A partnership with academia

Building knowledge for trade and development

Vi Digital Library - Text Preview

Are You Experienced? Survival And Recovery Of Trade Relations After Banking Crises

Working paper by Beverelli, Cosimo, Kukenova, Madina, Rocha, Nadia, 2011

Download original document (English)

We examine the impact of banking crises on the duration of trade relations. We also investigate the effect of product-level characteristics, such as the size of exports and exporting experience, and of sector-level financial dependence variables, on the time to recover after a banking crisis. Using highly disaggregated US import data from 157 countries between 1996 and 2009, we first provide evidence that banking crises negatively affect the survival of trade relations. On average, the occurrence of a banking crisis decreases the rate of survival of trade relations by 13 percent. Moreover, we find that both the size of exports and exporting experience matter for recovery of trade relations after banking crises. Sectoral financial dependence has an experience-specific effect. Relations with more experience recover faster in financially dependent sectors. There is instead no clear evidence indicating effects of size heterogeneity, neither in financially dependent sectors nor in non-financially dependent ones. The results are robust and consistent across alternative econometric models.

Staff Working Paper ERSD-2011-03 Date: 10 March 2011










World Trade Organization
Economic Research and Statistics Division
















ARE YOU EXPERIENCED?


SURVIVAL AND RECOVERY OF TRADE RELATIONS AFTER BANKING CRISES













Cosimo Beverelli


WTO


Madina Kukenova


University of Lausanne


Nadia Rocha


WTO




Manuscript date: March 2011



















Disclaimer: This is a working paper, and hence it represents research in progress. This paper
represents the opinions of the authors, and is the product of professional research. It is not meant to


represent the position or opinions of the WTO or its Members, nor the official position of any staff


members. Any errors are the fault of the author. Copies of working papers can be requested from the


divisional secretariat by writing to: Economic Research and Statistics Division, World Trade


Organization, Rue de Lausanne 154, CH 1211 Geneva 21, Switzerland. Please request papers by


number and title.













ARE YOU EXPERIENCED?


SURVIVAL AND RECOVERY OF TRADE RELATIONS AFTER BANKING CRISES






Cosimo Beverelli

, Madina Kukenova



and Nadia Rocha










Abstract


We examine the impact of banking crises on the duration of trade relations. We also investigate the


effect of product-level characteristics, such as the size of exports and exporting experience, and of


sector-level financial dependence variables, on the time to recover after a banking crisis. Using highly


disaggregated US import data from 157 countries between 1996 and 2009, we first provide evidence


that banking crises negatively affect the survival of trade relations. On average, the occurrence of a


banking crisis decreases the rate of survival of trade relations by 13 percent. Moreover, we find that


both the size of exports and exporting experience matter for recovery of trade relations after banking


crises. Sectoral financial dependence has an experience-specific effect. Relations with more


experience recover faster in financially dependent sectors. There is instead no clear evidence


indicating effects of size heterogeneity, neither in financially dependent sectors nor in non-financially


dependent ones. The results are robust and consistent across alternative econometric models.


Keywords: banking crises, financial dependence, export experience, duration models.


JEL classification: G01, C41 and F14.












Economic Research Division, World Trade Organization. Rue de Lausanne 154, 1211 Geneva,


Switzerland. E-mail: cosimo.beverelli@wto.org; nadia.rocha@wto.org.

Department of Economics, Université de Lausanne. E-mail: madina.kukenova@unil.ch.




We would like to thank seminar participants at the WTO Workshop and the European Trade Study Group


(ETSG) 2010 conference. We are also grateful to Andrei Levchenko, Logan Lewis and Linda Tesar for


providing us data on trade credit dependence measures.







1


I. Introduction




Survival matters in international trade. Besedes and Prusa (2006a) show that there is a remarkable


amount of entry and exit in the US import market. More than half of export relationships last only one


year, and 80 percent of them last less than five years. They argue that the inability to maintain export


relationships is a reason behind the lack of export growth at country level. Similarly, Brenton, Pierola


and von Uexkull (2009) show that poorly performing developing countries, despite having similar


rates of introduction of new trade flows, experience much lower rates of survival of trade relations


than stronger performing countries. These low survival rates undermine the expansion of export


flows. Nitsch (2008) finds that the stability of aggregate trade patterns may mask considerable


turnover at product level, with a large number of suppliers entering and exiting the market each year.




The literature on the duration of trade has considered the role of a wide range of product-, sector- and


country-level variables in determining survival rates.
1
To the best of our knowledge, there has been no


work that estimates how the occurrence of a banking crisis in an exporting country affects the survival


of its export relations. The first contribution of this paper is to provide evidence of the quantitative


impact of banking crises on trade survival. Using data on product-level exports to the US from 157


countries between 1996 and 2009, we first estimate a duration model à la Besedes and Prusa (2004) to


study how trade relations are affected by a banking crisis (survival analysis).
2
Banking crises are


found to have a negative and significant impact on the survival of export relations.




International trade has been rapidly recovering following a 12.2 percent fall in 2009, the biggest fall


in 70 years. The WTO forecasts a 13.5 percent rise in 2010 compared to the previous year.
3
In


addition, there is evidence that when recovery occurs it is fast. Descriptive statistics from the sample


of product-level export data to the US between 1996 and 2009 indicate that most of the relations that


recover after a banking crisis do so within two years (see Table 1).
4




Since recovery is well under way, it is as important as timely to draw lessons from past crises on the


factors that affect the probability of resuming trade relations that have been interrupted by the crisis.


Based on the previous evidence, the objective of the second part of this paper is to investigate which



1
See, among others, Besedes and Prusa (2006b), Besedes (2007), Brenton, Saborowski and von


Uexkull (2009), Fugazza and Molina (2009), and Volpe-Martincus and Carballo (2009) for a study using firm-


level export data.
2
The reason to have the US as destination country is that the original trade data we use (from Global


Trade Atlas) contains information at the 10 digit level of disaggregation only for trade flows in and out the US.
3
The World Bank's forecast is 15.7 percent, the OECD's is 12.3 percent.


