effect on the quantity of process innovations is also negative and statistically significant at the 5% level in Column 3; it is not statistically significant in Column 4 although it remains economically important. If the ratio

– 24 –

increases from 1 to 100, the quantity of process innovations decreases by 3% (Column 4).18 These results are consistent with the argument that the agreement increased the share of the profits U.S. firms capture post-1999 from their Chinese operations, effectively reducing their labor cost and thereby changing their process-product innovation mix.

VIII

Alternative Empirical Setting

In this section, we show that staggered removals of ownership restrictions on foreign investment imposed by the Chinese government across industries are associated with a lower share and quantity of process innovations, while there is no such association with the quantity of product innovations. The results from this alternative empirical approach thus reaffirm our baseline findings.

VIII.1

Ownership restrictions on foreign investment

Ownership restrictions on foreign investment, namely caps on the share of equity held by foreign investors in Chinese joint-ventures, constitute a major friction that affects how the profits of U.S. firms’ Chinese subsidiaries are split between the U.S. vis-à-vis the Chinese partners.

The restrictions are published in the

Catalogue of Industries Guiding Foreign Investment issued jointly by the National Development and Reform Commission (NDRC) and the Ministry of Commerce (MOFCOM), China’s governing bodies on economic development and trade and investment policy, respectively, and are an integral part of Chinese longstanding industrial policy. The 1999 U.S.-China bilateral agreement improved upon doing business in China, nevertheless, the ownership restrictions remain throughout the 2000s. The Catalogue was first published in 1995. Since then, it was revised five times: in 1997, 2002, 2004, 2007, and 2011. For each industry, the Catalogue indicates whether there are restrictions on foreign shareholdings by requiring specific types of foreign investments or by capping the percentage of equity held by foreign investors. Industry sectors not included in 18 We obtain similar results when we instead use the ratio of cumulated amount of invested capital at the time of the survey.

– 25 –

the Catalogue are “permitted”, as outlined in the Regulation on Guiding Foreign Investment Direction (State Council Order 346), and no ownership restrictions apply. Sectors included in the Catalogue are “encouraged”, “restricted”, or “prohibited”. “Restricted” sectors are subject to ownership restrictions. “Encouraged” sectors can be either “permitted”, and thus no ownership restrictions apply, or “restricted”, which are subject to restrictions, but enjoy easier approval procedures. No investment is allowed in “prohibited” sectors. Across all revisions, the structure of the Catalogue remains the same. We map industry descriptions used in the Catalogue into the 4-digit NAICS industry classification.19 We then group 4-digit NAICS industries into two categories: “permitted” industries that are not subject to ownership restrictions (permitted and encouragedpermitted sectors according to the Catalogue) and “restricted” industries that are subject to such restrictions (restricted, encouraged-restricted, and prohibited sectors according to the Catalogue). We create a dummy variable which takes a value of one if an industry is not subject to ownership restrictions for each year between the year of issue of the Catalogue and the year of issue of the next Catalogue, and zero if such restrictions are in effect. We end up with time-series information on ownership restrictions for 58 industries between 1995, the year of the first Catalogue, and 2012, the last year of our innovation data. Figure 3 presents the share of industries in our sample that are not subject to restrictions in each year the Catalogue was issued. Consistent with the opening of China to foreign investors, the share of industries not subject to restrictions is increasing over time with the biggest change observed between 1997 and 2002, the period around China’s entry into the WTO. A change in an industry’s status from our restricted to permitted category has two implications for U.S. firms. First, for those firms that increase equity shares in their Chinese subsidiaries, there is a direct increase in the share of profits from the subsidiaries. Second, there is an increase in the bargaining power of U.S. firms vis-à-vis their Chinese partners, 19

The descriptions that do not match with 4-digit NAICS industries are dropped from the analysis. Assuming instead that the non-matched NAICS industries are not included in the Catalogue and are thus permitted does not qualitatively change the results.

– 26 –

which indirectly allows them to extract a higher share of the profits. Consider a (typical) example of a sector where the Chinese partner holds (as per the Catalogue) a controlling stake in a subsidiary. Similarly to the effects of the 1999 U.S.-China bilateral agreement, lifting the majority ownership restriction allows the U.S. firm to gain more control over the subsidiary and thereby eliminate the potential for hold-up problems and/or resolve contract incompleteness to its benefit. The removal of the ownership restrictions on foreign investment therefore implies a reduction in effective labor costs.

VIII.2

Removal of ownership restrictions on foreign investment and the process-product innovation mix

To assess how an improved access to offshore labor due to the removal of ownership restrictions on foreign investment imposed by the Chinese government changes U.S. firms’ process-product innovation mix, we estimate a difference-in-differences regression similar to the one introduced in Section V.1. The regression specification is yi,t =αi + βt + δ1 · No Ownership Restrictionsj,t · Chinai + (2) +δ2 · No Ownership Restrictionsj,t + γ · Xi,t−1 + i,t , where i, j, and t index firms, industries, and years, respectively; yi,t stands for the Share of Process Innovationsit , Process Innovationsit , or Product Innovationsit ; No Ownership Restrictionsjt is an indicator variable that takes a value of one if an industry is not subject to ownership restrictions at year t, and is zero otherwise; Chinai is an indicator variable that takes a value of one for firms in the treated group;20 Xi,t−1 are time-varying firm-level control variables lagged by one year; αi and βt denote firm and year fixed effects, respectively; and i,t is the error term. Coefficient δ1 captures the change in the dependent variable at U.S. high-patenting firms with a subsidiary in China as of 1997 (treated) operating in industries where the restrictions on foreign investment are lifted as compared to years when these restrictions are in effect, relative to U.S. high-patenting firms with subsidiaries in low-wage Asian countries other than China (control). 20

Since regression (2) exploits variation across industries and over time, we alternatively estimate a specification with variable Chinait , an indicator variable which takes a value of one for firms we identify as having a subsidiary in China at year t using the 10-K fillings. The results do not change.

– 27 –

Table 14 presents the estimates of regression (2). Column 1 includes firm fixed effects and year fixed effects. We find that the share of process innovations decreases by 5 percentage points at treated firms following the ownership restrictions removal, as compared to control firms. This estimate is statistically significant at the 5% level. In Column 2, we additionally control for firm sales and the market-to-book equity ratio. The estimate of δ1 remains statistically significant at the 5% level, and its magnitude is similar. In Columns 3-6, we present the results on the quantities of process and product innovations. Process innovations decrease considerably and these results are statistically significant at the 1% level. On the contrary, there is no association with product innovations. Overall, the results in Table 14 support our main findings that greater ability of U.S. firms to benefit from cheap and abundant Chinese labor lead to less process innovations.

IX

Conclusion

China’s integration into the global economy has been one the most important economic developments of recent decades. Among other things, this integration is manifested by a large flow of investment by foreign companies into China that want to tap local labor market (China received $1.9 trillion FDI over the 1995-2012 period according to the World Bank). In this paper, we examine how the availability of abundant, cheap Chinese labor affects corporate innovation. To answer this question, we construct a novel firm-level data set on process and product innovations using text-based analysis of patent claims. We show that a better ability to tap cheap offshore labor due to the 1999 U.S.-China bilateral agreement leads firms to shift their innovation mix toward less process innovation. We obtain the same result using the inter-temporal variation in ownership restrictions on foreign investment in China across industries. Our findings are consistent with arguments that a better ability to utilize cheap offshore labor decreases return on investing in labor-saving innovation, namely innovation substituting for more expensive U.S. labor. Currently, there is a heated debate on the effects of globalization, with one argument being that it leads to U.S. jobs being relocated to low wage countries. Our results suggest that stopping the “globalization of work” may

– 28 –

not lead to more jobs in the U.S. as the firms will respond by investing more in process innovation and thus rely less on labor overall.

– 29 –

References [1] Acemoglu, D., and D. Autor, 2011, “Skills, Tasks and Technologies: Implications for Employment and Earnings”, Handbook of Labor Economics Volume, Amsterdam: Elsevier, Orley Ashenfelter and David E. Card (eds.), 1043-1171. [2] Acemoglu, D., V. M. Carvalho, A. Ozdaglar, and A. Tahbaz-Salehi, 2012, “The Network Origins of Aggregate Fluctuations”, Econometrica, 80, 1977-2016. [3] Ang, J., Y. Cheng, and C. Wu, 2014, “Does Enforcement of Intellectual Property Rights Matter in China? Evidence from Financing and Investment Choices in the High Tech Industry”, Review of Economics and Statistics, 96, 332-348. [4] Antràs, P., 2005, “Incomplete Contracts and the Product Cycle”, American Economic Review, 95, 1054-1073. [5] Antràs, P., 2014, “Grossman–Hart (1986) Goes Global: Incomplete Contracts, Property Rights, and the International Organization of Production”, Journal of Law, Economics, and Organization, 30, 118-175. [6] Atanassov, J., and E. H. Kim, 2009, “Labor Laws and Corporate Governance: International Evidence from Restructuring Decisions”, Journal of Finance, 64, 341-374. [7] Autor, D., L. F. Katz, and M. S. Kearney, 2006, “The Polarization of the U.S. Labor Market”, American Economic Review Papers and Proceedings, 96, 189 - 194. [8] Autor, D., L. F. Katz, and M. S. Kearney, 2008, “Trends in U.S. Wage Inequality: Revising the Revisionists”, The Review of Economics and Statistics, 90, 300 - 323. [9] Autor, D., F. Levy and R. J. Murnane, 2003, “The Skill Content of Recent Technological Change: An Empirical Exploration”, Quarterly Journal of Economics, 118, 1279-1334. [10] Bena, J., A. M. Ferreira, P. Matos, and P. Pires, 2016, “Are Foreign Investors Locusts? The Long-Term Effects of Foreign Institutional Ownership”, Journal of Financial Economics, forthcoming. [11] Bena, J., and K. Li, 2014, “Corporate Innovations and Mergers and Acquisitions”, Journal of Finance, 69, 1923-1960. [12] Benmelech E., N. Bergman, and A. Seru, 2015, “Financing Labor”, Working Paper. [13] Benmelech E., C. Frydman, and D. Papanikolaou, 2017, “Financial Frictions and Employment During the Great Depression”, NBER Working Paper. [14] Bernard, A. B., J.B. Jensen, and P. Schott, 2006, “Survival of the Best Fit: Low Wage Competition and the (Uneven) Growth of US Manufacturing Plants”, Journal of International Economics, 68, 219-237.

