Diversification, economies of scope, and exports growth of Chinese firms
DDiversification, economies of scope, and exportsgrowth of Chinese firms ∗ Mercedes Campi † , Marco Due˜nas ‡ , Le Li § , and Huabin Wu ¶ December 22, 2017
Abstract
In the 1990s, China started a process of structural reforms and of trade liberalization,which was followed by the accession to the World Trade Organization (WTO) in2001. In this paper, we analyze trade patterns of Chinese firms for the period2000-2006, characterized by a notable increase in exports volumes. Theoretically,in a more open economy, firms are expected to move from the production of a setof less-competitive products towards more internationally competitive ones, whichimplies specialization. We study several stylized facts on the distribution of Chinesefirms trade and growth rates, and we analyze whether firms have diversified orspecialized their trade patterns between 2000 and 2006. We show that Chineseexport patterns are very heterogeneous, that the volatility of growth rates dependson the level of exports, and that volatility is stronger after trade liberalization. Both,diversification in products and destinations have a positive impact on trade growth,but diversification of destinations has a stronger effect. We conclude that the successof Chinese exports is not only due to an increase in the intensive margin, relatedto the existence of economies of scale, but also due to an increase in the extensivemargin, related to the existence of economies of scope.
Keywords:
Industrial dynamics; Margins of trade; Diversification and specialization;Economies of scope
JEL Codes:
F14; F61; L25 ∗ We thank the Institute for Advanced Research of Shanghai, University of Finance and Economics forproviding access to the data. We also thank Giulio Bottazzi and Emanuele Pugliese for providing usefulcomments and discussions, and the participants of the Workshop “Crisis, Inequality and Development”2016 (Shanghai, China), EMAEE 2017 (Strasbourg, France), and MEIDE 2017 (Montevideo, Uruguay).Le Li gratefully acknowledges the research support by the IBIMET-CNR (grant CrisisLab-ProCoPe). † Corresponding author . CONICET - University of Buenos Aires, Faculty of Economics, IIEP-Baires,Buenos Aires, Argentina. [email protected] ‡ Department of Economics, International Trade and Social Policy – Universidad de Bogot´a JorgeTadeo Lozano. [email protected] § IBIMET-CNR & Sant’Anna School of Advanced Studies, Pisa, Italy. [email protected] ¶ The Institute for Advanced Research of Shanghai University of Finance and Economics, Shanghai,China. [email protected] a r X i v : . [ q -f i n . E C ] J a n Introduction
In the 1990s, China started a process of trade liberalization, along with several reformsacross a wide variety of sectors, which was finally followed by the accession to the WorldTrade Organization (WTO) in 2001. This event is expected to influence the behavior ofChinese exporting firms.Recent theories of international trade predict that facing trade liberalization, firmswill: (i) reduce the quantity of products that they export, (ii) intensify the volume ofexports of a limited number of products, and (iii) increase their market shares on thisreduced number of products (see, for example, Melitz, 2003; Melitz et al., 2008; Bernardet al., 2010, 2011). In addition, Di Clemente et al. (2014) argue that while competitivenessat the country level is mainly driven by diversification of productive systems, firms’competitiveness is mainly a matter of specialization. But also, at the country level, theeffect of liberalization on trade diversification is likely to depend on the income level ofcountries. For middle-income countries, some authors find a strong diversification trendafter trade liberalization, particularly strong in the five years following liberalization(Carrere et al., 2011).Along with the increasing interest in how liberalization affects diversification patternsand exports, a broad literature analyses how exports and diversification affect productivityand growth, both at the country level (see, for example, Hidalgo et al., 2007; Hausmannand Rodrik, 2003; Zaccaria et al., 2014) and at the firm level (see Wagner, 2007, for areview).Firms can diversify on their destinations because this can help them to stabilize exports(Kim et al., 1993). The effect of product diversification is less clear, although a theoreticalexplanation for the existence of multi-product firms is the reduction of risk that can bereached by diversifiying across product markets, which implies a negative relationshipbetween product diversification and the variability of sales the firm level (Lipczynski andWilson, 2001). However, Wagner (2014) finds that profits tend to be larger in Germanfirms with less diversified export sales over goods and in firms with more diversified exportsales over destination countries.There is less available evidence on how diversification affects the growth of exports.However, if there are economies of scope, diversification on both products and destinationscan lead to an increase in the level of sales. In addition, diversification as well ascompetitiveness depend on technological and organizational capabilities (Dosi et al., 1990).For a number of reasons, developing capabilities to export a new product might be moredifficult for a given firm than being able to arrive to a new destination with the sameproduct, particularly, if we consider that industries present stickiness in the reallocation ofresources from one sector to another. In this sense, we expect diversification in destinationsto have a stronger effect on export growth than product diversification.2n this paper, we investigate the links between diversification over products anddestinations and the growth of exports of Chinese exporting firms in a context of tradeliberalization and an impressive increase in total exports at a 25.39% annual growth rate.We use a data set from the Chinese Custom Office that contains information on thevolumes and values of exported goods and the destination of exports of Chinese firs forthe period 2000-2006.Firstly, we analyze the statistical properties of the distribution of Chinese firms exportsand their growth rates. Next, we study specialization and diversification patterns ofexports in the context of trade liberalization. To do this, we analyze how the intensiveand extensive margins of trade explain the evolution of trade during this period andwe investigate whether firms exhibit economies of scope both considering products anddestinations. Finally, we explore whether exports growth of Chinese exporting firms can beexplained by an increase in their traded products and in the total number of destinationsof exports, controlling for several firm characteristics.The statistical analysis revealed a high heterogeneity of Chinese trading firms. Manyrelatively small exporting firms coexist with a few very large exporters. Also, exportingfirms are often affected by extreme events (high negative and positive growth rates oftrade). Trading firms of different size are similarly affected by negative growth rates, butsmaller firms have higher probabilities of growing more