Natural disasters and countries' exports: New insights from a new (and an old) database

• Author: Isabelle Rabaud,Daniel Mirza,Hajare El Hadri
• Date: 01 September 2019
• Published date: 01 September 2019
• DOI: http://doi.org/10.1111/twec.12833

2668

|

wileyonlinelibrary.com/journal/twec World Econ. 2019;42:2668–2683.

© 2019 John Wiley & Sons Ltd

Received: 13 July 2017

|

Revised: 9 February 2018

|

Accepted: 6 March 2018

DOI: 10.1111/twec.12833

ORIGINAL ARTICLE

Natural disasters and countries' exports: New

insights from a new (and an old) database

HajareEl Hadri1

|

DanielMirza2,3

|

IsabelleRabaud1

1CNRS, LEO, UMR 7322,University of Orléans, Orléans Cedex 2, France

2CNRS, LEO, UMR 7322,University of Tours, Tours, France

3CEPII (Paris), Tours, France

KEYWORDS

agriculture, environment, natural disasters, trade

1

|

INTRODUCTION

In its Annual Disaster Review, the Center for Research on the Epidemiology of Disasters (CRED

Guha‐Sapir, Hoyois, Wallemacq, & Below, 2016) has reported an impressive frequency of natural

disasters with an average of 350 per year since the beginning of the years 2000s, and around 70,000

deaths per year, letting more than 15–20 millions persons injured or homeless and causing an average

yearly material cost of about $120 billions. How are disasters shaping exports of victim countries? Do

their effects on trade differ with respect to the types of disasters and their intensity? Do some sectoral

and countries' characteristics matter?

Some research has been focusing on sectoral and macroeconomic outcomes of natural disasters.

Dell, Jones, and Olken (2014) survey the literature devoted to the new climate economy. Among the

cited works, Toya and Skidmore (2007), Noy (2009), Felbermayr and Gröschl (2013, 2014) provide

examples. To synthesise their findings, agricultural output appears to be one of the sectors that is most

hit in the economy. Further, industrial output, energy demand, labour productivity, health, conflict

and political stability happen to be affected, all of these ending‐up curbing economic growth. Besides,

small and poor countries, being often associated with poor quality of institutions and low education

levels, appear to be much more sensitive to temperature shocks. Finally, the impact on outcomes is all

the more significant in the presence of severe disasters.

Nevertheless, the literature on the induced impact on trade flows has been relatively scarce. To the

best of our knowledge, four papers deal with this aspect. Jones and Olken (2010) look at the impact

of temperatures and precipitations on sectoral exports to the US and the world, respectively, by using

a cross‐country database from 1973 to 2001. They find that temperatures—albeit not precipitations—

reduce exports of poor countries: 1 Celsius degree warmer happens to reduce their exports by 2–5.7

percentage points. Agriculture happens to be, however, more affected than manufacturing. As for nat-

ural disasters, Oh and Reuveny (2010) rely on a gravity model to estimate their impact on international

trade for 116 countries over the period 1985–2003. They find an induced reduction in both imports

|

2669

EL HADRI EtAL.

and exports following disasters. Besides, the higher the political risk and the stronger this decrease.

Gassebner, Keck, and Teh (2010) analyse the impact on trade of major natural and technological

disasters. They rely on bilateral trade from 1962 to 2004 and use again a gravity equation à la Rose.

In their first series of results, the impact of the number of disasters does not appear to be significant

on bilateral imports. However, the number of disasters adjusted for the surface of the country deters

exports and imports. Besides, they find that the (negative) effect is stronger in autocratic and smaller

countries. Felbermayr and Gröschl (2013) look as well at the impact of disasters on trade but their

work was not centred on this issue. Their purpose was to employ natural disasters' variables as an

instrument for trade openness in order to measure more accurately the impact of the latter on GDP per

capita. They find, however, a positive impact of natural catastrophes on imports and a negative one

on exports.

Importantly for our paper, most of the studies use the Emergency Event Database (EM‐DAT) to

analyse the impact of disasters on outcome variables.1

Felbermayr and Gröschl (2014) note however

that the EM‐DAT database might not be suitable if one needs to exploit the data on the intensity of the

disasters. The reason mentioned by these authors is that the intensity of disasters, usually measured

from EM‐DAT by the amount of damage or number of victims, is itself correlated with the level of

development, which makes it difficult to use such measures as independent variables to explain vari-

ables such as GDP per capita but also migration or trade.

To avoid this caveat, Felbermayr and Gröschl (2014) compile and use a new alternative database

built on geophysical and meteorological measures (i.e. GeoMet database) to examine the impact of

disasters on growth. These measures could then offer the authors the possibility of assessing the link

to growth, derived from the physical strength of the disasters, a measure not reported by EM‐DAT. By

doing so, they could produce results that are quite different from the rest of the literature: they find

indeed a strong negative and statistically significant effect of disasters on growth.

This paper is the first to uncover the impact of different families of disasters, while testing for the

heterogeneity of their impact on exports. Besides, our paper is the first to compare in a quasi‐system-

atic way the results across the two data sets at hand, the standard EM‐DAT data and GeoMet data.

Our paper is mainly interested in the impact of disasters on trade through the supply of resources

channel. We ask here what is the impact of disasters on aggregate exports of a country when driven by

a reduction in its resource capacities (labour capacities, crops, public infrastructure). A disruption of

supply due to a disaster would affect a priori export flows across partners in the same fashion which

is why we do not consider here gravity equations, the latter being better suited to explain trade across

bilateral pairs of countries.

We begin by comparing the impact of the measures of occurrence or number of death‐type mea-

sures accessible via EM‐DAT with a composite measure of physical intensity of disasters computed

from GeoMet, encompassing the intensity of different types of disasters. When pooling all countries,

all products and all types of disasters, we do not find any statistical impact on exports, whichever the

database at hand (EM‐DAT or GeoMet). Besides, a concentration on agricultural products specifically

does not bring much to the picture. In a second phase, we focus on each type of disaster separately and

interact each of these with the size and the level of development of the countries, two of the characteris-

tics of the countries which are known to play a role in the expected influence of disasters on aggregate

outcomes. There, earthquakes appear to have a general and statistically robust depressive impact on

agricultural exports irrespective of the characteristics of the countries. Floods curb agricultural export

flows but only for small countries. Storms have mostly no impact. Changes in temperatures appear to

1 Through their use of other data sources, Jones and Olken (2010) and Felbermayr and Gröschl (2014) appear among few

exceptions, however.