– 30 –

[15] Bernard, A. B., S. J. Redding, and P. Schott, 2011, “Multi-Product Firms and Trade Liberalization”, Quarterly Journal of Economics, 126, 1271-1318. [16] Besley, T., and R. Burgess, 2004, “Can Labor Regulation Hinder Economic Performance? Evidence from India”, Quarterly Journal of Economics, 119, 91-134. [17] Blinder, 2006, “Offshoring: The Next Industrial Revolution?”, Foreign Affairs, 85, 113-128. [18] Blinder, A. S., 2009, “How Many U.S. Jobs Might Be Offshorable”, World Economics, 10, 41-78. [19] Blinder and Krueger, 2013, “Alternative Measures of Offshorability: A Survey Approach”, Journal of Labor Economics, 31, 97-128. [20] Bloom, N., M. Draka, and J. Van Reenen, 2015, “Trade Induced Technical Change: The Impact of Chinese Imports on Innovation, IT and Productivity”, Review of Economic Studies, forthcoming. [21] Cohen, W. M., and S. Klepper, 1996, “Firm Size and the Nature of Innovation within Industries: The Case of Process and Product R&D”, Review of Economics and Statistics, 78, 232-243. [22] Delgado, M., and K. G. Mills, 2017, “A New Categorization of the U.S. Economy: The Role of Supply Chain Industries in Performance”, Working Paper. [23] Devereaux, C., and R. Lawrence, 2004, “The Eagle and the Dragon: The November 1999 U.S.-China Bilateral Agreement and the Battle Over PNTR”, Harvard Case Study, HKS476. [24] Dixit, A. K., and R. S. Pindyck, 1994, “Investment Under Uncertainty”, Princeton University Press, Princeton. [25] Du, J., Y. Lu, and Z. Tao, 2008, “Economic Institutions and FDI Location Choice: Evidence from US Multinationals in China”, Journal of Comparative Economics, 36, 412-429. [26] Eswaran, M., and N. Gallini, 1996, “Patent Policy and the Direction of Technological Change”, The RAND Journal of Economics, 27, 722-746. [27] Fang, L. H., J. Lerner, and C. Wu, 2016, “Intellectual Property Rights Protection, Ownership, and Innovation: Evidence from China”, Harvard Business School Working Paper. [28] Giroud, X., and H. Mueller, 2015, “Firm Leverage, Consumer Demand, and Employment Losses during the Great Recession”Quarterly Journal of Economics, forthcoming. [29] Giroud, X., and H. Mueller, 2016, “Redistribution of Local Labor Market Shocks through Firms’ Internal Networks”, Working paper.

– 31 –

[30] Goos, M., A. Manning, and A. Salomons, 2014, “Explaining Job Polarization: RoutineBiased Technological Change and Offshoring”, American Economic Review, 104, 25092526. [31] Griliches, Z., 1990, “Patent Statistics as Economic Indicators: A Survey”, Journal of Economic Literature, 28, 1661-1707. [32] Grossman, S., and O. Hart, 1986, “The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Ownership”, Journal of Political Economy, 94, 691-719. [33] Grossman, G., and E. Helpman, 1991, “Quality Ladders and Product Cycles”, Quarterly Journal of Economics, 106, 557-586. [34] Grossman, G., and E. Helpman, 1992, “Innovation and Growth in the Global Economy”, Cambridge: MIT Press. [35] Hart, O., and J. Moore, 1990, “Property Rights and the Nature of the Firm”, Journal of Political Economy, 98, 1119-1158. [36] Hicks, J., 1932, “The Theory of Wages”, London: Macmillan. [37] Hombert, J., and A. Matray, 2016, “Can Innovation Help U.S. Manufacturing Firms Escape Import Competition from China?”, Working Paper. [38] Huang, Y., P. Loungani, and G. Wang, 2015, “Minimum Wages and Firm Employment: Evidence from China”, IMF Working Paper. [39] Klepper, S., 1996, “Association Entry, Exit, Growth, and Innovation over the Product Life Cycle”, American Economic Review, 86, 562-583. [40] Krugman, P., 1980, “Scale Economies, Product Differentiation, and the Pattern of Trade”, American Economic Review, 70, 950-959. [41] Levin, R. C., A. K. Klevorick, and R. R. Nelson, 1987, “Appropriating the Returns from Industrial R&D”, Brookings Papers on Economic Activity, 783-820. [42] Lileeva, A., and D. Trefler, 2010, “Improved Access to Foreign Markets Raises PlantLevel Productivity...for Some Plants”, Quarterly Journal of Economics, 125, 10511099. [43] Link, A. N., 1982, “A Disaggregated Analysis of Industrial R&D: Product versus Process Innovation”, in Devendra Sahal (ed.), The Transfer and Utilization of Technical Knowledge (Lexington, MA: Lexington Books). [44] Midler, P., 2009, “Poorly Made in China”, Hoboken, NJ: John Wiley & Sons, Inc. [45] Mitchell, M. and A. Skrzypacz, 2014, “A Theory of Market Pioneers”, Working Paper. [46] Moran, T. H., 2001, “Parental Supervision: The New Paradigm for Foreign Direct Investment and Development”, Institute for International Economics, Washington, D.C.

– 32 –

[47] Ouimet P., and R. Zarutskie, 2015, “Acquiring Labor”, Working Paper. [48] Pindyck, R. S., 1993, “Investments of Uncertain Costs”, Journal of Financial Economics, 31, 53-76. [49] Porter, M. E., and J. W. Rivkin, 2012, “Prosperity at Risk: Findings of Harvard Business School’s Survey on U.S. Competitivenes”, Report. [50] Rodrik, D., 2000, “How Far Will International Economic Integration Go?”, Journal of Economic Perspectives, 14, 177-186. [51] Scherer, F. M., 1982, “Inter-industry Technology Flows in the United States”, Research Policy, 11, 227-245. [52] Scherer, F. M., 1984, “Using Linked Patent and R&D Data to Measure Interindustry Technology Flows”, in Zvi Griliches (ed.), R&D, Patents, and Productivity (Chicago: University of Chicago Press for the National Bureau of Economic Research). [53] Schott, P., 2008, “The Relative Sophistication of Chinese Exports”, Economic Policy, 53, 5-49. [54] Tate, G., and L. Yang, 2016, “The Human Factor in Acquisitions: Cross-industry Labor Mobility and Corporate Diversification”, Working Paper. [55] Utterback, J. M., and W. J. Abernathy, 1975, “The Induced Innovation Hypothesis and Energy-Saving Technological Change”, OMEGA, 3, 639-656. [56] Williamson, O. E., 1979, “Transaction-Cost Economics: The Governance of Contractual Relations”, Journal of Law and Economics, 22, 233-261.

– 33 –

Share of Process Innovations

.03

.02

.01

0

-.01

-.02 1997

1998

1999

2000

Year

2001

Treated: Chinai = 1

2002

2003

2004

Control: Chinai = 0

Process Innovations

200

100

0

-100

-200 1997

1998

1999

2000

Year

Treated: Chinai = 1

2001

2002

2003

2004

Control: Chinai = 0

Figure 1. Share and quantity of process innovations around the 1999 U.S.China bilateral agreement This figure plots the within-firm variation in the average share of process innovations (top) and the average quantity of process innovations (bottom), net of changes in aggregate macroeconomic conditions, for the treated (solid line) and control (dotted line) firms. The vertical line indicates the time of the agreement.

– 34 –

.1

Fraction

.08

.06

.04

.02

0

-.04

-.03

-.02

-.01 0 .01 .02 Agreement(t>1999) . Pseudo-Chinai

.03

.04

-.2

-.15

-.1

-.05 0 .05 .1 Agreement(t>1999) . Pseudo-Chinai

.15

.2

.08

Fraction

.06

.04

.02

0

Figure 2. Histogram of placebo test coefficients This figure plots the histogram of estimated coefficients from 1,000 trials of our placebo test presented in Table 5. The top plot corresponds to the specification in Table 5 Column 1, while the bottom plot corresponds to the specification in Table 5 Column 4.

– 35 –

100% 90% 80%

70% 60% 50% 40% 30% 20% 10% 0% 1995

1997

2002

Restricted industries

2004

2007

2011

Permitted industries

Figure 3. Breakdown of “Permitted” and “Restricted” industries for each Foreign Investment Catalogue This figure shows the fraction of industries where foreign investment is subject to ownership restrictions (light grey) and those where investment is permitted without ownership restrictions (dark grey). The information is collected from the Catalogue of Industries Guiding Foreign Investment issued jointly by the National Development and Reform Commission (“NDRC”) and the Ministry of Commerce (“MOFCOM”) of China. The Catalogue was initially published in 1995 and was revised five times in 1997, 2002, 2004, 2007, and 2011.

– 36 –

Table 1: Process and product innovations This table reports summary statistics on patent claims for the set of patents applied for by the U.S. high patenting firms with subsidiaries in low-wage Asian countries (the baseline sample used in Table 3), and for the subsamples of the treated and control firms defined in the year prior to the U.S.-China bilateral agreement was signed. There are 282,337 patents in the baseline sample over the 1995-2004 period. Patent claims define, in technical terms, the scope of protection conferred by a patent and thus define what subject matter the patent protects. A process claim refers to an innovation that reduces production cost while a product claim refers to a new good. An independent claim stands on its own. In contrast, a dependent claim only has meaning when combined with a claim of the same patent it refers to.

