IMPACT OF GLOBAL FINANCIAL CRISIS ON FIRM GROWTH: AN … · 2019-11-30 · IMPACT OF GLOBAL FINANCIAL CRISIS ON FIRM GROWTH: AN ANALYSIS OF LARGE EXPORTERS IN INDIAN MANUFACTURING

IMPACT OF GLOBAL FINANCIAL CRISIS ON FIRM GROWTH: AN

ANALYSIS OF LARGE EXPORTERS IN INDIAN MANUFACTURING

Mitali Chatterjee

PhD Scholar

Indira Gandhi Institute of Development Research, Mumbai

Abstract: The paper studies the impact of the Global Financial Crisis of 2008-09 on Indian

manufacturing firms, using data from the Centre for Monitoring Indian Economy (CMIE)

Prowess database. While the crisis did not have its roots in India, the economy and firms were

affected mainly due to trade linkages with other affected countries. Since trade acts as a

transmission mechanism for the spread of crisis, it is expected that exporting firms will be

affected more severely than non-exporting firms as a result of crisis. We test this hypothesis at

the aggregate and disaggregate level and find that while crisis affects all firms, the magnitude of

this impact is higher for exporting firms. However, even within the group of exporters, this

impact varies significantly across firms and industries. Therefore, we analyze the firm level and

industry level characteristics that contribute to this differential performance. Specifically, we

analyze the role that trade exposure to crisis affected regions and degree of vertical specialization

play in explaining the differential firm and industry level performance during crisis.

JEL Classification: F10, F14, L25, L60

Keywords: Growth, Crisis, Exporting firms, Demand, Credit, Vertical Specialization,

Manufacturing Industries

2 | P a g e

I. INTRODUCTION

One of the most striking features of the recent Global Recession in 2008-09 was that it was

accompanied by a trade collapse of magnitude five times as much as global output.1 The shock to

the world trade was unprecedented, not only in terms of magnitude, but also in terms of its

spread. The drop in the value of world exports was 31 percent between first quarter of 2008 and

2009 and 17 percent between the second quarter of 2008 and 2009.2 Though the crisis originated

in the United States, it spread to most parts of the world quickly. Krugman argued that world

trade acted as an important transmission mechanism, affecting even those countries which had

relatively robust fundamentals.3

Eichengreen and O’Rourke show that during the 2008-2009 global recession, it took world trade

just 9 months to fall as much as it did during the Great Depression in a span of 24 months.

According to Oliveira Matins and Araujo (2009), past crises on an average lasted for 13 months,

with an average negative growth of -2 percent compared to the Global recession with a negative

growth rate of -25% registered between October 2008 and June 2009. Data from 52 nations

showed that 83 percent of the month-nation pairs reflected negative growth in imports and

exports as compared to 39 percent during the 2001 trade collapse (Baldwin, 2009).

In most of the available theoretical and empirical literature, the “sudden and synchronous” trade

collapse4 and the overreaction of trade compared to output is attributed to:- (i) compositional

effect (ii) supply chain effect and (iii) credit channel effect. There is little doubt that a decline in

demand was the major reason for the fall in trade. However, the reason why trade fell more than

output was because the fall in demand was mostly for durable consumer goods, which constitute

a larger part of trade than GDP. The largest contribution to GDP is from the service sector.

While trade in services is more resilient to crisis, trade in goods is the worst hit.

The second important reason for the trade collapse is attributed to the presence of global supply

chains. Intermediate inputs are exchanged between countries several times before the final

1 Freund (2009)

2 Eichengreen and O’Rourke (2010)

3 Other studies that highlighted the role of trade and financial linkages in spreading the crisis include Rose and

Spiegel (2010), Milesi-Ferrati and Lane (2010) among others. 4 Baldwin (2009)

3 | P a g e

product is assembled and sent to destination countries. Since all the different production stages

are interconnected through backward and forward linkages, this results in a magnification effect

on the collapse of world trade.

Yet another school of thought believes that given the financial origin of the crisis, credit

constraints are an important reason for the fall in trade. Exporting is a costly activity as it

involves undertaking risk, sunk costs and longer gestation periods between production, delivery

and payment receipts. Hence, those sectors which are heavily reliant on external finance are

likely to be more affected during times of crisis.5

Most empirical literature on trade and crisis has been undertaken from a macro perspective,

looking at aggregate bilateral trade flows between countries. The present paper will depart from

existing literature by analyzing the growth dynamics of exporting firms in India, in the context of

the global financial crisis. In particular, it will attempt to study firm behavior of Indian

manufacturing sector, using data from CMIE Prowess. While the India economy escaped the full

brunt of the crisis due to its strong macroeconomic fundamentals, increasing integration to the

global economy through trade and financial flows caused India to suffer from “second round”6

effects. A fall in foreign demand for Indian goods resulted in heavy decline in trade volumes, and

hence growth of exporting firms. Hence, we expect that exporting firms would behave

differently from domestic firms in the aftermath of the crisis. We try to ascertain the firm level

characteristics to which these differences can be attributed.

The paper is organized as follows. Section II discusses some theoretical underpinnings in the

related field while Section III reviews the empirical literature. Section IV establishes the linkages

between vertical specialization, trade and crisis. Section V discusses the firm level data obtained

from CMIE Prowess and trade data obtained from WITS, while Section VI proposes alternative

econometric estimation methodologies to study the impact of crisis on the subset of exporting

firms. Section VII discusses the main findings of our static analysis. Section VIII proposes

alternative dynamic panel estimation approach. Section IX discusses indirect impact on all the

Indian firms due to differing degree of domestic linkages with sectors that are involved in

international production sharing. Finally, Section X concludes the paper.

5 Amiti and Weinstein (2009), Iacovone and Zavacka (2009) and Chor and Manova (2010)

6 Kumar and Alex (2009)

4 | P a g e

II. THEORETICAL UNDERPINNINGS

2.1 Composition Effect of Demand

The composition effect of demand as an explanation towards trade collapse in the recent global

financial crisis can be attributed to Baldwin (2009). It is based on four fundamental ideas,

namely-

Consumer and capital goods constitute a larger share in international trade than GDP.

Services constitute a larger share in GDP than in international trade.

Services are less volatile as compared to manufacturing goods.

Consumer and capital goods are generally postponeable goods.

The 2008-09 trade collapse was a result of a sudden steep fall in the demand for postponeables,

i.e. consumer and investment electronics, transport equipment, machinery and their parts and

components. Hence, the fall in trade was perceived to be much larger than the fall in output.

McKibbin and Stoeckel (2009) model the global financial crisis and show that there is a sharp

fall in the demand for durable goods during crisis due to rise in risk-adjusted interest rates and

hence countries exporting mainly durable goods are more adversely affected.

2.2 Global Value Chain Effect

Trade economists like Batra & Casas (1973); Woodland (1977); Dixit & Grossman (1982) and

Helpman (1984) have worked to extend the original Heckscher-Ohlin (H-O) model- which

explains international trade in final consumer goods only, to build a more general model

involving the presence of traded intermediate inputs. Ethier (1979, 1982) highlighted that, in a

context where production process in different industries are globally dispersed, production of

intermediate goods are subjected to international economies of scale. Thus, specialization and

trade in differentiated intermediate products leads to efficiency gains. Similarly, Yi (2003) has

argued that global gains from trade may be enlarged with fragmentation because it allows more

finely defined production processes to be allocated across countries more efficiently.

The concept of global production network implies that trade involves not only the exchange of

end products but, increasingly, of P&C that go into making them. Each country specializes in a

5 | P a g e

particular fragment of the production process based on its comparative advantage, which in turn

is determined by factor intensity of fragments and differences in factor prices across countries. In

certain industries, such as electronics and automobiles, fragmentation of production process into

smaller and more specialized components allows firms to locate parts of production in countries

where intensively used resources are available at lower costs.

Initial traces of fragmentation were seen in the US- Mexico production sharing process whereby

a US firm would export to its subsidiary in Mexico, parts and components that were assembled

there and re-exported as final products to the US market7. Since then, fragmentation that was

mainly observed in electronics and textile industry has now spread to various other sectors.

These parts and components are mainly designed and produced in developed countries and then

later sent to low-wage developing countries, for assembly purposes. For example the raw

materials required for producing the Barbie doll were attained mainly from Taiwan and Japan

and later sent for assembly to low cost locations like Indonesia, Malaysia and China8.

In the light of the recent global financial crisis, it has been argued that the presence of global

value chains and fragmentation of trade are the major drivers of the synchronicity of the trade

collapse as well as the severity of its magnitude. Unbundling the production process would mean

imported inputs used for the exports of components and further used to assemble the final

product which is then exported again. Hence, goods are traded several times before reaching the

final consumer. Section IV of the paper discusses in details the nuances of the vertical

specialization channel and its ramifications in the context of the Global Financial Crisis.

2.3 Credit Channel

Both domestic producers as well as exporters have their individual credit requirements to cover

both fixed costs as well as variable costs. Fixed costs such as expenditure on R&D, advertising,

fixed capital equipment, product development are not dependent on the scale of operation.

However, exporters have to face an additional sunk costs as well as variable costs such as

creation and maintenance of distribution networks in foreign markets, learning the feasibility and

7 Kimura (2007)

8 Feenstra (1998)

6 | P a g e

potential of new markets, freight charges and shipping among others. Most of the expenditure

needs to be incurred before the receipt of payments.

Theoretically, financial frictions can be incorporated in the heterogeneous firm model proposed

by Melitz (2003)9. Using this framework, Manova et al. (2009) show that exporters can raise

external finance by pledging collateral. There is a probability associated with the exporter’s

ability to repay its investors. Firms belonging to certain industries are more dependent on

external finance than others and differ in their requirements of tangible assets10

.

Similar to the model proposed by Melitz, here, too, all firms fulfilling a certain productivity

threshold become exporters. However, since financial frictions also have to be taken into

consideration, the subset of firms who become exporters is even smaller. The most productive

firms are still able to export their optimal levels, but lower productivity firms may have to export

lower volumes. Some firms may not be able to explore the full breadth of the markets they would

like to cover, if credit constraints apply. Thus external finance is an important factor affecting

exports, both at intensive margins (trade volumes) as well as extensive margins (number of

firms/ products/ destination combinations)

In the context of a crisis, several studies have established a link between credit crunch and lower

trade flows. A banking crisis typically affects the health of banks and other financial institutions

and reduces the availability of trade finance. It is expected that those sectors and firms that have

greater dependence on external finance will be hit disproportionately during financial distress.

Greater access to trade credit means greater prospects of long term investments in international

markets and greater working capital. However, during crisis, trade finance becomes costlier as

banks are reluctant to lend due to increased credit risks. Chor and Manova11

(2012), Bricogne et

9 Melitz (2003) provides an extension to Krugman’s general equilibrium model, to incorporate firm level

productivity differences. According to this paper, only the more productive firms would be able to survive in the export market while others will be forced to exit. Also, increase in aggregate industry productivity will result in welfare gain. 10

Rajan and Zingales (1998) 11

This paper uses detailed US trade data during the 2008-09 crisis to show that countries with higher interbank rates and tighter credit markets imported less to US.

7 | P a g e

al.12

(2010), Amity and Weinstein13

(2009) and Iancovone and Zavacka14

(2009) all find an

important role for external finance in explaining the decline in trade.

Having discussed some of the important theoretical reasons why crisis can affect trade severely

at the macro level, we now turn to the body of available empirical studies at the firm level.

III. FIRM LEVEL EMPIRICAL LITERATURE REVIEW

Theoretical and Empirical literature on impact of crisis on trade often give a macroeconomic

perspective regarding how a country’s exports get affected in the aftermath of a crisis. Aggregate

data does not reflect the channels through which policy reforms can transform the economy at

the firm level.15

Firm level data contains detailed information on balance sheet and ownership,

enabling the investigation of a large number of variables such as sales, size, profitability and

assets. Thus it helps in the analysis of firm specific characteristics which affect performance

parameters such as growth and efficiency in the aftermath of a crisis. In this section, we discuss

some of the findings of such firm level studies.

Most studies focus on the role of demand side factors and financial constraints on individual

firms’ exports. Within the set of exporters, there are heterogeneities across firms on the basis of

factors such as size, age, financial dependence, productivity, industry, profitability, pre-crisis

growth rates, and product specific elasticities among others. The aim of most studies is to check

for the impact of crisis on growth and trade after controlling for these firm level heterogeneities.

Let us now discuss some of these papers in details.

Before examining the evidence in favor of impact of crisis on firm trade and growth, it is

important to look into the existing literature on some of the most important firm specific

12

The paper uses firm level data on French exports to show that sectors that were more dependent on external finance suffered more during the global financial crisis. 13

This paper worked on Japanese exporter firms of 1990s and showed that distress in the banking sector resulted in decline in exports. 14

This paper showed that in countries experiencing banking crises, exports were adversely impacted more for industries that were dependent on trade credit. 15

Alfaro and Chari (2009)

8 | P a g e

characteristics that affect firm growth. Given below, is a table summarizing some of these

attributes-

ATTRIBUTE RELATION STUDIES

Firm Size

Most studies find a negative

relation between firm size and

growth. However, it is

expected recessions will hit

small firms more than large

firms.

Dunne & Hughes (1994);

Almus & Nerlinger (2002);

Botazzi & Secchi (2003);

Jensen (2005); Calvo (2006);

Zhou et al. (2009); Yasuda

(2005); Bugamelli et al.

(2009)

Firm Age

In general, a negative

relationship between firm age

and growth. However, young

firms are more sensitive to

business cycle fluctuations.

Evans (1987); Dunne et al.

(1989); Barron et al. (1994);

Glancey (1998); Geroski and

Gugler (2004); Fort et al.

(2013)

Productivity

More productive firms will

grow faster and be more

resilient during economic

downturns.

Alvarez and Gorg (2009);

Coad (2009)

Export Propensity

Firms with higher export

propensity are expected to be

more productive, grow faster

and be more resilient during

recession. However, the GFC

provides evidence to show that

exporters were hit harder.

Bernerd and Jensen (1997,99);

Head & Ries (2003); Hahn

(2004); Bernard et al. (2005);

Burger et al. (2006);

Bugamelli et al. (2009)

Ownership

Foreign firms are expected to

be more productive than

locally owned firms and have

Dunning (1993); Head & Ries

(2003); Helpman et al. (2003);

Dunning & Lundan (2008)

9 | P a g e

access to better technology,

processes and more economies

of scale.

Financing

Firms which have lower level

of indebtedness and are less

dependent on external finance

are better able to cope with

economic crisis

Desai et al. (2004); Luzi

(2006); Kroszner et al. (2007);

Bugamelli et al. (2009);

Manova et al. (2009);

Bricogne et al. (2009)

Industry

Industry level factors may

affect a firm’s growth

patterns, both during normal

times and during recession

Geroski & Toker (1996); Coad

(2009); Bricogne et al. (2009);

Bugamelli et al. (2009); Chor

& Manova (2010)

Research & Development

R&D investment is generally

found to be procyclical.

During crisis, some firms may

increase innovative activities,

based on certain inherent

characteristics.

Aghion et al. (2005); Barlevy

(2007); Francois and Lloyd-

Ellis (2009); Correa and

Lootty (2011)

One of the earliest studies to analyze the impact of recession on firms was Geroski and Gregg

(1997). Their analysis is based on a large scale primary survey conducted in UK in 1993, with

participation from more than 600 firms. The core issues analyzed in this book are related to the

symptoms that a firm will be vulnerable to recession, mechanism to cope with recession,

feedback to the labor market among others. Firm size and export intensity of a firm are said to be

important factors indicating the exposure of firms to crisis. The book reveals that firms whose

growth is driven by acquisition and high levels of debt are more adversely affected during crisis.

Extremely high pre-crisis growth rates could be an indicator of increased vulnerability during

crisis. Heterogeneity among firms’ organization and structure may affect the choice of strategies

in the aftermath of a recession.

10 | P a g e

Bricogne et al. (2009) use French customs data to undertake a firm-destination level analysis on

the impact of Global Financial Crisis on trade. Specifically, they study the impact on both

intensive margins16

as well as extensive margins17

and conclude that while smaller firms were

affected more at the extensive margins whereby they cease to export some products or to some

destinations or altogether; larger firms generally suffer losses at intensive margins. Another

finding of this paper was that large exporters are concentrated more in sectors that are adversely

affected. The paper also argued that firms and sectors that were heavily dependent on external

finance were more adversely affected. However, since the share of these firms in total exports

was small, financial constraint played a minor role in aggregate trade contraction.

