[PDF] Searches related to indeed france filetype:pdf

View - Download Searches related to indeed france filetype:pdf

Previous PDF

Next PDF

NBER WORKING PAPER SERIESTHE MICRO ORIGINS OF INTERNATIONAL BUSINESS CYCLE COMOVEMENTJulian di GiovanniAndrei A. LevchenkoIsabelle MejeanWorking Paper 21885http://www.nber.org/papers/w21885NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts AvenueCambridge, MA 02138January 2016We would like to thank workshop participants at several institutions for helpful discussions, and EdaGulsen for expert research assistance. Di Giovanni gratefully acknowledges the CREI and the MarieCurie International Incoming Fellowship FP7-PEOPLE-2013-IIF (under Grant Agreement 622959)for financial support. Méjean gratefully acknowledges support from a public grant overseen by theFrench National Research Agency (ANR) as part of the "Investissements d'Avenir" program (IdexGrant Agreement No. ANR-11-IDEX-0003- 02/Labex ECODEC No. ANR-11-LABEX-0047 andEquipex reference: ANR-10- EQPX-17 Centre d'accès sécurisé aux données CASD). The views expressedherein are those of the authors and do not necessarily reflect the views of the National Bureau of EconomicResearch.NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies officialNBER publications.© 2016 by Julian di Giovanni, Andrei A. Levchenko, and Isabelle Mejean. All rights reserved. Shortsections of text, not to exceed two paragraphs, may be quoted without explicit permission providedthat full credit, including © notice, is given to the source.

The Micro Origins of International Business Cycle ComovementJulian di Giovanni, Andrei A. Levchenko, and Isabelle MejeanNBER Working Paper No. 21885January 2016JEL No. F44,F61,F62ABSTRACTThis paper investigates the role of individual firms in international business cycle comovement usingdata covering the universe of French firm-level value added, bilateral imports and exports, and cross-borderownership over the period 1993-2007. At the micro level, controlling for firm and country effects,trade in goods with a particular foreign country is associated with a significantly higher correlationbetween a firm and that foreign country. In addition, foreign multinational affiliates operating in Franceare significantly more correlated with the source economy. The impact of direct trade and multinationallinkages on comovement at the micro level has significant macro implications. Because internationallyconnected firms are systematically larger than non- internationally connected firms, the firms directlylinked to foreign countries represent only 8% of all firms, but 56% of all value added, and accountfor 75% of the observed aggregate comovement. Without those linkages the correlation between Franceand foreign countries would fall by about 0.091, or one-third of the observed average business cyclecorrelation of 0.29 in our sample of partner countries. These results are evidence of transmission ofbusiness cycle shocks through direct trade and multinational ownership linkages at the firm level.Julian di GiovanniDepartment of Economics and BusinessUniversitat Pompeu FabraRamon Trias Fargas 25-2708005 BarcelonaSpainjulian.digiovanni@upf.eduAndrei A. LevchenkoDepartment of EconomicsUniversity of Michigan611 Tappan StreetAnn Arbor, MI 48109and NBERalev@umich.eduIsabelle MejeanDépartement d'EconomieEcole Polytechnique91128 Palaiseau CedexFranceisabelle.mejean@polytechnique.edu

1 Introduction

Countries that exhibit greater bilateral trade and multinational production linkages have more correlated business cycles (