Panel A: All Firms Mean

Standard Deviation

25th Percentile

50th Percentile

75th Percentile

Number of claims per patent

20.4

14.20

11

18

25

Number of process claims

7.88

9.92

0

5

12

Number of product claims

12.50

11.80

4

11

18

Number of independent claims

3.46

2.63

2

3

4

Number of dependent claims

16.90

13.00

9

15

21

Panel B: Treated Firms

Number of claims per patent

Mean

Standard Deviation

25th Percentile

50th Percentile

75th Percentile

19.70

13.60

11

18

25

Number of process claims

7.51

9.19

0

5

11

Number of product claims

12.20

11.30

4

10

18

Number of independent claims

3.32

2.47

2

3

4

Number of dependent claims

16.30

12.50

8

15

21

Panel C: Control Firms Mean

Standard Deviation

25th Percentile

50th Percentile

75th Percentile

21.90

15.50

12

19

27

Number of process claims

8.69

11.40

0

6

13

Number of product claims

13.30

12.80

4

11

19

Number of claims per patent

Number of independent claims

3.79

2.95

2

3

5

Number of dependent claims

18.20

14.20

9

16

23

– 37 –

Table 2: Summary statistics This table reports summary statistics of characteristics for the U.S. high patenting firms with subsidiaries in low-wage Asian countries (the baseline sample used in Table 3), and for the subsamples of the treated and control firms measured in the year prior to the U.S.-China bilateral agreement was signed. Treated firms are firms with a subsidiary in China as of 1997, while control firms are firms with a subsidiary in a low-wage Asian country, but not in China, as of the same year. The sample period is 1995-2004. For the baseline sample, Column 1 reports means, Column 2 reports standard deviations, and Columns 3-5 report 25th, 50th, and 75th percentiles, respectively. Columns 6 and 7 present means and standard errors, respectively, for the treated and control firms, as measured in 1998. Column 8 reports p-values from the t-test for the difference in means between the treated and control firms.

Mean

Standard Deviation

25th Percentile

50th Percentile

75th Percentile

Mean

All Firm-years (N=2,278)

Share of Process Innovations

0.330

0.177

0.203

0.337

0.441

– 38 –

0.311

0.216

0.154

0.281

0.435

Citation-weighted Share of Process Innovations

0.329

0.193

0.185

0.324

0.448

Citation-weighted Share of Process Innovations_Pat

0.304

0.231

0.122

0.267

0.450

Assets (mil. $)

10,922

32,171

875

2,339

7,893

Sales (mil. $)

8,429

20,388

808

1,997

7,078

Market to Book

0.088

4.59

0.223

5.26

-0.005

2.07

0.081

3.21

p-value of Difference

In Year 1998

Share of Process Innovations_Pat

Sales Growth

Standard Error

0.176

5.49

CAPEX/Employment (thous. $)

16.96

17.15

6.69

12.04

20.56

EBITDA/Employment (thous. $)

47.38

46.91

20.49

34.35

62.60

treated

0.325

(0.016)

control

0.323

(0.018)

treated

0.280

(0.019)

control

0.311

(0.022)

treated

0.335

(0.018)

control

0.331

(0.020)

treated

0.288

(0.021)

control

0.312

(0.023)

treated

10,650

(2,489)

control

8,068

(3,035)

treated

9,442

(1,681)

control

5,971

(1,731)

treated

0.071

(0.017)

control

0.084

(0.023)

treated

6.03

(0.64)

control

4.92

(0.58)

treated

17.78

(1.26)

control

17.08

(1.41)

treated

45.97

(3.64)

control

43.75

(4.64)

0.95

0.28

0.88

0.44

0.51

0.15

0.64

0.20

0.71

0.71

Table 3: 1999 U.S.-China bilateral agreement and process and product innovations This table reports results of OLS regressions of the share of process innovations (Columns 1-3) and the quantity of process (Columns 4-6) and product (Columns 7-9) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample includes U.S. high patenting firms with a subsidiary in a low-wage Asian country as of 1997. The sample period is 1995-2004. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). The levels of process and product innovations are log-transformed. Sales is log-transformed and lagged by one year. Market to Book is the ratio of the market value of equity over the book value of equity, log-transformed and lagged by one year. Patents is one plus the total number of patents applied for by a firm in a given year and it is log-transformed. All regressions include firm and year fixed effects. Columns 3, 6, and 9 also include 2-digit SIC interacted with year fixed effects. All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

Agreement(t>1999) · Chinai

(2)

(3)

Process Innovations (4)

(5)

Product Innovations (6)

(7)

(8)

(9)

– 39 –

-0.0402

-0.0370

-0.0397

-0.256

-0.244

-0.284

-0.003

-0.007

-0.033

(0.0123)***

(0.0128)***

(0.0136)***

(0.0648)***

(0.0644)***

(0.0712)***

(0.0411)

(0.0421)

(0.0446)

-0.0189

-0.0164

-0.0676

-0.0584

0.0200

0.0111

(0.0110)*

(0.0117)

(0.0478)

(0.0495)

(0.0388)

(0.0394)

0.0006

-0.0045

0.0359

0.0238

0.0160

0.0256

(0.0059)

(0.0058)

(0.0287)

(0.0308)

(0.0204)

(0.0185)

Sales

Market to Book

Patents

Firm FE

Yes

Yes

Year FE

Yes

Yes

Industry×Year FE

Yes

1.141

1.135

1.119

1.145

1.136

1.135

(0.0316)***

(0.0397)***

(0.0405)***

(0.0223)***

(0.0253)***

(0.0273)***

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.69

0.72

0.78

0.92

0.93

0.94

0.95

0.95

0.96

Obs.

2,278

1,940

1,940

2,278

1,940

1,940

2,278

1,940

1,940

Table 4: Pre-treatment trends This table reports results of OLS regressions of the share of process innovations (Columns 1-2) and the quantity of process (Columns 3-4) and product (Columns 5-6) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 3, 6, and 9, respectively. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). dt is an indicator variable for year t. All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

Agreement(t>1999) · Chinai

(2)

Process Innovations (3)

(5)

-0.0318

-0.259

-0.0553

(0.0133)**

(0.0608)***

(0.0445)

d1997 · Chinai

d1998 · Chinai

(6)

-0.0102

-0.118

0.0741

(0.0198)

(0.113)

(0.0652)

-0.0175

-0.181

0.0635

(0.0238) d1999 · Chinai

(4)

Product Innovations

(0.134)

(0.0738)

0.0288

0.0193

0.0918

-0.0106

-0.0797

-0.0330

(0.0206)

(0.0252)

(0.125)

(0.144)

(0.0709)

(0.0834)

d2000 · Chinai

d2001 · Chinai

d2002 · Chinai

d2003 · Chinai

d2004 · Chinai

-0.0343

-0.279

-0.0312

(0.0210)

(0.121)**

(0.0750)

-0.0449

-0.403

0.00105

(0.0239)*

(0.116)***

(0.0719)

-0.0435

-0.285

0.0148

(0.0217)**

(0.124)**

(0.0784)

-0.0361

-0.408

-0.0177

(0.0226)

(0.124)***

(0.0833)

-0.0476

-0.437

-0.00982

(0.0237)**

(0.138)***

(0.0790)

Firm-level Controls

Yes

Yes

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Industry×Year FE

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.78

0.78

0.94

0.94

0.96

0.96

Obs.

1,940

1,940

1,940

1,940

1,940

1,940

– 40 –

Table 5: Placebo test This table reports results of a placebo test that randomly assigns firms into 1,000 pseudo treated groups (Pseudo Chinai ) and estimates the OLS regressions of Table 3 for each such pseudo treated group. The sample and regression specifications correspond to those in Table 3. The coefficients and standard errors reported in this table are the averages and standard deviations of the estimated 1,000 regression coefficients. All firm-level variables are winsorized at the 1% level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations

Process Innovations

Product Innovations

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

0.0026

-0.0003

0.0004

0.0125

0.0027

-0.0034

-0.0065

-0.0022

0.0004

(0.0119)

(0.0121)

(0.0125)

(0.0647)

(0.0639)

(0.0663)

(0.0387)

(0.0406)

(0.0397)

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Agreement(t>1999) · Pseudo Chinai

– 41 –

Firm-level Controls

Industry×Year FE

Repetitions (in times)

Yes

1,000

1,000

1,000

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

1,000

1,000

1,000

Yes

1,000

1,000

1,000

Table 6: Supply of product innovations This table reports results of OLS regressions of the share of process innovations (Columns 1-3) and the quantity of process (Columns 4-6) and product (Columns 7-9) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3, except that we include additional industry-level control variable Supply of Product Innovationsjt . Supply of Product Innovationsjt is the weighted sum of product innovations of firms in the originating industries (lagged by one year) computed for each using industry in each year. Originating and using industry linkages are defined using the 2002 U.S. Benchmark Input-Output Accounts with weights being the shares of outputs of using industries that are due to the supply of inputs from originating industries. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

Agreement(t>1999) · Chinai

(2)

(3)

Process Innovations (4)

(5)

Product Innovations (6)

(7)

(8)

(9)

-0.0408

-0.0450

-0.283

-0.270

-0.311

0.004

-0.005

-0.026

(0.0138)***

(0.0152)***

(0.0679)***

(0.0666)***

(0.0771)***

(0.0445)

(0.0448)

(0.0491)

-0.0033

-0.0082

-0.0364

0.0821

0.0747

0.0374

0.0580

0.0786

0.140

(0.0139)

(0.0129)

(0.0197)*

(0.0482)*

(0.0461)

(0.1110)

(0.0453)

(0.0410)*

(0.0618)**

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

– 42 –

-0.0444 (0.0134)***

Supply of Product Innovationsjt

Firm-level Controls

Industry×Year FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.68

0.72

0.77

0.92

0.92

0.94

0.94

0.95

0.96

Obs.

