Natural disasters, information/communication technologies, foreign direct investment and economic growth in developed countries

This paper investigates the causal relationship between natural disasters (DMS), information and communication technologies (ICT), foreign direct investment (FDI) and economic growth (GDP per capita) for 10 developed countries over the period 1990 to 2016. Panel DOLS and FMOLS results show that there is a positive relationship running from ICT to natural disasters and to foreign direct investment. In addition, ICT have a positive effect on GDP per capita. VECM Granger causality analysis results reveal a unidirectional causality in the short and long term from ICT to natural disaster and to FDI at the 5% and 10% levels. Therefore, one may note that there is a unidirectional relationship running from natural disaster to GDP and a bidirectional relationship between FDI and GDP.


Introduction 
Natural disasters are gaining ground in terms of frequency, duration and disastrous consequences. They affect more than 300 million people each year worldwide and are considered as complex threats involving several factors simultaneously (Rim et al., 2012), hold back the development of countries and increase poverty.
During the last decade, the risks and costs of natural and man-made disasters have significantly increased. For example, in 2010, a 7.3 magnitude earthquake on the Richter scale devastated Haiti, killing more than 222 000 people and leaving 300 000 injured, 1.2 million homeless in Port-au-Prince and more than 2 million displaced, especially in rural areas (United Nations Report, 2010).
In 2004, the most violent earthquake in the world after Chile in 1960 caused devastating tidal waves in part of the Indian Ocean, killed or disappeared more than 280 000 people (Red Cross Report 2010). In fact, from 2002 to 2011, there were 4 130 natural hazard disasters in the world, resulting in at least $ 1 195 billion in economic losses (UNISDR, 2015). Losses from natural disasters were estimated at $ 150 billion in 2010 (Becklumb, 2010).
In view of all this, governments around the world, civil society actors, scientists, development and humanitarian aid organizations, local communities This paper examines the relationships between Information Communication Technology, natural disasters, Foreign Direct Investment and GDP per capita for 10 developed countries over the period 1990 to 2016. Panel DOLS and FMOLS and Granger causality-VECM approach are used to investigate the short-and long-run relationship between variables and to reveal the direction of causality among them.
The paper is organized as follows. Section 1 presents an overview of literature, section 2 presents the data and the methodology, the empirical results are in Section 3 and the last section presents a summary of the results and draws conclusions.

Literature review
Skidmore and Toya (2002) revealed a positive correlation between natural disasters,human capital investment and factors of production. In addition, the occurrence of natural disasters encourages the adoption of new technologies and consequently leads to an increase in the factors of production in the long term. Cuaresma et al. (2008) and Hallegatte and Dumas (2009) showed that natural disasters do not affect long-term economic growth. The empirical results of Cavallo and Noy (2010) and Sawada et al. (2011) indicated that there is a negative relationship from natural disasters to economic growth. Cuñado and Ferreira (2014) used panel vector autoregression models, and showed that flood shocks have a positive impact on per capita GDP growth. In another study, Anuchitworawong and Thampanishvong (2015) investigated the effect of natural disaster on FDI. They found that an increase in severity of natural disaster leads to a decrease of FDI flows into Thailand. Benali et al. (2016 showed that there is a unidirectional relationship from natural disasters to budget deficit. Benali and Saidi (2017) have tested the relationship between natural disaster, economic growth, physical capital, labor and electricity for 41 countries. They showed that for African countries, natural disaster has a negative effect on all variables, for American countries, disaster measures have a negative and significant effect on economic growth and consumption electricity, and for European countries, there is a unidirectional relationship from the disaster measures to labor and from the disaster measures to electricity consumption. In reality, few studies have examined the role of ICT in the production of a common history of risk. In this regard, Shklovski et al. (2008) showed the importance of using ICT to solving the problems of catastrophic events.
Samarajiva and Waidyanatha (2009) indicated that using mobile application helps Asian government overcome difficulties caused by natural disasters. According to John et al. (2015), ICTs are instrumental in the recovery after the earthquake in Japon. Their use increases the level of social capital and civic participation. Toya and Skidmore (2015) examined the relationship between ICT and disaster fatalities. By using a panel data model over the 1980˗2013 period, they showed that ICTs help to minimize the number of fatalities following disaster events.

