The Private Credit Insurance Effect on Trade

DN

B W

or

kiN

g P

aP

er

DNB Working Paper

The Private Credit Insurance Effect

on Trade

Koen van der Veer

No. 264 / October 2010

Working Paper No. 264/2010

October 2010

De Nederlandsche Bank NV P.O. Box 98 1000 AB AMSTERDAM The Netherlands

The Private Credit Insurance Effect on Trade Koen van der Veer * * Views expressed are those of the author and do not necessarily reflect official positions of De Nederlandsche Bank.

The Private Credit Insurance Effect on Trade∗

Koen J.M. van der Veer†

October, 2010

Abstract

International trade relies on trade finance (credit or insurance) by financial insti-tutions. Data limitations, however, have made it diffi cult to quantify the impact ofchanges in the supply of trade finance on trade. This paper is the first to establisha causal link between the supply of private credit insurance and exports. I overcomeendogeneity issues by using a unique bilateral data set which covers the activities from1992 to 2006 of one of the world’s leading private credit insurers. This database en-ables me to use the insurer’s claim ratio — a primary determinant of the supply ofcredit insurance —as an instrument for insured exports. Subsequently, applying themethod of instrumental variables and a variety of trade models, I consistently find apositive and statistically significant effect of private credit insurance on exports. Theestimates are economically relevant and suggest that, depending on the decline in thesupply of private credit insurance during the 2008-09 international trade collapse, thereduction in private insurance exposure explains about 5 to 9 percent of the drop inworld exports and 10 to 20 percent of the drop in European exports.

JEL codes: F10, F14, G01, G20, G22.

Keywords: trade finance, private credit insurance, international trade, trade credit

∗I am especially grateful to Andrew Rose for his advice and numerous discussions throughout this project. I thankMartin Admiraal and Henk van Kerkhoff for their help with collecting the data. For helpful suggestions on an earlierdraft, I thank Peter Egger and Christoph Moser. I also thank Marco Hoeberichts, Eelke de Jong, Pierre Lafourcade,Iman van Lelyveld and seminar participants at the ECB workshop on Trade and Competitiveness, the 16th InternationalConference on Panel Data and De Nederlandsche Bank for comments. I thank the private credit insurer for kindlyproviding data and their staff for fruitful discussions. I bear full responsibility for any remaining errors. The viewsexpressed in this paper are those of the author and do not necessarily represent those of the Dutch central bank.†De Nederlandsche Bank, Economics and Research Division, P.O. Box 98, 1000 AB, Amsterdam, The Netherlands,

[email protected]

1 Motivation

Financial institutions play an important role in facilitating international trade. An estimated 80 to

90 percent of world trade relies on some form of credit, insurance or guarantee, issued by a bank

or other financial institution (Auboin, 2007). However, direct evidence on the link between trade

finance and trade is still missing, because detailed data on trade finance is hard to come by.1 As a

result, it is unclear to what extent changes in the supply of trade finance have an effect on trade.

A number of authors have studied the trade finance channel, but use inadequate proxies for trade

credit provided by banks or financial institutions. These studies examine whether the availability of

trade credit or dollar-denominated short-term credit affect exports (Ronci, 2005; Berman and Martin,

2009; Iacovone and Zavacka, 2009; Levchenko, Lewis, and Tesar, 2010). Short-term credit, however,

can be used for reasons other than trade financing and does not cover all trade transactions. Also, the

standard proxies for trade credit usage by firms —accounts receivable and payable —measure credit

extended between firms instead of a financial institution and a firm, and include credit for domestic

purchases. More fundamentally, the link between trade credit provided by a financial institution and

trade credit usage by firms is ambiguous, since institutional finance and trade credit can be substi-

tutes (Petersen and Rajan, 1997). Amiti and Weinstein (2009) overcome these measurement and

endogeneity issues by relating firms’export performance to the health of the institutions providing

trade finance. Their results show convincingly that financial shocks are transmitted from banks to

exporters, but provide only indirect evidence on the trade finance channel.

This study exploits a unique bilateral data set on the worldwide activities of a leading private

credit insurer to examine the effect of private credit insurance on exports. It is the first to focus

on private export credit insurance and to establish a causal link between the private supply of a

trade finance product and exports. Importantly, the data enable me to deal with reverse causality

and other potential endogeneity issues by using the private insurer’s claim ratio — received claims

over premium income —as an instrument for insured exports. Past and current claim ratios are a

primary determinant of the insurer’s supply of credit insurance. Subsequently, using the method of

instrumental variables, I find an average multiplier of private credit insurance of 2.3, implying that

every euro of insured exports generates 2.3 euro of total exports.

Trade credit insurance is a useful tool for firms to insure against the risk of non-payment. Es-

pecially in cross-border transactions where it is more costly to monitor risks and more diffi cult to

1

enforce payment, the provision of credit insurance could foster trade. The trade-promoting effect

of credit insurance is described in a formal model by Funatsu (1986), who shows that insurance

cover of trade credits will result in a higher output level as compared to the case without insurance.

Empirically, the evidence of a trade-promoting effect of credit insurance is limited to the case of

public guarantees. Two important contributions are Egger and Url (2006) and Moser, Nestmann and

Wedow (2008) who find that Austrian and German public export credit guarantees stimulate trade

in the long run.

For a number of reasons, however, the private credit insurance effect on trade can be expected to

differ from the impact of public guarantees. First, changes in the exposure of private credit insurance

are likely to affect exports immediately, whereas the short run impact of public guarantees is found

to be very small (see Egger and Url, 2006; and Moser, Nestmann and Wedow, 2008). This difference

follows from the varying maturities of private versus public credit insurance. Private credit insurers

usually cover short-term credits with a tenor of 60 to 120 days and medium- or long-term covers only

play a minor role (Swiss Re, 2006). Public guarantees, on the other hand, generally cover export

projects with a duration between two and five years. So the actual shipment of the good usually

follows a few years after the public provision of insurance cover. This difference in maturities is

especially clear in Europe, where offi cial export credit agencies have been restricted from providing

guarantees covering export risks to OECD core members with a maturity of less than two years.2

Second, relative changes in the supply of private credit insurance are likely to have a bigger

impact on total exports than changes in the supply of public guarantees. Obviously, this goes for

countries where the value of privately insured exports exceeds the value of public guarantees. For

example, private insurers covered an estimated 16.7 percent of Dutch exports in 2006, compared to

0.9 percent of exports insured by the Dutch State.3 Aside from a bigger impact due to the size of the

private credit insurance market, the greater effect on overall exports also stems from the potential

influence of private credit insurers on the export decision of non-insured firms. That is, private

credit insurers provide valuable information on the creditworthiness of firms via changes in their

policy stance vis-à-vis individual firms, their publicly available country ratings, and their detailed

firm-specific information services.

I test with a gravity model whether private credit insurance stimulates trade and consistently

find a positive and statistically significant effect. The bilateral data set includes data on privately

2

insured exports from 25 countries to 183 destination countries covering the period from 1992 to

2006. The data cover the insurance provided and claims and premiums received by one of the "Big

Three" private credit insurers.4 Importantly, the data enable me to identify in a first instance the

link between the claims received by the insurer and the supply of insurance, thereby overcoming

endogeneity issues. Also, I show results using more than one strategy to deal with "multilateral

resistance" to trade; the average barrier of two countries to trade with all their partners.

Finally, I shed some light on the role of private credit insurance during the world trade collapse in

2008-09. Anecdotal evidence suggests that private credit insurers reduced their exposure substantially

in reaction to the increased risk environment. I extrapolate the estimates of the insurance supply

elasticity of exports and calculate the contribution of the decline in the supply of private credit

insurance to the world trade collapse. Conditional on the actual decline in the supply of private

credit insurance, the estimates suggest that the reduction in private insurance exposure between the

third quarter of 2008 and 2009 explains about 5 to 9 percent of the drop in world exports and 10 to

20 percent of the drop in European exports.

In what follows, I describe the rise of private credit insurance since the early 1990s (Section 2),

briefly review the literature (Section 3), and examine empirically the private credit insurance effect

on trade (Section 4). In Section 5, I test the sensitivity of the benchmark results to sample changes,

endogeneity issues, the availability of public export credit guarantees, and possible misspecification

related to measuring "multilateral resistance" to trade. Section 6 examines the role of private credit

insurance in the 2008-09 world trade collapse. Section 7 concludes.

2 The Rise of Private Credit Insurance

Since the early 1990s, private trade credit insurance has registered strong growth and now dominates

the short term market.5 In 1999, more than 95 percent of the short term business from Europe was

underwritten by the private sector (Swiss Re, 2006). Private credit insurance is especially popular

in Europe, where an estimated 73 percent of worldwide premiums of US$6.6 billion were collected

in 2004 (Swiss Re, 2006). Likewise, around three-quarters of insured exports in 2005 were destined

to Europe (Swiss Re, 2006). Since the late 1990s, the American and Asian markets are growing in

importance, however, and Europe’s share of the world insurance market is declining.

In 2010, three private credit insurers —the "Big Three" —covered 87 percent of the world market:

3

Euler Hermes (36%), Atradius (31%) and Coface (20%). These private insurers usually cover short

term commercial and political risk. Commercial risk refers primarily to the risk of non-payment by

the importer due to default or insolvency, whereas political risk relates to non-payment as a result

of action by an importer’s government.6 More recently, the "Big Three" have also started to cover

longer maturities, but offi cial export credit agencies are still the primary players in this market.

The rise of private credit insurance followed a number of actions by OECD governments since the

debt crisis in the 1980s. The international debt crisis of the 1980s and 1990s had a profound impact

on how countries viewed their export credit agencies (Stephens, 1999). The crisis caused considerable

claims for offi cial export credit agencies that became a drain on government budgets. Offi cial export

credit agencies experienced a net cash flow deficit during the period from 1981 to 1995 (Wang et al.

2005). These losses led governments to rethink their role in the provision of export finance and their

competition and overlap with private sector insurers.

At the national level, OECD governments started to privatize their short-term activities. The

privatization trend began by the decision of the United Kingdom in 1991 to sell the short-term

business of its export credit agency (Stephens, 1999). The United States government followed suit in

1992, and Coface (one of the "Big Three" private insurers) of France was privatized in 1994. Instead

of sales or transfers removing the export credit agencies’short-term business to private hands, part of

the privatization has taken place more silently (Wang et al. 2005). For example, in the Netherlands,

Atradius (formerly NCM), acting as an agent of the government, has insured an increasing amount

of business on its own accounts.

At the international level, the European Union defined the concept of "marketable risks" to clarify

what type of business should be left to private insurers. As a result, since 1998, offi cial export credit

agencies have been restricted from providing guarantees covering export risks to OECD core members

with a maturity of less than two years. Public guarantees, therefore, generally cover business with a

credit period longer than two years.

3 Literature Review

Existing empirical studies focus exclusively on public credit insurance and predict an economically

significant effect on exports in the long run. For example, Egger and Url (2006) estimate that a

one percent increase of Austrian guarantees generates a .44 percent long-term increase in Austrian

4

exports. In turn, they compute an average multiplier of 2.8, implying that every euro spend on public

guarantees creates 2.8 euro worth of exports. Moser, Nestmann and Wedow (2008) conduct a similar

analysis for German guarantees, but account for possible endogeneity issues and trade dynamics. In

turn, they find a somewhat lower multiplier of 1.7.

The theoretical explanation for this trade-promoting effect of export credit insurance dates back

to Funatsu (1986). He shows that a government can aggressively promote exports by offering a

public guarantee against default by the importer and demanding a "more-than-favourable" premium

rate. By using a credit guarantee, a firm can reduce its profit uncertainty in the foreign market

thereby increasing the firm’s optimal output level. The reduction in risk increases exports to markets

where the firm would not sell otherwise. Abraham and Dewit (2000) demonstrate that government

guarantees can stimulate firms to export even without subsidisation by charging a fair premium.