2,119

1,809

1,809

2,119

1,809

1,809

2,119

1,809

1,809

Table 7: Capital investment This table reports results of OLS regressions of capital investment on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications are analogous to those in Table 3. Capital investment is defined as the logarithm of capital expenditure over employment in Columns 1-2 and as the logarithm of capital expenditure over total assets in Columns 3-4. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

CAPEX/Employment

CAPEX/Assets

(1)

(2)

(3)

(4)

-0.104

-0.124

-0.105

-0.119

(0.0568)*

(0.0657)*

(0.0597)*

(0.0690)*

Firm-level Controls

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Year FE

Yes

Agreement(t>1999) · Chinai

Yes

Industry×Year FE

Yes

Yes

R2

0.82

0.84

0.68

0.74

Obs.

1,793

1,793

1,803

1,803

– 43 –

Table 8: U.S. trade with China This table reports results of OLS regressions of the share of process innovations (Columns 1-4) and the quantity of process (Column 5) and product (Column 6) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 3, 6, and 9, respectively, except that we focus on manufacturing firms and include additional variables Importjt and Exportjt and interact those variables with Agreement(t>1999) . Following Bernard, Jensen, and Schott (2006), Importjt is measured, for each manufacturing 4-digit SIC industry, as the level of lagged Chinese import penetration in the U.S. in Column 1, and as the logarithmic growth rate of Chinese import penetration in the U.S. in Columns 2 and 4-6. Following Schott (2008), Exportjt is measured, for each manufacturing 4-digit SIC industry, as the logarithmic growth rate of exports from the U.S. to China. The growth rates of import penetration and exports are scaled by 1/100. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations

Agreement(t>1999) · Chinai

Agreement(t>1999) · Importjt

Product Innovations

(1)

(2)

(3)

(4)

(5)

(6)

-0.0426

-0.0428

-0.0443

-0.0429

-0.312

-0.0243

(0.0163)***

(0.0163)***

(0.0163)***

(0.0164)***

(0.0841)***

(0.0476)

0.0269

0.528

0.356

3.456

2.040

(0.0390)

(1.729)

(1.736)

(7.020)

(5.203)

-0.555

-0.200

6.401

5.166

(1.249)

(1.265)

(6.610)

(4.037)

-0.787

-4.115

0.466

Agreement(t>1999) · Exportjt

Importjt

Process Innovations

-0.204

-0.957

(0.164)

(0.496)*

Exportjt

(0.571)

(3.578)

(1.506)

1.038

0.696

-3.669

-3.788

(0.745)

(0.800)

(5.028)

(2.470)

Firm-level Controls

Yes

Yes

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Industry×Year FE

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.65

0.65

0.65

0.65

0.92

0.95

Obs.

1,281

1,259

1,267

1,259

1,259

1,259

– 44 –

Table 9: Trade secrets substituting for process innovations This table reports results of OLS regressions of the share of process innovations (Columns 1-2) and the quantity of process (Columns 3-4) and product (Columns 5-6) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 2-3, 5-6, and 8-9, respectively, except that we include the interaction of Agreement(t>1999) · Chinai with variable IP Ri as an additional explanatory variable. Following Ang, Cheng, and Wu (2014), IP Ri decreases in the degree of intellectual property rights enforcement across Chinese provinces. We collect information on geographic locations of U.S. firms’ Chinese subsidiaries from the 2001 Survey of Foreign Invested Enterprises conducted by the National Bureau of Statistics in China. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

Agreement(t>1999) · Chinai

Agreement(t>1999) · Chinai · IP Ri

(2)

Process Innovations (3)

(4)

Product Innovations (5)

(6)

-0.0291

-0.0286

-0.251

-0.286

-0.0434

-0.0690

(0.0168)*

(0.0166)*

(0.0786)***

(0.879)***

(0.0643)

(0.0567)

-0.0065

-0.0096

0.0057

0.0016

0.0296

0.0306

(0.0086)

(0.0090)

(0.0388)

(0.0452)

(0.0311)

(0.0275)

Yes

Yes

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Firm-level Controls

Yes

Industry×Year FE

Yes

Yes Yes

Yes

R2

0.72

0.78

0.93

0.94

0.95

0.96

Obs.

1,940

1,940

1,940

1,940

1,940

1,940

– 45 –

Table 10: Differential patent quality This table reports results of OLS regressions of the citation-weighted share of process innovations (Columns 1-2), the level of citation-weighted process innovations (Columns 3-4), the level of citation-weighted product innovations (Columns 5-6), the number of process citations (Column 7), and the number of product citations (Column 8) on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. We count citations received over a 3-year window starting with each patent’s application year. The levels of citation-weighted process and product innovations are log-transformed. The numbers of process and product citations are per patent and log-transformed. The sample and regression specifications correspond to those in Table 3 Columns 2-3, 5-6, and 8-9, respectively. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Citation-weighted Share of Process Innovations (1)

– 46 –

Agreement(t>1999) · Chinai

(2)

Citation-weighted Process Innovations (3)

(4)

Citation-weighted Product Innovations (5)

(6)

Citations of Process Innovations (7)

Citations of Product Innovations (8)

-0.0409

-0.0438

-0.237

-0.341

0.0453

-0.0325

-0.0364

-0.0077

(0.0144)***

(0.0151)***

(0.101)**

(0.107)***

(0.0713)

(0.0725)

(0.0670)

(0.0524)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Firm-level Controls

Industry×Year FE

Yes Yes

Yes Yes

R2

0.65

0.72

0.88

0.90

0.90

0.93

0.58

0.62

Obs.

1,940

1,940

1,940

1,940

1,940

1,940

1,919

1,936

Table 11: Technology focus This table reports results of OLS regressions of the share of process innovations (Columns 1-2) and the quantity of process (Columns 3-4) and product (Columns 5-6) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 2-3, 5-6, and 8-9, respectively, except that we exclude firms that applied for at least one patent in softwarerelated technology class in 2000 (Panel A), and we construct the dependent variables using technology-class-and-time-period adjusted counts of process and product innovations (Panel B). Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Panel A: Drop Software Patents Share of Process Innovations (1)

Agreement(t>1999) · Chinai

(2)

Process Innovations (3)

Product Innovations

(4)

(5)

(6)

-0.0552

-0.0689

-0.305

-0.363

0.0487

0.0280

(0.0233)**

(0.0242)***

(0.126)**

(0.153)**

(0.0777)

(0.0884)

Yes

Yes

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Firm-level Controls

Industry×Year FE

Yes Yes

Yes Yes

Yes

R2

0.73

0.82

0.87

0.90

0.87

0.93

Obs.

673

673

673

673

673

673

Panel B: Adjust for Technology Class and Time Period Adjusted Share of Process Innovations

Adjusted Process Innovations

Adjusted Product Innovations

(1)

(2)

(3)

(4)

(5)

(6)

-0.0469

-0.0533

-0.0858

-0.116

0.0276

-0.0162

(0.0149)***

(0.0159)***

(0.0401)**

(0.0416)***

(0.0365)

(0.0369)

Yes

Yes

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Agreement(t>1999) · Chinai

Firm-level Controls

Industry×Year FE

Yes Yes

Yes Yes

Yes

R2

0.68

0.74

0.94

0.95

0.94

0.95

Obs.

1,940

1,940

1,940

1,940

1,940

1,940

– 47 –

Table 12: Wage bill of U.S. firms’ Chinese subsidiaries This table reports results of OLS regressions of the share of process innovations (Columns 1-2) and the quantity of process innovations (Columns 3-4) on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 2-3 and 5-6, respectively, except that we include the interaction of Agreement(t>1999) · Chinai with variable Wage Billi as an additional explanatory variable. Wage Billi is an indicator variable equal to one if the number of employees at the U.S. firm’s subsidiary in China at the time of its registration is higher than the sample median and also the growth rate of the subsidiary’s county minimum wage in 1998 is larger than the sample median, and equal to zero otherwise. Information on minimum wages is from Huang, Loungani, and Wang (2015). Information on employment and location of U.S. firms’ subsidiaries in China is from the 2001 Survey of Foreign Invested Enterprises conducted by the National Bureau of Statistics in China. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

Agreement(t>1999) · Chinai

(2)

Process Innovations (3)

(4)

-0.0495

-0.0422

-0.313

-0.298

(0.0157)***

(0.0185)**

(0.0778)***

(0.0944)***

0.0413

0.0393

0.219

0.192

(0.0205)**

(0.0222)*

(0.102)**

(0.0933)**

Firm-level Controls

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Year FE

Yes

Agreement(t>1999) · Chinai · Wage Billi

Industry×Year FE

Yes Yes

Yes

R2

0.74

0.77

0.93

0.94

Obs.

1,449

1,449

1,449

1,449

– 48 –

Table 13: Equity share of U.S. firms in Chinese subsidiaries This table reports results of OLS regressions of the share of process innovations (Columns 1-2) and the quantity of process innovations (Columns 3-4) on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 2-3 and 5-6, respectively, except that we include the interaction of Agreement(t>1999) · Chinai with variable Equity Ratioi as an additional explanatory variable. Equity Ratioi is the U.S. firm’s capital invested in the Chinese subsidiary divided by the capital provided by the Chinese party to the subsidiary at the subsidiary’s registration. Information on capital investments at registration is from the 2001 Survey of Foreign Invested Enterprises conducted by the National Bureau of Statistics in China. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

Agreement(t>1999) · Chinai

Agreement(t>1999) · Chinai · Equity Ratioi

(2)

Process Innovations (3)

(4)

-0.0411

-0.0309

-0.229

-0.204

(0.0155)***

(0.0163)*

(0.0772)***

(0.0965)**

-0.0002

-0.0003

-0.0007

-0.0003

(0.0001)***

(0.0001)**

(0.0002)**

(0.0004)

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Year FE

Yes

Firm-level Controls

Industry×Year FE

Yes Yes

Yes

R2

0.75

0.79

0.92

0.93

Obs.