4
We have extrapolated all relations that were interrupted at the occurrence of a banking crisis in the


exporting country.




2


trade relations recover first and what distinguishes fast-recovering relations. In particular, we ask


whether the product-level characteristics, as opposed to characteristics of the sector they belong to,


matter more for recovery.




We use a duration model to examine how the size of export values and exporting experience have an


effect on the time to recover after a banking crisis (recovery analysis). We also investigate if certain


sector characteristics such as long- and short-term financial dependence have an impact on the time to


recover. The empirical literature on growth and finance
5
shows that firms operating in external


financially dependent industries rely heavily on bank loans, not only to support their capital


investment, but also to promote their export expansion. Therefore, we expect that after a banking


crisis the financial sector will decrease, if not stop, its credit operations. As a consequence, the


recovery of export relations will be slower for products belonging to financially dependent industries.




Finally, we are interested in whether the impact of sectoral characteristics has a product-specific


effect. Thus, we investigate if the interaction between sector level characteristics and either size or


experience has an impact on the time to recover. Throughout the paper, we follow some empirical


studies
6
in assuming that an individual trade flow represents a firm.




The novelty of our results is that while both size and experience matter for recovery of trade relations


after banking crises, experience is more significant in financially dependent sectors. This outcome is


consistent with some new empirical literature
7
showing that not all exporting firms are the same.


Firms that export for longer periods exhibit certain characteristics that differentiate them from


sporadic exporters. In this context, it is intuitive that, independently of size, products that have been


exported for longer time will have fewer difficulties in recovering after a negative shock such as a


banking crisis.




In addition, the fact that more experienced products enter first in financially dependent sectors is in


line with empirical studies on banks’ lending behavior such as Petersen and Rajan (1994). They show


that firm’s age, as well as the duration of its relationship with the financing bank, are an important


determinant of the cost of financing. In light of this evidence, it is not surprising that after a banking


crisis, when banks are faced with a lack of liquidity requiring them to restrict credit, only well


established and better known firms are likely to get access to credit from the banks, being able to


cover some of the cost of producing and exporting.



5
See papers such as Petersen and Rajan (1994), Berger and Urdell (1993), and Nilsen (2002).


6
See for instance Manova (2008).


7
This literature includes papers such as Alvarez (2007), Alvarez at al. (2009), Borgersen (2006).




3


The paper proceeds as follows. The next section discusses some stylized facts around which the


survival and the recovery analyses are organized. Section III discusses the data. Section IV presents


the estimation strategy. Section V describes the main results on the effects of banking crises on


survival. Section VI examines the recovery of trade relations and provides some robustness checks.


Section VII concludes and offers some policy implications.




II. Some stylized facts




Survival matters in trade, especially after banking crises




We have collected annual product-level exports, disaggregated at the HS-10 level, from 157 countries


to the US between 1996 and 2009. The dataset provides information on the duration of each export


relation, making it amenable to survival analysis. In this dataset, on average, 23 percent of trade


relations were interrupted by the occurrence of a banking crisis between 1996 and 2008 (see Table


2).
8
The stylized fact that banking crises negatively affect the survival of trade is confirmed by the


Kaplan-Meier survival estimates, shown in Figure 1. Trade relations hit by a banking crisis (BC


dummy equal to one) exhibit lower unconditional survival rates than trade relations not hit by a


banking crisis (BC dummy equal to zero).




Experience matters for recovery




Experience, defined as the number of years a relation was active before a banking crisis,


unambiguously helps firms to recover faster. As shown in Table 3, 58 percent of products with 18


years of experience re-entered the export markets after one year, while only 17 percent of products


with 1 year of experience re-entered after one year. Size, measured as value of exports at the spell that


ended with the crisis, does not matter as much as experience for recovery. A way to visualize this is


with Kaplan-Meier survival estimates (see respectively, Figure 2 and Figure 3). In Figure 2, products


have been ranked in three different groups (quantiles) according to their experience. It is evident that


the relations belonging to the third quantile (more experienced ones) recover faster than those



8
Table 2 lists all systemic banking crises that occurred between 1996 and 2008 in countries exporting to


the United States. It uses the definition of banking crisis from Leuven and Valencia (2008), and it includes crisis


episodes for 2008 for France, Germany, Luxemburg, Ireland, Belgium, Iceland, Netherlands and the UK. We


consider as "destroyed" all relations that were active the year before the crisis, and turned inactive on the year of


the crisis. Alternatively, in order to take into account the fact that the effects of banking crises can materialize


with a lag, we have counted as "destroyed" all relations that turned inactive on the year of the crisis or the year


after. The results are available upon request.




4


belonging to the second and first quantiles.
9
In Figure 3, products have been ranked in quantiles


according to the size of the relation. This figure shows only limited evidence that bigger relations


recover faster.
10






Sectoral financial dependence has an experience-specific effect




Statistical analysis also shows that measures of sectoral financial dependence have an experience-


specific effect. Consider the unconditional survival estimates graphed in Figure 4. The figure


represents the pattern of each quantile of experience, interacted with financial dependence. Within the


group of experienced relations (products belonging to the third quantile), the survival function is


lower in trade credit dependent sectors (TCD equal to one) than in non-trade credit dependent sectors


(TCD equal to zero). This implies that in the former type of sectors more experienced trade relations


re-enter faster than in the latter type of sectors.




This pattern is reversed for less experienced relations (products belonging to the first and second


quantiles). For these products, the survival function is higher in trade credit dependent sectors (TCD


equal to one) than in non-trade credit dependent sectors (TCD equal to zero). Therefore, in the former


type of sectors less experienced trade relations re-enter faster than in the latter type of sectors.