Thus, the rationale for the trade-promoting effect of export credit insurance seems to apply as well

to private insurers, who are unlikely to subsidize their clients.

These models, however, cannot explain the multiplier effect; the finding that the increase in

export value is greater than the value of insured exports. The rationale for such a multiplier effect

follows from the presence of sunk costs (Dixit, 1989). When firms face substantial entry costs, prior

export experience increases the probability of exporting by as much as 60 percentage points (Roberts

and Tybout, 1997). By providing insurance cover, public and private credit insurers reduce the

costs of insecurity and information related to the entry in foreign markets, allowing their clients

to learn about the creditworthiness of their trade partners (buyers). Subsequently, after repeated

transactions, the client may decide to export without costly export credit insurance.

A multiplier effect of private credit insurance could, however, also follow from the information on

foreign markets and firms that private insurers provide to non-insured firms.7 First, a private credit

insurer’s policy stance vis-à-vis a particular firm (buyer) or country could have a "signalling effect",

influencing the export decision of non-insured firms. Indeed, the news of an adjustment of a firm

rating or credit limit —the maximum exposure of the insurer in respect of a buyer —tends to travel

fast among all suppliers of the particular firm, potentially influencing all trade transactions of the

firm (Becue, 2009, p. 91). I.e. an upgrade generally improves the firm’s access to supplier credit,

and vice versa. Moreover, private credit insurers publish their country ratings. In principle, private

credit insurers let the country rating prevail over their sector- and firm-level rating when determining

5

premia and their maximum level of exposure. As a result, a firm can have a favourable rating but

still have a low or zero credit limit when the firm is situated in a weakly rated country (Becue, 2009,

p. 518). Finally, the "Big Three" insurers all offer some kind of information service, allowing firms to

get access to the insurers’detailed firm-level information on key customers, prospects or competitors

even without buying insurance cover.8 In short, private credit insurance could promote exports via

a reduction in export risk and information costs.

4 The Private Credit Insurance Effect on Exports

4.1 Specification and Data

To estimate the private credit insurance effect on exports, I rely on the standard "gravity" model of

bilateral trade. The gravity model explains trade between a pair of countries with the distance and

their economic "masses". I augment the basic specification with a number of conditioning variables

that might also affect bilateral trade, such as currency unions (Glick and Rose, 2002) and trade

agreements (Rose, 2004). I employ the following specification:

ln(Xijt) = β0 + β1 ln(Dij) + β2 ln(Popit) + β3 ln(Popjt) + β4 ln(GDPpcit) + β5 ln(GDPpcjt) + β6(CUijt)

+ β7(Langij) + β8(RTAijt) + β9(Borderij) + β10(Islandsij) + β11 ln(Areaij) + β12(ComColij)

+ β13(Colonyijt) + β14(EverColij) + β15(SameCtryijt) + γ1 ln(InsExpijt) + εijt.

where i denotes the exporting country, j denotes the importer, t denotes time, ln(.) denotes the

natural logarithm operator, and the variables are defined as:

• Xijt denotes real FOB exports from i to j, measured in euro,

• D is the distance between i and j,

• Pop is population,

• GDPpc is annual real GDP per capita,

• CU is a binary dummy variable which is unity if i and j use the same currency at time t,

• Lang is a binary variable which is unity if i and j have a common language,

• RTA is a binary variable which is unity if i and j have a regional trade agreement at t,

• Border is a binary variable which is unity if i and j share a land border,

• Islands is the number of island countries in the pair (0/1/2),

• Area is the log of the product of the areas of the countries,

6

• ComCol is a binary variable which is unity if i and j were both colonized by the same country,

• Colony is a binary variable which is unity if i colonizes j at time t (or vice versa),

• EverCol is a binary variable which is unity if i ever colonized j (or vice versa),

• SameCtry is a binary variable which is unity if i is part of the same country at time t (or vice

versa),

• InsExp denotes real privately insured exports from i to j, measured in euro,

• ε represents the omitted other influences on bilateral exports, assumed to be well behaved.

The parameter of interest is γ1. This represents the private credit insurance effect on exports

holding other export determinants constant through the gravity model. I estimate the equation with

OLS, using a robust covariance estimator (clustered by country-pair dyads) to handle heteroskedas-

ticity, adding year-specific fixed effects. I also adjust this specification in two important ways. First,

I add a comprehensive set of dyadic-specific fixed effects (i.e., a mutually exclusive and jointly ex-

haustive set of {βij} intercepts) to absorb any time invariant characteristics that are common to a

pair of countries. Second, I add comprehensive sets of exporter and importer fixed effects (i.e., sets

of {βi} and {βj}) to take account of any time invariant country-specific factors. I also show below

that the key results are insensitive to the use of other estimation strategies.

The sources of the bilateral data set are described in more detail in Appendix Table A1. This data

set includes annual observations between 1992 and 2006 (though with many missing observations) for

some 183 territories and localities (I refer to these as "countries" below). The countries themselves

are tabulated in Table A2. A correlation matrix for the variables used in the regression analysis is

presented in Table A3.

4.1.1 Data on Private Credit Insurance

The data on privately insured exports is the novel part of the data set and measures the real value of

exports insured (InsExpijt) by one of the "Big Three" private credit insurers.9 Summary statistics

for insured exports and the share of insured to total exports are presented in Table 1. Several features

regarding the data on insured exports are worth mentioning.

First, the data set on insured exports is constrained in two specific ways. That is, the number

of exporters is limited to 25 countries (all OECD members, except for Hong Kong) in which the

private insurer is active and data is available. Second, the number of observations per exporter

7

varies considerably (Table 1, Column 2). This reflects i) the entrance of the private insurer into new

markets (countries) over the years and ii) differences in the number of destination countries of each

exporter.

In addition, a special feature of the data is the variability in the private insurer’s share of insured

to total exports, varying from zero percent up to one hundred percent.10 Also, the mean (median)

share of insured exports for all exporters is 6.6 (1.5), but this figure varies by exporter from 0.1 in

Poland to 20.4 in Denmark.

Finally, the insurance data suffer from some measurement issues. Possible measurement errors

arise because i) clients of the insurer declare their turnover (value of insured exports) at different

frequencies; monthly, quarterly or yearly, ii) the amounts are allocated to periods when they were

invoiced by the insurer which does not always coincide with the period when the shipments took

place, and iii) data is migrated from systems used by acquired companies. Part of the measurement

errors is reduced by the yearly frequency of the data. More importantly, I apply the method of

instrumental variables which was pioneered to overcome measurement error problems in explanatory

variables (Angrist and Krueger, 2001; Hausman, 2001).11

4.2 Benchmark Results

The results of estimating the default specification are presented in Table 2. The model is estimated

with three different sets of fixed effects (none, dyad, and exporter/importer). Before I discuss the

private credit insurance effect on trade, I briefly discuss the other determinants of trade flows.

The model fits the data well. I obtain a high R-squared which is typical for gravity models and the

coeffi cient estimates are sensible. For instance, exports between a pair of countries fall with distance

and increase when countries share a currency, language, trade agreement or colonial heritage. In

addition, countries with a higher GDP per capita import more. The sign of the coeffi cient for the

importer’s population and exporter’s real GDP per capita changes, however, when including fixed

effects. Thus, larger and richer countries trade more (cross-sectional variation), but importers with

high population growth or exporters with high GDP per capita growth trade less, ceteris paribus.

Turning to the estimates of greatest interest; private credit insurance seems to stimulate exports.

The estimates are positive and statistically distinguishable from zero at all reasonable significance

levels. The size of the effect changes over the various specifications though, ranging between .02 and

.10. Hence, an increase of insured exports by 1 per cent causes additional exports by about .02 to

8

.10 per cent.

Two issues regarding these first regression results come up immediately. First, the benchmark

specification does not control for other insurers, either public or private. Basically, the regression

shows what happens to a country’s exports when the value of exports insured by "my" private insurer

increases, while the trading countries’GDP per capita, population size, transportation costs, and

trade costs related to various institutional settings, do not change. I show below that the results are

robust to the inclusion of public credit insurance in a sample with Dutch exports only, but I cannot

control for the activities of other private insurers.

Consequently, one could argue that an increase of coverage could simply reflect an increase in the

insurer’s share of the credit insurance market, making it unclear why this would stimulate trade. It is,

however, unlikely that substitution of credit insurance towards the private insurer drives the results.

For one thing, the credit insurance penetration rate (measured as premium income over GDP) has

risen steadily since 1990 in most of the large European markets, and credit insurance markets outside

Europe have grown even faster (Swiss Re, 2006). In addition, I show below that the results hold

for various reasonable changes in the sample. These robustness checks make it even more unlikely

that market share increases of the private insurer explain the findings. Another concern, however,

could be that I overestimate the credit insurance effect on trade because I include markets where

the private insurer is only a small player. In the sensitivity analysis below, I estimate the model for

various subsamples related to the share of insured to total exports covered by the private insurer.

Excluding the markets in which the private insurer is likely to have a small share does not reduce the

estimate of the private credit insurance effect on trade. On the contrary, the private credit insurance

effect increases with the share of insured to total exports.

A second, and related, issue is that the benchmark specification may suffer from an endogeneity

problem. Instead of some exogenous factor leading the insurer to extend more coverage (i.e. better

marketing of products, improvements in risk management practices reducing premia and maximum

exposure, lower risks in particular countries etc.), growth in trade could also explain growth in insured

exports. In the sensitivity analysis that follows below, I deal explicitly with the issue of endogeneity

by applying the method of instrumental variables.

9

5 Sensitivity Analysis

5.1 Robustness of the Private Credit Insurance Effect

I start the sensitivity analysis with a battery of robustness checks based on reasonable changes to

the sample. The purpose of this exercise is to show that the main results are not caused by some

small subset of the sample. The results are presented in Table 3. Each of the rows in the table

corresponds to a different sensitivity check, while the columns correspond to the benchmark model

estimated with three different sets of fixed effects, and also report the number of observations in each

subsample.

I check the sensitivity of the results by selectively dropping different sets of observations. Since I

am interested in exporter effects, I begin by dropping different sets of importer observations. First,

I drop all observations for importers that are industrial. I then successively delete observations

for developing countries from Latin America or the Caribbean, the Middle East, Asia, Africa, or for

(formerly) centrally managed economies.12 These robustness checks leave the basic results unchanged.

The same goes when dropping small importers (defined as a country with fewer than one million

people) or poor importers (those with real GDP per capita of less than 1000 euro per annum). I

then check the sensitivity of the results for some sets of exporter observations. Successively, I drop

non-European exporters and exporters not in the sample before 1995. Again, none of these changes

to the sample undermine the findings. Further, I check the sensitivity of the results with respect

to time. I separately drop the observations before and after 1999 respectively; the results remain

resilient. Finally, I successively delete observations in which the share of insured exports (by the

private insurer) to total exports is smaller than 1, 2, 5 and 10 percent. Again, all the results show

a positive and statistically significant effect of insured exports on total exports. The size of the

estimates, however, increases with the share of insured exports.

I conclude that the finding of a positive and statistically significant effect of private credit insur-

ance on trade is not due to some subset of the sample and is robust to reasonable changes in the

sample. Countries with a higher level of insured exports seem to have higher trade than others.

5.2 Endogeneity

Endogeneity is a primary concern in the estimates presented so far. For example, the total value

of exports could explain the value of insured exports instead of the other way around. Aside from

10

reverse causality, a third possibly omitted variable could arguably influence both total exports and

insured exports. The risk environment is a case in point. An increase in the risk perceived by

exporters might decrease total exports and increase demand for insurance. Note that the reverse

causality argument and the omitted variable bias story would result in opposite biases. Thus, the

direction of the bias in the benchmark results, if any, is unclear beforehand.13

I address the issue of endogeneity by using an instrumental variable for insured exports and apply

the two-stage least squares fixed effects estimator. The instrument is the private insurer’s claim

ratio (by exporter-importer-year), defined as claims over premium income. Claims are a primary

determinant of the supply of credit insurance.