Another recent paper on impact of crisis on exporting firms in France, Bellas and Vicard (2013)

focus on not just the Global Financial Crisis, but identify all episodes of banking and financial

crisis18

using data on residual GDP growth, after de-seasonalizing and de-trending. The paper

finds that there is overreaction of exports to GDP variations during crisis and elasticity of trade

to GDP almost trebles. Extensive margins are more responsive to demand during global

downturns than intensive margins. Separate decomposition of banking and financial crisis into

intensive margins and extensive margin adjustment reveal that the extensive margin of trade is

more dominant during a currency crisis while intensive margin adjustment takes place in case of

banking crisis.

Classens, Tong and Wei (2011) reiterate that firm level trade linkages are an important indicator

of firm performance during crisis. The paper argues that it is wrong to use trade data post crisis

to argue in favor of trade channel playing an important role in affecting a firm’s performance in

16

,16

Intensive margin is the share of trade that arises out of the surviving product-partner pairs for a country. It measures the ‘depth’ of a trade relationship by looking at the value of trade for each product-partner pair instead of the number of products traded. Extensive margin refers to growth attributable to new products and new partners (diversification). It is a measure of the ‘breadth’ of a trade relationship as its value depends on the number of products traded and the number of trade partners. Hence, a country’s trade growth can occur at the intensive and extensive margins. (Hummels & Klenow; 2005) 18

Banking crisis is said to occur when there are significant losses to the banking system of the economy, characterized by bank runs and bank liquidations (Leavan and Valencia, 2012).Currency crisis can be defined as a greater than thirty percent nominal depreciation of the domestic currency against the US dollar, which must also be ten percent higher than the rate of depreciation the year before (Frankel and Rose, 1996). Kaminsky and Reinhart (1999), in their paper, find that a banking crisis may be followed by a currency or sovereign debt crisis.

11 | P a g e

the aftermath of a crisis. Hence, the paper suggests the classification of firms on the basis of ex

ante characteristics regarding exposure to trade among other factors.

To analyze the relative impact of the demand and finance channels on a firm’s performance in

the aftermath of a crisis, Coulibaly et al. (2012) construct a global demand index that would

provide necessary weights to both domestic demand as well as demand arising out of exports, for

each firm grouped by sectors, summed across all destination countries.19

Measures of firms’

reliance on external and internal sources of finance before the crisis, measured by external

finance to total assets ratio and retained earnings to total assets ratio, are used along with

working capital to total assets ratio as a measure of financial liquidity. Results show that

exporters are affected more adversely as compared to non-exporters. This could possibly happen

because external finance as well as liquidity plays an important role in determining firm

performance and exporters are more reliant on external finance than domestic firms. The demand

channel, proxied by the global demand index also turns out to be significant. To account for

those firms which are exporters, but whose export data is not available; a logit model is

estimated with exporting status as the dependent variable and pre-crisis sales to assets ratio20

as

the predictor variable, for the sub-sample of exporters with zero or positive values only. Using

the logit estimates, the probability of exports for those firms with missing exports value is re

calculated. The paper also finds that firms with high inventory to sales ratio are less constrained

by liquidity crunch or external finance shortage during crisis.

Apart from the magnitude of the trade fall at the firm level, firm heterogeneities also play an

important role in determining the probability of survival of exporters in the aftermath of a crisis.

Melitz (2003), Das et al. (2007), and Bernerd et al. (2007) conclude that probability of survival

increases with size. This is mainly because the firms with higher productivity get self-selected as

exporters. Using Mexico customs transaction data, Giri, Seira and Teshima (2012) try to

ascertain whether size plays an important role in determining the survival of firms post crisis.

They find that contrary to the earlier studies, conditional on survival, small sized firms are not

more likely to exit in the event of a crisis. In fact, big exporters suffered more. This could be

19

(

)

∑ where I denotes the firm, s denotes

the sector and c denotes country of origin and d denotes destination. is the export share of sector s from country c to country d. 20

Proxy for Productivity

12 | P a g e

because of two reasons. Large exporters are generally highly integrated into global supply

chains, which have a limited ability to adjust during crisis.21

Another reason could be that crisis

results in substitution of high quality products with lower quality products, which would mean

large exporters suffer more, if quality and size are sufficiently positively correlated. The paper

shows that there is no statistical difference between exit probabilities of different sized exporters

during pre-crisis, crisis, and post-crisis periods.

Using firm level data during the Global Financial Crisis from UK, Gorg and Spaliara (2013) find

that export market exit increases during crisis. Their argument is that exporters are generally

more dependent on external finance and financial health of firms, as measured by their liquidity,

leverage and borrowing cost, affect exit probabilities. Their analysis also reveals that firms that

exit are generally either older or smaller and less productive firms. Also, unlike many earlier

studies, this paper finds that crisis has a positive impact on exit, over and above the impact

through the external finance channel.

Coming specifically to the Indian economy, the direction of impact of the crisis ran from the real

economy to the financial economy, unlike the developed countries of the West. Following the

crisis, the RBI followed an expansionary monetary policy for some time to pump in liquidity in

the economy. In spite of such measures, there was a severe decline in demand between 2007 and

2009, as measured by the private final consumption expenditure. The 2009 Economy View

report prepared by Crisil shows that in the third quarter of 2008-09, there was a severe fall in the

sales growth rate and profitability margins of most of the firms in the manufacturing sector.

Industrial growth decimated to 1.9% while manufacturing sector experienced a negative growth

rate of -0.3%.22

The pharmaceutical industry, gems and jewelry, four wheelers and commercial

vehicles were among the worst hit, followed by metal, minerals chemicals and textiles. Sharp fall

in the export demand was the main reason for the deterioration of many of these sectors,

particularly the gems and jewelry sector.23

Sengupta (2009) estimates that India’s exports in the

second quarter of 2008-09 shrunk by 15%.

21

Bems et al. (2011), Levchenko et al. (2010) and Alessandria et al. (2010) 22

Pavel Chakraborty (2012) 23

Gross profit margin in Q3 FY09 fell more than 100% as compared to Q3 FY08 for sectors such as pharmaceuticals, gems & jewelry, chemicals, textiles and non-metallic mineral products. Fall in sales margin was around 50% or more.

13 | P a g e

Table B1 in Appendix B shows that India’s real GDP, which was growing at above 9%, dipped

to 6.7% in FY 2008 before recovering to 8.6% in FY 2009. Real manufacturing growth dipped

from 10.3% in FY 2007 to 4.3% in FY 2008. In the recovery phase FY 2009 onwards, there was

a sudden jump of real manufacturing to 11.3% due to the base effect. However, thereafter

manufacturing growth declined on YoY basis. As far as real exports are concerned, there was a

lagged impact with growth turning negative in FY 2009. Similar picture arises for real

manufacturing sector growth (nominal manufacturing GDP deflated by manufacturing sector

WPI), except that the fall in growth of exports (-5.6%) is much higher than that of overall

manufacturing exports (-3.1%) in FY 2009.

Given that there is an apriori expectation that exporters may be hit harder than non-exporters

during crisis, Chakraborty (2012) studies the specific impact on Indian exporters due to exposure

to crisis hit zones like United States and European Union, in the aftermath of the Global

Financial Crisis. Since destination level firm export data is not available, the author matches the

NIC code of the firms to HS 6 digit level trade data at a sectoral level. Results show that, indeed,

firms belonging to sectors which export a large proportion of their total exports to US and EU

are more affected. Demand shock channel is found to be most important, with financial

constraints playing an insignificant role. Domestic credit (from banks), however, played a

positive role in easing the liquidity constraint. On segregating the firms on the basis of size, it is

found that smaller and medium sized firms are more affected by negative demand shocks than

the larger firms. The top 1 percentile exporters, in fact, seemed to have not been affected by

demand shocks at all. On decomposing the firms on the basis of end use categories, it is found

that trade in intermediate, durable and non-durable goods were most affected as a result of crisis.

From the above literature, it is clear that we expect exporters to behave differently from non-

exporters in the aftermath of a crisis. While many studies conclude that trade linkage is an

important factor affecting a firm’s performance, whether this effect is positive or negative differs

across countries and across crises. In general, if the origin of the crisis is the country which is the

subject of analysis, then exporters may be hit lesser than non-exporters because trade allows

firms to diversify across non affected regions and partially compensate for the losses at home.

Domestic firms are hit directly through the demand channel and may suffer more as they do not

have the trade channel to cushion their losses. On the other hand, if the crisis originates in a

14 | P a g e

country/countries where firms have strong trade linkages; or if the crisis is of a global nature; we

expect exporting firms to be on the receiving end of greater losses. However, demand is not the

only channel that determines the outcome of the crisis and extent of loss to the firms. There are

other factors such as financial constraints, availability of trade credit and degree of participation

in global production sharing which play an important part as well.

IV. VERTICAL SPECIALIZATION & CRISIS

A large number of manufacturing industries, in particular apparel, automobiles and electronics,

have been experiencing rapid growth in trade since the mid-1980s due to the emergence and

expansion of vertical production networks which encompass the different stages of production of

goods and services right from their inception to their final assembly and delivery, with multiple

entries and exits across countries24

. This has enabled many developing countries in Asia to

exploit their comparative advantage and attain greater specialization. A number of papers explain

the conceptual framework for fragmentation of trade including Krugman (1995), Feenstra and

Hansen (1996), Deardoff (1998), Jones and Kierzkowski (2001) and Grossman and Rossi-

Hansberg (2006) among others.

Theoretical underpinnings vary from standard Ricardian models to the new trade theory. Yi

(2003) uses a Ricardian framework to explain vertical specialization and the associated

“magnification” effect arising out of technological differences across countries. Deardoff (1998)

uses the Heckscher-Ohlin (H-O) model to explain vertical specialization arising out of

differences in factor endowment across countries. For example, US exports relatively skilled

labor intensive products to Mexico, who assembles them using unskilled labor intensively and

then re-exports them back to the US. The new trade theory brought with it the concept of love of

variety in order to explain the working of vertical specialization. According to this approach, for

every final good, there is an ideal intermediate good that fits its requirements perfectly.25

In

Burda and Dluhosch (2002), vertical specialization is understood through cost minimization and

economies of scale under monopolistic competition. Yi (2003) also talks about vertical

24

Refer Appendix D for illustration. 25

Dixit and Stiglitz (1977), Lancaster (1979), Krugman (1979 and 1981) and Eithier (1982)

15 | P a g e

specialization and fragmentation arising out of reduction of bilateral tariffs across preferential

partners leading to trade creation and trade diversion effects. Arndt and Kierzkowski (2001)

modeled trade in intermediates as an outcome of international division of labor. Baldwin (2006)

and Grosman and Hansberg differentiate between trade in goods from trade in tasks. While

traditional trade theory assumes close substitutability of goods, vertical specialization and trade

in tasks involves trade in complementary goods.26

Empirically, Pitigala (2008) showed that vertical trade between developing countries increased

from 8 percent in 1985 to 33 percent in 2005. Escaith, Lindernberg and Miroudot (2010) find

that while exports of final goods between Asia and United States increased by 7 percent annually

between 1990 and 2008, exports of inputs increased by more than 10 percent annually during the

same period. Milberg and Winkler (2010) study the historical trends in GVC structure across

sectors using the Herfindahl- Hirschman Index (HHI).27

Most of the sectors were found to

experience a dispersion of trade, as indicated by a falling value of HHI (hence rising GVC). This

was particularly the case in sectors like textiles, iron and steel, machinery and transportation.

Let us now look at the ramifications of increasing participation of countries in global production

networks and greater economic interdependency in the context of the Global Financial Crisis of

2008-09. The crisis, which originated in US resulted in a severe contraction of its import demand

which was reflected directly in its trade statistics. However this is the direct effect only. There

was direct and indirect impact on the import demand of different sectors across countries due to

the intertwined production networks. For example, a fall in US demand for automobiles would

result directly in the fall of Japanese car exports, but indirectly in the fall in export demand for

other parts and components required in the production of automobiles such as rubber from

Malaysia and tires from Korea.28

One way to capture the degree of a country’s participation in

global production networks is that employed by Hummels, Ishii and Yi (2001) who compute an

index of vertical specialization to measure the amount of imported inputs used in the production

of goods that is subsequently exported. Using data on input-output tables for 10 OECD countries

26

Gonsalez and Holmes (2011) 27

∑ where is the share of country I expressed as a percentage of total world exports of

product j. 28

Refer Appendix D for a graphic description on the different round of effects due to a fall in initial import demand of automobiles in the US (Inomata et. al. , 2010)

16 | P a g e

and 4 emerging market economies, the paper finds that vertical specialization increased by 30

percent between 1970 and 1990.

Given the increasing importance of vertical specialization, an important question that arises is

how exactly does greater production sharing exacerbate the adverse impact of the crisis? With

vertical specialization, a fall in final demand reduces trade in both intermediate and final goods.

According to Escaith, Lindenberg and Miroudot (2010), overreaction of trade to global financial

crisis is due to the rising elasticity of global trade to real GDP from 2 to over 3 in the recent

years due to production sharing. Among individual countries, Milberg and Winkler show that the

highest elasticities (with respect to United States) are for China and India. A 1 percent rise/fall in

US real output will increase/decrease import demand for China and India’s goods by 8.6 percent

and 8.1 percent, respectively.

In the context of production chains, it is interesting to note that a country, sector or a firm’s

ability to participate in a global value chain is closely linked with the availability of finance.

Trade finance, which comprises of instruments such as letters of credit, advance payments,

export credit insurance and guarantees dry up in times of crisis due to more stringent norms on

borrowing and greater wariness in the international banking community. A trade credit crunch is

more likely to adversely affect sectors with greater participation in production sharing. This is

because a bottleneck arising from the lack of credit in any part of the supply chain will have a

cascading effect for the rest of the sellers in the value chain. There is need for intra-period credit

as producers of final goods cannot pay the suppliers of intermediate inputs before they sell the

final good. Hence, the producer that is at the lowest rung of the ladder is the first one to produce

and supply, but the last one to receive his remuneration.29

According to Pitigala (2009), a 1

percent improvement in finance is associated with a 0.78 percent increase in vertical

specialization.

Having established the importance of vertical specialization as one of the strong contenders

within the demand channel that could worsen the impact of crisis on trade and growth, the next

few sections will develop an index of vertical specialization at the firm/industry level for the

purpose of this paper.

29

Leach (2008)

17 | P a g e

V. DATA

We use firm-data from the Prowess database. The sample period is from 2004 to 2012. The data

is collected by the Centre for Monitoring the Indian Economy (CMIE) from company balance-

sheets and income statements and covers both publicly-listed and unlisted firms from a wide

cross-section of manufacturing, services, utilities, and financial industries. About one-third of the

firms in Prowess are publicly listed firms. Prowess covers firms in the organized sector, which

according to the Government of India, comprises enterprises for which the statistics are available

from the budget documents or reports etc. Prowess represents more than 70 percent of the

economic activity in the organized industrial sector of India and accounts for 75 percent of the

corporate taxes and 95 percent of the excise duty collected by the Government of India. The

Indian National Industrial Classification (NIC) (1998) system is used to classify firms in the

Prowess dataset into industries. The data is also classified into ownership categories such as

state-owned, private business-group-affiliated firms, private stand-alone firms and foreign

firms.30

Data for the manufacturing firms has been obtained from the Prowess Database. The 2 digit NIC

(2008) codes considered for this purpose range from 10 to 32. After cleaning the data and

removing outliers, the total number of firms available for the analysis is 2379. Out of these 2379

firms, 14% of the firms belong to the manufacture of chemical and chemical products. 12.1% of

the firms are involved in manufacture of textiles and 10.6% in the manufacture of basic metals.31

Highest mean exports are by Coke, Refined Petroleum and Nuclear Fuel sector.32

Data on sales is obtained at quarterly frequency. It is not mandatory for firms to report their

interim financial statements which contain details on quarterly sales. In fact, only large firms

generate interim financial statements. Hence, the number of firms that we can use for our

analysis is lesser as compared to the number of firms we would have obtained had we used

annual data. However, as of 2012, 88% of the total annual net sales of all firms is captured using

quarterly sales. Hence, we can proceed with the analysis using quarterly data without the loss of

much information. Since data for every firm is not available for every year due to firm entry and

30

More than 80% of the firms in our dataset are privately owned. 31

Refer Table B2 for details in Appendix B 32

Refer Table B3 for details in Appendix B.