In contrast, as it can be observed in Figure 5, there is no clear descriptive evidence on the effect of


size heterogeneity on survival, neither in financially dependent sectors nor in non-financially


dependent ones.
11






III. Data




The analysis is based on US manufacturing imports from 157 countries, at the HS 10 digit level of


disaggregation.
12


Data from 1995 to 2009 are from the Global trade Atlas (GTA). To complement this


dataset, we also use HS 10 import flows between 1991 and1995, collected by the Centre of


International Data at UC Davis. As in Brenton, Saborowski and von Uexkull (2009), we use US



9
In the graph, higher survival rates imply longer periods of inactivity, therefore a lower probability of


re-entry.
10


From the graph it might seem that the variable size is not constant across time. We control for this in


the regressions by stratifying the sample (see below).
11


From the graph it might seem that the variable size is not constant across time. We control for this in


the regressions by stratifying the sample (see below).
12


The reason to have the US as destination country is that the original trade data uses (from Global


Trade Information Services) contains information at the 10 digit level of disaggregation only for trade flows in


and out the US.




5


import data as a mirror from exports to the US, because they tend to be more reliable than export data,


especially when the exporter is a developing country.


To calculate survival and recovery rates of trade relations, a key step involves converting the annual


data into periods (or spells) of service of each trade relationship. We define a trade relation as the


exports of a certain product k from country i. An export relation can have one or more non-


overlapping spells, depending on the number of times (if any) it is interrupted during the whole period


of analysis. If, for instance a country exports a product between 1997 and 2001, and again between


2005 and 2008, the export relation has two spells.




The survival analysis uses a database with a total of 921,960 spells. The database contains information


on the dates of exit and re-entry of products into the US export market, and on various product-,


sector- and country-specific characteristics for each spell.




The recovery analysis uses a database that contains information only on export relations that exit


during a banking crisis. The sample consists of 13,055 spells.




Data on banking crises are from Leaven and Valencia (2008). Their database contains information on


124 systemic banking crises over the period 1970- 2007. According to the authors' definition, a


systemic banking crisis is one in which "a country’s corporate and financial sectors experience a


large number of defaults and financial institutions and corporations face great difficulties repaying


contracts on time". We complement the dataset by adding information from the 2008 financial crisis.


Specifically, based on the criteria of bank default and government intervention, a banking crisis is


recorded in 2008 for the following countries: the UK, Germany, Belgium, Luxembourg, Iceland, the


Netherlands, and Ireland.
13


Overall, there are 23 systemic banking crises during the time span of our


sample.




Finally, we use variables for both short- and long-term financial dependence. For short-term financial


dependence, we use trade credit dependence (TCD) from Levchenko (2009), computed at NAICS four


digit level (the original measure is from Fisman and Love, 2003).The indicator of long-term financial


dependence is the external financial dependence (EFD) variable from Rajan and Zingales (1998). It is


computed at ISIC three-digit industry level. As shown in Table 4, the correlation between TCD and


EFD is very high, and equal to 0.7.



13


Each of these countries has experienced the failure of a significant Banking Institution. Northern


Rock for the UK, Fortis Bank in the case of the Benelux countries. Icesafe in Iceland. Hypo in Germany and


Bank of Ireland in Ireland.




6


IV. Empirical strategy




The empirical analysis is divided in two main parts. First, we estimate a duration model à la Besedes


and Prusa (2004) to study how trade relations are affected in times of crisis. Second, always using a


duration model but this time only for those products that exited with a banking crisis, we analyze how


certain exporter and sectoral characteristics have an effect on the time to recover after banking crises.


In this case, duration refers to the time during which a trade relation has been inactive, therefore the


shorter the duration, the faster the recovery.




The regression we estimate is a stratified Cox proportional hazard model of the form




)'exp()(),,( 0  xthxth cc 




where x denotes a series of explanatory variables and  is the vector of coefficients to be estimated.


The baseline hazard )(0 thc represents how the hazard function changes with time and is different for


each strata of the sample.




To allow for a different hazard function for each country and sector, the estimations are stratified by


exporting country and three-digit ISIC industry.
14


In addition, we cluster standard errors by sector


(ISIC three digit) and country, to allow for intra-industry and country correlation in the error terms.




The main explanatory variable for the survival analysis is a banking crisis dichotomous variable,


taking value one in those years in which a country has experienced a banking crisis. In addition, we


include various measures of size, recorded at the beginning of each spell.
15


All regressions also


include a common set of control variables, measured at the beginning of each spell.
16


. First, the total


number of countries exporting a certain product to the US and the total value of product exports


respectively serve as control for the extensive and the intensive margin of competition. Second, to


control for the fact that the banking crisis variable might be capturing a deterioration of demand in the


destination country, we introduce a product-specific measure of the growth of US imports. Finally, we



14


When sector-specific variables are included in the regression, we do not stratify the sample by sector.
15


Details are provided in Section V below. The variable experience cannot be included in a Cox


regression because it is highly correlated with the duration of a spell, which is the conditioning variable in


duration models. Alternative estimation methods that can accommodate both size and experience are discussed


in Section V.
16


Measurement at the beginning of the spell is in line with the literature. See for instance Besedes and


Prusa (2006b).




7


include fixed effects for the year in which a trade relation is started to control for the fact that certain


years might be more favorable than others when it comes to export decisions.




With respect to recovery, we test whether size and experience of export relations at the time of exit


have an impact on the number of years it takes to re-enter the export markets. In addition, to analyze


whether products that exit the export market during a crisis recover at different speeds depending on


the sector they belong to, we also include an interaction term between long- and short-term financial


dependence indicators and product characteristics. In this case, too, the total number of countries


exporting a certain product to the US and the total value of product exports are included as controls.


As it is not possible to compute these control variables for the sub-sample of products where exports


have never resumed, we calculate their averages between the first year after the banking crisis and


either the year of re-entry or the last year of the sample, depending on whether exports have resumed


or not.




There are some econometric issues related with the empirical methodology that are common to all


duration models. First and most important, in the survival analysis we do not want to artificially


record a banking crisis that occurred during a trade relation as happening at the beginning or at the


end of its duration. We solve this problem by splitting each export relationship at the time of the


banking crisis, and assuming that the crisis lasts for one year.