The link between claims and insured exports runs through two channels. First, past and current

claim ratios are important ingredients in the formula to calculate premia (Becue, 2009).14 In case of

a shock, i.e. a credit crisis or sovereign default, claims increase. The claim ratio also increases, as the

private insurer can only raise the premia of new contracts.15 The bulk of the contracts are fixed for

one year during which the premium charge cannot change, and about 25 percent of all contracts have

a duration of 2 or 3 years. A rise in the claim ratio reduces the profit of the private insurer, inducing

an increase of the premium charged in new insurance contracts, thereby lowering the demand for

insurance and hence the total value of insured exports.

The second channel linking claims and insured exports is more direct, and involves credit insurers’

right to reduce or remove the credit limit of a specific buyer at any given time (Swiss Re, 2006; Jones,

2010).16 While premium rates on contracts are fixed, credit insurers can manage their exposure (or

"cover limit") to mitigate claims. This way, credit insurers can react to problems that can affect a

foreign buyer’s credit quality even before they worsen. Thus, the mere expectation of rising claims

can immediately affect insured exports via a reduction in the maximum exposure of credit insurers.

The results for the First Stage regression on insured exports are presented in Table 4, Column 1.

The F-statistic for the excluded instruments exceeds the rule of thumb value of 10 for all specifications,

implying that the claim ratios are suffi ciently correlated with insured exports. I find a negative and

statistically significant effect of the claim ratio on insured exports up to two years ahead. The point

estimates indicate that a 1 percent increase in the claim ratio reduces insured exports by .11 percent

in the same year, exports are .06 percent lower in the following year, and .04 percent lower the year

thereafter. Importantly, estimation of the benchmark model including the claim ratios shows that

11

the claim ratio has no direct effect on total exports.17 Thus, the effect of the claim ratio on total

exports runs only via the insured exports.

The results for the Second Stage regression on exports are presented in columns 2 to 5 of Table

4. I estimate four specifications to check the sensitivity of the estimates to alternative instruments.

The first uses the contemporaneous, first and second lag of the claim ratio as instruments for insured

exports. The second up to fourth estimations use either the contemporaneous, first or second lag of

the claim ratio as instrument. All instruments are valid according to various statistical tests; none

of the models is under- or weakly identified and the first specification with three instruments is not

overidentified.18 The point estimate for the instrumented insured exports ranges between .02 and

.09, a slightly smaller range compared to the benchmark results.

Since I use the log of the claim ratio I lose all observations with zero claims, about two-third of

the sample. To see whether the results are sensitive to the sample size, I estimate the system with

the claim ratio in levels. The results are presented in the final column of Table 4. Reassuringly, the

point estimate of .06 for the instrumented insured exports is equal to the estimate of the smaller

sample in column 2, but the coeffi cient is not significant. Notice, however, that the F-statistic for

the excluded instruments is only 6.89, well below the threshold value of 10. This implies that the

claim ratio in levels is less fit as an instrument for the log insured exports.

Next, I examine whether the instrumental variable estimates are sensitive to various subsamples

related to the share of insured to total exports. The results are presented in Table 5. I estimate the

system using the contemporaneous log claim ratio as instrument for the log insured exports. This

way, I maximize the number of observations and the F-statistic for the excluded instruments, while

being conservative on the size of the estimated private credit insurance effect (compare Columns

2 to 5 of Table 4). Again, I find that the size of the estimate for insured exports increases when

successively dropping observations with a share of insured to total exports below a certain threshold.

The size of the effect ranges between .02 for the full sample (Table 5, Column 3) and .35 for the

subsample of observations with a share of insured exports above 10 percent.

I have argued that a possible omitted variable bias could come from an insuffi cient account of a

country’s risk environment in the benchmark model. Indeed, previous studies provide evidence that

corruption and low institutional quality limit trade of high risk countries (Anderson and Marcouiller,

2002; Meon and Sekkat, 2004; Moser, Nestmann and Wedow, 2008). I examine whether omitted

12

variable bias is still a concern in the instrumental variable estimates by including a measure of

importer country risk (see also Moser, Nestmann and Wedow, 2008). This composite risk indicator

is a combined measure of a country’s political, economic and financial risk. The political risk rating,

reflecting corruption, bureaucracy quality, and law and order among other things, contributes 50

percent of the composite rating, while the financial and economic risk ratings each contribute 25

percent (see the International Country Risk Guide for details). The original indicator runs from

0 (very high risk) to 100 (very low risk). I inverted the index to make the interpretation of the

coeffi cient more intuitive. Hence, a higher risk indicator implies higher risk and I expect a negative

correlation with exports. The results are presented in Table 6. As expected, exports are lower to

countries with a higher risk environment. Importantly, the results are robust to the inclusion of

importer country risk. The size of the private credit insurance effect ranges between .02 and .44, a

somewhat larger range compared to previous results. Controlling for country risk seems to correct

a slight negative bias in the estimates of the private credit insurance effect in the subsamples of

observations with a relatively high share of insured exports.

Further, all the results presented are based on static specifications of the gravity model. These

models allow only for contemporaneous effects of regressors on trade. Past trade patterns could,

however, affect current trade flows in the presence of sunk costs (Dixit, 1989; Roberts and Tybout,

1997). Therefore, some authors propose to extend the standard gravity model with lags of trade

(Eichengreen and Irwin, 1998; Bun and Klaassen, 2002). I examine whether the results hold up in

a dynamic specification of the instrumental variables model by including one lag of the dependent

variable.19 The results presented in Table 7 confirm that past exports affect current exports. Yet,

the main result is robust to this inclusion of trade dynamics. Insured exports stimulate total exports.

The range of the private credit insurance effect is again somewhat larger, ranging from .01 to .51.

So far, I examined whether private credit insurance stimulates trade by relating the value of

insured exports to the value of total exports. A possible concern of this set up is that insured

exports are part of total exports. In order to see whether the findings are due to this particular

approach, I re-estimated the instrumental variables model for the various subsamples, using the share

of insured to total exports instead of the value of insured exports. The results are reported in Table

8. Reassuringly, this alternative approach confirms the positive effect of private credit insurance on

exports. The point estimate for the instrumented share of insured exports is statistically significant

13

for all subsamples, although only at the 10 percent level for the subsamples with a share of insured

exports above 5 or 10 percent. Again, the range of the private credit insurance effect is somewhat

larger compared to the results of the preferred approach, ranging from .02 to .53.

Thus far, I have been largely concerned with the statistical significance and robustness of the

results, but I have given no attention to the economic significance. The instrumental variable models

estimate a private credit insurance effect on exports ranging between .01 and .53 (Tables 5 to 7).

These coeffi cients can be interpreted as the elasticity of exports to insured exports, implying that a

1 per cent increase in insured exports leads to an increase in exports in the range of .01 to .53 per

cent, depending on the threshold taken for the minimum share of insured exports. I calculate a world

share of private short term insured exports to total exports of 6.1 percent in 2007.20 Subsequently,

I calculate the median share of insured exports for each of the subsamples and find the sample of

observations with a share of insured exports above 1 percent to resemble the world share of short

term insured exports. The average estimate of the elasticity for this subsample is .14 (see Tables 5 to

7). Likewise, for the Euro area countries, I calculate a share of private short term insured exports to

total exports of 12.3 percent, and subsequently find an elasticity of .29.21 Using the sample average

of insured and total exports, I compute an average multiplier of private credit insurance of 2.3.22

This result is important for a number of reasons. First, it shows that private credit insurance

stimulates exports. Indeed, the short run impact of private credit insurance is bigger than the long

run multiplier of public guarantees found in Moser, Nestmann and Wedow (2008).23 Thus, private

insurance allows firms to learn about the creditworthiness of their trading partners by doing repeated

business. The recurring trade transactions help a trading partner to build up its reputation, thereby

reducing the need for the exporter to use costly insurance. Also, the impressive size of the multiplier

suggests that private credit insurers provide information on foreign markets and firms influencing

the export decision of non-insured firms. Finally, it demonstrates that credit insurance stimulates

exports even without subsidisation, assuming private insurers charge a fair premium.

5.3 Zero (Insured) Trade

All the results above are generated from a linear-in-logs specification that converts observations

with zero exports to missing and these observations drop out of the sample, potentially introducing

selection bias. Moreover, as I also take the log of insured exports, sample selection relating to the

availability of private credit insurance could have biased the results as well. Indeed, the data set has

14

65672 observations of which 5082 correspond to zero exports and 49515 to zero insured exports.24

I examine the sensitivity of the results when correcting for sample selection with respect to zero

exports and insured exports. I follow Wooldridge (1995) by applying a sample selection model that

is suitable for panel data with fixed effects.25 Accordingly, for each year, I estimate a probit model

where the dependent variable equals one if exports are positive. I derive the linear prediction of

this model for each year and calculate the inverse Mills ratio (IMR). Finally, I include the IMR as a

regressor in the instrumental variable model estimated with dyadic fixed effects. I repeat this exercise

for sample selection due to zero insured exports and estimate a probit model where the dependent

variable equals one if insured exports are positive. Notice that the second model corrects for sample

selection with respect to zero exports and zero insured exports simultaneously. This is the case as

zero exports imply zero insured exports.

The results are presented in Tables 9 and 10. The estimates for the inverse Mills ratio indicate

that there is significant selection into the sample for some subsamples. However, there is little impact

on the point estimates of the parameters of interest. The size of the private credit insurance effect

ranges between .02 and .35, similar to the results in Table 5.

5.4 Changes on the Extensive Margin

I have examined the effect of private credit insurance on exports conditional on export flows being

positive. These results can be interpreted as an increase in exports on the intensive margin. In this

section, I attempt to examine if private credit insurance also affects the extensive margin of exports,

that is, does the availability of private credit insurance increase the likelihood of trade between a

pair of countries.

I follow the approach taken by Head, Mayer and Ries (2010) and estimate a linear probability

model (LPM) where the dependent variable equals one if exports are positive. Contrary to a probit

model, the LPM allows for estimation with dyadic fixed effects.26 To evaluate to effect of private credit

insurance on the extensive margin, I cannot use the value of insured exports by dyad (InsExpijt)

as independent variable since positive insured exports imply positive exports.27 Instead, I use the

yearly sum of insured exports to all destination countries (∑j InsExpijt) for every exporter. Thus, I

examine whether an increase in an exporting country’s sum of insured exports increases the likelihood

of positive export flows between the exporter and any destination country. Additionally, I allow the

effect of the sum of insured exports on the probability of positive exports flows to vary with the level

15

of importer country risk. Hereto, I group countries according to the five categories of composite risk

as identified by the International Country Risk Guide.

The results are reported in Table 11. I only find evidence of a positive effect of private credit

insurance on the extensive margin for the very high risk group of destination countries. Thus, these

results seem to suggest that private credit insurance stimulates exports primarily on the intensive

margin. Nevertheless, while this might be true at the country-level, it is not unlikely that the impact

of private credit insurance on the extensive margin is much more prominent at the firm-level.

5.5 Public Credit Insurance

Next, I briefly examine whether the positive and significant effect of private export credit insurance

holds up when accounting for the public alternative. A priori, there is not much reason to expect

the results to change, since private and public credit insurance are —due to the legal framework —

generally complements instead of substitutes.

I examine the public and private insurance effect simultaneously by adding a variable for public

insurance premium income to the benchmark model.28 Since I only have data on public guarantees in

the Netherlands, the sample reduces to Dutch exports in the period 1992-2002. Results are reported

in Table 12. I do not find public insurance to stimulate exports, at least not in the short run.

More importantly, the private insurance effect remains positive and statistically significant with a

coeffi cient of .17 (no fixed effects) or .05 (dyadic fixed effects), larger even than the benchmark results.

5.6 Methodological Issues

In this section, I test the sensitivity of the results to two different specifications of the gravity model.

Both specifications deal with the possibility of misspecification in the benchmark model because of

"monadic" problems. These refer to omitted factors that are specific to a single country but may

vary over time, such a those associated with "multilateral resistance" to trade [e.g. Anderson and

Van Wincoop (2003)].