18 | P a g e

firm exits, we work with an unbalanced panel for the time period FY 2004 to FY 2012. We

delete firm time observations in which sales, total assets or compensation of employees are

missing from the dataset.

Firms are classified as exporters and non-exporters on the basis of their reported export of goods

(FOB). Since our objective is to study the impact of crisis33

on exporters vis a vis non exporters,

it will be incorrect to classify exporters and non-exporters on the basis of their current exports, as

we may face problems of endogeneity. Instead, we define a homogeneous criterion to list firms

as exporters and non-exporters. We look at all firms’ export intensity for five years before the

Global Financial Crisis, i.e. between the years 2003 and 2007. We classify a firm as an exporter

if the average export intensity over these five years is greater than 0.25, i.e. the firm is exporting

at least 25% of its total sales. Using this classification, we find on an average, the export

intensity is on an average- 0.5 between 2004 and 2008. However, there was a sharp drop in

export intensity in 2009 to 0.46 and though export intensity kept rising by small amounts, even

by 2012, it has not been able to recover back to its pre-crisis values. This indicates the

persistence of crisis34

.

However, the impact of crisis on exports seems to have taken place mainly in the lines of

reduced exports by existing exporters. Very few firms, which were labeled as exporters pre-

crisis, are seen exiting foreign markets completely. Those which do exit mainly belong to two

broad industry categories- textile and apparel, machinery and equipment. Going forward, we will

analyze whether the relatively higher degree of vertical specialization in these sectors could be

the reason for the severity of their impact. Total number of firms in our dataset that report

positive exports have increased from 1185 in FY 2005 to 1265 by FY 2008. Thereafter, the

number of firms decreased marginally in FY 2009 and FY 2010. 36% of all the firms in the

dataset report exports each year from 2005 to 2012. 57.5% of the firms report exports for 3 or

more years.

In order to calculate quarterly real net sales from quarterly nominal net sales, we construct a

concordance between 5 digit NIC codes and WPI sub categories of manufacturing goods. The 3

33

For the purpose of our analysis, we borrow from standard literature in defining the span of crisis. Five quarters, beginning from 2008q2 and ending at 2009q2 are labeled as crisis quarters. 34

Please refer to Appendix A for further discussion on persistence.

19 | P a g e

month average wholesale price indices are used as an estimate of quarterly WPI. Data on WPI is

obtained from the Database of Indian Economy, published by RBI. Using quarterly real sales (in

natural logarithmic form), growth in real sales in computed, on a YoY basis (between same

quarters).

Trade data, to calculate trade shares of industries and regions have been taken from WITS

COMTRADE database. All concordance tables, HS 1998/92 (6 digit) to ISIC Rev 3 (4 digit),

ISIC Rev 3 to ISIC Rev 3.1 (4 digit), ISIC Rev 3.1 to ISIC Rev 4 (4 digit) and HS 1998/92 (6

digit) to BEC categories have been obtain from the United Nations Statistical Divisions.

In order to compute the index of vertical specialization, we make use of the Input Output tables

of the United States, obtained from the Bureau of Labor Statistics, initially developed by Bureau

of Economic Analysis. In particular, we extract the annual files from the use matrix in order to

compute the direct requirements coefficients. The use matrix contains the inter-industry inputs

used in every industry. It is a 196x196 matrix. We obtain the direct requirement matrix for each

industry with each industry by dividing the (i,j)th

cell value with the column total (i.e. value of all

intermediates used in that particular industry). This is the US technology matrix that we

construct and use for other countries as well. It represents the amount of ith

industry’s

intermediate good required to produce one dollar worth of the jth

industry’s output. Sectoral

output, at the ISIC Rev. 3, 4 digit level for different countries is obtained from UNIDO

INDSTAT4 2015 ISIC Rev. 3 portal.

VI. METHODOLOGY

As already discussed in the literature, we hypothesize that crisis affects growth of exporters

mainly through three channels- the demand channel, the finance channel and vertical

specialization channel. In this paper we explore the two former channels before making a case

for the third. Our baseline specification is as follows:-

∑ ∑

∑ (1)

20 | P a g e

refers to the year-on-year quarterly growth rates of firm i. is a crisis dummy which takes

the value 1 for crisis quarters, viz. 2008 second quarter through 2009 second quarter. s are

the m explanatory variables covering the various channels discussed, acting both in isolation and

when interacted with the crisis dummy variable. is the industry dummy which takes the value

1 if firm i belongs to the NIC 2 digit industry k, and 0 otherwise. We interact each of the industry

dummy variable with the crisis dummy variable to check the differential impact of crisis on

different industries. In a separate specification, we also interact each industry dummy interacted

with crisis dummy with the trade share of that industry to the crisis affected regions, US and EU,

as a proportion of the total exports of that industry to the world. We include time dummies to

capture impact of macroeconomic factors such as bilateral exchange rates, or shocks to aggregate

demand. refers to firm fixed effects to take into account firm fixed effects help in controlling

for unobserved heterogeneity at the firm level such as managerial attributes, quality etc. Our

dataset contains the sample of exporting firms only, for the time period 2004 to 2012. The m

firm-time varying trade, financial and other explanatory variables are as follows-

Export Intensity- It is calculated at the exports of goods, free on board as a proportion of total

sales of the firm.

Elasticity- This paper makes use of the import elasticity estimates provided by Kee, Nicita and

Olarreaga (2004). They calculate elasticities for HS 1988/92 6 digit commodities. We first use

concordance tables to concord HS 1998/92 (6 digit) to ISIC Rev 3 (4 digit), then ISIC Rev 3 to

ISIC Rev 3.1 (4 digit) and finally, ISIC Rev 3.1 to ISIC Rev 4 (4 digit) which has exact

concordance with NIC 2008 at the 4 digit level. Since these import elasticities are available for

each country separately, we calculate weighted elasticity of each four digit industry for each

year, the weights being the share of India’s exports of that industry to the importing country as a

proportion of total Indian exports of that industry, in that particular year. Hence we obtain

weighted elasticity estimates at the four digit NIC level.

Liquidity- We measure this by the working capital required, which is the difference between the

firms current asset and current liability, measured as a proportion of its total assets. It is an

indication of financial dependence of firms. Data on total assets, current assets and current

liabilities is obtained from CMIE Prowess database.

21 | P a g e

Trade Credit- Trade credit is an arrangement that allows firms to buy goods and services without

making an immediate payment.35

It is an important source of funding for many manufacturing

firms. To capture trade credit, we use data on sundry creditors36

as a proportion of total sales,

from CMIE Prowess database.

Size- Size is captured by log of total assets. To control for endogeneity, we use the lagged value

of the variable in the specification.

Profitability- This is captured by the ratio of PBDITA to total assets, obtained from CMIE

Prowess database.

R&D- This is a dummy variable which takes the value 1 if the firm incurs research and

development expenditures, 0 otherwise.

Market Share- This is calculated for each firm as the share of its sales in the total sales of the

industry for that particular year.

Capacity Utilization- This is measured as the ratio of sales to total assets,37

computed from

CMIE Prowess database.

Age- Age of the firm is calculated as the difference between the year in question and the

incorporation year, computed from CMIE Prowess database.

Trade Exposure- This is calculated at the 4 digit NIC level, as the exports of that industry to US

and EU as a proportion of total exports of that industry from India. This variable is interacted

with the crisis dummy and the industry dummies to check if during crisis, firms which belonged

to sectors having higher trade exposure to US and EU were affected more. For this, we use

exports data at HS 6 digit from WITS Comtrade and use the concordance tables to arrive at a

measure for each 2 digit NIC industry, for each year.

Intermediate- This is a dummy variable which takes the value 1 when the firm belongs to a 4

digit NIC whose corresponding HS 6 digit code is mapped under intermediate goods, in

accordance to the Broad Economic Categories (BEC) and System of National Accounts (SNA).

35

Vaidya (2011), Akinlo (2012) 36

Grouped under Current Liabilities in CMIE Prowess Annual Financial Statements. 37

This variable is used in Cheung and Sengupta (2012)

22 | P a g e

Vertical Specialization Index- As discussed in the previous section, we first construct the direct

requirements matrix, comprising of This is the technological matrix that we assume

constant for all partner countries that India trades with. Let j stand for the product of the industry

that is traded, constructed at the ISIC Rev. 3, 4 digit level (i.e. NIC 1994 4 digit) and later

concorded with NIC 2008. Let c stand for India’s trading partner to whom it exports. At any

point of time t, the vertical specialization index for industry i is given by-

∑

∑

denotes the total amount of intermediate input i required to produce dollar worth

output for sector j at time t. ∑

denotes total intermediate input i required in country

c across all sectors j. We normalize it using output of sector i at t, , to obtain the proportion

of total output of sector i that is used as an intermediate input across the different sectors. Hence,

this measure tells us how intensively i is used as an intermediate input or the degree of vertical

specialization of sector i. As already noted, an important problem with the Indian firms’ exports

data drawn Prowess database is that the destination of the exports is not available. Since our

dataset is at the firm/industry level, we need to construct the vertical specialization index

independent of the trading partner. To do this we use trade weights, given by

where the

numerator represents the exports from India to country c in sector i in time t, and denominator

denotes the total exports of India in the year t to all countries. Hence this ratio gives us the

proportionate distribution of India’s exports of the product i to the different countries. We can

use this trade weight to sum up the index of vertical specialization computed above to obtain the

final index, which is independent of the trading partner.

thus indicates how intensively

India is trading various goods used in the production of other goods with differing intensity

across our various trading partners. To construct this, we make use of information on production,

trade and input output tables. To observe the variation of this measure at the firm level, we can

further vary it with the firm’s export intensity, to depict to what extent a firm is intensively

exporting a product which is intensively associated in production sharing. To our knowledge,

this is the first paper which undertakes such an exercise in the context of crisis and firm growth.

23 | P a g e

Similarly, to check how degree of vertical specialization matters for the imports in the Indian

context, we compute a similar index of vertical specialization using Input-Output Transactions

Table (IOTT), prepared by the Central Statistical Organization (CSO) These are not available on

an annual basis For our analysis, we use the tables available for the years 2003-04, 2006-07 and

2007-08. The index can be defined as-

∑

∑

Notice that the trade weight assigned here is the proportion of total Indian imports in industry i

that comes from country c. Since we are using the Indian input output tables, the associated

sectoral outputs are for India as well.

To make a choice of regression technique- between pooled OLS and Random Effects Panel

regression, we resort to Breusch-Pagan Lagrange multiplier test. The null hypothesis of OLS is

rejected. Hence we cannot use a pooled OLS regression. We then perform the Hausman test to

check whether we should use fixed effects or random effects model on our panel. The null

hypothesis of zero covariance, or random effects is rejected and hence we have used firm level

fixed effects to get consistent estimates with minimum variances. For comparison purpose, we

have also included pooled OLS with industry fixed effects. We also undertake a dynamic panel

estimation technique to account for endogeneity issues and as a robustness check.

VII. RESULTS

Firstly, we check whether crisis has an impact on growth at the aggregate level, with only two

groups- exporters and non-exporters. We find that not only does crisis have a negative and

significant impact on growth, but in accordance to our expectations, the negative impact is 50%

higher for exporters than for non-exporters. Table C138

summarizes the results. We further

hypothesize that among the group of exporters, there is differential impact of crisis on the

different industries. In order to check this, we control for industry fixed effects by using

38

Refer to Appendix C

24 | P a g e

industrial dummies39

. All the alternative specifications indicate that crisis affects growth in

different industries differently40

. For most industries, the impact of crisis is higher for exporters

than non-exporters. Also, the severity of the impact of crisis is highest for certain industries such

as textiles (NIC 13), computer & electronics (NIC 26), electrical equipment (NIC 27), machinery

and equipment (NIC 28) and transport equipment (NIC 30), sectors which display a higher

degree of vertical specialization than others. Within NIC 27, crisis causes a fall in growth of the

exporters by 62.8% as compared to non-exporters which experience a fall of 17.4%. In NIC 28

and NIC 30, exporters display a fall in exports by 23.8% and 28.8% respectively, while the

corresponding figures for non-exporters are 23.5% and 25.6% respectively. This, once again

motivates us to bring in the aspect of vertical specialization in our analysis. Table C3 summarize

these results.

Having looked at the two broad category of exporters and non-exporters and having established

that exporters are affected more adversely than non-exporters, we now move to a more granular

specification, involving firm level growth in a panel consisting of all exporting firms for the

period 2004 to 2012. The channels and variables in the analysis have already been explained in

the prvious section. In this section, we discuss the results obained for specification (1).

First, we check for differential impact of crisis on 2 digit NIC, by interacting the crisis dummy

variable with each industry dummy in a firm level fixed effect regression, along with variables

like export intensity, R & D, size, profitability, age, capacity utilization, market share, liquidity

and crisis dummy. Most of these interaction dummies are negative and significant showing that

the presence of crisis indeed has adverse growth implications for most industries, notably NIC 27

(Electrical Equipments) and NIC 19 (Coke and Refined Petroleum). To ensure that these results

are indeed owing to the crisis, and not industry fixed effects we compare with the model with

industry fixed effects, and find that individual industry dummies are not significant in the growth

regression. Further, in the firm fixed effects specification, we further interact the industry

dummies (interacted with the crisis dummies) with their corresponding trade shares to direct

crisis affected regions, US and EU, to find if greater industry level trade exposure to the crisis

affceted region further amplifies the effect of crisis on individual industries. Our expectations are

39

At 2 digit NIC level 40

We interact crisis dummies with industry dummies to check this.

25 | P a g e

proved correct, as for all the industries in which crisis had a significant negative impact, the

magnitude of this impact is found to be even higher now. For example, crisis results in a fall of

real sales growth by 65.6% in NIC 27. However, if this industry’s trade exposure to US and EU

was to increase, assuming US and EU are the most affected due to crisis, this industry’s real

growth would fall by a much larger margin. Results are summarized in Table C4 and Table C5.41

We have carried out two sets of regression analysis, one using liquidity as a measure for

financial dependence and another using trade credit. Our results are robust under both the

alternative specifications.

Having looked at industry specific growth dynamics during crisis, we now look at impact of

some important firm level characterestics during crisis. In this specification, crisis dummy, as

expected is negative and highly significant. While export intensity itself does not seem to have

an impact on firm growth, when interacted with the crisis dummy turn out be negative and

significant. This result is quite intuitive. If external demand through trade truly is a channel of

transmission of crisis, we would expect those firms which are more export intensive and have a

higher exposure to other regions of the world affected by crisis, to be more severely affected.

While we can expect that goods that have a higher import elasticity of demand will be more

severely affected during crisis, the four digit industry level computed weighted elasticities

interacted with the crisis dummy donot give us any evidence for the same. One major reason for

this could be the fact that elasticity of demand varies at the Indian firm-product level and is not

adequately captured at the four digit industry level. Also, the estimates for elasticity used here is

time invarient (although the weighted elasticities are time varying), and there is evidence to show

that crisis itself can cause product level import elasticity of demand to change.

In alternative specifications, we use liquidity and trade credit to measure the impact of the firm’s

financial dependence on growth, during crisis. While working capital requirements of firms, as

measured by the liquidity variable do not seem to have an impact on firm’s growth, both in

isolation and when interacted with the drisis dummy; the extent of trade credit plays an important

role in determining a firm’s growth. Specifically, higher dependence on trade credit by a firm has

negative impact on its growth. However, there seems to be no evidence to suggest that drying up

of trade credit during crisis has adverse implications for the firm’s growth. Alternatively, as a

41

Refer Appendix C

26 | P a g e

robustness check we have also used financial sector dependence index as computed by Gupta

Hasan and Kumar42

(2008). In this case too, we find that while in general financial sector

dependence plays an important role in explaining growth of exporting firms, it is not the case

that greater dependence on external finance during crisis imposes an additional constraint. This

result may seem surprising as there is a whole body of literature which tells us how finance is

one of the most important channels through which crisis affects trade and growth. One possible

reason why greater dependence on trade credit is not significantly hurting firm growth during

crisis may be the fact that we include in our sample only large exporters, with an export intensity

of greater than 25 percent. Malouche (2009) show that a trade credit crunch mainly impacts

small exporters who lose the support of their creditors when demand for their products reduce.