Frankel and Rose

,1998;Kleinert et al.,2015). While the empirical literature has repeatedly conrmed the trade-comovement relationship in the data, its meaning is not well-understood, either empirically or quantitatively. Taken at face value, the positive association between bilateral trade and multinational linkages and comovement is often interpreted as evidence of transmission of shocks across countries through those linkages. The empirical literature has faced two related challenges. The rst is the critique byImbs 2004
) that countries that trade more with each other are similar in other ways, and thus subject to common shocks. Under an extreme version of this view, the trade linkage variable in the Frankel-Rose specication does not re ect the intensity of transmission of shocks, but rather is simply a stand-in for the prevalence of common shocks. The second is that even if one accepts the transmission of shocks interpretation of the Frankel-Rose result, the coarse nature of the cross-country setting makes it dicult to learn about the micro underpinnings of the trade-comovement relationship. This lack of understanding is reinforced by the quantitative literature, which has struggled to capture the trade-comovement relationship. Kose and Yi(2006) andJohnson(2014) show that even quite sophisticated IRBC models fail to generate the observed positive association, dubbing it the \trade-comovement puzzle." 1 Until now the properties of international comovement at the rm level, or its aggregate implications, have by and large not been studied. This paper provides a forensic account of international comovement at both the micro and macro levels using data covering the universe of French rm-level value-added, destination-specic imports and exports, and cross-border ownership over the period 1993-2007. Examining cross-border comovement at the rm level has two advantages relative to the traditional approach of looking directly at GDP correlations. First, at the micro level, the data allow for precise measurement of trade and multinational linkages { by rmcountry { and to control for common shocks using appropriate xed eects. This overcomes the common shocks critique and lets us establish much more rmly that the positive trade-comovement relationship is due at least in part1 The literature on multinationals and international business cycle comovement is more limited, but

shares this feature.Kleinert et al.(2015) show that French regions that contain more multinationals from a

particular foreign country are more correlated with that country. However,Cravino and Levchenko(2015)

show that the observed multinational presence alone cannot generate the level of positive comovement found

in the data.Liao and Santacreu(2015) develop a model in which technology shocks are transmitted between

countries through changes in the mix of imported inputs, and show that allowing for the extensive margin

of trade yields more promising results. 1 to transmission of shocks at the rm level. Second, at the macro level, our approach allows us to capture the aggregate comove- ment implications of heterogeneity across rms in both size and the extent of international linkages. Larger rms are disproportionately more likely to trade internationally and own aliates in foreign countries. Indeed, in most countries international trade ows are dom- inated by only a handful of large rms. An emerging research agenda in closed-economy macro has argued convincingly that modeling and measuring shocks at the micro level (to rms and sectors), and linkages between them, is essential for understanding aggregate uctuations.

2If large rms and rm-to-rm linkages matter for aggregate

uctuations, a natural conjecture is that they will matter as much if not more for cross-border comovement. We begin by estimating a specication inspired by Frankel and Rose (1998), that relates the correlation of rm total value added growth with foreign GDP growth to rm-level direct linkages to that country. The data contain, for each rm and potential partner country, four types of direct linkages: (i) importing from it, (ii) exporting to it, (iii) being a France-based aliate of a multinational rm headquartered in that country; (iv) being a French rm with a foreign aliate in that country. Because the sample includes many rms and countries, estimation controls for both rm and country eects. Country eects in particular absorb the common aggregate shocks aecting France and each foreign country. In a sample of rm-level correlations with 10 large trading partners of France, trade linkages at the rm level are signicantly associated with increased comovement between an individual rm and the country with which it trades. An import link increases the correlation by 0.012, and an export link by 0.005. This is large relative to the average cor- relation between an individual rm and foreign GDP, which is 0.024 for directly connected rms, and essentially zero for non-directly connected ones. By a similar token, aliates of foreign multinationals operating in France have a 0.01 higher correlation with their source countries. At the same time, the empirical exercise reveals the importance of common shocks in the data. In a specication that omits the 10 country xed eects but still includes the

1 million rm eects, the coecients on the direct linkages variables are 2-5 times larger

in magnitude and all strongly statistically signicant. This underscores both the empirical2

Gabaix

2011
d iGio vanniand Lev chenko(2012), andCarvalho and Grassi(2015) develop models in which aggregate uctuations arise from shocks to individual rms, because the rm-size distribution is extremely fat-tailed (Zipf's Law).Acemoglu et al.(2012) andCarvalho and Gabaix(2013) argue that sectoral shocks lead to aggregate uctuations through interconnections between sectors.Di Giovanni et al. 2014
) andAtalay(2014) provide corresponding empirical evidence on the role of shocks to rms and sectors in aggregate uctuations. 2 relevance of common shocks, and how important it is to control for them in \gravity-macro" analyses of the eects of bilateral trade and capital ows linkages on aggregate outcomes. Nonetheless, the results when controlling for common shocks still provide clear evidence of transmission through direct linkages at the rm level. We then use the sector-level Input-Output table together with rm-level information on input purchases and domestic sales to construct proxies for indirect linkages between French rms and foreign markets. The measures, inspired by the \network eect" prop- agation terms inAcemoglu et al.(2015), capture the intensity with which a French rm interacts with internationally connected rms. The downstream indicator re ects whether a rm buys intermediate inputs from rms that import from a particular country. The up- stream indicator captures whether a rm sells its output to rms that export to a particular country. Both of these measures are rm-andforeign country-specic. We augment the main specication with these indirect linkage terms, and show that the downstream indirect linkages do matter signicantly for rm-level comovement with foreign markets. Firms that buy inputs from importers from a particular country are more correlated with that country. The evidence on upstream linkages is more mixed, with coecients diering in sign and signicance depending on specication. The second half of the analysis examines the macro implications of the micro-level ndings. We start with the observation that the aggregate business cycle correlation between France and another country is simply an appropriately weighted sum of the correlations of rm-level total value added with that country. The aggregate business cycle correlation between France and each country can thus be written as a sum of two terms: the part due to the directly connected rms, and the part due to the not directly connected rms. For the