5.6.1 Bonus Vetus OLS

Baier and Bergstrand (2009) propose a simple method for approximating trade-costs effects in the

presence of multilateral resistance. They call their approach "Bonus vetus (good old) OLS" since

they suggest a linear appromixation of the multilateral resistance terms, motivating a reduced-form

16

gravity equation that can be estimated using OLS. In practice, this involves estimation of each

country’s multilateral resistance to trade with other countries based on the simple or GDP-weighted

average of the indicators of trade barriers with all countries (such as distance, borders etc.); Baier

and Bergstrand (2009) provide more details.

I estimate the benchmark model twice; first including the transformed trade costs variables using

simple averages, and then using GDP weights. The results are presented in Table 13, Columns 1 to

4. I run both models for the full sample and the subsample of observations with a share of insured

exports above 10 percent. All estimations confirm the positive effect of insured exports on trade. The

point estimates are statistically significant, but with a range of .13 to .91 and .26 to .97 (respectively

simple and GDP-weighted average), much larger than any of the previous estimates.

5.6.2 Tetradic Estimates

Another way to deal with the presence of multilateral resistance is the "method of tetrads", advocated

by Head, Mayer and Ries (2010) (see also Rose and Spiegel, 2010). Under this method, consistent

estimates can be attained in the presence of multilateral resistance by comparing export observations

to exports for a pair of base countries for the same year (the technique is tetradic since one compares

trade flows for four countries). See Head, Mayer and Ries (2010) for more details.

The method presents two special issues. First, one needs to select a base exporter and importer

to do the tetradic calculations. To check the sensitivity of the results I use two different pairs of

countries: a) United Kingdom and The Netherlands; and b) France and Germany. Second, the

observations are likely to be dependent as the error terms in the tetrads appear repeatedly across

observations. I therefore use multi-way clustering to correct the standard errors, as proposed by

Head, Mayer and Ries (2010).

The results are presented in Table 13, Columns 5 to 8. Again, I obtain a positive and statistically

significant effect of private credit insurance on exports, regardless of the base exporter and importer

taken. The estimates range from .15 to .36 and .12 to .18 for the two respective sets of base countries.

6 Private Credit Insurance and the 2008-09 World Trade Collapse

The 2008-09 global financial crisis led to a dramatic drop in world trade. Real world exports fell

by 19 percent between the second quarter of 2008 and the second quarter of 2009. The failure of

standard gravity models to account for this unprecedented trade collapse has generated a search for

17

the "smoking gun" (Levchenko, Lewis and Tesar, 2010). In this section, I shed some light on the

issue by examining the role of private credit insurance.

Some observers suggested that a shrinking supply of trade finance contributed to the trade decline

(see Auboin, 2009; Dorsey, 2009; and OECD, 2009). But the lack of detailed data on trade-related

lending, insurance or guarantees issued by financial intermediaries precludes any strong conclusions

on the role of trade finance. Schmidt-Eisenlohr (2009) develops a theory of trade finance that explains

the co-existence of different trade finance products depending on enforcement and cost of financing.

Numerical experiments of his model show that limiting the choice between trade finance contracts

can reduce trade by up to 60 percent. Also, a few inventive studies do establish a link between a

shock to the financial sector and exports. For example, Amiti and Weinstein (2009) show that during

the Japanese financial crises in the 1990s, a firm’s export performance was related to the health of

the firm’s main bank. Their results suggest that trade finance accounted for about one-third of the

decline in Japanese exports. Chor and Manova (2010) find some evidence that countries with higher

interbank rates exported less to the United States during the recent crisis. Neither of these studies,

however, uses data on the actual supply of a trade finance product (credit, insurance or guarantee)

by a financial institution. Thus, they do not identify a direct link between trade finance and exports.

The results in this paper are evidence of a direct link between the supply of private credit insurance

and exports, but the data do not cover the 2008-09 global financial crisis. In order to get a sense of

the role of private credit insurance in the world trade collapse, I need to know if and by how much

private credit insurers reduced their supply of insurance during the crisis. As expected, anecdotal

evidence shows that private credit insurers reacted to the deteriorating economic environment at the

end of 2008 by reducing their exposure. For example, the 2008 annual report of Atradius —one of

the "Big Three" private insurers —shows that claims were rising fast in the second half of 2008 and

suggest that measures were taken to reduce exposure substantially:

"The net claims ratio for the second half of 2008 was 134.2% compared to 62.8% in the

first half of 2008. Dedicated and tailored measures in both buyer and policy underwriting

have been implemented in order to rebalance the risk-return ratio and to guide customers

through the economic cycle. These measures have included amongst others, price increases

to reflect the higher risk business environment and reductions of cover on buyers that are

deemed to represent imbalanced risks." 29

18

"The initial action was a review of our risk portfolio, protecting our customers from

the risk of default of buyers that carried unacceptable risks. This resulted in the reduction

or cancellation of those credit limits that showed imbalanced risks. These measures were

executed in a timeframe of two months and resulted in a substantial reduction of our

exposure." 30

In order to get a sense of the size of this "substantial" reduction of private insurance exposure

during the crisis, I use Berne Union data. As Figure 1 illustrates, world private short term export

credit insurance exposure declined by roughly 23 percent between the fourth quarter of 2008 and

third quarter of 2009. Public insurance exposure declined by less than 7 percent. Public export

credit agencies have generally increased their insurance supply during the crisis in order to mitigate

the impact of the trade finance crisis.31

The Berne Union figures give a first idea of the possible change in the supply of private credit

insurance, but there are two important caveats. First, the figures report insurance exposures at

quarter-end instead of actual insured exports during a quarter. Insurance exposures are available on

a quarterly basis, while only the yearly value of short term insured exports is reported by the Berne

Union. For the period considered, however, the change in insurance exposure seems a reasonable

approximation for the change in insured exports. Indeed, the Berne Union Yearbook 2010 reports

that short term insurance exposure and new business insured were both down by 13 percent in

2009. Second, demand factors are likely to have contributed to the reported decline in insurance

exposure, thus leading to an overestimation of the reduction when interpreting the Berne Union

figures strictly in terms of the supply of insurance. Since public export credit agencies are less likely

to have reduced their supply, one could use the difference between the decline in private versus public

insurance exposure (-16 percent) as a rough indication of the decline in the supply of private credit

insurance.

Either way, the actual decline in the supply of private credit insurance during the 2008-09 world

trade collapse is unknown. Therefore, I calculate the contribution to the world trade collapse of

various changes in the supply of private credit insurance, extrapolating the estimates of the insurance

supply elasticity of exports. The results are presented in Table 14. The first three rows estimate

the export decline of a 10, 15 or 20 percent reduction in the supply of private credit insurance for

various shares of privately insured to total exports. Recall from Section 5, that 6.1 percent of world

19

exports in 2007 were covered by private short term export credit insurance. Using this share, the

reduction in the supply of private insurance during the crisis can explain a decline of world exports

of 1.4 to 2.9 percent, or 5 to 9 percent of the total drop in world exports (Table 14, Column 1,

Rows 4 to 6). In the Euro area, where an estimated 12.3 percent of exports is covered by private

insurers, the reduction in the supply of private export credit insurance during the crisis can explain

a decline of exports of 2.9 to 5.8 percent, or 10 to 20 percent of the total drop in exports from the

Euro area countries (Table 14, Column 4, Rows 7 to 9). Thus, while macroeconomic factors played

an important role in the world trade collapse, these calculations suggest that the effect of private

credit insurance on exports identified in this paper can account for part of the world trade decline.

7 Conclusion

The main contribution of this paper is to estimate the private credit insurance effect on trade using

a unique data set on the insurance provided and claims received by one of the world’s largest credit

insurers. The matched insurance-claims data enable me to identify the link between claims and the

supply of export credit insurance, thus overcoming endogeneity issues. I find a short run average

multiplier of private credit insurance of 2.3, implying that every euro of insured exports generates

2.3 euro of total exports. This multiplier is impressive, especially considering that previous studies

find a long run multiplier of public guarantees of smaller size.

The paper is unique in its focus on the role of private export credit insurance and is the first to

establish a causal link between the supply of a trade finance product and exports. First, I provide

a number of arguments explaining why private export credit insurance is important to trade. In

particular, credit insurance stimulates exports to markets where firms would not sell otherwise,

allowing trade partners to build up reputation, thereby reducing the need for exporters to use costly

insurance. Moreover, private insurers are likely to influence the export decision of non-insured firms

via the "signalling effect" of their policy stance vis-à-vis individual firms, their publicly available

country ratings, and their detailed firm-specific information services.

Subsequently, I estimate an insurance supply elasticity of exports in the range of .01 to .53. The

elasticity increases with the share of insured to total exports. Accordingly, the estimates for the

relevant subsamples reveal an elasticity of .14 for world exports and .29 for exports from Europe.

These results suggest that conditional on the actual decline in the supply of private credit insurance,

20

the reduction in private insurance exposure during the 2008-09 world trade collapse explains about

5 to 9 percent of the drop in world exports and 10 to 20 percent of the drop in European exports.

21

Figure 1. World Exports of Goods and Short Term (ST) Export Credit Insurance Exposure, 2007Q12009Q3

Source: World exports in goods from the WorldTrade Monitor, CPB Netherlands Bureau for Economic Policy Analysis.Insured export credit exposure from the Berne Union, through the BISIMFOECDWB Joint External Debt Hub. The changein private and public short term exposure in the period from 2008Q4 to 2009Q3 is calculated assuming that the share of publicECAs increased from 25% to 30% (see ICC, 2010 p. 47) [25] at a constant rate. The figures are converted to Euro at thequarterultimodollar/euro exchange rate from the ECB. I use ultimos since the Berne Union secretariat also uses ultimos whenconverting the Euro values into US dollars. Short term exposure is comprised of short term commitments defined as totalamounts insured under all current policy limits for which premium has been paid or invoiced, including amounts overdue forpayment (arrears) until claims have been paid or rejected, and including uninsured percentages.

World exports World total shortterm export credit

insurance exposure

Private ST exposure

Public ST exposure

65

70

75

80

85

90

95

100

Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3

2007 2008 2009

Index 2008Q3=100, nominal values in Euro

22

Table 1: Summary Statistics for Insured Exports and Share of Insured to Total Exportsa

Insured Exports (millions) Share of Insured ExportsFirst yearin sample Obs. Mean Std.Dev. Min. Max. Mean Std.Dev. Min. Max.

All Exporters 17596 69 309 0 6220 6.6 13.6 0 100.0

By ExporterUnited Kingdom 1992 2713 131 428 0 6220 11.3 12.2 0 100.0The Netherlands 1994 1898 107 454 0 5760 8.4 11.6 0 100.0France 1992 1471 20 67 0 553 0.6 2.8 0 100.0Australia 1993 1387 16 61 0 793 10.7 20.6 0 100.0Germany 1994 1216 195 623 0 6030 2.5 6.5 0 100.0Belgium 1997 1041 70 261 0 2220 2.7 6.0 0 100.0Denmark 1999 995 74 255 0 2220 20.4 25.0 0 100.0United States 1997 897 41 156 0 1890 1.5 5.4 0 100.0Sweden 1998 824 73 260 0 3650 10.0 17.9 0 100.0Spain 1994 748 11 52 0 762 0.8 4.5 0 96.8Italy 1998 728 20 70 0 922 0.5 1.1 0 13.1Norway 1994 721 41 112 0 1150 9.1 17.9 0 100.0Mexico 1993 706 23 142 0 1990 3.0 6.7 0 100.0Ireland 1997 439 13 82 0 1410 1.5 5.6 0 83.2Luxembourg 1997 423 15 41 0 405 7.1 12.6 0 100.0Finland 1999 382 24 67 0 569 2.4 4.4 0 35.1Switzerland 2003 278 54 203 0 2010 4.1 10.5 0 14.9New Zealand 2004 241 8 28 0 270 6.9 14.7 0 100.0Austria 2003 171 23 64 0 473 1.3 3.2 0 28.0Czech Republic 2004 68 41 138 0 894 1.1 2.5 0 100.0Poland 2005 64 1 3 0 24 0.1 0.3 0 2.3Hungary 2005 59 3 6 0 24 0.4 1.4 0 10.6Greece 2004 58 20 29 0 108 11.7 25.2 0 100.0Slovak Republic 2004 49 9 27 0 135 0.6 0.8 0 4.1Hong Kong 2006 19 9 17 0 70 0.3 0.5 0 1.7aUnit of analysis: exporter-importer-year. Data on insured exports from one of the "Big Three" private

credit insurers.