Interestingly, we find evidence in support of higher growth of firms (by around 5%) which carry

out research and development activities during crisis than the ones that do not. Also, firms,

which have a higher market share in the industry, are better able to cope with crisis than firms

with a lower market share. The age and size variables are both negative and significant. Hence

bigger and older firms experience lower growth than smaller and younger firms, at all times, a

finding which is consistent with literature. Table C6 summarizes these results.

Moving one step ahead, we now use the same specification as above, but only for the subset of

intermediate products (end use category), as identified from BEC and SNA tables, after suitable

concordances. Interestingly, we find that crisis lowers the growth of exporting firms involved in

the manufacture of intermediate products by 15.7% as compared to the overall sample containing

intemediate, consumer and capitl goods, where crisis lowers growth by 14.6%. Also, while

liquidity constraints during crisis did not negatively affect firm growth in the overall sample, in

the sub-sample of intermediate products alone, there is evidence in support of liquidity

constraints adversely impacts growth during crisis. In the alternative model using trade credit,

the adverse impact of crisis on intermediate manufacturing firm growth comes out even more

42

Gupta, Hasan and Kumar construct a binary dummy variable to distinguish between manufacturing sectors (at a 3 digit NIC 98 level) which are more reliant on external finance than those that are not. They make use of two alternate estimation methods, one employing Rajan and Zingales (1998) and using CMIE Prowess data and the other using the ASI database, as the ratio of outstanding loans to estimated capital. In this analysis, we use their index of financial sector dependence based on ASI database, after making the necessary concordances to NIC 2008.

27 | P a g e

strongly (20.7%), as compared to the overall sample (17.7%). Table C7 contains detailed results

for this specification.

Since a large part of this paper is motivated by the impact of vertical specialization in influencing

differential impact of crisis in deifferent industries, we now ascertain how the inclusion of the

vertical specialization at the industry level affects our results. The vertical specialization index

has been defined in the previous section. Since this index is specific to each industry, in our

econometric specification we interact each two digit industry dummy variable with the vertical

specialization index. We then analyze the impact of crisis on firm growth in the same framework

as before with additional vertcal specialization variables. On interacting the industry specific

vertical specialization index with the crisis dummy, we find that a large number of industries

suffer a significant adverse impact during crisis through the production sharing channel. These

industries include NIC 19 (coke and refined petroleum products), NIC 26 (computer and

electronics), NIC 27 (electrical equipments) , NIC 30 (transport equipment) and NIC 32 (other

manufacturing). A large negative and significant impact of crisis on firm growth in NIC 19 (coke

and refined petroleum products) is mainly due to price effect and subsequent reduction in

demand. In particular, Column (1) of Table C843

shows that in general, there is a positive

relationship between growth and vertical specialization in many industries. Of particular interest

in NIC 27. While an increase in vertical specialization is by 1 percent is associated with a 34.7

percent increase in growth; during crisis, a rise in vertical specialization by 1 percent causes

growth to be lesser by 62.6% as compared to other industries during non crisis periods. Column

(4) of the same table shows that a credit crunch during crisis also affects industries such as those

belonging to NIC 19, NIC 27 and NIC 32 more disproportionately than the others. Hence, this

lends support to the theoretical hypothesis that trade credit crunch in vertically specialized

industries often result in a superimposed adverse impact of crisis on growth.

We repeat the same exercise with the vertical specialization channel for imports and find that

effects for certain industries such as NIC 19 and NIC 27 follow same pattern as that of the

vertical specialization for exports. However, lesser number of industries are affected in this case.

This could possibly be because of the fact that within the global production sharing arrangement,

Asia is more involved in the production and exports of parts and components.

43

Refer Appendix C

28 | P a g e

VIII. DYNAMIC PANEL ESTIMATION

The estimations and the results in the preceeding section using a fixed effect panel regression

was essentially carried out in a static framework. However, growth literature in recent times have

pointed out that a dynamic model may be more appropriate in order to address issues of

endogeneity arising from underlying theory. When a lagged dependent variable also appears as

an independent variable, exogeneity of regressors is no longer valid. Least square dummy

variables estimators become biased in such a case. In order to obtain consistent estimators, one

may resort to alternative techniques such as instrumental variable approach44

or the GMM

approach.45

The advantage of GMM is that it takes care of the econometric problems caused by

unobserved firm-specific effects and endogeneity of the independent variables in lagged-

dependent-variable models such as growth regressions. As an alternative to the static fixed effect

panel regression estimated in equation (1) in the previous scetion, in order to establish the

robustness of our results in a dynamic framework, we modify our model as follows-

∑ ∑

∑ (2)

The consistency of the GMM estimator depends on the absence of second-order serial correlation

in the residuals of the growth specifications. The Sargan test chi-square statistic allows us to

check for overidentification.46

In order to confirm adequate model specification, the first-order

serial correlation should be accepted whereas the second-order serial correlation should be

rejected. For our analysis, we choose system GMM estimator over difference GMM estimator as

the former produces more reasonable estimates of the autoregressive dynamics than the basic

first-differenced estimators. This is consistent with the analysis of Blundell and Bond (1998),

who show that in autoregressive models with persistent series, the first-differenced estimator can

be subject to serious finite sample biases as a result of weak instruments, and that these biases

can be greatly reduced by including the levels equations in the system estimator.47

Lastly, we use

the Windmeijer criterion (2005) small sample correction to have consistent standard errors.

44

Anderson and Hsiao (1982) 45

Aranallo and Bond (1985) 46

Arellano and Blond (1991); Arellano and Bover (1995) 47

Oliveira & Fortunato (2005)

29 | P a g e

The results of the dynamic panel estimation using GMM are summarized in tables C10 and C11

in Appnedix C. We find that all the important results and conclusions derived from the static

fixed effect panel model hold true for the dynamic panel estimation as well. After the inclusion

of lagged growth of the firms as an explanatory variable, we still find the crisis dummy to be

negative and highly significant across all specifications. Different industries are affected

differently as a result of crisis, with industries that are more involved in vertical specialization

and those having higher trade exposure to crisis affected regions getting impacted more severely.

Among the regressors that are interacted with the crisis dummy, we find eveidence in support of

higher export intensive firms performing worse during crisis, firms with greater market share

performing better during crisis and firms which incur research and development expenses

performing better during crisis. Like previously, the magnitude of impact of crisis on

intermediate goods is higher than that for the overall sample, once again reiterating the need for

investigating the vertical specialization channel further.

On incorporating the industry interacted vertical specialization index into the dynamic panel

model, our basic results obtained in the previous section hold. Though the impact is much more

muted in the dynamic panel estimation, nevertheless we do observe that both vertical

specialization channels and trade credit channel play an important role in exacerbating the

impact of crisis on growth at the industry level for specific industries like NIC 26, NIC 27 and

NIC 32.

IX. INDIRECT EFFECTS DUE TO DOMESTIC LINKAGES

What we have concentrated on, so far, has been the impact of the Global Financial Crisis on

exporting firms through the trade channel, both directly as a result of adverse demand shocks or

credit constraints, and also through vertical specialization linkages. However, what we have not

discussed, yet, is the second round impact of crisis on the Indian firms due to vertical production

linkages. The rationale for working extensively with the exporting firms has already been

provided in the preceding sections of the paper. As many studies suggest, during a crisis greater

exposure of the exporters to crisis affected regions makes them more vulnerable to a setback, as

compared to domestic suppliers within the home market. If countries are participating in global

30 | P a g e

supply chains, then a shock to any participating sector or country quickly spreads to other stages

of the supply chain through backward and forward linkages.

While looking at the vertical specialization channel, we have showed how exports from every

sector may be used as intermediate input at various stages of production for all sectors. In order

to ascertain the strength of these production linkages, we built an index of vertical specialization,

making use of Input-Output tables of importing countries. Our argument was that firms

belonging to sectors specializing in parts and components will be more affected as a result of

crisis. The same logic can be applied in the case of domestic firms during crisis, to check the

extent to which firms, providing intermediate inputs to exporting firms, get affected during crisis.

This is a more indirect effect, as our hypothesis is that when exporting firms experience a

setback due to the crisis, the impact trickles down to the non-exporting firms supplying them

with intermediate inputs. The extent of inter-sectoral dependence between domestic and

exporting firms is now established using the domestic input-output tables for India.

There is a huge volume of literature discussing, separately, the theoretical underpinnings as well

as empirical findings of both Vertical Specialization and backward and forward linkages in the

economy. Vertical specialization has been explained in trade literature both with the help of

traditional trade theories- such as the Ricardian model, where vertical specialization arises out of

technological differences as well as Hecksher-Ohlin model, which talks about vertical

specialization being driven by endowment differences across countries. The new trade theory

also strives to explain vertical specialization on the basis of product differentiation, intra-industry

trade in intermediate goods and monopolistic competition. There are also several studies on the

input-output based measures of vertical specialization, such as those proposed by Hummels et al

(2001) which defined the use of intermediate input goods for the production of goods that are

eventually exported.

While vertical specialization essentially highlights international production sharing, backward

and forward linkages measure the extent of domestic production sharing. Empirically, the earliest

works on linkage analysis to measure sectoral interdependence were done by Rasmussen

31 | P a g e

(1956)48

, Chenery and Watanabe (1958) and Hirschman (1958). Ramussen method is based on

the column sums of the Leontief inverse to measure inter sector linkages. Similary, the Chenery

and Watanabe backward linkage is the column sum of input coefficient matrices.

What we try to do in this paper is to essentially integrate the two concepts to see how an

international crisis such as the Global Financial Crisis indirectly affects sectors and sectors

which have strong domestic linkages. Several papers have tried to build a connection between

vertical specialization and forward and backward linkages. Prominent among them are- Kula

(2008)49

, Gonzalez & Holmes (2011), Heng & Yean (2011)50

, Nixon (2012), Timmer (2012),

Guo (2013)51

, among others. However, the way we have defined our index for domestic vertical

specialization based linkage is slightly different from these papers. The index has been computed

for the kth

sector as follows-

∑

Hence the domestic linkage index is a combination of the Vertical Specialization Index

developed in the preceding section of the paper, coupled with the domestic production structure

as reflected from the Indian Input-Output tables. What this means intuitively, is that the kth

sector

could be used as an input in every other ith

sector. Greater is the vertical specialization index for

the ith

sector and higher the intensity of average backward linkages, greater will be the extent of

domestic linkage. Our apriori notion is that greater the domestic linkages through vertical

specialization, i.e. higher the value of for any sector k, greater would be the adverse impact

on the firms belonging to this sector during crisis.

We maintain our usual fixed effects specification, adding the domestic linkage variable.

Surprisingly, results show that the sectors which were earlier most strongly affected as a result of

48 Rasmussen’s (1956) formula for Backward Linkage was as follows-

∑

∑ ∑

where n is the number of

sectors and are elements of the Leontief inverse matrix. An index of implies strong backward linkages. 49

The paper tries to analyze the production structure of the Turkish economy using input output tables and constructing backward and forward linkage indices. 50

The paper tries to decompose the Malaysian manufacturing sector into high VS-low BL, high BL-low VS, low BL-low VS and High VS-high BL sectors using Hummels VS Index and Rasmussen’s BL index. 51

The paper tries to measure Africa’s economic linkages with the rest of the world using WIOD tables and UNIDO INDSTAT database.

32 | P a g e

vertical specialization are typically not the sectors which are most impacted through the indirect

domestic linkage based vertical specialization52

. The impact is seen most strongly for sectors

such as wood and wooden products (NIC 16), Motor Vehicles, Trailers and Semi-Trailers (NIC

29), Other Transport equipment (NIC 30) and Manufacture of Furniture (NIC 31). This holds

true, irrespective whether we are doing this analysis for the entire gamut of firms, or only firms

classified as exporters or non-exporters. Hence, we can conclude from this analysis that the

adverse impact on these sectors during crisis, is more due to the high domestic linkages these

sectors share with the sectors displaying greater degree of international production sharing, and

not due to their direct involvement in the global production sharing.

X. CONCLUSION

This paper tries to analyze the impact of the Global Financial Crisis on Indian manufacturing

firms, using data from CMIE Prowess. We first study the impact of crisis at an aggregate level

for the group of exporters and non-exporters and conclude that exporters are impacted more than

non-exporters, since trade acts as a transmission mechansim for crisis among exporting firms.

We then try to ascertain, using a firm level fixed effects model, firm level and industry level

characterestics that could possibly explain the channels through which crisis impact the large

exporting firms. We find evidence in support of our hypothesis that crisis affects different

industries differently. In particular, we find that certain industries such as NIC 27, NIC 28, and

NIC 30, which are more involved in vertical specialization suffer more as a result of crisis. The

effect of crisis gets magnified for industries that have a higher trade exposure to crisis affected

regions such as US and EU. Intermediate goods suffer the most, which further motivates us to

investigate the role of global value chains and vertical specialization during crisis. To do this,

we compute an industry level index for vertical specialization, using trade and production data,

specifically input-output tables. The results for this exercise show that trade credit crunch is

generally felt in case of vertically specialized and the two together result in a far more adverse

impact of crisis on growth for industries such as NIC 26, NIC 27 and NIC 32 as compared to

other industries. A dynamic panel estimation is also undertaken apart from the fixed effects

model as a robustness check and to account for endogeneity. By and large, results obtained in the

alternative specifications are similar.

52

Refer Table C14, Appendix C

33 | P a g e

Certain firm level attributes play an important role in influencing firm growth during crisis.

Specifically, firms which are more export intensive are more badly hit. Firms conducting

research and development during crisis, or having a larger market share are expected to perform

better during crisis. We donot find much evidence in favour of the finance or trade credit channel

playing an important role during crisis. One important reason for the same could be the fact that

the Indian financial sector, particularly the domestic commercial banks were largely resilient

during crisis as a result of which Indian firms experienced a slowdown in growth mainly through

the external demand and the vertical specialization channel.

The paper also talks about indirect domestic linkages due to vertical specialization and how they

play an important role in determining the channel through which growth of firms belonging to

certain sectors are hampered during crisis. Results show that some sectors such as wood and

wooden products, furnitures motor vehicles etc. which themselves do not particpate in global

production sharing can still be affected during crisis due to their strong domestic linkages (as

they provide inputs) to sectors involved in global production sharing. Domestic input-output

tables are used in this analysis.

34 | P a g e

References

Alfaro, L., & Chari, A. (2009). India transformed? insights from the firm level 1988-2005 (No.

w15448). National Bureau of Economic Research.

Alfaro, L., & Chari, A. (2012). Deregulation, misallocation, and size: Evidence from India (No.

w18650). National Bureau of Economic Research.

Amiti, M., and D. Weinstein (2010). Exports and Financial Shocks. Quarterly Journal of

Economics, 126(4), 1841

Baldwin, R. (2009). The Great Trade Collapse: Causes, Consequences and Prospects. VoxEU

Ebook No. 27.

Bellas, D. & Vicard, V., (2013). French firms exports during downturns: evidence from past

crises, Working papers 467, Banque de France.

Bernard, Andrew B., & Jensen, J. N., (2004). Why some firms export. Review of Economics and

Statistics 86.2: 561-569.

Bond, S. R. (2002). Dynamic panel data models: a guide to micro data methods and

practice. Portuguese economic journal, 1(2), 141-162.

Bricongne, J. C., Fontagné, L., Gaulier, G., Taglioni, D., & Vicard, V. (2012). Firms and the

global crisis: French exports in the turmoil. Journal of international Economics, 87(1), 134-146.

Burda, M. C., & Dluhosch, B. (2002). Cost competition, fragmentation, an globalization. Review

of International Economics, 10(3), 424-441

Cattaneo, O., Gereffi, G., & Staritz, C. (Eds.). (2010). Global value chains in a postcrisis world:

A development perspective. World Bank Publications

Chakraborty, P. (2012). The Great Trade Collapse and Indian Firms. PhD Student, Graduate

Institute of International and Development Studies, Geneva.

Cheung, Y. W., & Sengupta, R. (2013). Impact of exchange rate movements on exports: an

analysis of Indian non-financial sector firms. Journal of International Money and Finance, 39,

231-245.

Chor, D., and K. Manova (2012). Off the Cliff and Back: Credit Conditions and International

Trade during the Global Financial Crisis. Journal of International Economics, 87(1), 117-133.

