Natural disasters, adverse climatic conditions and the level of global trade during the 19th century.
Megan Hayes, 15324627, hayesm6@tcd.ie
1. Aim
The occurrence of natural disasters and adverse climatic conditions is of huge historical and economic significance, impacting economic growth and overall global trade levels during the 19th and the early 20th century. Fluctuating temperatures and rainfall levels determine the number of crops for export that any given country can produce. Storms, hurricanes and earthquakes prevent the safe voyage of cargo ships, as well as causing damage to infrastructure such as ports, that are needed for trade. Volcanic eruptions have been a key determinant of crop and economic growth for centuries, specifically in the Pacific region. Ultimately, natural disasters commonly exacerbate or cause complex disasters, causing mass displacements of people, widespread famine, and weak economic, social and political institutions.
Trade is without a doubt, one of the major forces behind economic development, growth and the reduction of poverty. However, natural disasters have, in the past and continue to be, extremely disruptive to the level of trade with significant physical damages to infrastructure, and further impacts on export levels (Da Silva and Cernat, 2012). Natural disasters have the potential to negatively affect a countries’ trade level for multiple years after the event itself. Therefore, this paper hopes to find a decrease in trade levels globally or for any given country or continent, due to the occurrence of a natural disaster.
This study also takes into consideration the impact of climate, be it adverse or simply changing, on the level of global trade. Main climatic factors include temperature, rainfall and snow levels. This is especially interesting as it is extremely relevant in the world today, with global warming already a severe threat and climate changes’ recognised impact on the growth and production of certain crops and goods and the subsequent transport of them. An important and direct effect of changing climates is that the supply, transport and distribution chains can become more vulnerable to disruptions, thereby affecting current and future international trade patterns. Extreme weather events have the potential to impact the functionality of ports and other transport routes; they can cause lasting damage to infrastructure that is critical to trade, and have even further and longer-lasting effects. These various factors can cause significant delays and increase the costs of global trade. Additionally, they may cause a shift in trade patterns as companies involved in trade may seek alternatives or substitutes to increase reliability of transporting goods (OECD, 2016). According to reports prepared by the IPCC (2014), climate change will affect all forms of transport relevant to international trade, “including seaborne transportation, land-based transport, and aviation.” These current and future impacts of climate change are interesting, however, the fact that the effects of climate change on trade and transport has not properly been applied and analysed in a historic setting is surprising, especially in the 19th century where accurate data is readily available.
Another crucial factor to consider when looking at global and country specific trade levels, is that these natural disasters and adverse climatic conditions have, and will continue to have, a stronger negative impact on smaller, developing countries than larger, developed countries. This is due to the fact that small, developing countries cannot absorb the impact of these occurrences as efficiently due to weaker institutions and ultimately less capital to spend on disaster relief. Developing countries often also have a higher reliance on the export of agricultural products which generally are the most affected by weather changes. Exports of smaller developing countries fall by about 22% as a result of a natural disaster, however the exports of larger developed countries are not as significantly affected (Da Silva and Cernat, 2012).
2. Literature Review
This section provides a review of prior research of trade level determinants during the nineteenth century.
2.1 Determinants of trade levels
The study of the various factors that influence trade levels globally has long been an area of interest to researchers such as Williamson (2008) and Hynes, Jacks and O’Rourke (2012). These studies focus mainly on the impact and influence of trade costs, tariffs and exchange rates on overall trade levels. These factors have been proven time and time again to have significant and lasting impacts on the level of international trade. Academics including Magerman, Studnicka and Van Hove (2016) have focused on the impact that distance from neighbouring countries has on trade, finding that geographical location is a major determinant of export levels.
While it has been widely acknowledged that natural disasters generally have a negative impact on all aspects of trade, there is a significant lack of quantitative data regarding these studies, especially when looking at global economic history. However, Oh (2009) looked at the effects of natural disasters and political risk on bilateral trade from the 1950s, believing that the two variables needed to be analysed together. From the trade gravity model applied, a negative relationship was discovered, meaning that an increase in political risk and natural disasters led to a decrease in trade. Da Silva and Cernat (2012) looked at the impact of natural disasters on developing countries versus developed countries’ export levels. The authors used a simple gravity regression, in their attempts to discover whether there is significant evidence to prove that developing countries’ trade flows are most affected by natural disasters. The results show that both developed and developing countries will be negatively affected by natural disasters, as expected, however, the impact is significantly larger for developing countries and endures for a greater length of time.
Determinants of global trade levels, or even trade levels within or among continents has widely been studied. Multiple key factors have been identified and empirically proven to have an impact on trade. Important variables which have significant effects on trade levels include tariffs, transport costs, and the availability of infrastructure for transport. Key variables include employment rates, whether specific to any given country or globally, as well as real wage rates. Industrialisation clearly had a huge impact on trade levels with an increase in shipping speeds, making trade faster, cheaper and more efficient. The industrial revolution also spurred globalisation, allowing for increased international trade between continents and polities.
Exports by continent, constant prices, 1830-1913, log scale
(Source: Frederico and Tena-Junguito, 2016.)