than medium and large firms. Thisimplies the existence of a negative relation between size and volatility. Finally, we observethat, after trade liberalization, firms face more frequently stronger negative shocks.The analysis showed that exports growth at the firm level is due to increases inthe intensive and extensive margins of trade, and also to both. Chinese exports haveheterogeneous patterns in their diversification dynamics, both in products and destinations.In contrast to the expectations of recent international trade theory, there is no significantevidence of a general process towards specialization after trade liberalization. We observethat the exports of Chinese firms exhibit economies of scope both considering productsand destinations. Finally, we observe that diversifying in products and, especially, indestinations increase growth rates of exports, and that more diversified firms in bothaspects grow more compared to more specialized firms.The rest of the paper is organized as follows. Section 2 presents the motivation andthe literature review. Section 3 explains the data and presents a statistical analysis ofthe distribution of Chinese firms exports and the growth rates of exports. Section 4investigates the role of the intensive and extensive margins of trade and the existence ofeconomies of scope, and presents the econometric estimations of the effect of diversificationon trade growth. Finally, Section 5 concludes.3
Motivation and literature review
In the 1990s, China started a dramatic liberalization undertaking several measures, whichtook place in a period of extraordinary growth in trade and output. While the accessionto the WTO in 2001 was the result of this process, it also involved several other reformsacross a wide range of sectors in China (Ianchovichina and Martin, 2001). These structuralreforms are reflected in several important changes both at the macro and micro level.Between 2000 and 2006, Chinese exports increased at a 25.39% annual growth rate,private-owned enterprises increased their share in total exports from 0.84% to 17.56%,joint ventures and foreign-owned increased their share from 48.45% to 58.82%, while state-and collective-owned enterprises decreased their share in total exports from 50.71% to23.62%.At the micro level, considering the heterogeneity of firms, we might expect differenteffects of trade liberalization on firms performance. Several recent models of internationaltrade theory postulate that a more open economy will affect the capability of firms ofexporting new products as well as the range of products they specialized on (see, forexample Melitz, 2003; Melitz et al., 2008; Bernard et al., 2010, 2011).These models predict that trade liberalization will come along with an increasingcompetition and a decrease in trade costs. Facing an increasing competition, firms areexpected to move from the production of a set of less-competitive products towards moreinternationally competitive ones. Moreover, when trade costs decrease as a consequence oftrade liberalization, more productive firms are expected to enter international markets orincrease their export shares in their total sales (Melitz et al., 2008). In brief, theoreticalmodels predict that in the context of trade liberalization, firms will: (i) reduce the quantityof products that they export; (ii) intensify the volume of exports of a limited numberof products; and (iii) increase their market share on this reduced number of products.This implies a reduction in the extensive margin of trade and an increase in the intensivemargin of trade at the firm level.In these models, firms’ differential efficiency is considered as the main determinantof their participation in the international market (Bernard et al., 2007). Instead, anevolutionary perspective departs from the idea that there is persistent heterogeneityamong firms and countries, a systematic processes of competitive selection among them,and stickiness in the reallocation of resources from one sector to another (Dosi et al.,1990). Thus, international competitiveness responds to differences in variable costs butalso to wide differences in technological and organizational capabilities, which shapetrade patterns within sectors and countries. Dosi et al. (2015) showed that at the microlevel the probability of being an exporter as well as the capacity to increase exports arepositively correlated with investments and patents. In addition, several studies indicatethat differences in firms performance are highly correlated with their exporting activities4Roberts and Tybout, 1997; Mathew, 2017).At a macro level, there is no consensus on the effect of the WTO on trade. WhileRose (2004) found little evidence that countries becoming members or belonging to theGeneral Agreement on Tariffs and Trade (GATT) or the WTO changed their trade patternscompared with those who are not members, Subramanian and Wei (2007) found that theWTO has had a positive but uneven impact on trade.For China and Chinese firms trade patterns the evidence is limited and mixed.Ianchovichina and Martin (2001) found that China’s major trading partners gained fromaccession, while some competing countries suffered smaller losses. Rodrik (2006) arguesthat the success of China’s exports owes more to government policies than to comparativeadvantage and free markets. He claims that, as a result of these policies, China has shiftedtowards an export basket that is significantly more sophisticated than the one expectedfor countries at its income level. He also argues that this trade sophistication has been animportant determinant of China’s rapid growth.Amiti and Javorcik (2008) studied the over five-times export growth of China between1992 and 2005. They found that China’s export structure has changed dramatically, witha decline in agriculture and apparel, and growing shares of electronics and machinery.However, when they exclude processing trade, the content of China’s manufacturingexports remains practically unchanged. This implies that the seeming shift towards moresophisticated products is not verified once they consider the content of imported inputs thatare assembled and then exported. They also found evidence of an increasing specializationalong with export growth. Finally, they found that export growth derives mainly fromthe intensive margin (growth of existing products) rather than from the extensive margin(new products).Several other authors have highlighted that processing trade is behind the apparentsophistication of Chinese exports (see, for example Yao, 2009; Koopman et al., 2012; Xing,2014). However, other authors argue that although sophistication of exports is not deepif processing trade is excluded, there has still being a process of change that resulted ineconomic growth. Jarreau and Poncet (2012) studied the effect of export sophisticationon economic performance of different regions within China between 1997 and 2009. Theyfound substantial variation in export sophistication at the province and prefecture level.They showed that regions specializing in more sophisticated goods subsequently growfaster. But, their results suggest that gains are limited to