10 large trading partners of France in our sample, we show that the large directly connected

rms are important in accounting for aggregate comovement. For a typical foreign partner country, the directly connected rms represent only about 8% of all rms in our dataset, but account for 56% of total value added. The directly connected rms are also unconditionally more correlated with the foreign country. Together these two facts imply that the directly connected rms account for 75% of the aggregate business cycle correlation observed in the data for the median country. We then use the conditional relationship between direct linkages and rm-level correla- tions to compute the change in the aggregate correlation between France and each foreign country that would occur if direct linkages at the rm level disappeared. The aggregation exercise combines information on the change in the correlation at the rm level from the 3 regression estimates with rm-level weights. If direct linkages at the rm level were sev- ered, the aggregate correlation would fall by 0.091 on average in our sample of 10 partner countries. This is a non-negligible change relative to the observed correlations between France and its main trading partners of 0.29 on average over this period. Since our data allow us to estimate the coecients on trade and multinational links separately, we can also check which ones matter more for generating aggregate comovement. It turns out that the trade linkages are about 9 times more important in generating aggregate comovement than multinational linkages, accounting for 0.083 of the overall 0.091 eect. Augmenting the aggregate impact with the indirect linkage estimates, we show that indirect linkages are quantitatively important as well. Accounting for indirect linkages implies that aggregate correlation would fall by 0.13 on average in the whole economy if links to the foreign country were severed. Thus, direct and indirect linkages together can account for nearly half of the average 0.29 observed aggregate correlation. The results are even stronger in manufacturing. Direct and indirect linkages produce a correlation of 0.414 between French manufacturing and foreign GDP, accounting for the bulk of the

0.484 average correlation in the data. Indirect linkages have a much larger impact in the

manufacturing sector compared to the whole economy, alone accounting for 0.343 of the

0.414 total impact, more than 80%.

To summarize, on the one hand the data point clearly to the presence of common shocks, implying that it is imperative to control for them in the empirical exercise. On the other hand, even after controlling for common shocks, there is still substantial evidence of trans- mission of shocks through trade and multinational linkages. Among those linkages, trade linkages appear to matter more than multinational ones, especially in aggregate. Down- stream indirect input linkages are both statistically robust and quantitatively important as well. Our paper contributes to the empirical literature on international business cycle comove- ment. Studies building onFrankel and Rose(1998) have conrmed the positive association between trade and comovement and examined how it diers across types of goods trade and sub-samples of countries (see, e.g.Baxter and Kouparitsas,2005;Calderon et al.,2007;Ng, 2010
;Blonigen et al.,2014;Liao and Santacreu,2015). While the existing empirical litera- ture has almost exclusively used aggregate GDP correlations as the outcome variable, there has been comparatively little work on international comovement at more disaggregated lev- els.Di Giovanni and Levchenko(2010) estimate the relationship between bilateral trade, input linkages, and sector-level correlations. This paper's contribution is to examine the 4 trade-comovement relationship at the rm level, and to derive the aggregate implications based on micro-level estimates. In this respect, it shares some features with recent papers such asKurz and Senses(2015),Boehm et al.(2014), andCravino and Levchenko(2015), who perform related exercises. An important research agenda, going back toBackus et al.(1995), attempts to under- stand the positive GDP correlation across countries using representative rm models in which all shocks are aggregate. Later developments in this literature explored the role of the production structure, such as input-output linkages (

Burstein et al.