1,281

1,281

1,281

1,281

– 49 –

Table 14: Removal of ownership restrictions on foreign investment and process and product innovations This table reports results of OLS regressions of the share of process innovations (Columns 1-2) and the quantity of process (Columns 3-4) and product (Columns 5-6) innovations on firms that operate in industries with no restrictions on foreign ownership of Chinese subsidiaries imposed by the Chinese government (No Ownership Restrictionsjt ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 1-2, 4-5, and 7-8, respectively, except that the sample period is 1995-2012. No Ownership Restrictionsjt is an indicator variable equal to one if the U.S. firm’s industry, defined at the 4-digit NAICS level, is not subject to foreign ownership restrictions at a given year, and equal to zero otherwise. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

No Ownership Restrictionsjt · Chinai

– 50 – No Ownership Restrictionsjt

(2)

Process Innovations (3)

(4)

Product Innovations (5)

(6)

-0.0532

-0.0488

-0.444

-0.409

-0.0971

-0.0866

(0.0213)**

(0.0236)**

(0.128)***

(0.139)***

(0.0821)

(0.0858)

0.0330

0.0276

0.152

0.152

-0.0683

-0.0313

(0.0178)*

(0.0204)

(0.0948)

(0.107)

(0.0574)

(0.0629)

Firm-level Controls

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.56

0.59

0.89

0.89

0.90

0.91

Obs.

3,130

2,783

3,130

2,783

3,130

2,783

Appendix: Variable definitions

– 51 –

Variable

Definition

Process (Product) Innovationsit

Natural logarithm of one plus the number of process (product) claims included in patents applied for by firm i in year t. Source: The United States Patent and Trademark Office (USPTO).

Share of Process Innovationsit

Ratio of the number of process claims to the number of all claims included in patents applied for by firm i in year t. Source: USPTO.

Process (Product) Innovations_Independentit

Natural logarithm of one plus the number of independent process (product) claims included in patents applied for by firm i in year t. Source: USPTO.

Share of Process Innovations_Independentit

Ratio of the number of independent process claims to the number of all independent claims included in patents applied for by firm i in year t. Source: USPTO.

Process (Product) Innovations_Adjustedit

Natural logarithm of one plus the adjusted number of process (product) claims included in patents applied for by firm i in year t. The adjustment is done in three steps. First, for each technology class k and patent application year t, we compute the mean value of the number of process (product) claims in technology class k with application year t across all firms that were awarded at least one patent in technology class k with application year t. Second, we scale the number of process (product) claims of firm i in technology class k with application year t by the corresponding (technology class- and application year-specific) mean value from the first step. Third, for firm i, we sum the scaled number of claims from the second step across all technology classes in year t. We assign each claim proportionally to all technology classes listed on the claim’s patent. Technology classes are defined using the Cooperative Patent Classification (CPC) classification. Source: USPTO.

Share of Process Innovations_Adjustedit

Ratio of the adjusted number of process claims to the adjusted number of all claims included in patents applied for by firm i in year t. Source: USPTO.

Process Innovations_Patentit

Natural logarithm of one plus the sum of the number of process and process-apparatus patents (or, alternatively, the sum of the number of process, process-apparatus, and product-method patents) applied for by firm i in year t. Source: USPTO.

Share of Process Innovations_Patentit

Sum of the number of process and process-apparatus patents (or, alternatively, the sum of the number of process, process-apparatus, and product-method patents) divided by the total number of patents applied for by firm i in year t. Source: USPTO.

Citation-weighted Process (Product) Innovationsit

Natural logarithm of one plus the citation-weighted number of process (product) claims included in patents applied for by firm i in year t. For each claim, the weight is the number of citations received by the claim’s patent within a three-year period starting from the patent’s award year. We split patent citations proportionally across all claims included in the patent. Source: USPTO.

Citation-weighted Share of Process Innovationsit

Ratio of the citation-weighted number of process claims to the citation-weighted number of all claims included in patents applied for by firm i in year t. Source: USPTO.

Citations of Process (Product) Innovationsit

Natural logarithm of one plus the sum of citations received by process (product) claims, which are included in patents applied for by firm i in year t, within a three-year period starting from the patent’s award year, divided by the number of process (product) claims. We split patent citations proportionally across all claims included in the patent. Source: USPTO.

Patentsit

Natural logarithm of one plus the number of patents applied for by firm i in year t. Source: USPTO.

Appendix: Variable definitions (cont.)

– 52 –

Variable

Definition

Supply of Product Innovationsjt

Weighted sum of process claims of firms in the originating industries (lagged by one year) computed for each using industry j and each year t. Originating and using industry linkages are defined using the 2002 U.S. Benchmark Input-Output Accounts with weights being the shares of outputs of using industries that are due to the supply of inputs from originating industries. To construct these weights, we follow Acemoglu, Carvalho, Ozdaglar, and Tahbaz-Salehi (2012). We translate the I-O industry codes into NAICS industry codes available in Compustat. Source: USPTO and BEA Benchmark Input-Output Data.

Importjt

For each manufacturing 4-digit SIC industry j, we define Importjt as the level of lagged Chinese import penetration in the U.S. or as the logarithmic growth rate of Chinese import penetration in the U.S. Source: Bernard, Jensen, and Schott (2006).

Exportjt

For each manufacturing 4-digit SIC industry j, we define Exportjt as the logarithmic growth rate of exports from the U.S. to China. Source: Schott (2008).

IPRi

Degree of enforcement of intellectual property rights in the Chinese province where firm i’s subsidiary is located. We construct an ordinal variable decreasing in the intellectual property rights protection at the subsidiaries’ locations that takes values from 1 to 4. Source: Ang, Cheng, and Wu (2014). Information on subsidiaries’ locations is from the 2001 Survey of Foreign Invested Enterprises (FIEs) conducted by the National Bureau of Statistics in China.

Wage Billi

Indicator variable that takes a value of one if the number of employees at the U.S. firm’s subsidiary in China at the time of its registration is higher than the sample median and also the minimum wage growth rate in the county where the subsidiary is located in 1998 is higher than the sample median, and is zero otherwise. Wage Billi varies across treated firms and is zero for control firms. Source: Information on minimum wages is from Huang, Loungani, and Wang (2015). Information on subsidiaries’ locations is from the 2001 Survey of Foreign Invested Enterprises (FIEs) conducted by the National Bureau of Statistics in China.

Equity Ratioi

The U.S. firm’s capital invested in the Chinese subsidiary divided by the capital provided by the Chinese party at the subsidiary’s registration. Equity Ratioi varies across treated firms and is zero for control firms. Source: Information on equity stakes of U.S. and Chinese parties in Chinese subsidiaries is from the 2001 Survey of Foreign Invested Enterprises (FIEs) conducted by the National Bureau of Statistics in China.

No Ownership Restrictionsjt

Indicator variable that takes a value of one if industry j is not subject to ownership restrictions on foreign investment imposed by the Chinese government at year t, and is zero otherwise. We map industry descriptions used in the Catalogue into the 4-digit NAICS industry classification. Source: Catalogue of Industries Guiding Foreign Investment issued jointly by the National Development and Reform Commission (NDRC) and the Ministry of Commerce (MOFCOM).

Internet Appendix to Globalization of Work and Innovation: Evidence from Doing Business in China Jan Bena University of British Columbia

Elena Simintzi University of British Columbia

May 2017

A

Procedure to Distinguish Patent Claim Types

Patent grant publication documents available from the United States Patent and Trademark Office (USPTO) are structured using the Extensible Markup Language (XML), a markup language that defines a set of rules for encoding documents in a format that is both human readable and machine readable. Within a patent grant publication document, claims are numbered sequentially, with the first claim typically being the broadest and the most important one. Claims are of two basic types: product or process. Claims are written in a very legalistic and stilted way, which allows us to apply text analysis techniques to clearly determine the claim type. Claims that refer to process innovations begin with “A method for” or “A process for” (or minor variations of these two strings) followed by a verb (typically in gerund form), which directs to actions that are to take place as part of the process. We machine-read the text of all claims in USPTO patents and denote claims that begin in this way as process claims, while we denote the residual as product claims. Claims are also either independent or dependent. An independent claim stands on its own, while a dependent claim has meaning only when combined with a claim (in the same patent) it refers to. We machine-read the text of all claims in USPTO patents and identify references a claim makes to other claims. We denote claims that contain such references as dependent claims, while we denote the residual as independent claims. For example, USPTO patent grant document US 8317964 B2 titled “Method of manufacturing a vehicle” applied for on January 11, 2007 by Ford Motor Company has 11 process claims to protect a method of manufacturing a vehicle (Figure A1). The wording of claim 1 begins: “1. A method of manufacturing a vehicle comprising...”. The wording of claim 2 begins: “2. The method of claim 1 wherein the step of assembling the upper portion further comprises...”. We code claims 1 and 2 to be process claims, wherein claim 1 is an independent and claim 2 is a dependent claim. Taking a different example, USPTO patent grant document US 7535468 B2 titled “Inte-

–1–

grated sensing display” applied for on June 21, 2004 by Apple Inc. has 22 product claims to protect the invention of an integrated sensing display (Figure A2). The wording of claim 1 begins: “1. A device comprising a display panel...”. The wording of claim 2 begins: “2. The device of claim 1, wherein the image elements are located in a ...”. We code claims 1 and 2 to be product claims, wherein claim 1 is an independent claim and claim 2 is a dependent claim. Table IA-A1 reports summary statistics on claim types per patent. Panel A is based on the universe of 4,233,476 utility patents issued by USPTO with application dates between January 1976 and December 2012. On average, a patent has 15.2 claims, of which 4.6 are process, 10.7 are product, 2.7 are independent, and 12.5 are dependent. In this sample, process claims are 30% of total claims, while product claims are 70% of total claims. When we look at the patent decomposition, there are 15.4% process patents, 56% product patents, 11.3% process-apparatus patents, and 17.4% product-method patents. Panel B is based on 1,855,328 utility patents applied for at USPTO by firms matched to Compustat with application dates between January 1976 and December 2012. The innovation mix of Compustat firms is very similar to that of the patent universe. Specifically, on average, a patent has 16.0 claims, of which 5.3 are process, 10.7 are product, 2.9 are independent, and 13.1 are dependent. In this sample, process claims are 33% of total claims, while product claims are 67% of total claims. When we look at the patent decomposition, there are 16.7% process patents, 49% product patents, 14.5% process-apparatus patents, and 20.1% product-method patents.