Second, for some export relations it might be impossible to accurately observe their beginning and/or


their ending. We do not know if an export relation that is first observed at time t actually started at


time s<t (left censoring). Likewise, we do not know whether an export relation that is last observed at


time T was interrupted at T or continued after it (right censoring). To control for left censoring we


construct variables using trade data from 1991 until 2009. However, we exclude from estimations the


spells that started in the initial five years of the dataset (1991-1995). The Cox model, which we use


for estimations, controls for right censoring.




Third, there are products that exit more than once (multiple spells). The general approach of the


literature to control for multiple spells in duration models is to include in the regressions a multiple


spell dummy equal to one if the relation has at least one exit during the sample period. However, to


control for the fact that multiple spells are time-varying within a relation, a different definition of


multiple spell is considered, with the construction of a variable equal to the number of spells before


time t. This approach, the authors believe, is theoretically more correct than the standard approach of




8


the literature because it does not consider a relation to be characterized by multiple spells until its first


observed reentry, but only after it.
17






Due to the high level of disaggregation of the dataset, throughout all the analysis we make the


assumption that there is a representative firm for each trade relation. This allows referring to


"experience" and "size" as two measures of heterogeneity among exporters. As seen in Table 4, the


sample correlation between size of exports volumes and export experience is very low (0.07). This


means that these two variables are not the same and hence they capture different characteristics of


exporters.




V. Banking crises and the duration of trade relations




The Cox proportional hazard model results on the effect of a banking crisis on the survival of export


relations are presented in Table 5. All estimates in the table are expressed in terms of hazard ratios. A


hazard ratio greater than one indicates an increase in hazard and shorter duration, therefore meaning


that an export relation survives less.


As it can be seen in the first column, and in line with the stylized fact presented above, a banking


crisis raises the hazard ratio, thereby increasing the probability that a trade relation is interrupted by


11 percent. The total number of suppliers and the total exports of the products, in turn, have a positive


impact on the probability of survival. This result is consistent with the literature of trade survival, in


which both the extensive and the intensive margin of competition have a positive effect on survival.


The coefficient on demand shock also has the expected sign, since positive demand shocks reduce the


probability of exit. However, this coefficient is not significant in most of the regressions.


In column (2) we include a measure of the size of an export relation, recorded at the beginning of the


spell. Size increases survival. However, its inclusion does not affect the coefficient on BC – on the


contrary, it rises marginally.


An absolute measure of size might not be ideal if one deals with very heterogeneous products. As


mentioned by Besedes (2007), "for some products $15,000 may be big and for others $1 million could


be small". In column (3) we include a relative measure of size, "market share", to deal with the issue


of product heterogeneity. It is defined as the ratio of total product exports over its average world


exports. The results indicate that the effect of this alternative measure of size on survival is still


positive and significant.



17


Alternatively, a multiple spell dummy equal to one if the relation that is interrupted at time t has at


least one exit at time s<t has been included. Results are qualitatively the same.




9


Neither in the Leaven and Valencia (2008) dataset used for systemic banking crises, nor in other


similar datasets, is there systematic information on the final date of banking crises. Therefore, in the


previous regressions we have assumed a common duration of one year for all banking crises. In


column (4) we replicate column (2) regression, considering that the effect of a banking crisis lasts two


years instead of one. The banking crisis coefficient is still positive and significant, though it is


reduced by more than half. One intuition for this result is that for a significant number of products,


exports were resumed one year after a banking crisis (see Table 3). Hence assuming that banking


crises last for two years would make us consider that those products never exited the export markets.


Finally, in the last column of Table 5 we exclude banking crises that occurred in 2008 from the


sample. In this case, too, the coefficient on banking crisis remains positive and significant, but it


becomes smaller. This could be due to the fact that the 2008 crisis hit disproportionately larger


exporters from developed countries, which have a comparative advantage in financially dependent


sectors. Therefore, the banking crisis coefficient could be picking up some of the effect of financial


dependence.
18




Table 6 presents the results of alternative estimation techniques, a linear probability model (LPM,


columns (1) and (2)) and a Probit model (columns (3) and (4)), where the dependent variable is a


dichotomous variable equal to one if an export relation is interrupted.
19


The main reason to adopt


these techniques is that they can accommodate for the inclusion of product-level experience (the total


number of years that a relation was active) in the set of explanatory variables. As it has been argued in


footnote 18 above, this is not possible in the Cox model.


Consistently with the results obtained in the Cox regressions, the occurrence of a banking crisis


increases the probability of exit Moreover, the size of export flows and exporting experience decrease


the probability of exit of export relations. This last result is in line with studies such as Brenton,


Saborowski and von Uexkull (2009), which show that product specific characteristics such as initial


size of an export flow and experience positively affect survival.


VI. Are you experienced?


In this section we shift focus to the effect of the size of export values and export experience before a


banking crisis on the time to recover after the crisis. We also study the impact of sectoral financial


dependence variables, and whether it depends on product characteristics.



18


This issue is addressed in the recovery regressions of Section VI.
19


Notice that in the Probit regression we use robust standard errors, because clustering is not


computationally feasible.




10


In Table 7, all estimates are expressed in terms of hazard ratios. In this case, however, a spell begins


with the product exiting at the occurrence of a banking crisis, and it ends when exporting starts again.


An hazard ratio greater than one, which indicates a shorter duration, means that the export relation


recovers faster. The size of exports at exit and export experience at exit are included in the regressions


of column (1) and (2), respectively. Both experience and size are found to increase the probability of


recovery. Specifically, one extra year of export experience increases the probability of recovery by


almost 6 percent. A one percent increase in exports size increases the probability of recovery by about


3 percent. Results are qualitatively and quantitatively similar even after using a relative measure of


size such as product share of exports (column (3) of Table 7).