,2008;Arkolakis and Ramanarayanan ,2009), or rm heterogeneity (Ghironi and Melitz,2005;Alessandria and Choi ,2007) but have similarly been conned to considering only the role of aggregate productivity shocks in generating cross-country business cycle comovement. Our results suggest that to fully understand the impact of transmission of shocks for aggregate co- movement, a quantitative framework must feature a realistic micro-structure that combines granularity in the rm size distribution and systematic heterogeneity among rms in trade and multinational linkages. The rest of the paper is organized as follows.Section 2lays out the conceptual framework and the empirical exercises performed in the paper.Section 3describes the data, and

Section 4the results.Section 5concludes.

2 Conceptual Framework

Total value addedXtby all French rms in yeartis by denition given by:XtP f2Itxft, wherexftis dened as the value added of rmfin yeart, andItis the set of rms foperating att. The growth rate of aggregate value added is then dened simply as At=Xt=Xt11, where we assume thatXt1andXtare the aggregate value added of all rms that exist both att1 andt, i.e. we restrict attention to theintensive margin of aggregate value added growth.Appendix Adevelops a complete decomposition of the total value added growth into extensive and intensive margins, and presents the results for the relative contributions of the extensive and intensive margins to aggregate comovement between France and its main trade partners. The main result is that the large majority of aggregate comovement is accounted for by the intensive margin, with the extensive margin playing only a minor role. 33
These results are not inconsistent with the empirical ndings inLiao and Santacreu(2015), who show

that the extensive margin of trade is positively correlated with bilateral comovement. Those results relate

the cross-sectional variation in the number of products traded between country pairs to bilateral business

cycle comovement. Our extensive margin is the aggregate contribution of entry and exit of French rms 5 The growth rate of aggregate value added can be written as a function of individual rm growth rates and rm shares: At=X fw ft 1 ft;(1) where ftis the growth rate of value added of rmf, andwft1is the share off's value added in total French value added. The object of interest is the correlation between French aggregate growth and foreign

GDP growth. Let

Ctbe the GDP growth of a foreign countryCbetweent1 andt. This correlation is given by: At;

Ct) =Cov(

At; Ct)

AC;(2)

whereCis the standard deviation of countryCgrowth.

Combining

(1) and (2), the correlation between France andCat timetcan be written as At;

Ct) =CovP

fwft1 ft; Ct AC =X fw ft 1f A( ft;

Ct);(3)

wherefis the standard deviation of ft. While simply an identity,Equation (3)states the key premise of the paper: the aggregate correlation between the French economy and another country is an appropriately weighted sum of the rm-level correlations. The substantial literature on international comovement has studied empirically and theoretically the left-hand side of this equation { the aggregate business cycle comovement. This paper provides a picture of aggregate comovement by examining instead the components of the right-hand side. We proceed by analyzing rst the properties of the individual rm-level correlations( ft;

Ct), and then the consequences

of aggregation across rms.

2.1 Micro Evidence

Equation(3)emphasizes that the aggregate business cycle correlation between the French economy and foreign countries is a function of the individual rm-level correlations and

these rms' weights in the total French economy. We start by establishing whether thefrom year to year, an object that has no close relationship to the cross-country dierences in the number of

traded varieties. 6 direct trade and multinational linkages at the rm level to a particular foreign country are associated with a higher correlation between the rm and that foreign country. To that end, we estimate the following specication: ft;

Ct) =+1EXf;C+2IMf;C+3AFFf;C+4HQf;C+f+C+f;C:(4)

In equation

(4) , the correlation between a rm and a foreign marketCis related to binary indicators of whether the rm exports to there (EXf;C), imports from there (IMf;C), is a French multinational with aliates inC(HQf;C), or is an aliate of a foreign multinational headquartered inC(AFFf;C). Importantly, the specication admits both rm and country eects, allowing for a precise identication of transmission of shocks through direct linkages.