23

Table 2: Effect of Private Credit Insurance on Exports in Gravity Model

Fixed Effects: None Dyadic Exporter, ImporterLog Insured Exports .10∗∗∗

(.01).02∗∗∗(.00)

.08∗∗∗(.01)

Log Distance −.97∗∗∗(.03)

−1.36∗∗∗(.05)

Log Exp Population .81∗∗∗(.02)

2.00∗∗∗(.58)

−1.31∗∗(.68)

Log Imp Population .84∗∗∗(.02)

−1.03∗∗∗(.23)

−1.39∗∗∗(.24)

Log Exp Real GDP p/c 1.05∗∗∗(.09)

−.96∗∗∗(.28)

−.38(.31)

Log Imp Real GDP p/c 1.13∗∗∗(.03)

.48∗∗∗(.08)

.42∗∗∗(.08)

Currency Union .18∗∗(.08)

.17∗∗∗(.04)

−.21∗∗∗(.07)

Common Language .45∗∗∗(.07)

.39∗∗∗(.06)

RTA −.03(.06)

.15∗∗∗(.04)

.13(.07)

Common Border .06(.10)

−.38∗∗∗(.10)

No. Islands .27∗∗∗(.06)

−9.30∗∗∗(3.09)

Log Product Area −.05∗∗∗(.01)

2.16∗∗∗(.72)

Common Colonizer 1.59∗∗∗(.17)

1.62∗∗∗(.46)

Currently Colony .48∗∗∗(.13)

−.03(.03)

−.24(.22)

Ever Colony .51∗∗∗(.10)

.74∗∗∗(.08)

Common Country 1.50∗∗∗(.11)

.78∗∗(.41)

R2 .85 .98 .93RMSE .97 .35 .68

Data set includes 14,389 bilateral annual observations covering 183 countries, 1992 - 2006. Robust standard

errors (clustered by country-pairs) in parentheses. Year effects included but not recorded. Significance:

***1%, **5%, *10%.

24

Table 3: Sensitivity Analysis of Private Credit Insurance Effect on Exports

Fixed Effects: None Dyadic Exporter, Importer ObservationsDrop Industrial Importers .09∗∗∗

(.01).03∗∗∗(.00)

.07∗∗∗(.01)

9914

Drop Latin America, Caribbean Importers .11∗∗∗(.01)

.02∗∗∗(.00)

.07∗∗∗(.01)

11869

Drop Middle Eastern Importers .10∗∗∗(.01)

.02∗∗∗(.00)

.08∗∗∗(.01)

13081

Drop Asian Importers .09∗∗∗(.01)

.02∗∗∗(.00)

.07∗∗∗(.01)

12560

Drop African Importers .10∗∗∗(.01)

.02∗∗∗(.00)

.08∗∗∗(.01)

12261

Drop (Formerly) Centrally Managed Importers .09∗∗∗(.01)

.02∗∗∗(.00)

.08∗∗∗(.01)

12260

Drop Small Importers (Population <1 million) .09∗∗∗(.01)

.02∗∗∗(.00)

.07∗∗∗(.01)

12578

Drop Poor Importers (Real GDP p/c <1000) .10∗∗∗(.01)

.02∗∗∗(.00)

.08∗∗∗(.01)

13629

Drop Non-European Exporters .09∗∗∗(.01)

.02∗∗∗(.00)

.06∗∗∗(.01)

11790

Drop Exporters Not in Sample Before 1995 .12∗∗∗(.01)

.04∗∗∗(.00)

.10∗∗∗(.01)

8993

Drop Late Data (year>1999) .07∗∗∗(.01)

.02∗∗∗(.01)

.06∗∗∗(.01)

3447

Drop Early Data (year<1999) .12∗∗∗(.01)

.02∗∗∗(.00)

.10∗∗∗(.01)

9853

Drop if Share Insured to Total Exports <1% .50∗∗∗(.01)

.14∗∗∗(.01)

.37∗∗∗(.01)

8357


.18∗∗∗(.01)

.46∗∗∗(.02)

6870


.25∗∗∗(.02)

.58∗∗∗(.02)

4725


.35∗∗∗(.03)

.70∗∗∗(.02)

2816

Data set includes 14,389 bilateral annual observations covering 183 countries, 1992 - 2006. Robust standard

errors (clustered by country-pairs) in parentheses. Significance: ***1%, **5%, *10%. Regressors included

but not recorded: Log Distance; Log Exporter Population; Log Importer Population; Log Exporter Real

GDP p/c; Log Importer Real GDP p/c; Currency Union dummy; Common Language dummy; Regional

Trade Agreement dummy; Common Border dummy; # Islands; Log Product Area; Common Colonizer

dummy; Currently Colony dummy; Ever Colony dummy; and Common country dummy. Year effects

included but not recorded.

25

Table 4: Instrumental Variables, The "Insurance Supply Elasticity of Exports"

First StageDependent VariableLog Insured Exports

Second StageDependent Variable: Log Exports

Instrument(s)

Log Claim Ratio(s) Claim Ratiost, t− 1, t− 2 t t− 1 t− 2 t, t− 1, t− 2

Log Claim Ratiot −.11∗∗∗(.02)

Log Claim Ratiot−1 −.06∗∗∗(.01)

Log Claim Ratiot−2 −.04∗∗∗(.01)

Log Insured Exports, Instrumented .06∗∗(.02)

.02∗∗(.01)

.03(.02)

.09∗∗∗(.03)

.06(.03)

Log DistanceLog Exp Population 16.55∗∗∗

(3.38).40(.89)

1.45∗∗∗(.58)

1.08∗(.59)

−.06(.70)

1.25∗∗(.65)

Log Imp Population −3.03∗∗∗(.80)

−.70∗∗∗(.26)

−.53∗(.30)

−0.47(.38)

−.17(.36)

−.93∗∗∗(.23)

Log Exp Real GDP p/c −7.14∗∗∗(1.97)

−2.59∗∗∗(.48)

−1.71∗∗∗(.32)

−2.17∗∗∗(.37)

−2.13∗∗∗(.43)

−1.80∗∗∗(.36)

Log Imp Real GDP p/c .95∗∗∗(.28)

.77∗∗∗(.14)

.73∗∗∗(.10)

.66∗∗∗(.12)

.63∗∗∗(.12)

.34∗∗∗(.06)

Currency Union .12(.19)

.12∗∗∗(.04)

.14∗∗∗(.03)

.15∗∗∗(.03)

.15∗∗∗(.04)

.15∗∗∗(.04)

Common LanguageRTA −.16

(.12).13∗∗(.06)

.13∗∗∗(.04)

.11∗∗∗(.04)

.11∗∗∗(.04)

.15∗∗∗(.04)

Common BorderNo. IslandsLog Product AreaCommon ColonizerCurrently Colony .06

(.08)−.06(.04)

−.03(.04)

−.01(.04)

−.03(.04)

−.02(.04)

Ever ColonyCommon CountryF-statistic for excluded instruments 23.95 23.95 230.98 96.14 51.10 6.89RMSE .64 .18 .21 .21 .20 .29Observations 2974 2974 5210 4502 3851 9112

All models include dyadic fixed effects. Robust standard errors (clustered by country-pairs) in parentheses.

Year effects included but not recorded. Significance: ***1%, **5%, *10%.

26

Table 5: IV Estimates of the "Insurance Supply Elasticity of Exports" for Various Shares of Insured Exports

Second StageInstrument: Log Claim Ratio at t

Share of Insured to Total Exports > 1 > 2 > 3 > 4 > 5 > 6 > 7 > 8 > 9 > 10

Log Insured Exports, Instrumented .16∗∗∗(.05)

.20∗∗∗(.07)

.24∗∗∗(.08)

.29∗∗∗(.10)

.25∗∗(.10)

.23∗∗∗(.09)

.27∗∗∗(.09)

.32∗∗∗(.10)

.34∗∗∗(.11)

.35∗∗∗(.13)

Log DistanceLog Exp Population 1.47∗∗∗

(.76)2.20∗∗∗(.82)

1.58∗∗∗(.79)

.82(.81)

.82(.84)

.91(.87)

.87(.92)

1.19(1.30)

2.35(1.72)

1.94(1.77)

Log Imp Population −.31(.33)

−.39(.27)

−.56∗∗(.27)

−.54∗∗(.26)

−.51∗∗(.26)

−.47∗(.26)

−.36(.26)

−.20(.25)

−.16(.26)

−.15(.28)

Log Exp Real GDP p/c −1.94∗∗∗(.50)

−2.08∗∗∗(.54)

−2.25∗∗∗(.58)

−2.21∗∗∗(.63)

−2.24∗∗∗(.69)

−2.25∗∗∗(.66)

−1.98∗∗∗(.80)

−1.81∗∗(.89)

−2.09∗∗∗(.88)

−2.22∗∗∗(1.00)

Log Imp Real GDP p/c .51∗∗∗(.12)

.52∗∗∗(.12)

.49∗∗∗(.13)

.51∗∗∗(.13)

.53∗∗∗(.14)

.61∗∗∗(.12)

.57∗∗∗(.12)

.50∗∗∗(.13)

.50∗∗∗(.15)

.50∗∗∗(.15)

Currency Union .22∗∗∗(.04)

.22∗∗∗(.04)

.23∗∗∗(.04)

.26∗∗∗(.04)

.30∗∗∗(.04)

.29∗∗∗(.05)

.26∗∗∗(.04)

.24∗∗∗(.04)

.18∗∗∗(.05)

.23∗∗∗(.06)

Common LanguageRTA .13∗∗∗

(.04).13∗∗∗(.04)

.15∗∗∗(.04)

.16∗∗∗(.04)

.16∗∗∗(.04)

.19∗∗∗(.04)

.20∗∗∗(.04)

.19∗∗∗(.04)

.19∗∗∗(.05)

.17∗∗∗(.05)

Common BorderNo. IslandsLog Product AreaCommon ColonizerCurrently Colony −.04

(.04)−.07∗∗(.03)

−.10∗∗∗(.03)

−.12∗∗∗(.04)

−.10∗∗∗(.03)

−.12∗∗∗(.03)

−.13∗∗∗(.03)

−.13∗∗∗(.03)

−.15∗∗∗(.03)

−.17∗∗∗(.04)

Ever ColonyCommon CountryF-statistic for excluded instruments 63.56 49.40 33.65 20.76 17.47 23.39 18.81 16.70 14.72 9.90RMSE .20 .20 .20 .20 .19 .19 .19 .18 .18 .18Observations 3924 3383 3112 2821 2485 2174 1947 1767 1600 1459

All models include dyadic and year fixed effects. Robust standard errors (clustered by country-pairs) in

parentheses. Significance: ***1%, **5%, *10%.