–2–

Figure A1. Example of process innovations (US 8317964 B2) This figure shows an example of a (purely) process patent comprised of 11 claims.

–3–

Figure A2. Example of product innovations (US 7535468 B2) This figure shows an example of a (purely) product patent comprised of 22 claims.

–4–

Table IA-A1: Process and product innovations This table reports summary statistics on patent claims for the universe of utility patents (Panel A) and the utility patents matched to Compustat firms (Panel B) applied for at the USPTO with application dates from January 1976 till December 2012. Panel A statistics are based on 4,233,476 patents. Panel B statistics are based on 1,855,328 patents. Patent claims define, in technical terms, the scope of protection conferred by a patent and thus define what subject matter the patent protects. A process claim refers to an innovation that reduces production cost while a product claim refers to a new good. An independent claim stands on its own. In contrast, a dependent claim only has meaning when combined with a claim of the same patent it refers to.

Panel A: Universe of Patents issued by USPTO Mean

Standard Deviation

25th Percentile

50th Percentile

75th Percentile

Number of claims per patent

15.20

12.40

7

13

20

Number of process claims

4.56

8.16

0

0

7

Number of product claims

10.70

10.50

3

9

15

Number of independent claims

2.70

2.29

1

2

3

Number of dependent claims

12.50

11.40

5

10

17

Panel B: Compustat Firms’ Patents

Number of claims per patent

Mean

Standard Deviation

25th Percentile

50th Percentile

75th Percentile

16.00

12.60

8

14

20

Number of process claims

5.33

8.33

0

1

8

Number of product claims

10.70

10.60

3

9

15

Number of independent claims

2.93

2.43

1

2

4

Number of dependent claims

13.10

11.50

6

11

17

–5–

B

External Validity of Innovation Measures: Survey Evidence and Illustrative Correlations

Since we are the first to decompose innovations into products and processes using patent data for a broad sample of firms, we provide several validity checks on the main measures used in our analyses. First, we compare the process-product innovation mix computed using our data with that reported by other sources. The ‘Business Research and Development and Innovation Survey’ in the U.S., conducted by the National Science Foundation (NSF), reports the number of R&D performing firms that introduced new products or processes every year since 2006. On average, 42% of firms performing R&D over the 2006-2011 period, and 44% of firms with R&D activity over $100 million, report that they perform process innovation. Comparably, using our data, we find that 46% of Compustat firms patented process innovations over the same period. We also find that over the same period, on average, 39% of patented innovations are process innovations, albeit there is no question in the NSF survey that would allow us to make a direct comparison.21 Analogous statistics to those available in the NSF survey are also provided by the ‘European Firms in Global Economy: internal policies for external competitiveness’ (EFIGE) survey performed in 2010 in 8 European countries. Table IA-B1 shows that the percentage of firms active in process innovation ranges from 40 to 51 in these countries. Overall, both surveys confirm our finding that about 45% of R&D-active firms engage in process innovation. Next, we qualitatively validate our measures relying on the findings of the labor economics inequality literature. There are two prominent explanations in this literature for the displacement of middle-skilled jobs that we observe in the aggregate data. The first explanation is that technological progress allows firms to replace expensive labor that performs routine tasks with technology (Autor, Levy, and Murnane 2003; Acemoglu and Autor, 2011). To the extent that process innovations are aimed at reducing production cost (Scherer 1982, 1984; Link 1982; Cohen and Klepper, 1996; Eswaran and Gallini, 1996), 21

Estimates from earlier studies of the average process share in the manufacturing sector in the 1980s ranges between 25% to 30%. See Cohen and Klepper (1996) for a more detailed discussion.

–6–

this explanation suggests that process innovations displace routine labor tasks that can be more easily performed by technology. Due to this displacement, we should observe a negative correlation between process innovations and the subsequent change in the intensity of routine tasks in a given industry. The second explanation is that the globalization of labor markets allows firms to offshore part of their production to low-wage countries (Blinder 2009; Blinder and Krueger, 2013). This explanation implies that process innovations should be less beneficial if labor tasks are easily offshorable. We show evidence consistent with both predictions in Table IA-B2 and Table IA-B3. We classify intensity of routine tasks at the industry-year level. We use the Occupational Employment Statistics (OES), provided by the Bureau of Labor Statistics, to obtain information on total employment by occupation for each 4-digit NAICS industry over the 2002-2012 period. Using the classification of tasks’ routine intensity in Autor, Levy, and Murnane (2003) and Standard Occupational Classification (SOC) codes, we construct the average routine intensity of occupations in a given 4-digit NAICS industry-year, weighted by total employment for each occupation in a given industry-year.22 Consistent with our intuition, Table IA-B2 shows that higher share and quantity of process innovations are negatively associated with the change in an industry’s routine task intensity over the subsequent 5 years. We are also able to characterize the offshorability of labor tasks at the industry level. We match the classification of occupations by offshorability provided by Blinder (2009), available for about 290 SOC codes, to 4-digit NAICS industries. To do this, we use SOC crosswalks and information on occupations by industry available from the OES data. In Table IA-B3, we show that industries with an inherently higher degree of offshorability are associated with lower share and quantity of process innovations. In Table IA-B4, we show that industries with a high share and quantity of process innovations become subsequently more capital intensive. We measure capital intensity as 22

The BLS and the National Crosswalk Service Center in the U.S. provide crosswalks that allow us to match the SOC codes in the OES data with the Autor, Levy, and Murnane (2003) job title classifications.

–7–

the real capital invested in equipment over the total number of employees in a given 4digit SIC industry-year available from the NBER-CES Manufacturing Industry Database. Our sample starts in 1976, the first year in our patent data, and ends in 2011, the last year available in the NBER-CES Manufacturing Industry Database. Table IA-B4 shows that a higher share or quantity of process innovations in the industry are associated with increasingly higher capital intensity over the subsequent years. Finally, we rely on patent data to validate our measures. We search abstracts and background description sections of patent documents for texts directly mentioning labor cost reductions over the 1995-2012 period. Such keywords include, for example: reduce labor, save labor, decrease labor intensity, reduce wage costs, substitute manual workers, replace labor force, reduce manpower. We next aggregate the number of patents including such keywords at the firm-year level for Compustat firms and construct the variable Share of Patents with Labor Referencesit . In Table IA-B5, we show a positive and significant correlation at the firm-year level between the share of patents with specific references to labor cost reductions and the share or quantity of process innovations. The analogous correlation with the quantity of product innovations is zero. Note that, unlike our processproduct innovation measures, patent descriptions do not have to follow a specific set of language rules and are, therefore, less reliable.

–8–

Table IA-B1: Process innovations and survey comparisons This table reports the percentage of R&D performing firms which reported to have introduced process innovations at the National Science Foundation (NSF) survey for the U.S., and the EFIGE (European Firms in a Global Economy: internal policies for external competitiveness) survey for Europe. This number is compared to the universe of Compustat firms with process patents during the same time period. The reported number for the NSF is the average percentage of R&D performing firms doing process innovations over the period 2006-2011. The reported number for Compustat is the average number of firms which have patented process innovations over the 2006-2011 period. The EGIGE survey took place in early 2010 and covers 8 European countries.

Source

% of of R&D firms performing process innovation

U.S.

NSF

42

U.S.

Compustat

46

Austria

EFIGE

48

France

EFIGE

44

Germany

EFIGE

43

Hungary

EFIGE

40

Italy

EFIGE

45

Spain

EFIGE

51

UK

EFIGE

43

–9–

Table IA-B2: Process innovations and industry routine intensity This table shows the results of OLS regressions of a rolling window of 5-year changes of the industry’s j routine intensity between t and t + 5 on the share of process innovations (Column 1) and quantity of process and product innovations (Column 2) in a 4-digit NAICS industry j at time t. Innovation measures for each year and industry are computed from the universe of Compustat firms with patent data. To measure the routine intensity of a given occupation, we follow Autor, Levy, and Murnane (2003) and compute the ratio of routine tasks over the sum of all tasks. Routine tasks include the sum of routine cognitive and routine manual tasks, as defined by ALM. All variables are available by ALM and are matched to occupations at a given 4digit NAICS industry and year using the OES data and Crosswalks provided by BLS and the Crosswalk Service Center. For a given industry-year, we take the average of routine intensity of the industry’s occupations, weighted by employment of occupations at this industry and year. The sample period is 2002-2012. Standard errors are clustered at the 4-digit NAICS industry level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

∆(Industry Routine Share)t,t+5 (1)

Share of Process Innovationsjt

(2)

-0.907 (0.533)*

Process Innovationsjt

-0.132 (0.0805)*

Product Innovationsjt

0.153 (0.119)

Year FE

Yes

Yes

R2

0.06

0.06

Obs.

685

685

– 10 –

Table IA-B3: Process innovations and offshorability This table reports the results from OLS regressions of the industry share of process innovations (Column 1), and the industry quantity of process innovations (Column 2) on the offshorability of occupations at a given 4-digit NAICS industry. The offshorability of occupations is based on the index provided by Blinder (2009) classifying the offshorability of 291 SOC occupations in the 2004 U.S. workforce. Using crosswalks provided by BLS and the Crosswalk Service Center, we match the index to occupations provided by OES for each 4-digit NAICS-year level. Since the offshorability index is time-invariant, we collapse the innovation measures at the industry level (over the period 20022012). Standard errors are robust. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Offshorability

Share of Process Innovations

Process Innovations

(1)

(2)

-0.0161

-0.0450

(0.00605)***

(0.0263)*

Patentsj

1.370 (0.0340)***

R2

0.05

0.88

Obs.