The specication is inspired by

F rankeland Rose (1998), who were the rst to establish the robust empirical benchmark in this literature: country pairs that trade more with each other exhibit higher business cycle correlation. However, the interpretation of this stylized fact and the mechanisms responsible for generating it are still not well-understood. Taken at face value, the positive association between trade intensity and comovement suggests transmission of shocks through international trade linkages. However, as argued byImbs 2004
), an alternative interpretation is that countries that trade more with each other are similar in other ways, and thus subject to common shocks. Under an extreme version of this view, the trade linkage variable in the Frankel-Rose specication does not re ect the intensity of transmission of shocks, but rather is simply a stand-in for similarity between countries. In addition, even if one accepts the transmission of shocks interpretation of the Frankel-Rose result, the coarse nature of the cross-country setting makes it dicult to learn about the micro underpinnings of the trade-comovement relationship. 4 Our rm-level specication leads to qualitatively new insights relative to the traditional cross-country empirical model. First, comparing rms within the same country and includ- ing country and rm xed eects addresses theImbs(2004) common shock critique. Since all rms in this specication are in France, country eects will absorb the common shocks aecting France and countryC. As a result, we can establish convincingly that trade and multinational linkages are indeed a source of transmission of shocks, rather than simply a stand-in for the presence of common shocks.Di Giovanni and Levchenko(2010) adopt a related strategy and use data on sector-level trade and comovement together with country- pair eects that control for common aggregate shocks. At the industry level, a link between4 These diculties in interpreting the cross-country relationship are underscored by the nding that traditional international business cycle models have trouble reproducing it quantitatively (

Kose and Yi

2006

Jo hnson

2014
7 sectoral correlation and sectoral trade intensity does not have a tight interpretation, since it is not clear that exports from one of the sector-countries in the pair are used by the other sector-country. Extending this approach to rm-level data is a signicant improvement in measurement, since we observe exactly which rms have direct links with which countries. While the country eects are useful at absorbing the macro common shocks, the rm eects allow us to control for heterogeneity at the rm level that varies by rm but not partner country: size, primary industry of operation, capital or skill intensity, access to external nance, R&D intensity, and so on. Second, estimation at the rm level reveals the micro underpinnings of the aggregate relationship. Observing cross-border links at the rm level allows us to establish with foren- sic precision the role of each type of trade and multinational relationship in international comovement. With very few exceptions (e.g.

Kleinert e tal.

2015
), existing papers do not combine information on both trade and multinational linkages in the same specication. This may be important: if both types of linkages potentially matter, not including them will lead to omitted variable bias. Information at the rm level paired with rm eects also enables us to control for heterogeneity across rms in volatility and comovement, as well as for forces that shape a rm's average comovement with the rest of the world (such as its domestic linkages, for instance).

2.1.1 Indirect Linkages

It may be that even rms not directly connected to a particular foreign country comove with that country through indirect linkages, that is, interactions inside the French economy induced by the directly connected rms. A complete account of all indirect linkages is not possible in this empirical setting, as those linkages can take a variety of forms, from purchases/sales of intermediate inputs by the directly connected rms to changes in factor and goods prices in general equilibrium. Nonetheless, we attempt to capture one type of indirect linkage, that can be potentially measured: indirect linkages through input-output relationships inside the French economy. To that end, we construct the following rm- specic indices of indirect linkages: DS f;j;

C=INPUTINTfX

iIO ijNIMi;CN i(5) US f;j;

C=DOMINTfX

iIO jiNEXi;CN i:(6) In these expressions,iandjindex sectors, and rmfbelongs to sectorj.IOijis the 8 domestic direct requirement coecient of the 1995 French Input-Output matrix, dened as the share of spending on domestically-produced sectoriinputs in production in sectorj. NIM i; Cis the number of French rms in sectorithat import fromC,NEXi;Cis the number of French rms in sectorithat export toC, andNiis the total number of rms in sectori. Each of these numbers is computed excluding rmfitself (which is obviously only relevant ifi=j), and thus are in that sense rm-specic, but we suppress the dependence of those values onfto economize on notation. Finally,INPUTINTfis the rm's total input usage intensity, dened as the total material input spending divided by material input spending plus wage bill, averaged across years.DOMINTfis the domestic sales intensity, dened as the share of rmfsales that goes to the domestic market, averaged across years. These indices are heuristic, but inspired by the formulation of the \network eect" prop- agation of terms in

Acemoglu et al.