the ordinary export activitiesundertaken by domestic firms, given that no direct gains result from either processingtrade activities or foreign firms.Also, other authors suggest that not all the sophistication of Chinese exports is dueto an increase in processing trade. Wang and Wei (2010) found that there are relevantregional variations in the use of processing trade. They argue that there exist cross-citydifferences in human capital, which are linked to cross-city differences in the sophistication5f the export structure. They also argue that the increasing sophistication owes to thegovernment promotions through high-tech and economic development zones, and thatforeign investment has been conducive to greater product sophistication in China.Manova and Zhang (2009) analyzed Chinese trade flows at the firm level for the years2003 to 2005. They confirmed several stylized facts that also characterize firms trade inother countries. They showed that the bulk of exports and imports are captured by a fewmulti-product firms that transact with a large number of countries. They also observethat, compared to private domestic firms, foreign affiliates, and sino-foreign joint venturesimport more products from more countries, but export fewer products to fewer destinations.Moreover, they found that the relationship between the intensive and extensive marginsof trade is non-monotonic, differs between exporters and importers, and depends on theownership type of the firm. They also found that firms frequently exit and re-enter intotrade and regularly change their product mix and trade partners, but foreign firms exhibitless churning. Finally, they showed that the growth in Chinese exports between 2003 and2005 was mainly driven by deepening and broadening of trade relationships by survivingfirms.While the composition of trade and diversification patters have been analyzed by alarge number of studies, to our knowledge, there are no analysis of how specializationor diversification of Chinese exporting firms affect the growth of exports. Recent tradetheories have focused on the exploration of the linkages between productivity and trade,highlighting complex relationships between trade diversification and productivity (seeCarrere et al., 2011, for a review). But, diversification or specialization can also have aneffect on the growth of trade, particularly if there exist economies of scale or economies ofscope.
Our database contains exports and imports of the universe of Chinese firms for the period2000 to 2006, which are collected from trade records in the Chinese Customs Office. Thecompleteness of the data is confirmed with a simple exercise. According to the NationalStatistics Office, total exports of China in 2001 was 266.10 and total imports 243.55 billiondollars. The aggregation of our firm level data provides a value of 266.66 billion dollarsfor total exports and 243.57 billion dollars for total imports. We have used deflators fromthe US Bureau of Labor Statistics to have the data in constant US dollars. The data set includes information on the volume of exported and imported products andthe value in US dollars, the date of the shipment, the HS code at 8 digits of disaggregation, See (inChinese), accessed on December 2017. See , accessed on April 2016. Ordinary trade refers to the export of a product mainly produced with Chineseinputs. Instead, processing trade includes products with a high content of imported inputs(raw materials, parts and components, accessories, and packaging materials) that areprocessed or assembled by firms and then exported. For our period of study, processingtrade is more than half of total Chinese exports (57.81% in 2000 and 54.50% in 2006).Finally, the data have 8 different ownership types.We grouped these types into threemain categories:(i) stated-owned and collective-owned enterprises, (ii) private-ownedenterprises (which include private enterprises, and individual industrial and commercialhouseholds), and (iii) foreign-owned enterprises (including sino-foreign contractual jointventures and sino-foreign equity joint ventures). The shares of these three groups in totalexports has changed between 2000 and 2006, increasing the shares of private-owned andforeign-owned enterprises and decreasing the share of state-owned enterprises.We define X i,t as total exports of firms, where X stands for the total value of exportsfor firm i in year t . The growth of firms exports ( r ) is the log difference of total exports inthe consecutive year: g i,t = ln ( X i,t ) − ln ( X i,t − ) . (1)We define the number of exported products using a 6-digit code within the HarmonizedSystem classification. We take the number of different products exported and the numberof foreign destinations as a proxy for firm scope as in Bee et al. (2017).Table 1 reports the shares of firms and exports for the years 2000 and 2006 for thefull sample and for different sub-samples based on the number of exported products anddestinations of firms.The full sample of firms included 46,279 firms in 2000 and 166,930 in 2006, with 27,415surviving firms, which indicates a process of entry and exit of exporting firms. We do notobserve changes in the shares of single and multi-products firms. In both years, almost aquarter of firms export only one product and around 6-7% of total exports while 75% offirms export multiple products and around 93-94% of total exports. Conversely, the shareof firms that export to multiple destinations increased from 68.46% in 2000 to 72.63% in2006, while firms exporting to only one destination decreased from 31.54% to 27.37% of Although there are 19 types of trade, these two broad categories include between 96 and 98% of totaltrade each year. The categories that we excluded are not useful for our purposes because they includespecial deals such as compensation trade, duty free foreign exports, or donated materials. able 1: Summary statistics: share (in %) of the number of firms and total exports (in millions)for the full sample and different sub-samples for 2000 and 2006 Number of firms Total Exports2000 2006 2000 2006
Shares by product diversification
Single product 24.22 24.20 6.93 6.05Multi-product 75.78 75.80 93.07 93.95
Shares by destination diversification
Single destination 31.54 27.37 7.65 4.45Multiple destination 68.46 72.63 92.35 95.55
Shares by joint diversification types