In column (4), both the size of exports and export experience, are contemporaneously included in the


regression. Also in this case both variables decrease the time to recover. However, while the


magnitude of the experience coefficient remains unchanged, the coefficient of size becomes slightly


smaller.


In column (5), regressions are performed ranking products in 3 different quantiles of export size and


experience. Statistical tests of equality of coefficients indicate that the effect on the time to recover


does not vary across different export size groups. The effect of export experience, in turn, is


statistically different across quantiles. Specifically, while for products with two to five years of export


experience the probability of recovery is about 35 percent higher than for products with low export


experience belonging to the reference quantile, for products with more than five years of experience


the increase in the probability of recovery with respect to products in the reference quantile almost


doubles. This last result implies that export experience might have a non-linear effect on recovery.


With respect to the control variables, both the number of suppliers and the total exports of a certain


product have a positive effect on the probability of recovery.
20


These results indicate the presence of a


pro-competitive effect both at the extensive and the intensive margin of competition. Finally, in all


regressions, the higher the frequency a product has exited and entered the export market, the lower is


the probability of recovery. A possible intuition for this result is as follows: relationships with


multiple spells before the crisis might be low-productivity ones, with productivity levels close to the


cut-off that makes exporting profitable. These trade flows will therefore tend to re-enter later after a


banking crisis.


The regressions in Table 8 include a set of variables capturing sectoral financial dependence. Columns


(1) and (2) include a dichotomous variable proxying long- term external finance dependence (EFD).


Columns (3) and (4) include a dichotomous variable equal to one for trade credit dependent sectors



20


Recall that these variables are computed as averages after banking crises.




11


(TCD). In neither of the regressions do the indicators of financial dependence have a significant


effect. The coefficients of size and experience remain roughly the same, and heterogeneity across


groups of the latter still persists.


To examine whether financial dependence variables have an experience-specific effect, we re-


estimate the Cox proportional model separately for each of the three quantiles of export experience.


Results are presented in Table 9.
21


Reading across columns, it is possible to observe that the


coefficients of long-term financial dependence and trade credit dependence change across quantiles,


implying that they have an experience-specific effect. While for products with least experience


financial dependence has a negative impact on the time to recovery, products with more experience


enter faster in financially dependent sectors. The sign and the magnitude of the other explanatory


variables do not vary significantly across different groups of export experience.
22




An alternative approach to investigate whether financial dependence has a product-specific effect is


presented in Table 10. Always using a Cox proportional hazard model, we interact where long-term


financial dependence and trade credit dependence with exporting experience and size of exports. The


results confirm the fact that in both long- and short-term financially dependent sectors, products with


more experience recover faster than products with less experience (see columns (1) and (2)). More


specifically, in financial dependent sectors, more experienced products enter first. In contrast, there is


no clear evidence indicating the effects of size heterogeneity on the time to recover, neither in


financially dependent sectors nor in non-financially dependent ones (see columns (3) and (4)).


Due to the fact that the interpretation of interaction terms in not an easy task in Cox proportional


models, similar regressions have been estimated using a linear regression model (LPM) and a Tobit


model. Results are in Table 11). For both methodologies, the dependent variable is the number of


years it takes an export relation to re-enter the foreign market after a banking crisis. In addition, the


Tobit model takes into account that some export relations are right censored and hence have not been


resumed yet.
23


As in the Cox proportional model, experience and size reduce the time to recover..


Moreover, as it can be seen in columns (1)-(4), more experienced exporters enter first in financially


dependent sectors. Once again, the interaction between exports size and long- or short-term financial


dependence is not significant (see columns (5)-(8)).



21


In columns (1) and (4), the variable total number of previous spells is dropped from the regressions


due to the very high collinearity with the experience variable.
22


A similar Cox regression has been estimated for different groups of export size. Results, available


under request, show that neither financial dependence variables nor other control variables have a size-specific


effect.
23


The maximum value of time to recover in the sample is 12 years. We have hence assumed that the


products that have not re-entered the export market yet will enter after 15 years. We have also experimented


with re-entry after 20 and 30 years, respectively. Results do not change.




12




From these results we can conclude that, independently of size, products with more years of


experience might have an advantage in obtaining external finance, thereby recovering faster after a


banking crisis.
24






VII. Conclusion




Measuring the effects of a banking crisis and understanding the patterns of the recovery of trade


relations is a very important question for policy makers when reacting to financial shocks. This study,


based on a duration model using export data disaggregated at product level, has presented a set of


results with relevant policy implications.




First, we have shown that a banking crisis negatively affects export trade relations. In addition, we


have found that bigger and more experienced exporters are less adversely hit by a banking crisis than


smaller and less experienced exporters, which may not be productive enough to overcome a sharp


drop in foreign demand. This result is consistent with some empirical studies that have found a


positive effect of initial size of an export flow, and of exporting experience, on trade survival.


Second, while on average size and experience have a significant impact on the recovery after banking


crises, only the latter matters for the recovery of products belonging to industries that highly depend


on external finance. Consistently with the idea that within-sector heterogeneity matters, we find that


long- and short-term sectoral financial dependence has an experience-specific effect. In particular,


more experienced exporters re-enter faster in financially dependent sectors. This result has very


important policy implications. If the objective of a policy is to help trade recover faster after financial


disruption, relatively un-experienced exporters should be targeted to restart foreign operations,


independently of their size.



24


In order to sharpen these conclusions, we are planning to perform the same analysis using firm-level


data.




13


Bibliography




Alvarez, Roberto (2007), "Explaining Export Success: Firm Characteristics and Spillover Effects,"


World Development 35(3): 377-393.




Alvarez, R., Hasan Faruq and Ricardo Lopez (2009). "New Products in Export Markets: Learning


from Experience and Learning from Others", mimeo.




Berger, Allen N. and Gregory F. Udell (1993), "Lines of credit, collateral, and relationship lending in


small firm finance," Finance and Economics Discussion Series 93-9, Board of Governors of the


Federal Reserve System (U.S.).