Claessens, S., Tong, H., & Wei, S. J. (2012). From the financial crisis to the real economy:

Using firm-level data to identify transmission channels. Journal of International

Economics, 88(2), 375-387.

35 | P a g e

Coulibaly, B., Sapriza, H., & Zlate, A. (2013). Financial frictions, trade credit, and the 2008-09

global financial crisis. International Review of Economics and Finance, 26, 25-38.

Dixit, A.K. and Grossman, G. (1982). Trade and Protection with Multistage Production. Review

of Economic Studies.

Eichengreen, Barry and Kevin H. O’Rourke (2009). A Tale of Two Depressions, VoxEU.org,

CEPR, April 2009.

Ethier, W. J. (1979). Internationally Decreasing Costs and World Trade. Journal of International

Economics, 9(1), pp. 1–24.

Ethier, W. J. (1982). National and International Returns to Scale in the Modern Theory of

International Trade. American Economic Review, 72(3), pp. 389–405.

Feenstra, R.C. (1998). Integration of Trade and Disintegration of Production in the Global

Economy. Journal of Economic Perspectives, 12 (Fall), 31-50.

Freund, C. (2009). The Trade Response to Global Downturns. Historical Evidence, World Bank

Working Papers 5015, August 2009.

Geroski, P. A., & Gregg, P. (1997). Coping with recession: UK company performance in

adversity (Vol. 38). Cambridge University Press.

Giri, R., Seira, E., & Teshima, K. (2013). Did trade crisis affect different exporters differently?

Case of Mexico. Case of Mexico (March 31, 2013).

González, J. L., & Holmes, P. (2011). The Nature and Evolution of Vertical Specialisation: What

is the Role of Preferential Trade Agreements? (No. 2011-41). NCCR TRADE Working Paper

Görg, H., & Spaliara, M. E. (2013). Export market exit, financial pressure and the crisis (No.

1859). Kiel Working Paper.

Guo, D. (2013), Comparison of economic linkages between China and Africa: Applying the

WIOD database. Development Policy, Statistics and Research Branch, UNIDO, WP 4/2013

Helpman, E. & Krugman, P.R. (1985). Market Structure and Foreign Trade: Increasing Returns,

Imperfect Competition, and the International Economy. Cambridge, Mass.: MIT Press.

Hummels, D and P. J. Klenow (2005). The Variety and Quality of a Nation's Exports, American

Economic Review, 95(3), pp.704-723.

Hummels, D., Ishii, J., & Yi, K. M. (2001). The nature and growth of vertical specialization in

world trade. Journal of international Economics, 54(1), 75-96

36 | P a g e

Iacovone, Leonardo and Zavacka, Veronika, (2009). Banking crises and exports: lessons from

the past, Policy Research Working Paper Series 5016, The World Bank

Inomata, S., & Uchida, Y. (2009). Asia beyond the crisis. visions from international input–output

analyses. Institute of Developing Economies IDE-JETRO Spot Survey 31

Kaminsky, Graciela L., and Carmen M. Reinhart. (1999). The Twin Crises: The Causes of

Banking and Balance-of-Payments Problems. American Economic Review, 89(3): 473-500.

Khadraoui, N., & Smida, M. (2012). Financial development and economic growth: Static and

dynamic panel data analysis. International Journal of Economics and Finance, 4(5), p94.

Kimura, F., Takahashi, Y., and Hayakawa, K. (2007). Fragmentation and Parts and

Components Trade: Comparison between East Asia and Europe. North American Journal of

Economics and Finance’ 18(1), 23-40.

Loke, W.H. and Tham, S.Y. (2011), Vertical specialization and backward linkages:

Reconsidering Malaysian manufacturing development, paper presented at the Singapore

Economic Review Conference 2011

Malouche, M. (2009). Trade and trade finance developments in 14 developing countries post

September 2008-A World Bank Survey. World Bank Policy Research Working Paper Series, Vol.

(2009)

Mathew, N. (2012). Drivers of firm growth: Micro-evidence from Indian manufacturing (No.

2012/24). LEM Working Paper Series.

McKibbin, Warwick J. & Stoeckel, Andrew (2009). The potential impact of the global financial

crisis on world trade, Policy Research Working Paper Series 5134, The World Bank.

Mehmet, K. U. L. A., & Expert, T. (2008). Supply-use and input-output tables, backward and

forward linkages of the Turkish economy. In16th

INFORUM WORLD Conference in Northern

Cyprus (pp. 01-05).

Melitz, M. J. (2003). The Impact of Trade on Intra-industry Reallocations and Aggregate

Industry Productivity. Econometrica 71(6), 1695-1725.

Mhadhbi, K. (2014). Financial Development and Economic Growth: A Dynamic Panel Data

Analysis. International Journal of Econometrics and Financial Management, 2(2), 48-58.

Oliveira, B., & Fortunato, A. (2006). Firm growth and liquidity constraints: A dynamic

analysis. Small Business Economics, 27(2-3), 139-156.

P. L. Beena (2011). Financing Pattern of Indian Corporate Sectors under Liberalization: with

focus on Acquiring firms Abroad. Working Paper 440 – Centre for Development Studies, 20-22.

37 | P a g e

Pitigala, Nihal. (2009). Global Economic Crisis and Vertical Specialization in Developing

Countries. PREM Notes 133. http://www1.world bank.org/prem/PREMNotes/prem note133.pdf

Poonam Gupta & Rana Hasan & Utsav Kumar (2008). What Constrains Indian

Manufacturing?, Macroeconomics Working Papers 22162, East Asian Bureau of Economic

Research

Rajan, R., & Zingales, L. (1998), Financial dependence and growth, American Economic

Review 88, 559–586.

Rajiv Kumar & Mathew Joseph & Dony Alex & Pankaj Vashisht & Debosree Banerjee (2009).

Indian Economic Outlook 2008-09 and 2009-10, Macroeconomics Working Papers 22916, East

Asian Bureau of Economic Research

Roodman, D. (2006). How to do xtabond2: An introduction to difference and system GMM in

Stata. Center for Global Development working paper, (103).

Srinivasan, T.N. And Archana, V. (2009), India in the global and regional trade: determinants

of aggregate and bilateral trade flows and firms’ decision to export, Working Paper No. 232,

Indian Council for Research on International Economic Relations (ICRIER)

Vaidya, R. R. (2011). The Determinants of trade credit: Evidence from Indian manufacturing

firms. Modern Economy, 2(05), 707.

Wu, Y. (2012). Performance of Publicly Listed Chilean firms during the 2008-2009 global

financial crisis. IMF Working paper, WP/12/261

Yi, K.M. (2003). Can Vertical Specialization Explain the Growth of World Trade?Journal of

Political Economy, vol. 111, No.1

38 | P a g e

Appendix A

Transitional Probability Matrix

The transitional probability matrix helps is studying the extent of persistence between entities

over time. For the present analysis, TPMs have been constructed to study the relative

performance and mobility of the group of exporters and non-exporters. Persistence refers to the

probability of remaining in the state in which the firm initially is. State here is defined as the

average quarterly growth rate of all firms under three scenarios- Pre-Crisis, which is defined to

be the average growth53

of each firm from 2007q1 to 2008q1; Crisis, which is defined to be the

average growth of each firm from 2008q2 to 2009q2; and Post-Crisis or Recovery, which is

defined to be the average growth of each firm from 2009q3 to 2010q3. Firms in each period are

distributed along the four quartiles, on the basis of average growth, for exporters and non-

exporters separately. Quartile 1 will have the lowest growth firms and quartile 4 the highest. We

then trace the movement of both exporters as well as non-exporters between pre-crisis and crisis

in the first phase, and crisis and recovery in the second phase. The diagonal elements of the

matrix represent the proportion of firms which remain in the same quartile in which they started,

i.e. persistence. Table A reports the TPM for each group over each period.

There are three important observations that emerge from these probability matrices.

1. The proportion of firms along the diagonal is consistently higher for non-exporters than

exporters. Hence, non-exporters show greater persistence while exporters display higher

volatility in performance. Hence, on one hand we expect exporters may be more

adversely affected due to crisis but on the other hand, they are likely to recover faster as

well.

2. As a result of crisis, a larger proportion of exporting firms (70.5%) move to lower growth

quartiles as compared to non-exporting firms (67.9%). This is again consistent with our

hypothesis that exporting firms are more adversely affected as a result of crisis.

3. Post crisis, in the recovery phase only 15.9 % of those exporting firms which were in the

lowest growth quartile during crisis remained there, whereas an overwhelming 29.2% of

the non-exporting firms remain stuck at the lowest growth quartile.

53

Logarithmic difference of sales, quarter on quarter

Exporters

Non Exporters

Crisis

Crisis

Transitional

Probability Matrix:

Pre Crisis to Crisis

Pre

Cri

sis

1 2 3 4

Pre

Cri

sis

1 2 3 4

1 29.3% 28.0% 18.3% 24.4%

1 30.3% 24.1% 20.4% 25.2%

2 23.0% 25.3% 27.6% 24.1%

2 20.5% 27.5% 31.2% 20.8%

3 22.8% 27.8% 29.1% 20.3%

3 21.5% 28.5% 27.6% 22.4%

4 22.7% 23.9% 23.9% 29.5%

4 26.9% 20.1% 20.9% 32.1%

Recovery

Recovery

Transitional

Probability Matrix:

Crisis to Recovery Cri

sis

1 2 3 4

Cri

sis

1 2 3 4

1 15.9% 20.7% 22.0% 41.5%

1 29.2% 18.8% 15.0% 37.0%

2 27.6% 25.3% 25.3% 23.0%

2 24.4% 23.4% 28.1% 21.8%

3 34.2% 30.4% 27.8% 12.7%

3 19.3% 31.8% 30.7% 22.1%

4 20.5% 27.3% 25.0% 21.6%

4 27.9% 23.5% 25.1% 22.2%

Table A: Transitional Probability Matrix for Exporters and Non-Exporters

Appendix B

Table B1: Annual Real Economic Activity Growth Rates

Year FY04 FY05 FY06 FY07 FY08 FY09 FY10 FY11 FY12 FY13

Manufacturing Growth 7.4% 10.1% 14.3% 10.3% 4.3% 11.3% 8.9% 7.4% 1.1% -0.7%

Agg. growth 7.1% 9.5% 9.6% 9.3% 6.7% 8.6% 8.9% 6.7% 4.5% 4.7%

Mfg. Real Exports Growth 15.2% 15.0% 13.2% 2.9% 28.7% -5.6% 24.6% 15.1% 6.2% 13.4%

Agg. Real Exports Growth 20.2% 16.4% 17.5% 9.6% 18.6% -3.1% 23.4% 17.7% 3.8% 9.4%

Source: RBI Handbook of Statistics

Table B2: NIC 2008 Manufacturing Goods

NIC Code Product Details Distribution

10 Mfr. Of food products 10.1%

11 Mfr. Of beverages 1.6%

12 Mfr. Of Tobacco products 0.2%

13 Mfr. Of Textiles 12.1%

14 Mfr. of Wearing apparels 1.2%

15 Mfr. Of leather and related products 1.0%

16 Mfr. Of wood and products of wood and cork 0.5%

17 Mfr.. Of paper and paper products 3.4%

18 Printing and reproduction of recorded media 0.2%

19 Mfr. Coke and refined petroleum products 1.2%

20 Mfr. Of Chemicals and chemical products 14.0%

21 Mfr. Of Pharmaceuticals, medicinal, chemical and botanical products 7.5%

22 Mfr. Of Rubber and plastic products 8.2%

23 Mfr. Of other nonmetallic mineral products 4.6%

24 Mfr. Of Basic Metals 10.6%

25 Mfr. Of Fabricated metal products except machinery and equipments 3.2%

26 Mfr. Of Computer, electronics and optical products 2.8%

27 Mfr. Of Electrical Equipments 4.5%

28 Mfr. Of Machinary and equipment n.e.c. 5.6%

29 Mfr. Of Motor Vehicles, Trailers and Semi Trailers 0.3%

30 Mfr. Of Other Transport equipment 5.3%

31 Mfr. Of Furniture 0.1%

32 Other Manufacturing 1.9%

Source: CMIE Prowess Database using author’s own calculations.

41 | P a g e

Table B3: 2 Digit Industry Avg. Exports (In Rs. Millions)

NIC Code 2005 2006 2007 2008 2009 2010 2011 2012

10 670.09 694.21 1032.74 1142.93 1006.76 1625.92 1864.28 2359.98

11 307.02 556.03 500.34 422.01 353.68 592.56 687.31 645.6

12 3585.05 4638.38 4423.53 5222.85 6089.48 6674.68 6490.85 9276.43

13 601.34 807.02 924.25 1004.82 1066.86 1460.77 1749.39 1940.74

14 328.65 889.8 981.46 1047.58 1128.06 899.31 975.09 1090.35

15 511.56 638.94 648.69 751.02 799.44 1133.52 1078.48 1119.97

16 108.96 181.6 189.76 173.95 270.45 345.1 495.5 681.82

17 244.09 226.01 189.28 199.72 199.48 248.2 311.25 349.64

19 44552.78 73606.64 95439.43 98198.79 115212 137183.2 195036.8 240771.9

20 612.17 695.34 757.44 942.65 944.08 1047.26 1373.39 1423.82

21 1243.02 1707.02 1910.34 2301.75 2593.72 2847.37 3588.1 4122.86

22 504.78 595.73 636.62 690.42 806.4 1147.25 1463.46 1580.46

23 478.7 586.14 651.64 699.76 673.11 659.02 791.36 782.47

24 2550.59 3890.86 3616.91 3489.87 3027.18 3902.19 4482.64 5014.62

25 314.97 452.73 604.62 781.75 697.86 625.47 742.17 925.54

26 620.36 723.7 702.16 743.52 771.23 811.01 892.3 745.03

27 280.21 425.61 526.07 626.21 633.37 735.54 679.15 700.44

28 464.88 777.47 876.77 1293.14 752.76 713.7 883.99 850.57

29 5464.2 5657.1 7206.08 7710.08 11861.5 13343.1 14844.2 15765.54

30 478.1 619.06 668.28 911.81 900.06 1422.27 2036.04 1995.87

31 4.1 8 2.9 4.4 9.9 8.5

32 2840.06 3108.07 3363.63 3352.64 4808.49 6779.06 8003.93 8973.23

Source: CMIE Prowess Database

Table B4: Size Wise Distribution of Average Sales (In Rs. Millions)

Exporters Non Exporters

Year 1st

Quartile 2nd

Quartile 3rd

Quartile 4th

Quartile 1st

Quartile 2nd

Quartile 3rd

Quartile 4th

Quartile

2005 263.5 1367.9 1822.7 12866.7 219.7 622.6 1550.8 11295.8

2006 315.4 1509.6 2217.8 14934.9 240.8 805.8 1988.5 13193.0

2007 394.0 1524.0 2883.8 16502.8 268.3 1025.1 2276.6 14647.8

2008 383.2 1413.3 3460.9 17610.7 278.5 1158.0 2653.5 16022.7

2009 412.6 1803.4 3300.5 18581.4 300.0 1310.6 2890.3 17316.4

2010 482.7 2351.6 3802.7 21932.0 366.1 1446.2 4024.7 20045.4

2011 520.5 2637.1 4167.5 23832.9 406.0 1637.6 4700.0 22305.3

2012 579.8 2289.4 4679.0 25658.6 459.9 1857.5 5394.8 23772.8

Source: CMIE Prowess Database using author’s own calculations. If a firm’s total asset lies below the 25th

percentile of total assets of that particular industry, the firm belongs to the lowest or the 1st quartile

42 | P a g e

Table B5: Size Wise Proportional Export Distribution for Firms Classified as Exporters

Year 1st

Quartile 2nd

Quartile 3rd

Quartile 4th

Quartile

2005 1.40% 7.85% 7.12% 83.64%

2006 1.14% 5.29% 6.24% 87.33%

2007 1.20% 4.64% 6.61% 87.56%

2008 1.06% 3.56% 8.28% 87.10%

2009 0.99% 4.87% 6.27% 87.87%

2010 0.77% 4.39% 5.68% 89.15%

2011 0.62% 3.57% 5.42% 90.39%

2012 0.59% 2.73% 4.58% 92.10%

Source: CMIE Prowess Database using author’s own calculations. If a firm’s total asset lies below the 25th

percentile of total assets of that particular industry, the firm belongs to the lowest or the 1st quartile. The

values reflect the proportional contribution of each size band to the total exports.