Multiple product-destination 58.05 62.51 87.42 90.97Single product-destination 13.82 14.08 2.01 1.47Multi-product and single destination 17.73 13.29 5.64 2.98Single product and multiple destination 10.40 10.12 4.92 4.58Total 46,279 166,930 225,769 865,038 total firms. Likewise, if we classified firms according to both diversification in products anddestinations, we observe that highly-diversified firms that export more than one productto multiple destinations increased from 58.05% to 62.51%, while firms specialized in oneproduct and one destination remained unchanged (14%) exporting a very low share of totalexports (1-2%). Finally, single-product firms shipping to multiple destinations are around10% in both years, while single destination firms exporting multiple products decreasedfrom 17.73% in 2000 to 13.29% in 2006. Interestingly, highly-diversified firms accountedfor 87.42% of total sales in 2000 and 90.97% in 2006.A number of studies have documented the presence of regularities in empirical data oftrade flows and bilateral trade flows (see, for example Melitz et al., 2008; Easterly et al.,2009; Fagiolo et al., 2009; Sufrauj et al., 2015). The main stylized facts of trade andbilateral trade flows are: (i) sparsity, which implies the presence of many zeros, (ii) skeweddistribution at the extensive margin of trade (power law distribution), (iii) concentrationat the intensive margin (log-normal distribution), and (iv) high volatility in the growthrates (fat tailed distributions).Some of these statistical regularities have been also confirmed at the firm level (see,for example Bee et al., 2017). Bellow, we carry on a statistical analysis in order to explorewhether our firm level data present these empirical regularities. Our time period allows usto study trade patterns in the context of liberalization, so we present the analysis for 2000and 2006.Figure 1 shows the distribution of firms exports for the years 2000 and 2006. Firstly,we observe for both years a wide variability in the size of firms measured by total exports.8 D e n s i t y ln(X) fi t 2000Kernel fi t 2006 Figure 1: Distribution of firms’ total exports
Both distributions resemble a log-normal, which seem somehow left-skewed but still quitesymmetric. These distributions evidence the broad heterogeneity of Chinese firms, with alarge number of relatively small exporters that coexist with a few very large exporters.The empirical literature on trade has shown that firms and also countries grow anddecline driven by extreme episodes of expansion and contraction that are relatively frequent(Fagiolo et al., 2009). This is reflected in fat tailed distributions of growth rates acrosscountries and firms, different levels of sectoral disaggregation, and for other measuresof size (see, for example, Lee et al., 1998; Bottazzi, 2001; Bottazzi and Secchi, 2003).This fat tailed ubiquity in the distribution of growth stresses the existence of stronginterdependence or correlating mechanisms that, ultimately, determine growth patterns.
Density g Figure 2: Distribution of exports growth of Chinese firms for 2001 and 2006. AEP: asymmetricexponential power This implies that firms were morefrequently affected by negative shocks after 2000, and also, as the left tail is wider, thatnegative growth rates were higher.
Density g BigMediumSmall
Figure 3: Kernel estimation of the empirical firm growth rate distributions for the year 2006 bysize binned in three equipopulated categories: big, medium, and small
Figure 3 shows the distributions of growth rates for 2006 and for three differentequipopulated size bins of exporting firms: small, medium and large. We observe in theright side (positive growth), that small firms (that export little) grow more than mediumand, especially, than large firms. In the left side (negative growth), the probability of facingnegative shocks is more uniform for different size bins. Therefore, firms of different size aresimilarly affected by negative growth rates, but smaller firms have higher probabilities ofincreasing their exports at high rates compared to medium and large firms. It is interesting For the sake of simplicity, we do not present the estimations for every year, but the distributionspresent more volatility for all the years compared to the distribution of growth rates for 2000-2001, this isbefore China became a member of the WTO.
10o note that the distance between the left and right tails increases with size, which impliesthat large firms have less volatility in their trade growth rates.Regaring the ubiquity in the fat tailed distribution of growth rates of firms, Dosi (2007)claims that the variance-scale relation depends on the relation between diversification andsize. The growth of firms can be explained by both the increase in the existing lines ofbusiness, as well as through the diversification of the existing ones. This is also interestingfor the growth of trade and refers to the effect of the intensive and extensive margins oftrade.
In this section, we investigate how product and trade partners diversification have affectedChinese exporting firms. We claim that diversification in products and destinations canpositively affect firms trade growth. Firstly, we analyze the intensive and extensive marginsof trade. Next, we investigate whether exporting firms present economies of scope relatedto both product and destination diversification. Finally, we study how diversification inproducts and destinations affect the growth of firms’ exports.
In the first place, we analyze the extensive and intensive margins of exports in bothdestinations and products, which can provide a broad picture of the diversification patterns.For the case of diversification in destinations, the extensive margin for a firm i is definedas the number of exporting destinations N d,i , and the intensive margin as the averagetotal exports by destination X i /N d,i . In a similar way, in the case of diversification inproducts, the extensive margin is defined as the number of exporting products N p,i , andthe intensive margin is defined as the average trade by products X i /N p,i .We are interested in determining whether firms have specialized or diversified theirexports between 2000 and 2006. The question we want to address is whether the evolutionof exports and the creation/destruction of markets (products and destinations) have beengoverned by the intensive margin, the extensive margin, or both. As mentioned, from atheoretical point of view, a consequence of competing in international markets is that firmsare expected to specialize in more productive products (Melitz et al., 2008). Therefore, apossible outcome of trade liberalization is that firms will intensify their exports in someproducts and destinations. This means that, in the context of increasing competition, theextensive margin will be reduced while the intensive margin will be expanded.We select the sample of surviving firms from 2000 to 2006, and for each of themwe calculate the ratio of the extensive margin and the ratio of the intensive margin for11006 over 2000. Correspondingly, when these ratios are greater than one, the extensiveand intensive margins of a firm have enlarged. Conversely, if the ratio is less than one,the extensive and intensive margins have shrunk. We do this for both diversificationin products and in destinations of exports. Table 2 shows the percentage of firms withchanges in their intensive and extensive margins between 2000 and 2006. Table 2: Changes in the intensive margin (IM) and the extensive margin (EM) of exportsbetween 2000 and 2006 for surviving firms in percentage
Product (%) Destination (%)IM enlargement 59.14 55.25EM enlargement 47.81 54.76
Joint changes of margins
EM and IM shrink 20.23 21.31EM shrinks and IM enlarges 31.96 23.94EM enlarges and IM shrinks 20.63 23.44EM and IM enlarge 27.17 31.31
Notes:
The number of firms was 46,279 in 2000 and 166,930 in2006. We used 27,415 surviving firms to estimate the margins oftrade.