Data description and methodology
(2) Total population killed = ( total people killed ijt total population where i denotes the country, j represents the natural disaster (drought, floods, earthquake and storms) and t = 1, ..., N indicates the year. The disaster measures (DMS) are calculated as follows: To consider the panel unit root, one can apply the following autoregressive model: The IPS test takes the following form: where Δ is the first-difference operator, p_i is the ag order in the ADF regression.
Null hypotheses and alternatives can be written as follows:  To better take into account the degree of heterogeneity of the panel, Pedroni (1997Pedroni ( , 1999 suggests seven tests: four are based on the intraindividual dimension and three on the interindividual dimension. In the inter-individual dimension, alternative hypothesis is spelled 1 : In contrast to Pedroni tests, Kao considers the special case, in which co-integration vectors are supposed to be homogeneous among individuals. In other words, these tests do not make it possible to consider heterogeneity under the alternative hypothesis and are otherwise valid only for a bivaried system.

Panel DOLS and FMOLS estimates.
Having proved that all variables are stationary in first differences and the long-term cointegration in the preceding steps exists, one can apply the estimation tests of these long-term panel relationships using the methods of FMOLS and DOLS estimators proposed by Pedroni (2001) and Mark and Sul (2002). The FMOLS and DOLS estimates generally give different results.

Panel causality tests.
After establishing the existence of cointegration relationship, in order to study the long-term causal relationship between variables, one can use the Granger causality test (Granger, 1988). If the cointegration relationship is confirmed, Vector Error Correction (VECM) Granger causality test can be applied.
where ∆ is the first difference operator, ECT presents the error correction term.

Panel unit root results.
The results of the unit root tests from Table 2 show that the GDP, DMS, FDI, MOB and INT are not stationary in level, but stationary in first difference ( Table 3). Given that all variables are integrated for order 1 I (1), the long-term relationship between these variables is possible Note: *, **, and *** represent significance at the 1%, 5%, and 10% levels, respectively. Note: *, **, and *** represent significance at the 1%, 5%, and 10% levels, respectively.    Table 6. Note: * indicates statistical significance at the 5% level.

Panel cointegration test results.
In Table 6, FMOLS test shows a positive relationship running from mobile cellular and internet users to GDP per capita, at 5% of significance. This means that, a 1% increase in mobile cellular and internet leads to increase in GDP per capita by 9.6% and 12%, respectively. The results are consistent with Stanley et al. (2015). Mobile cellular and internet users are further stimulate economic growth. This is due to the role of ICT in improving the functioning of markets, reducing transaction costs and increasing productivity through better management. Besides, 1% increase in disaster measure leads to a decrease in GDP by 4%. Natural disasters tend to cause a series of major economic upheavals. It reduces production and number of hours worked. Reconstruction efforts compensate part of these losses and, paradoxically, stimulating effect on economic growth. In addition, it is shown that there is a positive link between foreign direct investment and GDP per capita. This implies that foreign direct investment has an important part to play in the acceleration of economic growth in developed countries. Furthermore, the effects of Internet and mobile cellular on natural disaster event and foreign direct investment are positive and statistically significant at the 5% level. Using mobile cellular and Internet can help people in preventing and moderating the serious impact of disasters. In addition, ICT stimulates foreign investment; these results are similar to Fakher (2016). Panel DOLS results indicate that a 1% increase in DMS leads to a decrease in GDP per capita by 2%. In addition, an increase in ICT leads to an increase in FDI.

Panel causality tests results.
Results are reported in Table 7. One can deduce the meaning of causal relationships that may appear between the variables at the critical level of 5%.

Conclusion
In this paper, the objective has been to use the Panel DOLS and FMOLS and Granger causality-VECM approach to characterize the relationship between natural disasters (DMS), information and communication technologies (ICT), foreign direct investment (FDI) and economic growth (GDP per capita). The study was carried out on a sample of 10 developed countries over the period 1990 to 2016. Results of FMOLS and DOLS showed that mobile cellular and internet users have a positive effect on GDP per capita and natural disaster event. In addition, natural disaster has a negative impact on GDP per capita. Results of VECM Granger causality test indicated that there is a unidirectional causality in the short and long term from ICT to natural disaster and to FDI at the 5% and 10% levels. Therefore, it was noted that there is a unidirectional causality from natural disaster to GDP and a bidirectional causality between FDI and GDP.
To avoid the consequences of disaster events, the ICT is an important learning phenomenon in the occurrence of disasters by reducing uncertainty of natural hazards. When they represent a simple information dissemination technology, digital tools are mainly used to configure alerts, establish diagnoses and record activity traces. Similarly, ICTs play a key role in accelerating the potential for economic growth, generating productivity gains of their own.