176

176

– 11 –

Table IA-B4: Process innovations and industry equipment intensity This table shows the results of OLS regressions of equipment intensity in a 4-digit SIC industry j at time t on contemporaneous and lagged share of process innovations (Column 1), quantity of process innovations (Column 2), and quantity of process and product innovations (Column 3). Equipment intensity is measured as the real capital invested in equipment over industry employment (log-transformed) available from the NBER-CES Manufacturing Industry Database. All regressions include (4-digit SIC) industry and year fixed effects. The sample period is 1976-2011. Standard errors are clustered at the 4-digit SIC industry level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Equipment Intensity (1)

Share of Process Innovationsjt

(2)

(3)

0.0328 (0.0436)

Share of Process Innovationsj,t−1

0.0795 (0.0436)*

Share of Process Innovationsj,t−2

0.154 (0.0425)***

Share of Process Innovationsj,t−3

0.184 (0.0416)***

Process Innovationsjt

Process Innovationsj,t−1

Process Innovationsj,t−2

Process Innovationsj,t−3

0.0235

0.0151

(0.0076)***

(0.0084)*

0.0248

0.0214

(0.0078)***

(0.0087)**

0.0271

0.0198

(0.0077)***

(0.0087)**

0.0429

0.0195

(0.0073)***

(0.0091)**

Product Innovationsjt

0.0122 (0.0118)

Product Innovationsj,t−1

0.0037 (0.0127)

Product Innovationsj,t−2

0.0050 (0.0128)

Product Innovationsj,t−3

0.0448 (0.0117)***

Industry FE

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

R2

0.91

0.92

0.92

Obs.

3,051

2,550

2,527

– 12 –

Table IA-B5: Process innovations and patents with references to labor This table reports the results from OLS regressions of the share of patents with references to labor costs on the share of process innovations (Columns 1, 3) and quantity of process and product innovations (Columns 2, 4). The sample includes Compustat firms for the period 1995-2012. Columns 1-2 include all firm-years, while Columns 3-4 include firm-years for which total number of claims is greater than the sample median. Our dependent variable is based on the count of patents which include keywords indicating reduction of labor costs. Such keywords include, for example: reduce labor, save labor, decrease labor intensity, reduce wage costs, substitute manual workers, replace labor force, reduce manpower. All regressions include firm and year fixed effects. Standard errors are robust and clustered at the firm level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Patents with Labor References (1)

(2)

(3)

(4)

Firm Patent Claims> Sample Median

Share of Process Innovationsjt

0.00245

0.00536

(0.0021)

(0.00306)*

Process Innovationsjt

Product Innovationsjt

0.0009

0.0129

(0.0004)**

(0.0005)**

-0.0001

-0.0007

(0.0004)

(0.0006)

Firm FE

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

R2

0.41

0.41

0.45

0.45

46,078

46,078

26,068

26,068

Obs.

– 13 –

C

Robustness checks

In this Appendix, we report results of a number of additional analyses and robustness checks on our baseline findings.

– 14 –

Table IA-C1: Robustness: Different thresholds for defining high-patenting firms This table reports results of OLS regressions of the share of process innovations (Columns 1-3), and the quantity of process (Columns 4-6) and product (Columns 79) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The regression specifications correspond to those in Table 3, except that we use different cutoffs to define high-patenting firms. Panel A imposes no restriction, and Panels B, C, D, and E include all firms with 40, 60, 80, and 90 patents or more during our sample period. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

(2)

(3)

Process Innovations (4)

(5)

Product Innovations (6)

(7)

(8)

(9)

Panel A: All firms

– 15 –

Agreement(t>1999) · Chinai

-0.0226

-0.0226

-0.0246

-0.186

-0.179

-0.227

-0.0350

-0.0215

-0.0573

(0.0123)*

(0.0130)*

(0.0139)*

(0.0610)***

(0.0625)***

(0.0694)***

(0.0404)

(0.0415)

(0.0432)

R2

0.65

0.68

0.73

0.91

0.92

0.93

0.94

0.94

0.96

Obs.

2,661

2,256

2,256

2,661

2,256

2,256

2,261

2,256

2,256

Panel B: Firms with 40 patents or more

Agreement(t>1999) · Chinai

-0.0222

-0.0218

-0.0240

-0.186

-0.178

-0.226

-0.0361

-0.0249

-0.0591

(0.0124)*

(0.0130)*

(0.0139)*

(0.0612)***

(0.0626)***

(0.0694)***

(0.0404)

(0.0415)

(0.0433)

R2

0.65

0.68

0.73

0.91

0.92

0.93

0.94

0.95

0.96

Obs.

2,654

2,250

2,250

2,654

2,250

2,250

2,654

2,250

2,250

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Firm-level Controls Firm FE

Yes

Yes

Year FE

Yes

Yes

Industry×Year FE

Yes

Yes

Yes

Share of Process Innovations (1)

(2)

Process Innovations (3)

(4)

Product Innovations

(5)

(6)

(7)

(8)

(9)

Panel C: Firms with 60 patents or more

Agreement(t>1999) · Chinai

-0.0230

-0.0233

-0.0241

-0.192

-0.184

-0.227

-0.0416

-0.0283

-0.0640

(0.0124)*

(0.0130)*

(0.0141)*

(0.0618)***

(0.0633)***

(0.0699)***

(0.0406)

(0.0418)

(0.0434)

R2

0.65

0.68

0.73

0.91

0.92

0.93

0.94

0.94

0.96

Obs.

2,618

2,218

2,218

2,618

2,218

2,218

2,618

2,218

2,218

Panel D: Firms with 80 patents or more

– 16 –

Agreement(t>1999) · Chinai

-0.0329

-0.0321

-0.0333

-0.214

-0.215

-0.256

-0.0124

-0.0146

-0.0458

(0.0124)***

(0.0128)**

(0.0137)**

(0.0633)***

(0.0639)***

(0.0701)***

(0.0404)

(0.0413)

(0.0432)

R2

0.67

0.70

0.75

0.91

0.92

0.93

0.94

0.95

0.96

Obs.

2,483

2,110

2,110

2,483

2,110

2,110

2,483

2,110

2,110

Panel E: Firms with 90 patents or more

Agreement(t>1999) · Chinai

-0.0350

-0.0349

-0.0382

-0.237

-0.234

-0.275

-0.0117

-0.0112

-0.0345

(0.0123)***

(0.0127)***

(0.0135)***

(0.0638)***

(0.0635)***

(0.0703)***

(0.0407)

(0.0413)

(0.0437)

R2

0.68

0.72

0.77

0.92

0.93

0.94

0.94

0.95

0.96

Obs.

2,356

2,000

2,000

2,356

2,000

2,000

2,356

2,000

2,000

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Firm-level Controls Firm FE

Yes

Yes

Year FE

Yes

Yes

Industry×Year FE

Yes

Yes

Yes

Table IA-C2: Robustness: Alternative definitions of process innovations This table reports results of OLS regressions of the share of process innovations (Columns 1-2), and quantity of process innovations (Columns 3-4) on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 2-3, and 5-6, respectively, except we use alternative definitions for our dependent variables. In Panel A, we construct our measures based on independent claims, i.e. we exclude claims that are subordinate to other claims. In Panels B and C, we define our measures at the patent-level (instead of claim-level). In Panel B, we define process patents as the sum of process and process-apparatus patents divided by the total number of patents. In Panel C, we define process patents as the sum of process, process-apparatus, and product-method patents divided by the total number of patents. In Panel D, we use citation-weighted measures of the patent-level variables in Panel B. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1. Share of Process Innovations

Process Innovations

Panel A: Using only independent claims (1) Agreement(t>1999) · Chinai

(2)

(3)

(4)

-0.0324

-0.0296

-0.145

-0.168

(0.0111)***

(0.0123)**

(0.0513)***

(0.0563)***

R2

0.67

0.73

0.95

0.95

Obs.

1,940

1,940

1,940

1,940

Panel B: Using patent-level measures

Agreement(t>1999) · Chinai

(1)

(2)

(3)

(4)

-0.0425

-0.0367

-0.151

-0.132

(0.0155)***

(0.0161)**

(0.0533)***

(0.0551)**

R2

0.74

0.78

0.95

0.95

Obs.

1,940

1,940

1,940

1,940

Panel C: Using patent-level measures (alt.)

Agreement(t>1999) · Chinai

(1)

(2)

(3)

(4)

-0.0346

-0.0255

-0.0810

-0.0537

(0.0170)**

(0.0182)

(0.0403)**

(0.0425)

R2

0.75

0.79

0.97

0.97

Obs.

1,940

1,940

1,940

1,940

Panel D: Using citation-weighted patent-level measures

Agreement(t>1999) · Chinai

(1)

(2)

(3)

(4)

-0.0487

-0.0401

-0.201

-0.217

(0.0179)***

(0.0194)**

(0.0958)**

(0.101)**

R2

0.66

0.71

0.89

0.91

Obs.

1,940

1,940

1,940

1,940

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Year FE

Yes

Firm-level Controls

Industry×Year FE

Yes Yes

– 17 –

Yes

Table IA-C3: Robustness: Normalize innovation output by R&D and employment This table reports results of OLS regressions of the quantity of process (Columns 1, 2, 5, 6) and product (Columns 3, 4, 7, 8) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 5-6, and 8-9, respectively, except that process and product innovations are normalized by R&D expenses in Columns 1-4 and by number of employees in Columns 5-8. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Log(Process/R&D) (1)

Agreement(t>1999) · Chinai

(2)

Log(Product/R&D) (3)

(4)

Log(Process/Employment) (5)

(6)

Log(Process/Employment) (7)

(8)

– 18 –

-0.241

-0.250

-0.0279

-0.0392

-0.103

-0.129

0.0364

0.0046

(0.0619)***

(0.0699)***

(0.0528)

(0.0554)

(0.0432)**

(0.0479)***

(0.0416)

(0.0386)

Firm-level Controls

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Industry×Year FE

Yes Yes

Yes Yes

Yes Yes

Yes

R2

0.92

0.93

0.94

0.95

0.92

0.94

0.96

0.97

Obs.