Table 6: Importer Country Risk and IV Estimates of the "Insurance Supply Elasticity of Exports"


Share of Insured to Total Exports All >1 >2 >3 >4 >5 >6 >7 >8 >9 >10Log Insured Exports, Instrumented .02∗

(.01).13∗∗(.05)

.16∗∗(.07)

.21∗∗∗(.08)

.25∗∗(.11)

.27∗∗(.12)

.26∗∗∗(.10)

.29∗∗∗(.11)

.36∗∗∗(.12)

.40∗∗∗(.13)

.44∗∗∗(.16)

Importer Country Risk −.01∗∗∗(.00)

−.01∗∗∗(.00)

−.01∗∗∗(.00)

−.01∗∗∗(.00)

−.01∗∗(.00)

−.00(.00)

−.01∗(.00)

−.00(.00)

−.00(.00)

−.00(.00)

−.00(.00)

F-statistic for excluded instruments 218.11 54.39 41.23 27.03 16.51 13.00 18.22 14.37 13.16 11.82 7.67RMSE .20 .19 .19 .19 .18 .18 .18 .18 .18 .18 .18Observations 4993 3729 3198 2933 2652 2331 2026 1809 1633 1472 1334


parentheses. Significance: ***1%, **5%, *10%. Regressors included but not recorded: Log Distance; Log

Exporter Population; Log Importer Population; Log Exporter Real GDP p/c; Log Importer Real GDP

p/c; Currency Union dummy; Common Language dummy; Regional Trade Agreement dummy; Common

Border dummy; # Islands; Log Product Area; Common Colonizer dummy; Currently Colony dummy;

Ever Colony dummy; and Common country dummy.

27

Table 7: Trade Dynamics and IV Estimates of the "Insurance Supply Elasticity of Exports"


Share of Insured to Total Exports All >1 >2 >3 >4 >5 >6 >7 >8 >9 >10Log Insured Exports, Instrumented .01∗∗

(.01).14∗∗∗(.04)

.18∗∗∗(.05)

.23∗∗∗(.07)

.31∗∗∗(.10)

.35∗∗∗(.12)

.32∗∗∗(.09)

.37∗∗∗(.10)

.40∗∗∗(.12)

.46∗∗∗(.14)

.51∗∗∗(.17)

Importer Country Risk −.01∗∗∗(.00)

−.01∗∗∗(.00)

−.01∗∗∗(.00)

−.01∗∗∗(.00)

−.01∗∗(.00)

−.00(.00)

−.01∗(.00)

−.00(.00)

−.00(.00)

−.00(.00)

−.00(.00)

Log Exportst−1 .59∗∗∗(.02)

.50∗∗∗(.04)

.47∗∗∗(.04)

.45∗∗∗(.05)

.39∗∗∗(.06)

.38∗∗(.06)

.39∗∗∗(.05)

.36∗∗∗(.06)

.34∗∗∗(.07)

.30∗∗∗(.08)

.24∗∗∗(.09)








Table 8: Share of Insured to Total Exports and IV Estimates of the "Insurance Supply Elasticity of Exports"


Share of Insured to Total Exports All >1 >2 >3 >4 >5 >6 >7 >8 >9 >10Log Share of Insured Exports, Instrumented .02∗∗

(.01).20∗∗∗(.08)

.25∗∗(.11)

.32∗∗(.14)

.41∗∗(.21)

.33∗(.18)

.30∗∗(.15)

.37∗∗(.18)

.47∗∗(.21)

.53∗∗(.25)

.53∗(.30)








28

Table 9: Zero Exports, Endogenous Sample Selection and the "Insurance Supply Elasticity of Exports"


Share of Insured to Total Exports All >1 >2 >3 >4 >5 >6 >7 >8 >9 >10Log Insured Exports, Instrumented .02∗∗

(.01).16∗∗∗(.05)

.20∗∗∗(.07)

.24∗∗∗(.08)

.29∗∗∗(.10)

.24∗∗(.10)

.23∗∗∗(.09)

.26∗∗∗(.10)

.32∗∗∗(.10)

.34∗∗∗(.11)

.35∗∗∗(.13)

Inverse Mills Ratio −15.35∗∗(6.91)

−13.54∗(8.00)

−13.06∗(7.32)

−4.45(6.51)

−4.12(6.91)

−9.27(6.69)

−9.75(7.22)

−3.70(6.90)

−3.77(6.69)

−1.62(6.74)

−1.77(7.09)


The inverse Mills ratio is calculated from the linear prediction of a probit model on P(Xijt>0), estimated for

each year, following Wooldridge (1995). All models include dyadic and year fixed effects. Robust standard

errors (clustered by country-pairs) in parentheses. Significance: ***1%, **5%, *10%. Regressors included

but not recorded: Log Distance; Log Exporter Population; Log Importer Population; Log Exporter Real

GDP p/c; Log Importer Real GDP p/c; Currency Union dummy; Common Language dummy; Regional

Trade Agreement dummy; Common Border dummy; # Islands; Log Product Area; Common Colonizer

dummy; Currently Colony dummy; Ever Colony dummy; and Common country dummy.

Table 10: Zero Insured Exports, Endogenous Sample Selection and the "Insurance Supply Elasticity of Exports"


Share of Insured to Total Exports All >1 >2 >3 >4 >5 >6 >7 >8 >9 >10Log Insured Exports, Instrumented .02∗

(.01).16∗∗∗(.05)

.20∗∗∗(.07)

.24∗∗∗(.08)

.29∗∗∗(.10)

.25∗∗∗(.10)

.23∗∗∗(.09)

.26∗∗∗(.10)

.31∗∗∗(.10)

.34∗∗∗(.11)

.34∗∗∗(.13)

Inverse Mills Ratio −.21∗∗∗(.05)

−.14∗∗(.06)

−.13∗∗(.07)

−.11(.08)

−.07(.09)

−.11(.09)

−.14∗(.08)

−.12(.08)

−.09(.08)

−.06(.08)

−.05(.08)


The inverse Mills ratio is calculated from the linear prediction of a probit model on P(InsExpijt>0), esti-

mated for each year, following Wooldridge (1995). All models include dyadic and year fixed effects. Robust

standard errors (clustered by country-pairs) in parentheses. Significance: ***1%, **5%, *10%. Regressors

included but not recorded: Log Distance; Log Exporter Population; Log Importer Population; Log Ex-

porter Real GDP p/c; Log Importer Real GDP p/c; Currency Union dummy; Common Language dummy;

Regional Trade Agreement dummy; Common Border dummy; # Islands; Log Product Area; Common

Colonizer dummy; Currently Colony dummy; Ever Colony dummy; and Common country dummy.

29

Table 11: Linear Probability Model Estimates of the Private Credit Insurance Effect on the Extensive Margin

Dependent Variable: Dummy Equal to 1 if Xijt >0(1) (2) (3) (4) (5) (6)

Sum Insured Exportsit (billion) −.0004∗∗∗(.0001)

−.0004∗∗∗(.0001)

−.0004∗∗∗(.0001)

−.0004∗∗∗(.0001)

−.0004∗∗∗(.0001)

−.0004∗∗∗(.0001)

Sum Insured Exportsit * Very High Riskjt .0001∗∗∗(.0000)

.0001∗(.0001)

Sum Insured Exportsit * High Riskjt .0000(.0000)

.0000(.0001)

Sum Insured Exportsit * Moderate Riskjt −.0000∗∗(.0000)

−.0000(.0000)

Sum Insured Exportsit * Low Riskjt .0000(.0000)

.0000(.0000)

Sum Insured Exportsit * Very Low Riskjt .0000(.0000)

R2 .27 .27 .27 .27 .27 .27RMSE .07 .07 .07 .07 .07 .07Observations 41600 41600 41600 41600 41600 41600







30

Table 12: Accounting for Public Export Credit Insurance, Sample with Dutch Exports 1992-2002

Fixed Effects: None DyadicLog Insured Exports .17∗∗∗

(.03).05∗∗∗(.02)

Log Publicly Insured Exports .03(.02)

.01(.01)

Log Distance −.72∗∗∗(.09)

Log Exp PopulationLog Imp Population .63∗∗∗

(.09)−1.16(.99)

Log Exp Real GDP p/cLog Imp Real GDP p/c .97∗∗∗

(.13).69∗(.40)


.25∗∗∗(.10)

Common Language −.07(.16)

RTA −.19(.14)

.25∗∗∗(.08)

Common Border .16(.18)

No. Islands −.27(.21)

Log Product Area −.06(.07)

Common ColonizerCurrently Colony 1.33∗∗∗

(.27)


Common CountryObservations 357 357R2 .92 .99RMSE .54 .21

Robust standard errors (clustered by country-pairs) in parentheses. Year effects included but not recorded.

Significance: ***1%, **5%, *10%.

31

Table 13: Bonus Vetus OLS and Tetradic Estimates of the Private Credit Insurance Effect on Exports

Transformation of trade costs variables Tetradsà la Baier and Bergstrand (2009) à la Head, Mayer and Ries (2010)Using Using

Simple Averages GDP WeightsBase Exporter UK GERBase Importer NLD FRAShare of Insured to Total Exports All > 10 All > 10 All > 10 All > 10

Log Insured Exports .13∗∗∗(.01)

.91∗∗∗(.02)

.26∗∗∗(.01)

.97∗∗∗(.01)

.15∗∗∗(.01)

.36∗∗∗(.07)

.12∗∗∗(.02)

.18∗∗∗(.03)

Log Distance −1.16∗∗∗(.06)

−.36∗∗∗(.05)

−.05∗∗∗(.01)

−.02∗∗(.01)

−.69∗∗∗(.06)

.05(.13)

−1.13∗∗∗(.06)

−.88∗∗∗(.14)

Log Exp Population .71∗∗∗(.02)

.19∗∗∗(.02)

.70∗∗∗(.02)

.17∗∗∗(.03)

Log Imp Population .76∗∗∗(.01)

.15∗∗∗(.02)

.65∗∗∗(.02)

.11∗∗∗(.02)

Log Exp Real GDP p/c 0.23∗∗(.10)

.31∗∗(.14)

.51∗∗∗(.12)

.16(.13)

Log Imp Real GDP p/c 1.36∗∗∗(.03)

.24∗∗∗(.03)

1.08∗∗∗(.03)

.15∗∗∗(.03)


.14(.11)

−.44∗∗∗(.08)

−.07(.05)

−.88∗∗∗(.10)

−.31(.29)

−.26∗∗∗(.09)

−.29∗∗∗(.11)

Common Language .30∗∗∗(.09)

−.07(.07)

.07∗∗∗(.02)

.04∗∗∗(.01)

.41∗∗∗(.07)

.70∗∗∗(.20)

.24∗∗∗(.07)

.46∗∗∗(.12)

RTA .38∗∗∗(.09)

.16∗∗(.07)

.23∗∗∗(.03)

.02(.02)

.18(.16)

1.00∗∗∗(.36)

.01(.15)

−.21(.42)

Common Border −.15(.11)

−.18∗(.10)

.28∗∗∗(.07)

.04(.05)

−.20∗(.10)

.27(.27)

−.12(.07)

.12(.13)

No. Islands −26.03∗∗∗(3.25)

1.58(2.50)

.17∗∗∗(.02)

.03∗∗(.01)

Log Product Area .38(.27)

−.03(.21)

.01∗∗∗(.00)

.01∗∗∗(.00)

Common Colonizer 1.69∗∗∗(.58)

−.25(1.14)

.08(.87)

.38(1.32)

2.58∗∗∗(.42)

1.61∗∗∗(.24)

Currently Colony −.37∗(.19)

−.30∗∗∗(.08)

−.42(.55)

−.30(.22)

.25(.35)

1.14∗∗∗(.09)

.82∗∗∗(.17)


.19∗∗∗(.07)

.34∗∗∗(.04)

.00(.03)

.41∗∗∗(.07)

.19(.18)

.72∗∗∗(.09)

.67∗∗∗(.13)

Common Country 1.07∗∗(.44)

.72∗∗∗(.09)

.39(.71)

.56∗∗∗(.22)

R2 .83 .96 .78 .96RMSE 1.03 .52 1.16 .53Observations 14389 2816 14389 2816 9959 1212 8780 1774

Robust standard errors (clustered by country-pairs in the Bonus Vetus OLS estimation and (multiway)

clustered by dyad, exporter and importer for the Tetrad estimation) in parentheses. Year effects included

but not recorded. Significance: ***1%, **5%, *10%.