Table B6: Avg. Financial Health of Exporters and Non Exporters (in Rs. Millions)

Exporters Non Exporters

Year

Avg. Tot. borrowings

Avg. Sec. Bank

Borrowings

Avg. Finanial Instns.

Borrowing

Avg. of ECB

Avg. Tot. borrowings

Avg. Sec. Bank

Borrowings

Avg. Finanial Instns.

Borrowing

Avg. of ECB

2005 2752.5 1011.4 411.8 94.4 1537.1 589.7 424.8 78.1

2006 3413.0 1444.8 481.9 270.3 1892.6 785.4 463.4 103.8

2007 4128.6 1655.0 417.4 370.5 2451.4 988.9 595.6 366.9

2008 5528.0 1807.4 483.3 520.6 3087.0 1307.5 557.3 356.5

2009 5365.4 1972.3 417.2 539.6 3437.4 1492.7 732.3 336.5

2010 6382.4 2201.6 897.4 322.3 3905.5 1597.2 887.6 299.5

2011 7209.5 2618.4 496.1 722.8 4672.9 1937.6 550.5 410.8

2012 8524.4 3314.8 415.1 789.0 5591.2 2559.8 631.4 660.1

Source: CMIE Prowess Database using author’s own calculations.

43 | P a g e

Table B7: Percentage of Firms carrying out Research & Development Expenditure

Year Exporters Non Exporters

2005 40.7% 32.1%

2006 38.6% 32.3%

2007 41.4% 32.1%

2008 40.2% 32.6%

2009 41.4% 32.5%

2010 39.0% 32.7%

2011 39.1% 33.5%

2012 40.3% 35.1%

Source: CMIE Prowess Database using author’s own calculations.

Table B8: Summary Statistics Of Some Important Variables (Whole Sample)

Variable (Rs. Crores) Mean Std.

Deviation Median Min. Max.

Sales 4774.882 60939.53 510.7 0.09 4757501

Gross Fixed Assets 2468.792 26131.82 251.5 0.1 2212530

Export of Goods 1296.721 22419.7 95 0 2278830

Total Income 4898.867 61857.09 525.6 0.09 4795096

Profits 598.9869 5769.902 42.3 -8303 412680

Compensation of Employees

205.5037 1509.374 27.6 -0.8 86429.9

R&D Expenses 60.52611 403.1088 3.6 0 16425.3

Unsecured Borrowing 1558.081 12923.4 154.2 0 808944.9

Source: CMIE Prowess Database using author’s own calculations.

44 | P a g e

Table B9: Summary Statistics Of Some Important Variables (Exporters Only)

Variable (Rs. Crores) Mean Std.

Deviation Median Min. Max.

Sales 6170.911 12590.35 1621.6 1.8 66578.7

Gross Fixed Assets 9816.975 95637.3 1005.7 9.4 2200000

Export of Goods 6968.138 74435.44 722.1 0 2300000

Total Income 14681.75 133403.2 1681.33 2.2 3800000

Profits 2171.141 17914.14 196.4 -1390.9 412680

Compensation of Employees

433.4648 1655.897 103.2 0.1 33540

R&D Expenses 193.3944 724.4475 8.5 0 7098.5

Unsecured Borrowing 5427.579 34782.23 648.8 0 739045

Source: CMIE Prowess Database using author’s own calculations.

45 | P a g e

Appendix C

Table C1

(1) (2) (3)

Variables Full Sample Exporters Non-Exporters

Crisis_dummy -0.184*** -0.314*** -0.201***

(0.0298) (0.0343) (0.0351)

Export_dummy 0.0753***

(0.0189)

Interaction_dummy -0.128***

(0.0463)

Observations 88 44 44

R-squared 0.579 0.601 0.504

Table C2

(1) (2) (3)

Variables Full Sample Exporters Non-Exporters

Crisis_dummy -0.0641*** -0.0976*** -0.0637***

(0.00776) (0.0134) (0.00776)

Export_dummy 0.0170***

(0.00579)

Interaction_dummy -0.0329**

(0.0155)

NIC_10 -0.00312 0.0525 -0.0367

(0.0346) (0.0380) (0.0594)

NIC_13 -0.0153 0.00137 -0.0417

(0.0335) (0.0361) (0.0588)

NIC_15 0.00100 0.0280 -0.0321

(0.0380) (0.0397) (0.0690)

NIC_16 -0.0624 0.0220 -0.106

(0.0409) (0.0471) (0.0650)

NIC_17 0.0101 0.0606 -0.0208

(0.0352) (0.0501) (0.0595)

NIC_19 0.109** 0.102** 0.0839

(0.0452) (0.0472) (0.0684)

NIC_20 0.0296 0.0695* -0.00347

(0.0336) (0.0369) (0.0587)

NIC_21 0.0550 0.0878** 0.0212

(0.0338) (0.0362) (0.0592)

NIC_22 0.0462 0.0977*** 0.0101

(0.0339) (0.0362) (0.0590)

NIC_23 0.0194 0.0282 -0.00845

(0.0347) (0.0376) (0.0595)

NIC_24 -0.0359 -0.00506 -0.0672

46 | P a g e

(0.0340) (0.0367) (0.0591)

NIC_25 0.0258 -0.0794* 0.0217

(0.0369) (0.0423) (0.0613)

NIC_26 -0.0145 0.0639 -0.0586

(0.0362) (0.0412) (0.0610)

NIC_27 0.0166 0.0366 -0.0134

(0.0353) (0.0510) (0.0596)

NIC_28 0.0171 0.0461 -0.0140

(0.0341) (0.0400) (0.0591)

NIC_30 0.0425 0.0453 0.0152

(0.0338) (0.0390) (0.0588)

NIC_32 -0.0630 -0.0892* -0.0523

(0.0410) (0.0508) (0.0669)

Constant 0.0485 0.0399 0.0787

(0.0333) (0.0350) (0.0584)

Observations 57,477 10,608 46,869

R-squared 0.036 0.038 0.037

Table C3

(1) (2) (1) (2)

Variables Non-Exporters Exporters Variables Non-Exporters Exporters

crisis*NIC10 -0.199*** -0.173*** crisis*NIC24 -0.209*** -0.162***

(0.0272) (0.0558) (0.0263) (0.0417)

crisis*NIC13 -0.174*** -0.198*** crisis*NIC25 -0.273*** -0.225***

(0.0268) (0.0293) (0.0462) (0.0697)

crisis*NIC26 -0.135*** -0.178***

crisis*NIC15 -0.290*** -0.129* (0.0483) (0.0692)

(0.105) (0.0688) crisis*NIC27 -0.174*** -0.628***

crisis*NIC16 -0.0832 -0.217 (0.0360) (0.125)

(0.107) (0.154) crisis*NIC28 -0.235*** -0.238***

crisis*NIC17 -0.0373 -0.266* (0.0344) (0.0481)

(0.0421) (0.161)

crisis*NIC19 0.277*** -0.346*** crisis*NIC30 -0.256*** -0.288***

(0.0732) (0.124) (0.0343) (0.0657)

crisis*NIC20 -0.127*** -0.136***

(0.0232) (0.0355) crisis*NIC32 -0.302*** -0.176***

crisis*NIC21 -0.124*** -0.0722** (0.0761) (0.0588)

(0.0321) (0.0350) (0.249)

crisis*NIC22 -0.179*** -0.133*** Constant 0.0499 0.0401

(0.0296) (0.0437) (0.0350) (0.0250)

crisis*NIC23 -0.0840** -0.0593 Observations 54,985 12,101

(0.0373) (0.0696) R-squared 0.014 0.032

47 | P a g e

Table C4

(Pooled) FE(1) FE(2) FE(3)

VARIABLES Growth growth growth growth

Crisis dummy -0.150*** -0.148***

(0.0175) (0.0155)

Export intensity 0.00545* 0.00868 0.00765 0.00771

(0.00300) (0.00640) (0.00641) (0.00641)

R&D 0.0129 0.0477** 0.0506** 0.0511**

(0.00937) (0.0205) (0.0205) (0.0205)

Lagged size 0.00805** -0.103*** -0.104*** -0.104***

(0.00377) (0.0138) (0.0138) (0.0138)

Capacity utilization 0.0880*** 0.318*** 0.319*** 0.319***

(0.00895) (0.0153) (0.0153) (0.0153)

Liquidity 0.0494 0.00329 0.00717 0.00940

(0.0304) (0.0443) (0.0442) (0.0443)

Mkt share 0.00358*** 0.00660** 0.00581** 0.00577**

(0.000632) (0.00261) (0.00262) (0.00262)

Age -0.00117*** -0.00596** -0.00589** -0.00586**

crisis*NIC10 -0.171***

(0.0473)

crisis*NIC13 -0.0829***

(0.0254)

crisis*NIC14 0.00563

(0.0537)

crisis*NIC15 -0.141**

(0.0577)

crisis*NIC16 -0.198

(0.127)

crisis*NIC17 -0.153

(0.161)

crisis*NIC19 -0.359***

(0.104)

crisis*NIC20 -0.168***

(0.0308)

crisis*NIC21 -0.0578*

(0.0297)

crisis*NIC22 -0.153***

(0.0362)

crisis*NIC23 -0.0926

(0.0580)

crisis*NIC24 -0.168***

(0.0349)

crisis*NIC25 -0.251***

(0.0599)

48 | P a g e

crisis*NIC26 -0.170***

(0.0574)

crisis*NIC27 -0.656***

(0.103)

crisis*NIC28 -0.192***

(0.0403)

crisis*NIC30 -0.327***

(0.0544)

crisis*NIC32 -0.336***

(0.0549)

exposure*crisis*NIC10 -0.746***

(0.212)

exposure*crisis*NIC13 -0.178***

(0.0557)

exposure*crisis*NIC14 0.0119

(0.0719)

exposure*crisis*NIC15 -0.189**

(0.0793)

exposure*crisis*NIC16 -1.100

(0.716)

exposure*crisis*NIC17 -0.961

(1.243)

exposure*crisis*NIC19 -1.463***

(0.447)

exposure*crisis*NIC20 -0.498***

(0.0958)

exposure*crisis*NIC21 -0.140*

(0.0739)

exposure*crisis*NIC22 -0.393***

(0.0949)

exposure*crisis*NIC23 -0.348

(0.226)

exposure*crisis*NIC24 -0.823***

(0.174)

exposure*crisis*NIC25 -0.508***

(0.121)

exposure*crisis*NIC26 -0.350**

(0.137)

exposure*crisis*NIC27 -1.435***

(0.225)

exposure*crisis*NIC28 -0.554***

(0.119)

exposure*crisis*NIC30 -1.899***

(0.366)

exposure*crisis*NIC32 -1.426***

(0.237)

49 | P a g e

R-squared 0.051 0.081 0.089 0.088

Table C5

(Pooled OLS) FE(1) FE(2) FE(3)

VARIABLES growth growth growth Growth

Crisis dummy -0.150*** -0.148***

(0.0175) (0.0154)

Export intensity 0.00500* 0.00862 0.00761 0.00769

(0.00298) (0.00639) (0.00640) (0.00640)

R&D 0.0173* 0.0475** 0.0504** 0.0509**

(0.00946) (0.0205) (0.0205) (0.0205)

Lagged size 0.00432 -0.103*** -0.104*** -0.104***

(0.00346) (0.0137) (0.0137) (0.0137)

Capacity utilization 0.0781*** 0.308*** 0.310*** 0.310***

(0.00881) (0.0158) (0.0157) (0.0158)

Trade credit -0.169*** -0.105** -0.108** -0.108**

(0.0422) (0.0434) (0.0434) (0.0434)

Mkt share 0.00396*** 0.00675*** 0.00599** 0.00595**

(0.000634) (0.00261) (0.00262) (0.00262)

Age -0.00134*** -0.00596** -0.00589** -0.00586**

(0.000277) (0.00269) (0.00269) (0.00269)

crisis*NIC10 -0.166***

(0.0473)

crisis*NIC13 -0.0812***

(0.0254)

crisis*NIC14 0.00616

(0.0536)

crisis*NIC15 -0.142**

(0.0577)

crisis*NIC16 -0.201

(0.127)

crisis*NIC17 -0.156

(0.161)

crisis*NIC19 -0.358***

(0.104)

crisis*NIC20 -0.170***

(0.0307)

crisis*NIC21 -0.0588**

(0.0297)

crisis*NIC22 -0.153***

(0.0362)

crisis*NIC23 -0.0900

(0.0580)

50 | P a g e

crisis*NIC24 -0.170***

(0.0349)

crisis*NIC25 -0.251***

(0.0599)

crisis*NIC26 -0.173***

(0.0574)

crisis*NIC27 -0.660***

(0.103)

crisis*NIC28 -0.192***

(0.0403)

crisis*NIC30 -0.326***

(0.0544)

crisis*NIC32 -0.333***

(0.0549)

mktexposure*crisis*NIC10 -0.725***

(0.212)

mktexposure*crisis*NIC13 -0.174***

(0.0557)

mktexposure*crisis*NIC14 0.0126

(0.0719)

mktexposure*crisis*NIC15 -0.191**

(0.0793)

mktexposure*crisis*NIC16 -1.114

(0.715)

mktexposure*crisis*NIC17 -0.981

(1.242)

mktexposure*crisis*NIC19 -1.457***

(0.447)

mktexposure*crisis*NIC20 -0.504***

(0.0958)

mktexposure*crisis*NIC21 -0.143*

(0.0739)

mktexposure*crisis*NIC22 -0.395***

(0.0949)

mktexposure*crisis*NIC23 -0.338

(0.226)

mktexposure*crisis*NIC24 -0.833***

(0.174)

mktexposure*crisis*NIC25 -0.507***

(0.121)

mktexposure*crisis*NIC26 -0.355***

(0.137)

mktexposure*crisis*NIC27 -1.443***

(0.225)

mktexposure*crisis*NIC28 -0.553***

(0.119)

mktexposure*crisis*NIC30 -1.893***

51 | P a g e

(0.366)

mktexposure*crisis*NIC32 -1.415***

(0.237)

Observations 10,198 10,198 10,198 10,198

R-squared 0.055 0.082 0.090 0.089

Table C6

(1) (2)

VARIABLES growth growth

Crisis dummy -0.146*** -0.177***

(0.0293) (0.0249)

Export intensity 0.00764 0.00751

(0.00645) (0.00644)

Expint*crisis -0.0283** -0.0285**

(0.0130) (0.0130)

Elasticity 0.00404 0.00358

(0.00432) (0.00432)

Elasticity*crisis -0.00203 -0.00339

(0.00553) (0.00547)

Liquidity 0.0113

(0.0450)

Liquidity*crisis -0.0750

(0.0604)

Trade credit -0.123***

(0.0468)

Tradecredit*crisis 0.0546

(0.0572)

R&D 0.0415** 0.0413**

(0.0207) (0.0207)

R&D*crisis 0.0481** 0.0494**

(0.0207) (0.0206)

Mkt share 0.00639** 0.00667**

(0.00261) (0.00261)

Mktshare*crisis 0.00468** 0.00481**

(0.00190) (0.00190)

Capacityutilization 0.318*** 0.308***

(0.0154) (0.0158)

Laggedsize -0.103*** -0.103***

(0.0138) (0.0137)

Age -0.00594** -0.00596**

(0.00270) (0.00270)

Constant 0.599*** 0.631***

(0.0809) (0.0795)

52 | P a g e

Observations 10,166 10,166

R-squared 0.083 0.083

Table C7

FE(1) FE(2)

VARIABLES Growth Growth

Crisis dummy -0.157*** -0.207***

(0.0313) (0.0266)

Export intensity 0.00520 0.00484

(0.00619) (0.00618)

Expint*crisis -0.0361*** -0.0359***

(0.0126) (0.0126)

Elasticity 0.00315 0.00208

(0.00547) (0.00547)

Elasticity*crisis -0.00269 -0.00517

(0.00579) (0.00572)

Liquidity -0.0279

(0.0483)

Liquidity*crisis -0.139**

(0.0654)

Trade credit -0.168***

(0.0503)

Tradecredit*crisis 0.0616

(0.0580)

R&D 0.0657*** 0.0665***

(0.0219) (0.0218)

R&d*crisis 0.0846*** 0.0879***

(0.0223) (0.0222)

Mkt share 0.0161*** 0.0163***

(0.00403) (0.00403)