Bernard, Andrew B. and J. Bradford Jensen (1999), "Exporting and Productivity," NBER Working


Paper 7135.


Besedes, Tibor (2007), "A Search Cost Perspective on Duration of Trade," Departmental Working


Paper 2006-12, Department of Economics, Louisiana State University.


Besedes, Tibor and Thomas J. Prusa (2004), "Surviving the U.S. Import Market: The Role of Product


Differentiation," NBER Working Paper 10319.


Besedes, Tibor and Thomas J. Prusa (2006a), "Ins, outs, and the duration of trade," Canadian Journal


of Economics 39(1): 266-295.




Besedes, Tibor and Thomas J. Prusa (2006b), "Product differentiation and duration of US import


trade," Journal of International Economics 70: 339-358.




Borgersen, T-A (2006). "When Experience Matter: The Export Performance of Developing Countries'


SMEs", Journal of Sustainable Development in Africa, 8 (1): 106-118.




Brenton, Paul , Christian Saborowski and Erik von Uexkull (2009), "What explains the low survival


rate of developing country export flows ?," Policy Research Working Paper Series 4951, The World


Bank.


Brenton, P., Pierola, M. D. and von Uexküll, E. (2009), “The Life and death of trade flows:
understanding the survival rates of developing-country exporters”, in R. Newfarmer, W. Shaw and P.
Walkenhorst (eds.), Breaking Into New Markets: Emerging Lessons for Export Diversification,


Washington D.C.: The World Bank.


Fisman, Raymond and Inessa Love (2003), "Financial Dependence and Growth Revisited," NBER


Working Paper 9582.


Fugazza, Marco and Ana Cristina Molina (2009), "The determinants of trade survival," Working


Paper 05-2009, Graduate Institute, Geneva.




Laeven, Luc and Fabian Valencia (2008), "Systemic Banking Crises: A New Database," IMF


Working Paper 08/224, International Monetary Fund.


Levchenko, Andrei A., Logan Lewis and Linda L. Tesar (2009), "The Collapse of International Trade


During the 2008-2009 Crisis: In Search of the Smoking Gun," Working Paper 592, Research Seminar


in International Economics, University of Michigan.


Manova, Kalina (2008), "Credit Constraints, Heterogeneous Firms and International Trade," NBER


Working Paper14531.




14


Nilsen, Jeffrey H, (2002), "Trade Credit and the Bank Lending Channel," Journal of Money, Credit


and Banking 34(1): 226-53.


Organization of Economic Cooperation and Development (2009), OECD Economic Outlook No. 85,


June.


Petersen, Mitchell A. and Raghuram G. Rajan (1994), "The Benefits of Lending Relationships:


Evidence from Small Business Data," Journal of Finance 49(1): 3-37.




Rajan, Raghuram G. and Luigi Zingales (1998), "Financial Dependence and Growth," American


Economic Review 88(3): 559-86.




Therneau, Terry M. and Patricia M. Grambsch (2000), "Modeling Survival Data: Extending the Cox


Model," New York: Springer.




Volpe-Martincus, Christian and Jerónimo Carballo (2009), "Survival of New Exporters in Developing


Countries: Does it Matter How They Diversify?," IDB Working Paper 140, Inter-American


Development Bank.




15


Figures and Tables


Figure 1: Banking Crisis and survival of trade relations


0


50


25


75


100


K
a


p
la


n
-M


e
ie


r
s


u
rv


iv
a


l
e


s
ti


m
a


te
s


,
p


e
rc


e
n


t


0 5 10 15 20


time, years


BC = 0 BC = 1




Note: BC = banking crisis


Figure 2: Experience and recovery of trade relations


0


25


75


50


100


K
a


p
la


n
-M


e
ie


r
s


u
rv


iv
a


l
e


s
ti


m
a


te
s


,
p


e
rc


e
n


t


0 5 10 15
time, years


q1: 1 yr. of experience q2: 2-4 yrs. of experience q3: 5-18 yrs. of experience




Note: q = quantile. In the graph, higher survival rates imply longer periods of


inactivity, therefore a lower probability of reentry.




16




Figure 3: Size and recovery of trade relations


50


0


75


100


25


K
a


p
la


n
-M


e
ie


r
s


u
rv


iv
a


l
e


s
ti


m
a


te
s


,
p


e
rc


e
n


t


0 5 10 15
time, years


exports at exit q1 exports at exit q2 exports at exit q3




Note: From the graph, it might seem that the variable size is not constant across time. This


is controlled for in the regressions by stratifying the sample (see below).






Figure 4: Experience, trade credit dependence and recovery of trade relations


0


50


100


75


25


K
a


p
la


n
-M


e
ie


r
s


u
rv


iv
a


l
e


s
ti


m
a


te
s


,
p


e
rc


e
n


t


0 5 10 15
time, years


experience q1=1 & EFD=0 experience q1=1 & EFD=1


experience q2=1 & EFD=0 experience q2=1 & EFD=1


experience q3=1 & EFD=0 experience q3=1 & EFD=1




Note: EFD external finance dependence. q = quantile.




17


Figure 5: Size, trade credit dependence and recovery of trade relations


0


50


100


75


25


K
a


p
la


n
-M


e
ie


r
s


u
rv


iv
a


l
e


s
ti


m
a


te
s


,
p


e
rc


e
n


t


0 5 10 15
time, years


size at exit q1=1 & EFD=0 size at exit q1=1 & EFD=1


size at exit q2=1 & EFD=0 size at exit q2=1 & EFD=1


size at exit q3=1 & EFD=0 size at exit q3=1 & EFD=1




Note: EFD external finance dependence. q = quantile.




Table 1: Recovery Time after Banking Crises, 1996–2009


Recovery


time (yrs.)


No. of


products


% of


products



1 3,640 49.48


2 1,193 16.22


3 695 9.45


4 444 6.04


5 387 5.26


6 278 3.78


7 220 2.99


8 199 2.7


9 132 1.79


10 90 1.22


11 57 0.77


12 22 0.3


Total 7,357 100





18


Table 2: Survival of Trade Relations after Banking Crises, 1996–2008


Country
Year of


crisis a
Tot relations (no.)