32

Table 14: The Supply of Private Credit Insurance and the World Trade Collapse, 2008Q4-2009Q2

Share of privately insured to total exports6 8 10 12 14 16

(World ≈ 6.1 in 2007) (Europe ≈ 12.3 in 2007)Percent decline in thesupply of private

export credit insurance Estimated export decline in percent...10 1.4 1.8 2.3 2.9 3.1 4.315 2.2 2.7 3.4 4.4 4.7 6.520 2.9 3.6 4.5 5.8 6.2 8.7

...percent of World export decline 2008Q4-2009Q2 a

10 5 6 715 7 9 1120 9 12 15

...percent of European export decline 2008Q4-2009Q2 a

10 10 11 1515 15 16 2320 20 22 31

aThe nominal Euro value of World (Euro area countries) exports declined by 31% (28.4%).

33

References

[1] Abraham, F. and G. Dewit (2000). "Export Promotion via Offi cial Export Insurance." Open

Economies Review, 11(1), pp. 5-26.

[2] Amiti, M. and D.E. Weinstein (2009). "Exports and Financial Shocks." NBER Working Paper

No. 15556.

[3] Anderson, J.E. and D. Marcouiller (2002). "Insecurity and the Pattern of Trade: An Empirical

Investigation." Review of Economics and Statistics, 84(2), pp. 342-352.

[4] Anderson, J.E. and E. Van Wincoop (2003). "Gravity with Gravitas: A Solution to the Border

Puzzle." American Economic Review, 93(1), pp. 170-192.

[5] Angrist, J.D. and A.B. Krueger (2001). "Instrumental Variables and the Search for Identification:

From Supply and Demand to Natural Experiments." Journal of Economic Perspectives, 15(4),

pp. 69-85.

[6] Angrist, J.D and J. Pischke (2009). "Mostly Harmless Econometrics. An Empiricist’s Compan-

ion." Princeton University Press: Princeton.

[7] Auboin, M. (2007). "Boosting Trade Finance in Developing Countries: What Link With The

WTO?" Economic Research and Statistics Division Discussion Paper, World Trade Organiza-

tion. Geneva: WTO.

[8] Auboin, M. (2009). "Restoring Trade Finance: What the G20 Can Do." In R. Baldwin and S.J.

Evenett (ed.) The Collapse of Global Trade, Murky Protectionism, and the Crisis: Recommen-

dations for the G20, London: Center for Economic Policy Research.

[9] Baier, S.L. and J.H. Bergstrand (2009). "Bonus vetus OLS: A Simple Method for Approxi-

mating International Trade-Cost Effects Using the Gravity Equation." Journal of International

Economics, 77(1), pp. 77-85.

[10] Becue, P. (2009). "Handboek Kredietverzekering. De Onzichtbare Bank." Antwerpen: Intersen-

tia.

34

[11] Berman, N. and P. Martin (2009). "The Vulnerability of Sub-Saharan Africa to the Financial

Crisis: The Case of Trade." Sciences Po, mimeo.

[12] Bernard, A.B. and J.B. Jensen (2004). "Why Some Firms Export." Review of Economics and

Statistics, 86(2), pp. 561-569.

[13] Blundell, R. and S. Bond (1998). "Initial Conditions and Moment Restrictions in Dynamic Panel

Data Models." Journal of Econometrics, 87(1), pp. 115-143.

[14] Bun, M. and F. Klaassen (2002). "The Importance of Dynamics in Panel Gravity Models of

Trade." Discussion Paper: University of Amsterdam.

[15] Chauffour, J. and T. Farole (2009). "Trade Finance in Crisis. Market Adjustment or Market

Failure?" World Bank Policy Research Paper 5003.

[16] Chor, D. and K. Manova (2009). "Of the Cliff and Back? Credit Conditions and International

Trade during the Global Financial Crisis." NBER Working Paper No. 16174.

[17] Dixit, A. (1989). "Entry and Exit Decisions Under Uncertainty." Journal of Political Economy,

97(3), pp. 620-638.

[18] Dorsey, T. (2009). "Trade Finance Stumbles." Finance and Development, 46(1), pp. 18-19.

[19] Egger, P. and Nelson, D. (2010). "How Bad is Antidumping?: Evidence from Panel Data." The

Review of Economics and Statistics, forthcoming.

[20] Egger, P. and T. Url (2006). "Public Export Credit Guarantees and Foreign Trade Structure:

Evidence from Austria." The World Economy, 29(4), pp. 399-418.

[21] Eichengreen, B. and D.A. Irwin (1998). "The Role of History in Bilateral Trade Flows." In J.A.

Frankel (ed.) The Regionalization of the World Economy, Chicago: University of Chicago Press.

[22] Funatsu, H. (1986). "Export Credit Insurance." Journal of Risk and Insurance, 53(4), pp. 680-

692.

[23] Glick, R. and A.K. Rose (2002). "Does a Currency Union Affect Trade? The Time Series

Evidence." European Economic Review, 46(6), pp. 1125—1151.

35

[24] Hausman, J. (2001). "Mismeasured Variables in Econometric Analysis: Problems from the Right

and Problems from the Left." Journal of Economic Perspectives, 15(4), pp. 57-67.

[25] Head, K., T. Mayer and J. Ries (2010). "The Erosion of Colonial Trade Linkages After Inde-

pendence." Journal of International Economics, 81(1), pp.1-14.

[26] Helpman, E., M. Melitz and Y. Rubinstein (2008). "Estimating Trade Flows: Trading Partners

and Trading Volumes." Quarterly Journal of Economics, 123(2), pp. 441-87.

[27] Iacovone, L. and V. Zavacka (2009). "Banking Crises and Exports: Lessons from the Past."

World Bank Policy Research Working Paper 5016, Washington, D.C.: World Bank.

[28] ICC (2010). "Rethinking Trade Finance 2010: Global Survey." ICC Banking Commission Market

Intelligence Report, April.

[29] IMF (2009). "Survey of Private Sector Trade Credit Developments." Washington, D.C.: Inter-

national Monetary Fund.

[30] Jones, P.M. (2010). "Trade Credit Insurance." Primer Series on Insurance, 15, Washington,

D.C.: World Bank.

[31] Kleibergen, F. and R. Paap (2006). "Generalized Reduced Rank Tests Using The Singular Value

Decomposition." Journal of Econometrics, 127(1), pp. 97-126

[32] Levchenko, A.A., L. Lewis and L.L. Tesar (2010). "The Collapse of International Trade during

the 2008—2009 Crisis: In Search of the Smoking Gun." NBER Working Paper No. 16006.

[33] Meon, P.-G. and K. Sekkat (2004). "Does the Quality of Institutions Limit the MENA’s Inte-

gration in the World Economy?" The World Economy, 27(9), pp. 1475-1498.

[34] Moser, C., T. Nestmann and M. Wedow (2008). "Political Risk and Export Promotion: Evidence

from Germany." The World Economy, 31(6), pp. 781-803.

[35] OECD (2009). "OECD Economic Outlook (85)." Chapter 1. General Assessment of the Macro-

economic Situation, Paris: OECD.

[36] Petersen, M. and R. Rajan (1997). "Trade Credit: Theories and Evidence." The Review of

Financial Studies, 10(3), pp. 661-691.

36

[37] Roberts, M.J. and J.R. Tybout (1997). "The Decision to Export in Colombia: An Empirical

Model of Entry with Sunk Costs." American Economic Review, 87(4), pp. 545-564.

[38] Ronci, M. (2005). "Trade Finance and Trade Flows: Panel Data Evidence from 10 Crises." In

J. Wang and M. Ronci (eds.) Access to Trade Finance in Times of Crisis, Washington, D.C.:

International Monetary Fund.

[39] Rose, A.K. (2004). "Do We Really Know that the WTO Increases Trade?" American Economic

Review, 94(1), pp. 98-114.

[40] Rose, A.K. and M.M. Spiegel (2010). "The Olympic Effect." Economic Journal, forthcoming.

[41] Schmidt-Eisenlohr, T. (2009). "Towards a Theory of Trade Finance." EUI Working Paper 43.

[42] Stephens, M. (1999). "The Changing Role of Export Credit Agencies." Washington, D.C.: In-

ternational Monetary Fund.

[43] Stock, J.H. and M. Yogo (2005). "Testing for Weak Instruments in Linear IV Regression."

In D.W. Andrews and J.H. Stock (eds.) Identification and Inference for Econometric Models:

Essays in Honor of Thomas Rothenberg, New York, Cambridge University Press.

[44] Swiss Re (2006). "Credit Insurance and Surety: Solodifying Comments." Sigma 6/2006, Zurich.

[45] Wang, J., M. Mansilla, Y. Kikuchi and S. Choudhury (2005). "Offi cially Supported Export

Credits in a Changing World." Washington, D.C.: International Monetary Fund.

[46] Wooldridge, J.M. (1995). "Selection Corrections for Panel Data Models under Conditional Mean

Independence Assumptions." Journal of Econometrics, 68(1), pp. 115-132.

37

Appendix

Table A.1: Data Sources

• FOB exports in US dollars are taken from IFS Direction of Trade CD-ROM. The figuresare converted to euros at the average annual exchange rate. Pre-1999 exchange rates werecalculated as the weighted bilateral dollar exchange rate of the 11 countries participat-ing at the start of the euro in 1999 (Source: FT/Reuters). All figures are deflated bythe Harmonised Index of Consumer Prices (HICP), overall index, taken from Eurostat,2000=1.• Population and real GDP per capita (rgdpl) taken from PWT Mark 6.2. If PWT dataare unavailable, I use World Development Indicators. The figures are converted to eurosat the average annual exchange rate.• Country-specific data (on location, area, island-nation status, contiguity, language, col-onizer, and independence) taken from CIA World Factbook website.• Currency-union data taken from Glick-Rose (2002).• Regional trade agreements taken from WTO websitehttp//www.wto.org/english/tratop_e/region_e/eif_e.xls• The credit insurance data comes from one of the "Big Three" internationally activeprivate credit insurers; company details are confidential.

38

Table A.2: Country List

Afghanistan, Albania, Algeria, Angola, Antigua & Barbuda, Argentina, Armenia, Aus-tralia, Austria, Azerbaijan, Bahamas, Bahrain, Bangladesh, Barbados, Belarus, Belgium,Belize, Benin, Bhutan, Bolivia, Bosnia & Herzegovina, Botswana, Brazil, Brunei, Bulgaria,Burkina Faso, Burundi, Cambodia, Cameroon, Canada, Cape Verde, Central African Re-public, Chad, Chile, China, P.R.: Mainland, China, P.R.: Macao, Colombia, Comoros,Congo, Dem. Rep., Congo, Republic of, Costa Rica, Cote D’Ivoire, Croatia, Cuba, Cyprus,Czech Republic, Denmark, Djibouti, Dominica, Dominican Republic, Ecuador, Egypt, ElSalvador, Equatorial Guinea, Eritrea, Estonia, Ethiopia, Fiji, Finland, France, Gabon,Gambia, Georgia, Germany, Ghana, Greece, Grenada, Guatemala, Guinea, Guinea-Bissau,Guyana, Haiti, Honduras, Hong Kong, Hungary, Iceland, India, Indonesia, Iran, Iraq, Ire-land, Israel, Italy, Jamaica, Japan, Jordan, Kazakhstan, Kenya, Kiribati, Korea, Rep,Kuwait, Kyrgyzstan, Laos, Latvia, Lebanon, Lesotho, Liberia, Libya, Lithuania, Lux-embourg, Macedonia, Madagascar, Malawi, Malaysia, Maldives, Mali, Malta, Maurita-nia, Mauritius, Mexico, Moldova, Mongolia, Morocco, Mozambique, Myanmar, Namibia,Nepal, Netherlands, Netherlands Antilles, New Zealand, Nicaragua, Niger, Nigeria, Nor-way, Oman, Pakistan, Palau, Panama, Papua New Guinea, Paraguay, Peru, Philippines,Poland, Portugal, Qatar, Romania, Russian Federation, Rwanda, Samoa, Sao Tome &Principe, Saudi Arabia, Senegal, Seychelles, Sierra Leone, Singapore, Slovakia, Slovenia,Solomon Islands, Somalia, South Africa, Spain, Sri Lanka, St. Kitts & Nevis, St. Lucia, St.Vincent & Grens., Sudan, Suriname, Swaziland, Sweden, Switzerland, Syria, Tajikistan,Tanzania, Thailand, Togo, Tonga, Trinidad & Tobago, Tunisia, Turkey, Turkmenistan,Uganda, Ukraine, United Arab Emirates, United Kingdom, United States of America,Uruguay, Uzbekistan, Vanuatu, Venezuela, Vietnam, Yugoslavia, Zambia, Zimbabwe.