Mktshare*crisis 0.00274 0.00276

(0.00221) (0.00222)

Capacity utilization 0.282*** 0.265***

(0.0167) (0.0171)

Lagged size -0.108*** -0.107***

(0.0157) (0.0157)

Age -0.00700** -0.00694**

(0.00305) (0.00305)

Constant 0.712*** 0.737***

(0.0904) (0.0887)

Observations 7,752 7,752

R-squared 0.088 0.089

53 | P a g e

(1) (2) (3) (4)

VARIABLES growth growth growth growth

Export intensity 0.0492** 0.0459** 0.00806 0.00695

(0.0198) (0.0181) (0.00640) (0.00642)

R&D 0.0600** 0.0634** 0.0365* 0.0506**

(0.0301) (0.0298) (0.0205) (0.0205)

Lagged size -0.0996*** -0.104*** -0.0940*** -0.102***

(0.0201) (0.0199) (0.0138) (0.0137)

Crisis dummy -0.152*** -0.148*** -0.148*** -0.137***

(0.0154) (0.0211) (0.0154) (0.0188)

Lagged profitability -0.222*** -0.231*** -0.265*** -0.259***

(0.0517) (0.0516) (0.0417) (0.0416)

Capacity utilization 0.381*** 0.386*** 0.307*** 0.316***

(0.0230) (0.0233) (0.0159) (0.0153)

Trade credit -0.115* -0.103

(0.0655) (0.0651)

Mkt share 0.0137*** 0.0112** 0.00729*** 0.00642**

(0.00464) (0.00456) (0.00271) (0.00262)

Age firm 0.0260*** 0.0256*** -0.00627** -0.00599**

(0.00560) (0.00536) (0.00270) (0.00269)

VS*NIC10 -0.000875

(0.0173)

VS*NIC13 -0.00142

(0.0110)

VS*NIC14 -0.0460

(0.185)

VS*NIC15 -0.00620

(0.00884)

VS*NIC16 1.080**

(0.538)

VS*NIC17 -0.0188

(0.0210)

VS*NIC19 0.263

(0.210)

VS*NIC20 0.0271

(0.0166)

VS*NIC21 0.0511

(0.0712)

VS*NIC22 0.000689

(0.0327)

VS*NIC23 0.239***

(0.0736)

VS*NIC24 0.0909

(0.0748)

VS*NIC25 -4.62e-06

(8.93e-06)

54 | P a g e

VS*NIC26 0.00129

(0.0104)

VS*NIC27 0.347***

(0.0831)

VS*NIC28 0.00518

(0.00331)

VS*NIC30 -0.126**

(0.0578)

VS*NIC32 0.213***

(0.0509)

VS*crisis*NIC10 -0.0577***

(0.0194)

VS*crisis*NIC13 0.00132

(0.00350)

VS*crisis*NIC14 10.10*

(5.738)

VS*crisis*NIC15 0.00730

(0.00789)

VS*crisis*NIC16 -0.0450

(0.118)

VS*crisis*NIC17 0.0157

(0.0455)

VS*crisis*NIC19 -0.563**

(0.240)

VS*crisis*NIC20 0.00460

(0.00543)

VS*crisis*NIC21 0.143***

(0.0486)

VS*crisis*NIC22 0.00309

(0.0177)

VS*crisis*NIC23 0.0455

(0.0284)

VS*crisis*NIC24 0.00453

(0.0180)

VS*crisis*NIC25 -4.62e-06

(8.92e-06)

VS*crisis*NIC26 -0.0230***

(0.00756)

VS*crisis*NIC27 -0.626***

(0.0988)

VS*crisis*NIC28 0.00151

(0.00298)

VS*crisis*NIC30 -0.0994**

(0.0404)

VS*crisis*NIC32 -0.0552***

(0.0178)

tradecredit*NIC10 -0.0730

55 | P a g e

(0.132)

tradecredit*NIC13 -0.0575

(0.0702)

tradecredit*NIC14 0.137

(0.478)

tradecredit*NIC15 -0.469

(0.311)

tradecredit*NIC16 -0.167

(0.629)

tradecredit*NIC17 -2.681

(1.878)

tradecredit*NIC19 1.042

(0.815)

tradecredit*NIC20 0.414**

(0.187)

tradecredit*NIC21 -1.305***

(0.143)

tradecredit*NIC22 0.270

(0.178)

tradecredit*NIC23 0.857**

(0.337)

tradecredit*NIC24 -0.452***

(0.124)

tradecredit*NIC25 1.077***

(0.315)

tradecredit*NIC26 1.109***

(0.365)

tradecredit*NIC27 -0.0938

(0.431)

tradecredit*NIC28 -0.0545

(0.144)

tradecredit*NIC30 0.126

(0.235)

tradecredit*NIC32 0.324

(0.233)

tradecredit*crisis*NIC10 0.115

(0.140)

tradecredit*crisis*NIC13 0.0557

(0.0658)

tradecredit*crisis*NIC14 0.188

(0.425)

tradecredit*crisis*NIC15 -0.310

(0.415)

tradecredit*crisis*NIC16 -0.0690

(0.672)

tradecredit*crisis*NIC17 0.922

(1.359)

56 | P a g e

tradecredit*crisis*NIC19 -1.537**

(0.731)

tradecredit*crisis*NIC20 -0.549**

(0.216)

tradecredit*crisis*NIC21 0.204

(0.144)

tradecredit*crisis*NIC22 0.560**

(0.219)

tradecredit*crisis*NIC23 0.650**

(0.268)

tradecredit*crisis*NIC24 -0.215

(0.174)

tradecredit*crisis*NIC25 -0.552*

(0.316)

tradecredit*crisis*NIC26 -0.0133

(0.413)

tradecredit*crisis*NIC27 -4.090***

(0.702)

tradecredit*crisis*NIC28 0.0802

(0.164)

tradecredit*crisis*NIC30 -0.213

(0.214)

tradecredit*crisis*NIC32 -0.768***

(0.187)

Constant -0.407*** -0.292** 0.573*** 0.590***

(0.156) (0.143) (0.0789) (0.0780)

Observations 6,428 6,428 10,198 10,198

R-squared 0.098 0.104 0.094 0.089

Number of cmiecompanycode 351 351 352 352

(1) (2) (3) (4)

VARIABLES growth growth growth growth

Export intensity 0.0464** 0.0464** 0.00806 0.00695

(0.0186) (0.0182) (0.00640) (0.00642)

R&D 0.0589* 0.0602** 0.0365* 0.0506**

(0.0302) (0.0299) (0.0205) (0.0205)

Lagged size -0.108*** -0.109*** -0.0940*** -0.102***

(0.0202) (0.0200) (0.0138) (0.0137)

Crisis dummy -0.147*** -0.187*** -0.148*** -0.137***

(0.0154) (0.0246) (0.0154) (0.0188)

Lagged profitability -0.239*** -0.238*** -0.265*** -0.259***

(0.0519) (0.0518) (0.0417) (0.0416)

Capacity utilization 0.373*** 0.374*** 0.307*** 0.316***

(0.0231) (0.0230) (0.0159) (0.0153)

57 | P a g e

Trade credit -0.0955 -0.107*

(0.0654) (0.0652)

Mkt share 0.0153*** 0.0158*** 0.00729*** 0.00642**

(0.00474) (0.00470) (0.00271) (0.00262)

Age firm 0.0270*** 0.0263*** -0.00627** -0.00599**

(0.00555) (0.00538) (0.00270) (0.00269)

VS*NIC10 0.101

(0.0620)

VS*NIC13 -0.0721

(0.0535)

VS*NIC14 -0.0215

(1.523)

VS*NIC15 -0.0111

(0.0520)

VS*NIC16 -0.219

(0.320)

VS*NIC17 -0.0363

(0.310)

VS*NIC19 2.132*

(1.170)

VS*NIC20 -0.0105

(0.0116)

VS*NIC21 0.864

(1.056)

VS*NIC22 0.372

(0.596)

VS*NIC23 0.117

(0.205)

VS*NIC24 0.247

(0.168)

VS*NIC25 -2.87e-06

(5.11e-06)

VS*NIC26 -0.0237

(0.0621)

VS*NIC27 1.471***

(0.383)

VS*NIC28 -0.00585

(0.0149)

VS*NIC30 -0.209

(0.160)

VS*NIC32 0.0811

(0.101)

VS*crisis*NIC10 0.122***

(0.0460)

VS*crisis*NIC13 0.0561

(0.0394)

VS*crisis*NIC14 1.962**

58 | P a g e

(0.820)

VS*crisis*NIC15 0.0276

(0.0337)

VS*crisis*NIC16 -0.0342

(0.342)

VS*crisis*NIC17 0.0762

(0.166)

VS*crisis*NIC19 -1.438**

(0.651)

VS*crisis*NIC20 0.0276

(0.0294)

VS*crisis*NIC21 0.538***

(0.147)

VS*crisis*NIC22 0.0981

(0.0775)

VS*crisis*NIC23 0.0681

(0.0774)

VS*crisis*NIC24 0.0548

(0.0459)

VS*crisis*NIC25 -2.00e-06

(5.11e-06)

VS*crisis*NIC26 0.0156

(0.0446)

VS*crisis*NIC27 -1.419***

(0.261)

VS*crisis*NIC28 0.00336

(0.00987)

VS*crisis*NIC30 -0.0846

(0.140)

VS*crisis*NIC32 -0.245

(0.156)

tradecredit*NIC10 -0.0730

(0.132)

tradecredit*NIC13 -0.0575

(0.0702)

tradecredit*NIC14 0.137

(0.478)

tradecredit*NIC15 -0.469

(0.311)

tradecredit*NIC16 -0.167

(0.629)

tradecredit*NIC17 -2.681

(1.878)

tradecredit*NIC19 1.042

(0.815)

tradecredit*NIC20 0.414**

(0.187)

59 | P a g e

tradecredit*NIC21 -1.305***

(0.143)

tradecredit*NIC22 0.270

(0.178)

tradecredit*NIC23 0.857**

(0.337)

tradecredit*NIC24 -0.452***

(0.124)

tradecredit*NIC25 1.077***

(0.315)

tradecredit*NIC26 1.109***

(0.365)

tradecredit*NIC27 -0.0938

(0.431)

tradecredit*NIC28 -0.0545

(0.144)

tradecredit*NIC30 0.126

(0.235)

tradecredit*NIC32 0.324

(0.233)

tradecredit*crisis*NIC10 0.115

(0.140)

tradecredit*crisis*NIC13 0.0557

(0.0658)

tradecredit*crisis*NIC14 0.188

(0.425)

tradecredit*crisis*NIC15 -0.310

(0.415)

tradecredit*crisis*NIC16 -0.0690

(0.672)

tradecredit*crisis*NIC17 0.922

(1.359)

tradecredit*crisis*NIC19 -1.537**

(0.731)

tradecredit*crisis*NIC20 -0.549**

(0.216)

tradecredit*crisis*NIC21 0.204

(0.144)

tradecredit*crisis*NIC22 0.560**

(0.219)

tradecredit*crisis*NIC23 0.650**

(0.268)

tradecredit*crisis*NIC24 -0.215

(0.174)

tradecredit*crisis*NIC25 -0.552*

(0.316)

tradecredit*crisis*NIC26 -0.0133

60 | P a g e

(0.413)

tradecredit*crisis*NIC27 -4.090***

(0.702)

tradecredit*crisis*NIC28 0.0802

(0.164)

tradecredit*crisis*NIC30 -0.213

(0.214)

tradecredit*crisis*NIC32 -0.768***

(0.187)

Constant -0.369** -0.273* 0.573*** 0.590***

(0.149) (0.143) (0.0789) (0.0780)

Observations 6,428 6,428 10,198 10,198

R-squared 0.093 0.099 0.094 0.089

Number of cmiecompanycode 351 351 352 352

Table C10: Dynamic Panel Estimation- Crisis, Exposure and Industries

(1) (2) (3)

VARIABLES growth growth growth

Crisis_dummy -0.126***

(0.0188)

L.growth -0.228*** -0.231*** -0.232***

(0.0263) (0.0256) (0.0257)

Exportintensity 0.00747* 0.00669 0.00784

(0.00446) (0.00436) (0.00506)

Randd 0.0574*** 0.0518*** 0.0536***

(0.0172) (0.0171) (0.0170)

Laggedsize 0.00949 0.00872 0.00861

(0.00638) (0.00638) (0.00639)

Laggedprofitability 0.0739 0.0770 0.0739

(0.0920) (0.0905) (0.0944)

Capacityutilization 0.122*** 0.125*** 0.120***

(0.0270) (0.0272) (0.0268)

Tradecredit -0.230** -0.229** -0.234**

(0.100) (0.0988) (0.101)

Mktshare 0.00326** 0.00387** 0.00348**

(0.00150) (0.00162) (0.00157)

Age_firm -0.00149*** -0.00136*** -0.00142***

(0.000534) (0.000510) (0.000508)

crisis*NIC10 -0.152***

61 | P a g e

(0.0576)

crisis*NIC13 -0.0799**

(0.0324)

crisis*NIC14 -0.0161

(0.0862)

crisis*NIC15 -0.187***

(0.0672)

crisis*NIC16 -0.219***

(0.0607)

crisis*NIC17 -0.180*

(0.107)

crisis*NIC19 -0.293***

(0.0570)

crisis*NIC20 -0.123**

(0.0586)

crisis*NIC21 0.00685

(0.0402)

crisis*NIC22 -0.101**

(0.0402)

crisis*NIC23 -0.0358

(0.111)

crisis*NIC24 -0.179***

(0.0523)

crisis*NIC25 -0.227***

(0.0534)

crisis*NIC26 -0.157*

(0.0854)

crisis*NIC27 -0.554**

(0.217)

crisis*NIC28 -0.179**

(0.0734)

crisis*NIC30 -0.237**

(0.120)

crisis*NIC32 -0.381***

(0.119)

exposure*crisis*NIC10 -0.718***

(0.258)

exposure*crisis*NIC13 -0.194***

(0.0734)

exposure*crisis*NIC14 -0.0227

(0.117)

exposure*crisis*NIC15 -0.237**

(0.0994)

exposure*crisis*NIC16 -1.416

(1.034)

exposure*crisis*NIC17 -0.448

(1.501)

62 | P a g e

exposure*crisis*NIC19 -1.223**

(0.523)

exposure*crisis*NIC20 -0.362**

(0.176)

exposure*crisis*NIC21 0.0216

(0.0947)

exposure*crisis*NIC22 -0.248**

(0.103)

exposure*crisis*NIC23 -0.116

(0.397)

exposure*crisis*NIC24 -0.912***

(0.257)

exposure*crisis*NIC25 -0.484***

(0.112)

exposure*crisis*NIC26 -0.308

(0.220)

exposure*crisis*NIC27 -1.173**

(0.476)

exposure*crisis*NIC28 -0.540**

(0.225)

exposure*crisis*NIC30 -1.442*

(0.826)

exposure*crisis*NIC32 -1.615***

(0.516)

Observations 8,872 8,872 8,872

Number of cmiecompanycode 348 348 348

Table C11: Dynamic Panel Estimation- Full Sample vs.