Relations


destroyed (no.) b


Relations


destroyed (%)


Argentina 2001 2534 636 25


Belgium 2008 6596 1450 22


Bulgaria 1996 726 246 34


China 1998 9382 949 10


Colombia 1998 2239 573 26


Czech Republic 1996 2382 610 26


Denmark 2008 11116 1128 10


Dominican Republic 2003 2210 494 22


Ecuador 1998 1059 321 30


Great Britain 2008 10585 1350 13


Honduras 1998 573 180 31


Indonesia 1997 3619 649 18


Ireland 2008 3280 833 25


Iceland 2008 610 235 39


Jamaica 1996 786 245 31


Japan 1997 10014 985 10


Republic of Korea 1997 7013 1118 16


Malaysia 1997 3420 721 21


Nicaragua 2000 386 96 25


The Netherlands 2008 6856 1295 19


Philippines 1997 3334 704 21


Russian Federation 1998 2415 667 28


Slovakia 1998 807 263 33


Thailand 1997 4632 870 19


Turkey 2000 3323 693 21


Ukraine 1998 752 235 31


Uruguay 2002 715 171 24


Viet Nam 1997 825 186 23


Yemen 1996 23 5 22




19


Table 3: Recovery time by Experience level


Experience


(yrs.)


Total no. of


products


Product


reentry after


1yr. (no.)


Product reentry


after 1 yr. (%)


1 3,939 654 17


2 1,978 512 26


3 1,371 377 27


4 986 307 31


5 795 237 30


6 707 245 35


7 554 165 30


8 385 119 31


9 364 104 29


10 368 125 34


11 351 139 40


12 350 126 36


13 263 100 38


14 221 96 43


15 196 94 48


16 172 77 45


17 159 75 47


18 151 88 58












Table 4: Correlation of Explanatory Variables




Experience


at exit


Exports at


exit


N. of


supplier


reentry


Tot


exports


reentry


N.


previous


spells


Ext. Fin.


Dep.


Trade


Credit


Dep.


Experience at exit 1



Exports at exit 0.07 1


N. of supplier reentry -0.09 -0.09 1



Tot exports reentry -0.04 0.19 0.62 1



N. previous spells 0.63 -0.03 -0.11 -0.09 1



Ext. Fin. Dep. -0.01 0.09 0.04 0.21 -0.01 1



Trade Credit Dep. 0.01 0.12 0.06 0.25 -0.001 0.67 1











20


Table 5: The effect of banking crises on trade relations survival


(Cox proportional hazard estimates)




BC


(1 yr length)


BC


(1 yr length)


BC


(1 yr length)


BC


(2 yrs length)


BC excl. 2008


(1 yr length)


Variables (1) (2) (3) (4) (5)



Banking crisis 1.112*** 1.133*** 1.135*** 1.052*** 1.031***




[0.013] [0.013] [0.014] [0.013] [0.011]



Size at spellbegin




0.906***




0.906*** 0.903***




[0.001]




[0.001] [0.001]



Market share at spellbegin




0.902***






[0.001]





Number of suppliers at spellbegin 0.990*** 0.988*** 0.991*** 0.988*** 0.988***




[0.000] [0.000] [0.000] [0.000] [0.000]



Total product exports at spellbegin 0.966*** 0.992*** 0.903*** 0.992*** 0.992***




[0.001] [0.001] [0.002] [0.001] [0.001]



Demand shock 0.991** 0.993** 0.993** 0.994* 0.994*




[0.003] [0.003] [0.003] [0.003] [0.003]



N previous spells 0.951*** 0.945*** 0.946*** 0.944*** 0.945***




[0.002] [0.002] [0.002] [0.002] [0.002]



Observations 921960 921960 921960 889208 908854


Note: Standard errors (in brackets) clustered by country and by ISIC three digit industry.


*** p<0.01, ** p<0.05, * p<0.1. Sample stratified by country and ISIC three digit


industry. Other controls: year FE.





21


Table 6: The effect of banking crises on product exit


LPM PROBIT


Dependent variable: Exit (1) (2) (3) (4)





Banking crisis 0.017*** 0.034*** 0.018*** 0.037***




[0.003] [0.003] [0.001] [0.001]



Size t-1




-0.342***




-0.364***




[0.007]




[0.001]



Experience t-1




-0.022***




-0.019***




[0.000]




[0.000]



N suppliers t-1 -0.003*** -0.003*** -0.004*** -0.004***




[0.000] [0.000] [0.000] [0.000]



Total product exports t-1 -0.212*** -0.112*** -0.165*** -0.060***




[0.007] [0.010] [0.002] [0.002]



Demand shock -0.005*** -0.003*** -0.009*** -0.009***




[0.001] [0.001] [0.000] [0.000]



Multiple spell dummy 0.146*** 0.145*** 0.159*** 0.165***




[0.001] [0.002] [0.000] [0.000]



Observations 3,873,513 3,083,889 3,873,513 3,083,889


R-squared 0.114 0.299


Note: In columns (1) and (2) standard errors (in brackets) clustered by country and


by ISIC three digit industry. In columns (3) and (4) robust standard errors in


brackets. *** p<0.01, ** p<0.05, * p<0.1. Other controls: ISIC three digit, country


and year FE.