The first thing to notice is the high increase in the number of firms and the relativelysmall number of surviving firms. In 2000, there were 46,279 exporting firms , whichreached, in 2006, 166,930 exporting firms, increasing 3.61 times. The number of survivingfirms is 27,415, 59.24% of the existing firms in 2000 and 16.42% of total firms in 2006. Itis worth mentioning that not necessarily the missing amount of firms actually died, theycould have merged with other international affiliates, changed their business name, or alsothey could have not traded in the international markets but only in the domestic market.However, this reflects a dynamic market in which there is exit and entry of exporting firms.Analyzing diversification in products, we observe that 59.14% of firms were able toincrease their intensive margins of exports. In the case of destinations, the conclusionis quite similar, we observe that 55.25% of firms increase their intensive margins. Tosome extent, this supports the hypothesis that trade liberalization leads to specialization.However, we also observe a high percentage of firms that diversify their exports throughthe enlargement of their extensive margin both in products (47.81%) and destinations(54.76%). Thus, there is no conclusive evidence of a general process towards specialization.Also, we analyze the joint changes in the margins of exports. If we consider products,a firm can: (i) shrink both the extensive and intensive margins, meaning that it specializesin a lower number of products and exports less per product (20.23% of firms), (ii) shrinkthe extensive margin while enlarging the intensive margin, therefore, specializing in a lowernumber of products and exporting more per product (31.96%), this implies specialization–in line with the theoretical expected outcome we discussed above–, (iii) enlarge theextensive margin while shrinking the intensive margin, meaning that it diversifies the12xported products but exports less per product (20.63%), and (iv) enlarge both marginsof trade, which means that the firm diversifies its exported products and increases thequantity of exports per product (27.17%). Also, in the case of destinations, we observethat a relatively similar percentage of firms appear in each of the three first possibilities(21.31, 23.94, and 23.44%, respectively), but a slightly higher percentage of firms (31.31%)diversify their destinations and increase the volume of exports to those destinations,enlarging both the intensive and extensive margins.Notice that, an increase in the intensive margin can indicate the existence of economiesof scale. While there is a percentage of firms that specialize and intensify their exports, astheoretically expected in the context of liberalization, the proportion of firms that manageto diversify their exports is not negligible at all. The expansion of the extensive marginderived from an increase in the number of products and destinations for an important partof the population of Chinese exporters might indicate the existence of economies of scope.Overall, the analysis of the changes on the margins of trade between 2000 and 2006evidences heterogeneous patterns in the dynamics of exporting firms, both in productsand destinations, as expected from an evolutionary perspective. However, it is importantto highlight that these heterogeneous changes could be also related with changes made byChina in its industrial and foreign trade policies, such as attracting more foreign capitaland integrating to global value chains.
We are interested in studying how diversification affects the growth of exports. Thus,we test whether exports of Chinese firms exhibit economies of scope. In the context ofour study, and given the limitations of the data, the economies of scope mean that thenumber of existing markets of a firm (exported products or destinations of exports) canbe determined by the total volume of exports of the firm itself. In addition, the existenceof economies of scope imply that the average total cost of production decreases as a resultof increasing the number of different products exported or of increasing the number ofdestinations. If Chinese exports have economies of scope, we expect firms to increase theirnumber of products and destinations as their exports grow.The diversification behavior of firms can be modelled by a birth process, this is bydescribing how the probability of having a given number of markets changes as thefirm grows. Bottazzi and Secchi (2006) showed that the Yule process provides a goodcharacterization of the diversification patterns of the pharmaceutical industry. As presentedin Feller (1968), in this process it is considered that a product can give birth to a newproduct but cannot die. Therefore, given a small expansion of exports of length h for a firm,each of its markets (products or destinations) has probability λh + o ( h ) of creating a new13arket, where the parameter λ determines the rate of increase of the population. Assumingthat at size s there is no interaction between markets (no instantaneous branching), thenfor a firm with n -markets the probability of diversifying in one market between s and s + h is nλh + o ( h ). The probability P n ( s ) that the number of markets is exactly n satisfies thefollowing differential equation:˙ P n ( s ) = − nλP n ( s ) + ( n − P n − ( s ) with: n ≥ n ˙ P n ( s ) = − n λP n ( s ); (2)where n > s . It can be verified that thesolution of this equation for n ≥ n is P n ( s ) = (cid:18) n − n − n (cid:19) e − n λ ( s − s ) (1 − e − λ ( s − s ) ) n − n . (3)This is called Yule distribution and the average number of markets (products anddestinations) is: η ( s ) = n e λ ( s − s ) . (4)Given that the parameter λ is strictly positive, the number of markets is expected to growexponentially with firm size. ND ln(X) Exp. fit NP ln(X) Exp. fit
Figure 4: Scatter plots of binned statistics: total exports versus number of destinations (ND)and versus number of products (NP), and exponential fit for 2006
Figure 4 (left) shows the average number of destinations of the firms belonging todifferent size bins against the bin average (ln) total exports. Similarly, Figure 4 (right)shows the average number of products of firms belonging to different size bins against thebin average (ln) total exports. The evidence suggests the above mentioned exponentialpositive relation between the number of markets and total exports. The continuous line in We present the estimations for the year 2006, but similar relations are estimated for all years. Resultsare available upon request.