1,930

1,930

1,930

1,930

1,940

1,940

1,940

1,940

Table IA-C4: Robustness: Negative binomial model; Drop observations when innovation is zero This table reports results of regressions of the quantity of process (Columns 1-3) and product (Columns 4-6) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 4-6, and 7-9, respectively, except the estimation is implemented by Negative binomial model (Panel A) and dropping the zeros from our innovation measures by not adding one before taking the natural logarithm of process and product claims (Panel B). Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Process Innovations

Product Innovations

Panel A: Negative binomial model

Agreement(t>1999) · Chinai

(1)

(2)

(3)

(4)

(5)

(6)

-0.181

-0.195

-0.232

0.0046

0.0031

-0.0323

(0.0517)***

(0.0510)***

(0.0524)***

(0.0341)

(0.0349)

(0.0351)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Firm-level Controls

Yes

Yes

Firm FE

Yes

Yes

Yes

Year FE

Yes

Yes

Industry×Year FE

Obs.

Yes

2,278

1,940

1,940

Yes

2,278

1,940

1,940

Panel B: Drop observations with zero process or product claim counts

Agreement(t>1999) · Chinai

(1)

(2)

(3)

(4)

(5)

(6)

-0.209

-0.204

-0.252

-0.0120

-0.0147

-0.0431

(0.0628)***

(0.0635)***

(0.0710)***

(0.0405)

(0.0430)

(0.0452)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Firm-level Controls

Yes

Yes

Firm FE

Yes

Yes

Yes

Year FE

Yes

Yes

Industry×Year FE

Yes

Yes

R2

0.92

0.93

0.94

0.95

0.96

0.97

Obs.

2,246

1,919

1,919

2,271

1,936

1,936

– 19 –

Table IA-C5: Robustness: Fractional response model This table reports results of regressions of the share of process innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 1-3, except the estimation is implemented by Fractional response model. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations

Agreement(t>1999) · Chinai

(1)

(2)

(3)

-0.040

-0.038

-0.039

(0.012)***

(0.012)***

(0.012)***

Yes

Yes Yes

Firm-level Controls Firm FE

Yes

Yes

Year FE

Yes

Yes

Industry×Year FE

Obs.

Yes

2,278

– 20 –

1,940

1,940

Table IA-C6: Mean-reversion in firms’ innovation activities This table reports results of OLS regressions of the share of process innovations (Columns 1-2) and the quantity of process (Columns 3-4) and product (Columns 5-6) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3 Columns 3, 6, and 9, respectively, except that we additionally control for firm-specific innovation trends. In Columns 1, 3, and 5 we interact year fixed effects with the dependent variable defined pre-treatment in 1997 and in Columns 2, 4, and 6 we interact year fixed effects with the number of patents (log-transformed) defined pre-treatment in 1997. Chinai is an indicator variable equal to one if a U.S. firm has a subsidiary in China (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations

Agreement(t>1999) · Chinai

Year FE × P rocessRatio1997

Process Innovations

Product Innovations

(1)

(2)

(3)

(4)

(5)

(6)

-0.0380

-0.0417

-0.211

-0.277

0.0109

-0.0294

(0.0132)***

(0.0139)***

(0.0605)***

(0.0708)***

(0.0432)

(0.0453)

Yes

Year FE × P rocess1997

Yes

Year FE × P roduct1997

Yes

Year FE × P atents1997

Yes

Yes

Yes

Firm-level Controls

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Industry×Year FE

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.78

0.77

0.93

0.94

0.95

0.96

Obs.

1,900

1,900

1,900

1,900

1,900

1,900

– 21 –

Yes

Yes

Yes

Yes

Yes

Table IA-C7: Robustness: Allowing for entry into China This table reports results of OLS regressions of the share of process innovations (Columns 1-3) and the quantity of process (Columns 4-6) and product (Columns 79) innovations on treated firms following the 1999 U.S.-China bilateral agreement (Agreement(t>1999) ) as compared to control firms. The sample and regression specifications correspond to those in Table 3, except that we use a time-varying measure of treatment. Chinait is an indicator variable equal to one if a U.S. firm has a subsidiary in China in year t (treated), and equal to zero if it has a subsidiary in a low-wage Asian country except China (control). All firm-level variables are winsorized at the 1% level. Standard errors are clustered at the firm-level. *** indicates p< 0.01, ** indicates p< 0.05, and * indicates p< 0.1.

Share of Process Innovations (1)

Agreement(t>1999) · Chinait

(2)

(3)

Process Innovations (4)

(5)

Product Innovations (6)

(7)

(8)

(9)

– 22 –

-0.0477

-0.0411

-0.0419

-0.286

-0.274

-0.304

-0.0081

-0.0099

-0.0433

(0.0131)***

(0.0138)***

(0.0140)***

(0.0664)***

(0.0702)***

(0.0747)***

(0.0454)

(0.0460)

(0.0473)

0.0043

0.0048

0.0119

0.0937

0.0905

0.142

0.0256

0.0122

0.0352

(0.0158)

(0.0161)

(0.0186)

(0.0857)

(0.0798)

(0.0872)

(0.0532)

(0.0559)

(0.0625)

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Chinait

Firm-level Controls

Industry×Year FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

R2

0.69

0.72

0.78

0.92

0.93

0.94

0.95

0.95

0.96

Obs.

2,278

1,940

1,940

2,278

1,940

1,940

2,278

1,940

1,940

References [1] Acemoglu, D., and D. Autor, 2011, “Skills, Tasks, and Technologies: Implications for Employment and Earnings”, In D. Card and O. Ashenfelter (Eds.), Handbook of Labor Economics, Volume 4, Chapter 12, 1043-1171. Elsevier. [2] Autor, D. H., F. Levy, and R. J. Murnane, 2003, “The Skill Content of Recent Technological Change: An Empirican Exploration”, Quarterly Journal of Economics, 118, 1297-1333. [3] Blinder, A. S., 2009, “How Many U.S. Jobs Might Be Offshorable”, World Economics, 10, 41-78. [4] Blinder, A. S., and A. B. Krueger, 2013, “Alternative Measures of Offshorability: A Survey Approach”, Journal of Labor Economics, 31, 97-128. [5] Cohen, W. M., and S. Klepper, 1996, “Firm Size and the Nature of Innovation within Industries: The Case of Process and Product R&D”, Review of Economics and Statistics, 78, 232-243. [6] Eswaran, M., and N. Gallini, 1996, “Patent Policy and the Direction of Technological Change”, The RAND Journal of Economics, 27, 722-746. [7] Link, A. N., 1982, “A Disaggregated Analysis of Industrial R&D: Product versus Process Innovation”, in Devendra Sahal (ed.), The Transfer and Utilization of Technical Knowledge (Lexington, MA: Lexington Books). [8] Scherer, F. M., 1982, “Inter-industry Technology Flows in the United States”, Research Policy, 11, 227-245. [9] Scherer, F. M., 1984, “Using Linked Patent and R&D Data to Measure Interindustry Technology Flows”, in Zvi Griliches (ed.), R&D, Patents, and Productivity (Chicago: University of Chicago Press for the National Bureau of Economic Research).

– 23 –

Evidence from Doing Business in China - SSRN papers

Affiliations: Sauder School of Business, The University of British Columbia. ... landscape, lifted U.S. firms' restrictions on doing business in China, such as: the ...

1MB Sizes 1 Downloads 291 Views

Recommend Documents

Is Advertising Informative? Evidence from ... - SSRN papers
Jan 23, 2012 - doctor-level prescription and advertising exposure data for statin ..... allows advertising to be persuasive, in the sense that both E[xat] > δa.

Evidence from Doing Business in China
control for potential differential trends between the treaded and control firms by ... and Seru 2011; Giroud and Mueller 2015, 2016) and investment (e.g. Besley ...

budget procedures and fiscal performance: evidence ... - SSRN papers
Oct 2, 2016 - IESE Business School-University of Navarra - 1. BUDGET ... Pompeu Fabra and Public-Private Sector Research Center, IESE Business School.

Technology Shocks and Hours Revisited: Evidence ... - SSRN papers
Jun 11, 2017 - I exploit heterogeneous impulse responses at the household level due to limited stock market participation to provide novel evidence on the degree of nominal rigidities. A number of studies show that positive technology shocks reduce a

Vietnam's Lesson for China: An Examination of the ... - SSRN papers
supports the big bang approach. In particular, alluding to Vietnam's 1989 reforms, Sachs and Woo (1997, 2000) put forward the hypothesis that China.

Are Voters Sensitive to Terrorism? Direct Evidence ... - SSRN papers
Candidate, Master of Public Policy (MPP), Georgetown University, Expected ... supportive of the policy voting hypothesis, according to which “parties benefit from ...

Blaming Youth - SSRN papers
Dec 14, 2002 - Social Science Research Network Electronic Paper Collection ... T. MacArthur Foundation Research Network on Adolescent Development and.

law review - SSRN papers
Tlie present sentencing debate focuses on which decisionmaker is best .... minimum sentences even after a sentencing guideline system is in place to control ...

Labor-induced Technological Change: Evidence from Doing Business ...
landscape, lifted U.S. firms' restrictions on doing business in China, such as: the ... with the definition of process innovations being labor-saving technologies.

Optimism and Communication - SSRN papers
Oct 10, 2010 - Abstract. I examine how the communication incentive of an agent (sender) changes when the prior of the principal (receiver) about the agent's ...

yale law school - SSRN papers
YALE LAW SCHOOL. Public Law & Legal Theory. Research Paper Series by. Daniel C. Esty. This paper can be downloaded without charge from the.

from institutions to financial development and growth - SSRN papers
Contract enforcement; Economic growth; Financial development; Financial .... Enforcement and other transac on costs. Ins tu ons. Growth. Property. Rights.

[PDF BOOK] Doing Business in China For Dummies
Online PDF Doing Business in China For Dummies, Read PDF Doing Business in China For Dummies, ... Simply Sign Up to one of our plans and start browsing.