22


Table 7: Time to recover and exporter characteristics (Cox proportional hazard estimates)


Variables
(1) (2) (3) (4) (5)





Years of experience at exit 1.058***




1.057***






[0.005]




[0.005]




Exports at exit




1.027***




1.020***






[0.008]




[0.007]





Product Market share at exit




1.032***






[0.007]





Years of experience at exit q2




1.352***




[0.049]



Years of experience at exit q3




1.692***




[0.084]



Exports at exit q2




1.043




[0.029]



Exports at exit q3




1.082**




[0.036]



N of suppliers at re-entry 1.032*** 1.034*** 1.034*** 1.032*** 1.032***




[0.002] [0.002] [0.002] [0.002] [0.002]



Total product exports at re-entry 1.020** 1.019** 1.046*** 1.017** 1.019**




[0.008] [0.008] [0.010] [0.008] [0.008]



N previous spells 1.032** 1.090*** 1.088*** 1.034** 0.982




[0.013] [0.015] [0.015] [0.013] [0.015]



Observations 13055 13055 12974 13055 13055


Note: Standard errors (in brackets) clustered by country and by ISIC three digit industry. *** p<0.01, **


p<0.05, * p<0.1. Sample stratified by country and ISIC three digit industry.















23


Table 8: Time to recover and financial dependence


(Cox proportional hazard estimates)




Variables (1) (2) (3) (4)





Years of experience at exit 1.064***




1.069***






[0.004]




[0.005]





Exports at exit 1.023***




1.020***






[0.007]




[0.007]





Years of experience at exit q2




1.416***




1.445***




[0.050]




[0.053]



Years of experience at exit q3




1.827***




1.893***




[0.089]




[0.102]



Exports at exit q2




1.046*




1.047*




[0.028]




[0.029]



Exports at exit q3




1.102***




1.098***




[0.035]




[0.035]



EFD 0.986 0.997






[0.034] [0.035]





TCD




0.953 0.968




[0.033] [0.034]



Observations 12,928 12,928 11,628 11,628


Note: Standard errors (in brackets) clustered by country and by ISIC three digit


industry. *** p<0.01, ** p<0.05, * p<0.1. Sample stratified by country. Other


controls: total product exports at re-entry, total number of suppliers at re-entry,


demand shock, number of previous spells.















24


Table 9: Time to recover and financial dependence


(Cox proportional hazard estimates with group varying characteristics)



Experience q1 Experience q2 Experience q3 Experience q1 Experience q2 Experience q3


Variables
(1) (2) (3) (4) (5) (6)



Exports at exit 1.014 1.024** 1.022** 1.002 1.023** 1.024**




[0.013] [0.011] [0.010] [0.013] [0.011] [0.010]



EFD 0.893* 1.017 1.066*






[0.055] [0.049] [0.041]





TCD




0.844*** 0.987 1.053




[0.053] [0.050] [0.043]



Observations 3,744 4,257 4,927 3,253 3,861 4,514


Note: EFD = external financial dependence. TCD = trade credit dependence. Standard errors (in


brackets) clustered by country and by ISIC three digit industry. *** p<0.01, ** p<0.05, * p<0.1.


Sample stratified by country. Other controls: total product exports at re-entry, total number of suppliers


at re-entry, demand shock, number of previous spells.







































25


Table 10: Financial dependence, exporter characteristics and recovery


(Cox proportional hazard estimates)


FD=EFD FD=TCD FD=EFD FD=TCD


Variables (1) (2) (3) (4)



Years of experience at exit q2 1.322*** 1.354***






[0.060] [0.060]




Years of experience at exit q3 1.635*** 1.693***






[0.085] [0.093]




FD 0.894** 0.857*** 0.98 0.977




[0.051] [0.050] [0.044] [0.044]



Years of experience at exit q2 x FD 1.129** 1.135**






[0.065] [0.067]




Years of experience at exit q3 x FD 1.216*** 1.256***






[0.082] [0.091]




Exports at exit q2




1.017 1.036




[0.041] [0.040]


Exports at exit q3




1.105** 1.130***




[0.046] [0.047]



Exports at exit q2 x FD




1.034 0.994




[0.054] [0.054]



Exports at exit q3 x FD




0.993 0.931




[0.059] [0.055]




Observations 12,928 11,628 12,928 11,628


Note: FD= Financial dependence. EFD = external financial dependence. TCD = trade credit


dependence. Standard errors (in brackets) clustered by country and by ISIC three digit industry.


*** p<0.01, ** p<0.05, * p<0.1. Sample stratified by country. Other controls: total product exports


at re-entry, total number of suppliers at re-entry, demand shock, number of previous spells, exports


at exit (in columns (1) and (2)) and years of experience at exit (in columns (3) and (4)).




26


Table 11: Financial dependence, exporter characteristics and recovery


(OLS and Tobit estimates)


OLS Tobit OLS Tobit OLS Tobit OLS Tobit


Variables (1) (2) (3) (4) (5) (6) (7) (8)



Years of experience at exit -0.208*** -0.443*** -0.230*** -0.488*** -0.266*** -0.542*** -0.289*** -0.590***



[0.024] [0.044] [0.025] [0.046] [0.020] [0.036] [0.021] [0.039]


Exports at exit -0.146*** -0.223*** -0.133*** -0.196*** -0.178*** -0.269*** -0.175*** -0.255***




[0.031] [0.056] [0.032] [0.057] [0.046] [0.081] [0.043] [0.076]


EFD 0.311 0.626




-0.736 -0.958






[0.442] [0.732]


[0.748] [1.284]


TCD




0.765*** 1.177**




-0.478 -0.681



[0.281] [0.481]



[0.622] [1.088]


Years of experience at exit x EFD -0.104*** -0.178***






[0.030] [0.050]


Years of experience at exit x TCD




-0.117*** -0.197***





[0.030] [0.049]


Exports at exit x EFD




0.053 0.072





[0.062] [0.107]


Exports at exit x TCD




0.073 0.099



[0.062] [0.109]


Observations 12,928 12,928 11,628 11,628 12,928 12,928 11,628 11,628


R-squared 0.283 0.284 0.282 0.282


Note: OLS = ordinary least squares. EFD = external financial dependence. TCD = trade credit dependence. Standard errors


(in brackets) clustered by country and by industry.*** p<0.01, ** p<0.05, * p<0.1. Other controls: Industry and country FE,


total product exports at re-entry, total number of suppliers at re-entry, demand shock, number of previous spells




Login