39

Table A.3: Correlation Matrix

Xijt InsExp Dis Pop1 Pop2 GDPpc1 GDPpc2 CU Lang

Xijt 1.00InsExp .62 1.00Dis −.45 −.32 1.00Pop1 .19 .06 .19 1.00Pop2 .49 .19 .04 −.05 1.00GDPpc1 −.03 .04 −.04 −.36 −.03 1.00GDPpc2 .48 .33 −.28 −.17 −.13 .05 1.00CU .26 .13 −.35 −.10 .04 .07 .23 1.00Lang −.04 .06 .14 .12 −.14 −.07 −.11 −.07 1.00RTA .39 .25 −.63 −.18 .12 −.06 .30 .28 −.16Border .31 .16 −.43 −.01 .06 −.04 .16 .24 .06Isl −.28 −.10 .37 −.05 −.33 .05 .06 −.12 .19Area .27 .04 .26 .37 .62 −.15 −.19 −.08 .00CCol .03 .02 −.05 −.02 −.00 −.03 .01 −.01 −.00Col .01 .02 .02 .00 −.05 −.01 .02 −.01 .07ECol .01 .17 .07 .17 −.12 −.05 −.14 −.01 .48SameC .00 .02 .01 −.01 −.04 .00 .01 −.01 .06

RTA Border Isl Area CCol Col ECol SameC

RTA 1.00Border .21 1.00Isl −.26 −.11 1.00Area −.09 −.01 −.10 1.00CCol .02 .09 −.01 −.02 1.00Col .01 −.01 −.02 −.06 −.00 1.00ECol −.13 .02 .04 −.07 −.01 .11 1.00SameC .03 .01 −.01 −.06 −.00 .78 .08 1.00

40

Notes1For example, knowledge on changes in the supply and price of trade loans, letters of credit and export credit

insurance, during the 2008-2009 global financial crisis came only from bank surveys (see IMF, 2009). While informative,it is diffi cult to separate supply and demand in these surveys (Dorsey, 2009).

2See EU Council Directive 98/29/EC at http://eur-lex.europa.eu/JOIndex.do?ihmlang=en, last accessed on October14, 2010

3The estimate of private insurance is based on the value of exports insured by one private insurer combined with confi-dential data on its market share. The value of insured exports by the Dutch government is available in the 2007 annual re-port of Atradius Dutch State Business at http://www.atradiusdutchstatebusiness.nl/publicaties/jaaroverzichten/index.html,last accessed on October 14, 2010

4The "Big Three" private credit insurers cover 87 percent of the world private credit insurance market: Euler Hermes(36 percent), Atradius (31 percent) and Coface (20 percent).

5Short-term trade finance business is usually defined as business with a maximum credit length of one year, althoughin practice most short-term business involves 180 days or less (Stephens, 1999).

6Such action may include intervention to prevent the transfer of payments, cancellation of a license, or acts of waror civil war. Non-payment by sovereign buyers is also a political risk.

7Bernard and Jensen (2004) examine empirically whether public export promotion expenditures promote exportparticipation by gathering information on foreign markets, but find no significant contemporaneous effect in theirsample of large plants.

8For example, Atradius offers an information service called "Observa News" with current business information onkey customers, prospects or competitors. The service is charged as a flat annual fee of 24 euro per company that ismonitored and a reduced rate of 16 euro for Atradius insured customers (see www.atradius.com). Coface also offersvarious rating and business information services (see www.coface.com). Euler Hermes offers similar services in somecountries, see http://www.eulerhermes.com/en/products-solutions/eolis-online-service.html. Last accessed on October14, 2010.

9Company details are confidential.10The raw data includes 114 observations with a share of insured to total exports above one hundred. These 114

observations seem to be randomly distributed over 15 different exporters and 61 destination countries. I adjusted thevalue of insured exports in these 114 observations to equal the value of total exports. The main results are insensitiveto these adjustments.11 Instrumental variables provide a consistent estimate even in the presence of measurement error if the instrument is

uncorrelated with the measurement error and the equation error, but correlated with the correctly measured variable.12 I use country codes from the IMF’s International Financial Statistics for these classifications.13Hausman tests cannot reject the null that insured exports may be treated as exogenous.14 In general, premia are calculated as the sum of the expected loss (due to claims), administrative and capital costs.15For example, the ICC Global Survey report (2010) reports that "Total claims paid to insured customers by all

Berne Union members more than doubled from 2008 to 2009 and reached USD2.4 billion. As the total premium stayedroughly the same at an estimated USD2.8 billion, the loss ratio jumped from 40 to 87 percent. The Berne Union is theleading international organisation of public and private sector providers of export credit and investment insurance.16This ability to set and manage exposures distinguishes credit insurance from other kinds of insurance and many

other credit instruments (Swiss Re, 2006).17Results not recorded. Also, a test of the hypothesis that the conditional elasticity of the log claims ratio is equal

to zero cannot be rejected by any reasonable significance levels.18 I test for underidentification by applying Anderson’s canonical correlations test and using the LM version of the

Kleibergen-Paap (2006) rk statistic, weak identification using the Wald version of the Kleibergen-Paap (2006) rk statisticand the critical values calculated by Stock and Yogo (2005), and the Hansen’s J test of overidentifying restrictions.19 Including more lags of the dependent variable did not change the results. Moreover, these additional lags were

statistically insignificant. Also, it is well known that fixed effects regression including lagged dependent variables mayyield biased estimates. I examined this potential bias by estimating the benchmark model (Table 1, Column 2) includinglags of the dependent variable and compared the results with estimates of the same model and sample using the systemgeneralised method of moments (GMM) estimator proposed by Blundell and Bond (1998). Both estimators give astatistically significant estimate for insured exports of .02, identical to the benchmark model.20The estimate is calculated using the 2007 world value of "short term new business insured" from the Berne Union

website, available at http://www.berneunion.org.uk/bu-total-data.html, and world exports from the World Trade Mon-itor of the CPB Netherlands Bureau for Economic Policy Analysis.21This elasticity is the average of the estimates for the subsample of observations with a share of insured exports

above 5 percent (see Tables 5 to7). The private insurer’s raw data reveal that 60 percent of the total value of its turnoveron exports in 2007 related to exports from the Euro area countries (excl. Cyprus, Malta, Portugal and Slovenia). Iused this share to calculate the world value of private short term insured exports from the Euro area countries. The

41

2007 value of Euro area countries’exports in goods is taken from the World Trade Monitor, CPB Netherlands Bureaufor Economic Policy Analysis.22A 1 per cent increase in insured exports (2.62 million euro on average) leads to an increase of exports amounting

to 5.96 million euro. The average multiplier increases somewhat with the share of insured exports ranging between 2.3and 3.2 (for the subsamples), with a mean (median) of 2.7 (2.8).23The size of the multiplier is comparable to the long run multiplier of Austrian guarantees in Egger and Url (2006),

allthough they do not account for the endogeneity problem.24The benchmark gravity model regressions lose 1768 observations due to missing data on the gravity controls.25See also Egger and Nelson (2010). Cross-section procedures as in Helpman, Melitz and Rubinstein (2008) are not

applicable in this case, as pointed out by Wooldridge (1995).26See Angrist and Pischke (2009, pp. 102-107, 197) for additional reasons for using LPM instead of probit or logit.27 Insured exports would perfectly predict the probability of positive export flows.28The measure for public insurance relates to premium income and thus differs from the measure for private insurance.29See "Atradius reports 2008 results", available at http://global.atradius.com/corporate/pressreleases/atradius-reports-

2008-results.html, last accessed on October 14, 2010.30See Annual review 2008 Atradius N.V. available at http://global.atradius.com/corporate/aboutus/annualreportspage.html,

last accessed on October 14, 201031See Chauffour and Farole (2009) for an overview of trade finance measures taken by governments to mitigate the

impact of the trade finance crisis.

42

Previous DNB Working Papers in 2010 No. 242 Leo de Haan and Jan Kakes, Momentum or Contrarian Investment Strategies: Evidence

from Dutch institutional investors No. 243 Ron Berndsen, Toward a Uniform Functional Model of Payment and Securities

Settlement Systems No. 244 Koen van der Veer and Eelke de Jong, IMF-Supported Programs: Stimulating Capital to

Solvent Countries No. 245 Anneke Kosse, The safety of cash and debit cards: a study on the perception and

behaviour of Dutch consumers No. 246 Kerstin Bernoth, Juergen von Hagen and Casper de Vries, The Forward Premium Puzzle

and Latent Factors Day by Day No. 247 Laura Spierdijk, Jacob Bikker and Pieter van den Hoek, Mean Reversion in International

Stock Markets: An Empirical Analysis of the 20th Century No. 248 F.R. Liedorp, L. Medema, M. Koetter, R.H. Koning and I. van Lelyveld, Peer monitoring

or cantagion? Interbank market exposure and bank risk No. 249 Jan Willem van den End, Trading off monetary and financial stability: a balance of risk

framework No. 250 M. Hashem Pesaran, Andreas Pick and Allan Timmermann, Variable Selection,

Estimation and Inference for Multi-period Forecasting Problems No. 251 Wilko Bolt, Leo de Haan, Marco Hoeberichts, Maarten van Oordt and Job Swank, Bank

Profitability during Recessions No. 252 Carin van der Cruijsen, David-Jan Jansen and Jakob de Haan, How much does the

public know about the ECB’s monetary policy? Evidence from a survey of Dutch households

No. 253 John Lewis, How has the financial crisis affected the Eurozone Accession Outlook in Central and Eastern Europe?

No. 254 Stefan Gerlach and John Lewis, The Zero Lower Bound, ECB Interest Rate Policy and the Financial Crisis

No. 255 Ralph de Haas and Neeltje van Horen, The crisis as a wake-up call. Do banks tighten screening and monitoring during a financial crisis?

No. 256 Chen Zhou, Why the micro-prudential regulation fails? The impact on systemic risk by imposing a capital requirement

No. 257 Itai Agur, Capital Requirements and Credit Rationing No. 258 Jacob Bikker, Onno Steenbeek and Federico Torracchi, The impact of scale, complexity,

and service quality on the administrative costs of pension funds: A cross-country comparison

No. 259 David-Jan Jansen and Jakob de Haan, An assessment of the Consistency of ECB Communication using Wordscores

No. 260 Roel Beetsma, Massimo Giuliodori, Mark Walschot and Peter Wierts, Fifty Years of Fiscal Planning and Implementation in the Netherlands

No. 261 Jan Marc Berk, Beata Bierut and Ellen Meade, The Dynamic Voting Patterns of the Bank of England’s MPC

No. 262 Maria Demertzis, An Operational Measure of Riskiness: A Comment No. 263 Klaus Abbink, Ronald Bosman, Ronald Heijmans and Frans van Winden, Disruptions in

large value payment systems: An experimental approach

DN

B W

OR

KIN

GP

AP

ER

DNB Working PaperNo. 35/April 2005

Jan Kakes and Cees Ullersma

Financial acceleration of booms

and busts

De Nederlandsche BankDe Nederlandsche Bank

The Private Credit Insurance Effect on Trade

Documents

Transcript of The Private Credit Insurance Effect on Trade