2015
). TheDSf;j;Cindicator, short for \downstream," is meant to capture the fact that rms in sectorjbuy inputs from other sectorsiin the economy. To the extent that sectoriis populated by rms that import from countryC, countryCshocks will propagate to input-buying rms injthrough their input purchases of i. For any individual sectori, the term in the summation,IOijNIM i; CN i, will be high either if juses a lot of sectoriinputs (IOijis high), or if a high fraction of sectorirms import from C. The summation aggregates this information across all the input-supplying sectors ofj, and multiplies it by the rm-specic input intensity, since the importance of downstream linkages will be higher for rms that spend a lot on inputs. TheUSf;j;C(\upstream") indicator is meant to capture the fact that rms in sectorj supply inputs to other sectorsi, and thus will be aected by whether the sectoribuying its inputs has a large population of directly connected rms. Export opportunities in sector ito countryCwill propagate to sectorjas an increase in exports fromitoCwill raise demand for sectorjinputs. For an individual output sectori, the termIOjiNEX i; CN iwill be high if eitheriuses a lot ofjinputs, or if a large share of rms iniexport toC. The summation across sectors is multiplied by the share of rmf's sales in the domestic market, since iffdoes not sell much of its output in France, by construction it must be a relatively unimportant supplier of inputs to the French market. The indices are constructed using sector-level information by necessity, as there is yet no rm-level Input-Output matrix available for France. If we had rm-level information, these indices would have a much simpler form, and would exploit information on whether rmfsources inputs from directly connected rms, or supplies inputs to directly connected 9 rms. 5 Note that while these formulations appear the most natural to us, one can think of other transmission mechanisms that might be at work. For example, one can build an alternative DSindicator that instead of counting the number of importing rms in the input-supplying sector, counts the number of exporting rms. This indicator would be relevant if there are capacity constraints, and thus greater export opportunities in the input-supplying sector ireduce those rms' domestic supply of inputs. We view those alternative indicators as less compelling and most likely second-order relative to those set out above. An additional question is whether we should also build propagation indices for multinationals and aliates. In this case, it is also unclear in which way do shocks to multinationals propagate to non- directly connected rms. To avoid a proliferation of regressors, we favor a more parsimonious specication with only the two indices (5) (6) We add these two variables to the baseline specication(4). Thus, we include the indirect linkage indicators alongside the direct linkage indicators and country and rm xed eects: ft;

Ct) =+1EXf;C+2IMf;C+3AFFf;C+4HQf;C+

5DSf;j;C+6USf;j;C+f+C+f;C:(7)

2.2 From Micro to Macro

Next, we investigate the macroeconomic implications of these micro ndings. The aggre- gate correlation as written in(3)is additive in the individual rm-level correlations with foreign GDP, and thus can be decomposed easily into the various components. Since we are interested in the impact of individual rms on aggregate correlations, we can decompose the aggregate growth rate into the contribution of two sets of rms: the directly connected and the not directly connected to a particular country: At=X fw ft 1 ft=X f2ICw ft 1 ft+X f2IcCw ft 1 ft; whereICis the set of rms that satisfy at least one of the four criteria included in estimating equation(4): (i) export toC; (ii) import fromC; (iii) is a French aliate of a multinational based inC; or (iv) is part of a French multinational that has aliates inC. Correspondingly,5 If we had a rm-to-rm IO matrix, we could construct the simple indexDSf;C=P gIOgfIMg;C, where IO gfif the share of spending by rmfon inputs supplied by rmginf's total output (the rm-to-rmquotesdbs_dbs14.pdfusesText_20

[PDF] indeed netherlands

[PDF] indeed spain

[PDF] indeed suisse

[PDF] indeed traduction

[PDF] indeed uk

[PDF] indemnisation catastrophe naturelle inondation

[PDF] indemnisation degat des eaux maif

[PDF] indemnisation maif accident

[PDF] indemnisation maladie professionnelle fonction publique

[PDF] indemnité d'éloignement nouvelle calédonie

[PDF] indemnité de caisse agent comptable eple

[PDF] indemnite de deplacement a l'interieur du france

[PDF] indemnité de licenciement ile maurice

[PDF] indemnité de licenciement france imposable

[PDF] indemnité de résidence nouvelle calédonie