N D ) ∼ λ d ln( X ) + θ d , (5)ln( N P ) ∼ λ p ln( X ) + θ p . (6)The estimated values are: λ d = 0 . .
01) and λ p = 0 . .
01) and θ d = − . . θ p = − . . In order to further investigate the effect of diversification of products and destinations ofexports on firms trade growth, we carry out an econometric exercise. Given the evidenceof economies of scope on the exports of Chinese firms, we expect diversification to havea positive effect on exports growth rates. Given that diversification depends on thedevelopment of capabilities, diversification of destinations might be relatively easier toachieve for a firm compared to diversification of products, which implies the developmentof a new exportable product. In addition, we include in the estimations several controlvariables that aim to capture structural changes experienced between 2000 and 2006.We estimate the following model: g i,t = α + φ · g i,t − + β · ln( X i,t − ) + θ p · ∆ ln( N P i,t ) + θ d · ∆ ln( N D i,t )+ ω u · F irmT ype i,t + ρ · P rocT rade i,t + ν v · Ownership i,t + γ v · Ownership i,t · P rocT rade i,t + τ t + (cid:15) i,t ; (7)where the dependent variable g is the growth rates of exports as defined in equation (1),and g i,t − is the growth rates of exports in t − X i,t − ) is thevolume of exports in t − N P i,t ) is thechange in the number of products a firm exports from period t to t −
1, ∆ ln(
N D i,t ) is thechange in the number of destinations of exports from t − t , ω u controls for a set offour firm diversification types, where ω is a firm that exports multiple product to multipledestinations, this is, highly-diversified firms, ω is a highly-specialized firm, this is, singleproduct and destination firms, ω is a multi-product and single destination firm, and ω is a single product and multiple destination firm, ρ controls if the firm i does processingtrade at time t , ν v controls for a set of three ownership types: with ν includes state-ownedand collective-owned enterprises, ν private-owned enterprises, and ν are foreign-owned15nterprises, including joint ventures, γ v controls for the interaction between ownershiptype and processing trade, τ t is a set of time-dummies. Finally, (cid:15) i,t is the residual term.We estimate the model with ordinary least squares (OLS) and an asymmetric leastabsolute deviation (ALAD) estimation methods. The error term is assumed to be normallydistributed in the OLS estimations and Laplacian distributed in the ALAD estimations.The ALAD is preferred to OLS when there are outliers and when the distribution ofthe residuals is non-normal, asymmetric, and has high kurtosis. The assumptions of theALAD better agree with what we observed in Figure 2. Table 3 shows the results of theestimations of the growth of exports. Models (1-4) show the results of the OLS estimationsand models (5-8) report the results of the ALAD estimations.In the different specifications estimated with OLS, the growth rate of the previousperiod has a negative effect in the growth of exports. Conversely, when considering thenon-normality of the distribution of the data (models 5-8), the growth rates in t − t . This is the main differencebetween the results of the OLS and the ALAD estimation methods. However, despite theestimations of the autoregressive term report opposite signs using OLS and ALAD, theestimated coefficients are relatively small.In all the estimated models, smaller firms (that trade low quantities) are expected togrow more compared to larger firms (exports ( t − ). This agrees with the fact that theright tail of the growth rates distribution allows for larger events for smaller firms (seeFigure 3).Model (1) and (5) report the results for our benchmark model. We obtain robustestimations that indicate that diversification in both products and destinations havepositive effects on the growth of exports in all the estimated models (1-8). Also, in allthe estimated models, diversification in the destinations of exports increases trade growthrates more than diversification of products.The variables indicating the type of firm in terms of diversification and specializationin products and destinations can shed further light on these effects. Models (2) and(6) present the results using highly-diversified firms (multiple products and destinations)as the baseline. We observe that being a highly-specialized firm has a negative impacton growth rates compared to being a highly-diversified firm and also compared to firmsthat are specialized in products or destinations but diversified in either destinations orproducts. In general, all types of less diversified firms are associated with lower growthrates of exports compared to highly-diversified firms. Also, it is interesting to note thatthe difference is lower for the case in which firms are diversified in trade partners butspecialized in products, which enriches the idea that diversifying in destinations generateshigher growth opportunities than diversifying in products.Models (2) and (6) also includes a dummy indicating whether the firm does processingtrade. We estimate significant and positive coefficients for the dummy, which means that16 able 3: The effect of product and trade partner diversification on the growth of exports.Econometric estimations Estimation Method OLS ALADModel (1) (2) (3) (4) (5) (6) (7) (8)Growth ( t − -0.026*** -0.026*** -0.028*** -0.025*** 0.012*** 0.020*** 0.017*** 0.019***(0.002) (0.002) (0.002) (0.002) (0.003) (0.001) (0.001) (0.001)Exp. ( t − -0.063*** -0.100*** -0.102*** -0.103*** -0.049*** -0.057*** -0.060*** -0.060***(0.001) (0.001) (0.001) (0.001) (0.002) (0.001) (0.001) (0.001)∆ ln products t t × private-owned 0.211*** 0.116***(0.008) (0.005)PT × foreign-owned 0.100*** 0.121***(0.004) (0.006)PT × state-owned 0.252*** 0.027***(0.007) (0.002)Constant 0.777*** 1.299*** 1.343*** 1.337*** 0.791*** 0.931*** 0.975*** 0.976***(0.017) (0.019) (0.020) (0.019) (0.021) (0.009) (0.009) (0.009)Year dummies yes yes yes yes no no no noObservations 311,034 311,034 311,034 311,034 311,034 311,034 311,034 311,034 Notes:
The dependent variable is the growth rate of exports. The number of products is at 6-digit of the HS. Robust standarderrors are in parentheses. Significance level: *** p < < < firms that do processing trade grow more than firms doing only ordinary trade.In models (3) and (7), we also include variables indicating the ownership type offirms, using state-owned enterprises as the baseline. The results show that private-ownedenterprises grow more than state-owned enterprises but foreign-owned enterprises grow lesscompared to both state-owned and private-owned enterprises. This is surprising to someextent but, foreign-owned enterprises are highly involved in processing trade (57.69%),while only 27.05% of state-owned enterprises and 9.80% of private-owned enterprises doprocessing trade. Moreover, the shares of processing trade in total trade of each type offirms is quite different. While foreign-owned enterprises have a share of between 78 and 84%17f processing trade in their total exports, this share reaches only between 28 and 34% forstate-owned enterprises, and between 8 and 17% for private-owned enterprises, dependingon the year. Considering this, the dummy for processing trade might be capturing to agreater extent the behavior of foreign-owned enterprises. Thus, in models (4) and (8), weconsider the interaction between ownership type and processing trade. The estimatedcoefficients show that private-owned and foreign-owned enterprises that do processingtrade grow more than state-owned enterprises that do processing trade.The lack of data characterizing firms prevents us from making a more thorougheconometric analysis. However, the estimations using different specifications show thatdiversification in products and destinations of exports increase growth rates of exports.This is confirmed for different types of firms in terms of diversification in products anddestinations: more diversified firms show higher growth rates while more specialized firmshave lower grow rates. In particular, diversification in destinations has a stronger effectin exports growth rates. This implies that reaching a new destination leads to highergrowth opportunities than starting a new production line. Moreover, facing a shock in theinternational market (for example, a global or regional crisis), it could be easier for a firmto redirect exports to a different destination than to reallocate resources to develop newor different products, especially, considering that industries usually present stickiness inthe reallocation of resources from one sector to another. This paper studies trade patterns of Chinese firms for the period 2000 to 2006. Thestatistical analysis reveals that there exists high heterogeneity in Chinese exporting firms.The distribution of exports resembles a log-normal distribution. Many relatively smallexporting firms coexist with a few very large exporting firms. Also, we observe fat tails inboth sides of the distribution of growth rates, which implies that firms are often affectedby extreme events. In addition, there is a negative relation between size and volatility ofthe growth rates. The distribution of exports growth rates shows that firms of differentsize (in terms of exports) are similarly affected by negative growth rates, but smaller firmshave higher probabilities of growing at higher growth rates than medium and large firms.Finally, after trade liberalization, firms face more frequently stronger negative shocks.The analysis of the changes in the intensive and extensive margins of exports showedthat more than half of the firms intensified their exports between 2000 and 2006. But also,an important part of the firms diversified products (47.81%) and destinations (54.76%)enlarging their extensive margins. We showed that the exports of Chinese firms presenteconomies of scope, which can explain, to a certain extent, the increase in the diversificationof exported products and destinations of exports.The econometric estimations showed a negative relation between size and growth, a18ositive impact of diversification in products and destinations on trade growth rates, asignificant effect of different ownership types of firms, and that firms involved in processingtrade also exhibit higher growth rates. The results are robust to different specification ofthe model and to different estimation methods.Interestingly, highly-diversified firms grow more than highly-specialized firms and alsothan firms that diversified in products or destinations but specialized in destinations ofproducts, respectively. Diversification in destinations have a stronger positive effect thandiversification in products. This implies that facing more competition, firms might benefitmore from exporting to a new destination rather than starting a new production line orreallocating resources to produce and export new products. In fact, we can consider thatthe development of a new exportable product can require more capabilities than reachinga new destination with an old product. Also, this suggests that, facing a negative shock,it could be easier for a firm to redirect exports to a different country than reallocatingresources to develop new products. 19 eferences
Amiti, M. and B. S. Javorcik (2008). Trade costs and location of foreign firms in China.
Journal of Development Economics 85 (1), 129–149.Bee, M., M. Riccaboni, and S. Schiavo (2017). Where Gibrat meets Zipf: Scale and scopeof French firms.
Physica A: Statistical Mechanics and its Applications 481 , 265–275.Bernard, A., J. Jensen, S. Redding, and P. Schott (2007). Firms in international trade.
The Journal of Economic Perspectives 21 (3), 105–130.Bernard, A., S. Redding, and P. Schott (2010). Multiple-product firms and productswitching.
American Economic Review 100 (1), 70–97.Bernard, A., S. Redding, and P. Schott (2011). Multiproduct firms and trade liberalization.