Intermediates

(Full Sample) (Intermediate)

VARIABLES Growth growth

L.growth -0.224*** -0.182***

(0.0265) (0.0272)

Crisis dummy -0.183** -0.334***

(0.0864) (0.113)

Export intensity 0.00748 0.00823*

(0.00458) (0.00481)

Expint*crisis -0.0258** -0.0319***

(0.0103) (0.0112)

Tradecredit -0.347*** -0.350***

(0.112) (0.128)

Tradecredit*crisis 0.0487 0.0597

(0.139) (0.151)

63 | P a g e

R&D 0.0483*** 0.0475**

(0.0164) (0.0188)

R&D*crisis 0.0664** 0.0981***

(0.0335) (0.0377)

Mkt share 0.00270* 0.00360

(0.00141) (0.00256)

Mktshare*crisis 0.00496* -0.00130

(0.00257) (0.00365)

Lagged profitability 0.123 0.0963

(0.0954) (0.115)

Lagged size -0.00404 -0.0150**

(0.00562) (0.00722)

Size*crisis 0.00373 0.0217

(0.0122) (0.0160)

Age firm -0.00140*** -0.00156***

(0.000468) (0.000549)

Observations 8,872 6,783

Number of cmiecompanycode 348 262

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Table C12: Dynamic Panel Estimation- Vertical Specialization (Exports)

(1) (2) (3) (4)

VARIABLES growth growth growth growth

Lagged growth -0.247*** -0.248*** -0.226*** -0.231***

(0.0396) (0.0391) (0.0263) (0.0259)

Export intensity 0.0131 0.0169** 0.00594 0.00849*

(0.0107) (0.00845) (0.00716) (0.00507)

Crisis dummy -0.124*** -0.121*** -0.127*** -0.118***

(0.0177) (0.0244) (0.0187) (0.0262)

R&D 0.0594** 0.0619*** 0.0428* 0.0511***

(0.0248) (0.0219) (0.0221) (0.0172)

Lagged size 0.0133 0.0163* 0.0132* 0.0106*

(0.00926) (0.00837) (0.00723) (0.00641)

Lagged profitability 0.0105 0.0181 0.0668 0.0639

(0.127) (0.129) (0.0919) (0.0957)

Capacity utilization 0.139*** 0.123*** 0.121*** 0.120***

(0.0311) (0.0314) (0.0265) (0.0259)

Trade credit -0.176* -0.198** -0.126 -0.234**

(0.0992) (0.0967) (1.080) (0.106)

Mkts hare 0.00426 0.00362 0.00313* 0.00310**

(0.00326) (0.00240) (0.00177) (0.00151)

Age firm -0.00203*** -0.00212*** -0.00141*** -0.00152***

(0.000735) (0.000724) (0.000529) (0.000508)

64 | P a g e

VS*NIC10 -0.0225

(0.0401)

VS*NIC13 0.00326

(0.00313)

VS*NIC14 0.119

(0.258)

VS*NIC15 -0.0141**

(0.00664)

VS*NIC16 0.0825

(0.0752)

VS*NIC17 -0.0116

(0.0192)

VS*NIC19 -0.00323

(0.166)

VS*NIC20 -0.00130

(0.00461)

VS*NIC21 0.144***

(0.0415)

VS*NIC22 0.0181

(0.0134)

VS*NIC23 0.0477*

(0.0285)

VS*NIC24 -0.0174

(0.0238)

VS*NIC25 -4.54e-06**

(1.84e-06)

VS*NIC26 -0.00389**

(0.00186)

VS*NIC27 0.0488

(0.0592)

VS*NIC28 0.00313

(0.00290)

VS*NIC30 -0.0390

(0.0415)

VS*NIC32 -0.0223

(0.0220)

VS*crisis*NIC10 -0.0349

(0.0280)

VS*crisis*NIC13 0.00191

(0.00472)

VS*crisis*NIC14 12.29

(8.562)

VS*crisis*NIC15 -0.00310

(0.0113)

VS*crisis*NIC16 -0.0763

(0.0944)

VS*crisis*NIC17 -0.0205

65 | P a g e

(0.0184)

VS*crisis*NIC19 -0.588***

(0.194)

VS*crisis*NIC20 0.00407

(0.00435)

VS*crisis*NIC21 0.195***

(0.0627)

VS*crisis*NIC22 0.00755

(0.0170)

VS*crisis*NIC23 0.0206

(0.0630)

VS*crisis*NIC24 -0.0273

(0.0310)

VS*crisis*NIC25 -4.61e-06***

(1.68e-06)

VS*crisis*NIC26 -0.0257***

(0.00492)

VS*crisis*NIC27 -0.494***

(0.0944)

VS*crisis*NIC28 -8.99e-05

(0.00232)

VS*crisis*NIC30 -0.0649

(0.0642)

VS*crisis*NIC32 -0.0810**

(0.0383)

tradecredit*NIC10 -0.164

(1.085)

tradecredit*NIC13 -0.111

(1.097)

tradecredit*NIC14 -0.101

(1.109)

tradecredit*NIC15 -0.329

(1.116)

tradecredit*NIC16 0.304

(1.407)

tradecredit*NIC19 0.207

(1.784)

tradecredit*NIC20 0.0719

(1.115)

tradecredit*NIC21 0.0399

(1.172)

tradecredit*NIC22 0.183

(1.125)

tradecredit*NIC23 0.339

(1.121)

tradecredit*NIC24 -0.485

(1.118)

66 | P a g e

tradecredit*NIC25 -0.185

(1.091)

tradecredit*NIC26 0.226

(1.134)

tradecredit*NIC27 -0.509

(1.161)

tradecredit*NIC28 -0.0970

(1.105)

tradecredit*NIC30 0.266

(1.105)

tradecredit*NIC32 -0.0865

(1.079)

tradecredit*crisis*NIC10 0.0823

(0.108)

tradecredit*crisis*NIC13 0.0741

(0.163)

tradecredit*crisis*NIC14 0.0966

(0.873)

tradecredit*crisis*NIC15 -0.772

(0.548)

tradecredit*crisis*NIC16 -0.860

(0.704)

tradecredit*crisis*NIC17 -0.261

(1.395)

tradecredit*crisis*NIC19 -1.902

(1.241)

tradecredit*crisis*NIC20 -0.302

(0.422)

tradecredit*crisis*NIC21 0.535**

(0.254)

tradecredit*crisis*NIC22 0.611***

(0.178)

tradecredit*crisis*NIC23 0.810***

(0.283)

tradecredit*crisis*NIC24 -0.382

(0.242)

tradecredit*crisis*NIC25 -0.721*

(0.380)

tradecredit*crisis*NIC26 -0.149

(0.722)

tradecredit*crisis*NIC27 -3.773***

(1.260)

tradecredit*crisis*NIC28 -0.0187

(0.194)

tradecredit*crisis*NIC30 0.142

(0.420)

tradecredit*crisis*NIC32 -1.025*

67 | P a g e

(0.537)

Constant -0.0387 -0.0307 -0.175** -0.149**

(0.0862) (0.0814) (0.0740) (0.0667)

Observations 5,441 5,441 8,872 8,872

Number of cmiecompanycode 346 346 348 348

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Table C13: Dynamic Panel Estimation- Vertical Specialization (Imports)

(1) (2) (3) (4)

VARIABLES growth growth growth growth

Lagged growth -0.245*** -0.245*** -0.226*** -0.231***

(0.0391) (0.0394) (0.0263) (0.0259)

Export intensity 0.0114 0.0170* 0.00594 0.00849*

(0.0116) (0.00873) (0.00716) (0.00507)

Crisis dummy -0.123*** -0.146*** -0.127*** -0.118***

(0.0177) (0.0343) (0.0187) (0.0262)

R&D 0.0549** 0.0626*** 0.0428* 0.0511***

(0.0257) (0.0223) (0.0221) (0.0172)

Lagged size 0.0114 0.0156* 0.0132* 0.0106*

(0.00996) (0.00857) (0.00723) (0.00641)

Lagged profitability -0.00464 0.00621 0.0668 0.0639

(0.125) (0.128) (0.0919) (0.0957)

Capacity utilization 0.137*** 0.124*** 0.121*** 0.120***

(0.0353) (0.0330) (0.0265) (0.0259)

Trade credit -0.191* -0.187* -0.126 -0.234**

(0.103) (0.0954) (1.080) (0.106)

Mkts hare 0.00524* 0.00438* 0.00313* 0.00310**

(0.00295) (0.00228) (0.00177) (0.00151)

Age firm -0.00204*** -0.00220*** -0.00141*** -0.00152***

(0.000717) (0.000727) (0.000529) (0.000508)

VS*NIC10 -0.00573

(0.0543)

VS*NIC13 0.0506

(0.0487)

VS*NIC14 0.578

(0.708)

VS*NIC15 -0.0568

(0.0365)

VS*NIC16 0.137

(0.111)

VS*NIC17 -0.111

(0.114)

VS*NIC19 0.159

(0.497)

68 | P a g e

VS*NIC20 -0.00360

(0.0102)

VS*NIC21 0.488***

(0.133)

VS*NIC22 0.125

(0.0886)

VS*NIC23 0.0983*

(0.0566)

VS*NIC24 -0.0152

(0.0509)

VS*NIC25 -2.56e-06**

(1.16e-06)

VS*NIC26 -0.0170

(0.0332)

VS*NIC27 -0.00790

(0.214)

VS*NIC28 -0.000188

(0.00615)

VS*NIC30 0.0471

(0.132)

VS*NIC32 -0.0421

(0.126)

VS*crisis*NIC10 0.0297

(0.0449)

VS*crisis*NIC13 0.0380

(0.0573)

VS*crisis*NIC14 1.941

(1.278)

VS*crisis*NIC15 -0.0134

(0.0551)

VS*crisis*NIC16 -0.153

(0.308)

VS*crisis*NIC17 -0.0667

(0.0616)

VS*crisis*NIC19 -1.521***

(0.514)

VS*crisis*NIC20 0.0208

(0.0327)

VS*crisis*NIC21 0.597***

(0.196)

VS*crisis*NIC22 0.0981

(0.106)

VS*crisis*NIC23 0.0212

(0.122)

VS*crisis*NIC24 -0.0381

(0.0765)

VS*crisis*NIC25 -2.22e-06**

69 | P a g e

(1.01e-06)

VS*crisis*NIC26 -0.00162

(0.0589)

VS*crisis*NIC27 -1.037**

(0.413)

VS*crisis*NIC28 -0.00965

(0.0158)

VS*crisis*NIC30 0.0773

(0.272)

VS*crisis*NIC32 -0.591

(0.359)

tradecredit*NIC10 -0.164

(1.085)

tradecredit*NIC13 -0.111

(1.097)

tradecredit*NIC14 -0.101

(1.109)

tradecredit*NIC15 -0.329

(1.116)

tradecredit*NIC16 0.304

(1.407)

tradecredit*NIC19 0.207

(1.784)

tradecredit*NIC20 0.0719

(1.115)

tradecredit*NIC21 0.0399

(1.172)

tradecredit*NIC22 0.183

(1.125)

tradecredit*NIC23 0.339

(1.121)

tradecredit*NIC24 -0.485

(1.118)

tradecredit*NIC25 -0.185

(1.091)

tradecredit*NIC26 0.226

(1.134)

tradecredit*NIC27 -0.509

(1.161)

tradecredit*NIC28 -0.0970

(1.105)

tradecredit*NIC30 0.266

(1.105)

tradecredit*NIC32 -0.0865

(1.079)

tradecredit*crisis*NIC10 0.0823

(0.108)

70 | P a g e

tradecredit*crisis*NIC13 0.0741

(0.163)

tradecredit*crisis*NIC14 0.0966

(0.873)

tradecredit*crisis*NIC15 -0.772

(0.548)

tradecredit*crisis*NIC16 -0.860

(0.704)

tradecredit*crisis*NIC17 -0.261

(1.395)

tradecredit*crisis*NIC19 -1.902

(1.241)

tradecredit*crisis*NIC20 -0.302

(0.422)

tradecredit*crisis*NIC21 0.535**

(0.254)

tradecredit*crisis*NIC22 0.611***

(0.178)

tradecredit*crisis*NIC23 0.810***

(0.283)

tradecredit*crisis*NIC24 -0.382

(0.242)

tradecredit*crisis*NIC25 -0.721*

(0.380)

tradecredit*crisis*NIC26 -0.149

(0.722)

tradecredit*crisis*NIC27 -3.773***

(1.260)

tradecredit*crisis*NIC28 -0.0187

(0.194)

tradecredit*crisis*NIC30 0.142

(0.420)

tradecredit*crisis*NIC32 -1.025*

(0.537)

Constant -0.0333 -0.0373 -0.175** -0.149**

(0.0954) (0.0842) (0.0740) (0.0667)

Observations 5,441 5,441 8,872 8,872

Number of cmiecompanycode 346 346 348 348

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

71 | P a g e

Table C14: Sectoral Domestic Linkage Index

(Exporters) (Non-exporters) (All firms)

VARIABLES growth growth growth

Export intensity 0.0477** 0.299** 0.0812*

(0.0200) (0.126) (0.0457)

R&D 0.0690 -0.00852 0.00605

(0.0563) (0.0207) (0.0227)

Lagged size -0.0995*** -0.120*** -0.111***

(0.0322) (0.0399) (0.0302)

Lagged profitability -0.223* -0.206* -0.216**

(0.113) (0.112) (0.0901)

Capacity utilization 0.386*** 0.315*** 0.335***

(0.0651) (0.0306) (0.0280)

Liquidity -0.0195 0.00933 0.00542

(0.138) (0.0889) (0.0758)

Mkt share 0.0107 0.0137* 0.0120*

(0.00836) (0.00795) (0.00671)

Age firm 0.0253*** 0.0212*** 0.0228***

(0.00693) (0.00567) (0.00457)

DLVS_crisis10 -0.0420 -0.0999*** -0.0769***

(0.0325) (0.0235) (0.0238)

DLVS_crisis13 -0.00264** -0.00433*** -0.00356***

(0.00106) (0.00132) (0.000865)

DLVS_crisis14 -0.00438 0.00514 -0.00386

(0.0550) (0.122) (0.0548)

DLVS_crisis15 -0.00218 -0.00713*** -0.00313

(0.00845) (0.000643) (0.00759)

DLVS_crisis16 -0.152*** -0.134* -0.138**

(0.0301) (0.0787) (0.0556)

DLVS_crisis17 -0.00209 9.53e-05 7.50e-05

(0.00373) (0.00713) (0.00684)

DLVS_crisis19 -0.0470*** -0.0156 -0.0226**

(0.00771) (0.0144) (0.0111)

DLVS_crisis20 -0.0152 -0.0142*** -0.0141***

(0.00965) (0.00239) (0.00240)

DLVS_crisis21 -0.0454 -0.620** -0.369*

(0.356) (0.274) (0.222)

DLVS_crisis22 -4.70e-05 -5.43e-05*** -5.09e-05***

(2.91e-05) (1.85e-05) (1.57e-05)

DLVS_crisis23 -0.0308 -0.0576 -0.0518

(0.0792) (0.0552) (0.0468)

DLVS_crisis24 -4.96e-06* -1.13e-05*** -9.43e-06***

(2.80e-06) (1.97e-06) (1.65e-06)

DLVS_crisis25 -4.82e-05** -4.80e-05*** -4.74e-05***

72 | P a g e

(2.32e-05) (1.08e-05) (1.01e-05)

DLVS_crisis26 -4.81e-05* -4.62e-05 -4.67e-05*

(2.90e-05) (3.49e-05) (2.60e-05)

DLVS_crisis27 -0.264*** -4.59e-06 -4.74e-06

(0.0263) (2.86e-05) (2.89e-05)

DLVS_crisis28 -4.77e-05* -5.44e-05*** -5.27e-05***

(2.82e-05) (1.60e-05) (1.40e-05)

DLVS_crisis30 -0.253** -0.184*** -0.190***

(0.113) (0.0359) (0.0348)

DLVS_crisis32 -0.0971** -0.114*** -0.108***

(0.0474) (0.0375) (0.0328)

DLVS_crisis11 -1.019 -1.029

(0.917) (0.911)

DLVS_crisis12 0.691 1.092

(1.058) (1.029)

DLVS_crisis29 -0.198*** -0.191***

(0.0382) (0.0414)

DLVS_crisis31 -2.252*** -2.195***

(0.178) (0.150)

Constant -0.324* -0.172 -0.243*

(0.174) (0.163) (0.129)

Observations 6,428 16,870 23,298

R-squared 0.092 0.068 0.068

Number of cmiecompanycode 351 1,118 1,469

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Appendix D

The following flow chart, prepared by Inomata, Meng and Uchida (2010) illustrates how the vertical specialization channel generates a

multiplier effect and propagates shocks across industries and countries. As a result of the global financial crisis in US, there is an initial

shock to the automobile industry which translates into reduction in demand in automobile and other industries in US and other countries

due to complex production networks.

Automobiles

Car Chassis

Glass Products

Tyres and Tubes

Engines

Car Parts

Cold-finished Steel

Steel shar Slit

Paint

Composite Rubber

Carbon Black

Silk and Rayon

Imp. Raw Rubber

Wholesale

Steel

Road freight

transport

Sea shipping

Dye

Financing

Chemical Fibers

Wholesale

Rough Steel

Electricity

Coal Products

Petroleum

Products

Intermediate

Chemical Products

Electricity

Wood Pulp

Wholesale

Composite plastics

Initial Shock

1st Round Impacts

2nd Round Impacts

3rd Round Impacts

4th Round Impacts