The Quarterly Journal of Economics 126 (3), 1271–1318.Bottazzi, G. (2001). Firm diversification and the law of proportionate effect. arXiv preprintcond-mat/0103575 .Bottazzi, G. and A. Secchi (2003). Why are distributions of firm growth rates tent-shaped?
Economics Letters 80 (3), 415–420.Bottazzi, G. and A. Secchi (2006). Gibrat’s law and diversification.
Industrial andCorporate Change 15 (5), 847–875.Carrere, C., O. Cadot, and V. Strauss-Kahn (2011). Trade diversification: drivers andimpacts. In M. Jansen, R. Peters, and J.-M. Salazar-Xirinachs (Eds.),
Trade andEmployment: from Myths to Facts , pp. 253–307. Geneva: ILO-EC International LabourOffice - European Commission.Di Clemente, R., G. L. Chiarotti, M. Cristelli, A. Tacchella, and L. Pietronero (2014).Diversification versus specialization in complex ecosystems.
PloS one 9 (11), e112525.Dosi, G. (2007). Statistical regularities in the evolution of industries. a guide throughsome evidence and challenges for the theory. In F. Malerba and S. Brusoni (Eds.),
Perspectives on Innovation , pp. 153–186. Cambridge, MA: Cambridge University Press.Dosi, G., M. Grazzi, and D. Moschella (2015). Technology and costs in internationalcompetitiveness: from countries and sectors to firms.
Research Policy 44 (10), 1795–1814.Dosi, G., K. Pavitt, and L. Soete (1990).
The economics of technical change andinternational trade . New York [etc.]: Harvester Wheatsheaf.Easterly, W., A. Reshef, and J. M. Schwenkenberg (2009). The power of exports. WorldBank Policy Research Working Paper Series, Vol. 5081.Fagiolo, G., J. Reyes, and S. Schiavo (2009). World-trade web: Topological properties,dynamics, and evolution.
Physical Review E 79 (3), 036115.Feller, W. (1968).
An introduction to probability theory and its applications: volume I ,Volume 3. John Wiley & Sons New York.Hausmann, R. and D. Rodrik (2003). Economic development as self-discovery.
Journal ofDevelopment Economics 72 (2), 603–633. 20idalgo, C. A., B. Klinger, A.-L. Barab´asi, and R. Hausmann (2007). The product spaceconditions the development of nations.
Science 317 (5837), 482–487.Ianchovichina, E. and W. Martin (2001). Trade liberalization in China’s accession toWTO.
Journal of Economic Integration 16 (4), 421–445.Jarreau, J. and S. Poncet (2012). Export sophistication and economic growth: Evidencefrom China.
Journal of Development Economics 97 (2), 281–292.Kim, W. C., P. Hwang, and W. P. Burgers (1993). Multinationals’ diversification and therisk-return trade-off.
Strategic Management Journal 14 (4), 275–286.Koopman, R., Z. Wang, and S.-J. Wei (2012). Estimating domestic content in exportswhen processing trade is pervasive.
Journal of Development Economics 99 (1), 178–189.Lee, Y., L. Nunes Amaral, D. Canning, M. Meyer, and H. Stanley (1998). Universal featuresin the growth dynamics of complex organizations.
Physical Review Letters 81 (15), 3275.Lipczynski, J. and J. Wilson (2001).
Industrial organisation: an analysis of competingmarkets . Financial Times Prentice Hall.Manova, K. and Z. Zhang (2009). China’s exporters and importers: Firms, products andtrade partners.
NBER Working Paper Series , 15249.Mathew, N. (2017). Drivers of firm growth: micro-evidence from Indian manufacturing.
Journal of Evolutionary Economics 27 (3), 585–611.Melitz, M., E. Helpman, and Y. Rubinstein (2008). Estimating trade flows: Tradingpartners and trading volumes.
Quarterly Journal of Economics 123 (2), 441–487.Melitz, M. J. (2003). The impact of trade on intra-industry reallocations and aggregateindustry productivity.
Econometrica 71 (6), 1695–1725.Roberts, M. J. and J. R. Tybout (1997). The decision to export in Colombia: an empiricalmodel of entry with sunk costs.
The American Economic Review , 545–564.Rodrik, D. (2006). What’s so special about China’s exports?
China & WorldEconomy 14 (5), 1–19.Rose, A. (2004). Does the WTO really increase trade.
American Economic Review 94 (1),98–114.Subramanian, A. and S.-J. Wei (2007). The WTO promotes trade, strongly but unevenly.
Journal of International Economics 72 (1), 151–175.Sufrauj, S. B., S. Schiavo, and M. Riccaboni (2015). Big hits, export concentration andvolatility.
Economia Politica 32 (2), 135–166.Wagner, J. (2007). Exports and productivity: A survey of the evidence from firm-leveldata.
The World Economy 30 (1), 60–82.Wagner, J. (2014). Is export diversification good for profitability? First evidence formanufacturing enterprises in Germany.
Applied economics 46 (33), 4083–4090.Wang, Z. and S.-J. Wei (2010). What accounts for the rising sophistication of China’sexports? In R. C. Feenstra and S.-J. Wei (Eds.),
China’s Growing Role in World Trade ,pp. 63–104. Chicago: University of Chicago Press.21ing, Y. (2014). China’s high-tech exports: The myth and reality.
Asian EconomicPapers 13 (1), 109–123.Yao, S. (2009). Why are Chinese exports not so special?
China & World Economy 17 (1),47–65.Zaccaria, A., M. Cristelli, A. Tacchella, and L. Pietronero (2014). How the taxonomy ofproducts drives the economic development of countries.