2.2 The omission of natural disasters and climatic conditions as an explanatory variable
The omission of the occurrence of natural disasters and adverse weather conditions as a determinant of trade in the early 19th century from previous studies is puzzling. Researchers such as Skidmore and Toya (2002) and Gassebner, Keck and Teh (2006), who did look at trade and natural disasters, failed to look at the overall impact that climate and natural disasters together can have on every aspect of determining the level of trade, from crop growth, mining or production through the supply chain to transportation. Moreover, these researchers have looked at the effects in today’s world, rather than examining history, with the earliest study going back to 1964. There is a significant difference between the impact that natural disasters have had in the last 50 years and that in the 19th century. This comparison has not yet been explored by researchers
Gessebner, Keck and Teh (2006), found that the impact of these conditions on trade is definitely significant but also dependent on countries size, institution strength, whether or not they are a democracy and if they are an importing or exporting country. However, there were only 13 democratic countries in 1939 (Max Roser, 2018), in comparison to the 35 listed in their study. Comparisons between historical and modern results need to be made throughout economics in order to capture possibility of changing variables and their fluctuating impacts on trade levels.
The number of disasters per year was taken from EM – DAT, the Natural disasters database, for the purpose of this study and for past studies including Gessebner, Keck and Teh (2006). According to this database, certain criteria must be reached in order for natural disasters to be recorded. These criteria include; “ten or more people reported killed; hundred or more people reported affected; a declaration of a state of emergency or; a call for international assistance” (em-dat.be, 2018). However, this does not allow for discrimination between the actual impact of each disasters. For example, the number of natural disasters per year does not differentiate between 5 small natural disasters a year, with 10 deaths each, and one large disaster with hundreds of deaths and a large impact on the the level of trade. This could seriously affect the results achieved from these studies. Therefore, the sum of the magnitude of natural disasters should really be looked at in the year, possibly measured by people affected, or by the number of deaths.
3. Objective
Using a panel data model, this research proposal aims to answer the following question;
“Was the occurrence of natural disasters and changing climatic conditions a significant determinant of the level of global trade from 1800 – 1939?”
The number of natural disasters continued to increase during the 19th century and into the 20th century, and the effects of climate change and adverse weather conditions only became more severe. While these variables are a topic of much discussion in the world today when analysing trade levels, they have yet to be applied to the trade levels in the 19th century. There are a multitude of factors that have been proven to influence and impact trade levels as mentioned in section 2.1, however it is clear that they are not the sole factors at play. This research proposal hopes to fill a niche in the existing literature by explicitly investigating if the frequency of both natural disasters and fluctuating climatic conditions significantly influenced the level of trade in the 19th an early 20th century and, if so, by how much.
This study aims to improve upon existing research and to bring a new dimension to this subject by using climatic conditions or weather data in addition to that of natural disasters. This methodology hopes to overcome or make up for the problem that has effected this area of research until now, that of a lack of data for natural disasters before 1900. However, data for the frequency and level of adverse weather conditions will be used, with some possible correlations between the occurrence of natural disasters and that of negative climatic conditions. Therefore, this research proposal aims to bring a fresh approach to the existing field of study.
4. Data
This section outlines the details of the data required for the model.
Variable Source Geographic coverage Period covered Frequency
Global Trade Levels Frederico – Tena World Trade Historical Database. Entire world 1800-1939 Yearly
Independent Variable 1: Natural Disaster Frequency EM-DAT | The International Disasters Database Entire world 1900 – 1939 Yearly
Independent Variable 2: Climate Conditions NOAA | National Centres for Environmental Information Entire world 1800 -1939 Yearly
Global wage rates How Was Life? Global Well-being since 1820 Entire world, split by continent 1820 – 1939 Yearly
Global Employment rates Labour Force and Employment, 1800 – 1960. Split by country 1800 – 1939 Yearly
Trade Costs Trade Costs, 1870 – 2000. Split by country 1870 -1913 Yearly
Democracy Introducing the Historical Varieties of Democracy Dataset: Political Institutions in the Long 19th Century Split by country 1789 – 1920 Yearly
4.1 Dependent Variable Data
Y: The level of global trade during the 19th century –
Data on the level of global trade during the 19th century has been taken from the Frederico – Tena World Historical Database. This database provides comprehensive global trade figures from 1800 – 1939 (Appendix A). The data and trade statistics are also provided split by import levels and export levels, in addition to separating it by continent for the purpose of this study.
4.2 Key Independent Variable Data
X1: Natural Disaster Frequency –
Data on the number of disasters per year was taken from EM-DAT, The International Disasters Database. It provides data on the occurrence, as well as the effects of mass disasters globally (Appendix B). This database shows the number of natural disasters per year, separated by continent, as well as the number of deaths caused by the event. However, these figures only go back until 1900. Unfortunately, the lack of data from before 1900 ensures that a regression could only be run on analysing the effects on natural disasters on trade from 1900 – 1939. The prediction for this model is that this independent variable would have a negative coefficient. This would mean that an increase in the frequency of natural disasters would lead to a decrease in trade.
X2: Rainfall and Temperature Data –
Figures for global weather was taken from the World Weather Records dataset, from the National Centres for Environmental Information, NOAA (Appendix C). Data was available from 1755 – 1950, with sparse figures for certain continents until the start of the 1800s. The coefficient for this variable is unknown, difficult to predict and perhaps dependent on the region affected. For example, lack of rain could benefit crops that are meant to be grown in Mediterranean climates (such as olives and citrus fruits), but lead to the failure of crops that require water. Ireland’s experience with the potato crop failures of 1845-1849 is a striking example of the impact of weather on an economy and its people.
4.3 Control Variables Data
X3: Global Wage Rates
Data for global wages rates were taken from ‘Real Wages Since 1820’ (de Zwart et al, 2014). Data are provided for a large number of countries from 1820, allowing for their use in the model for this paper. However, one criticism of this dataset is that it does not include all of the countries in the world, meaning that the figures could be off. A positive coefficient is expected as according to Paresh and Russell (2009), as wage rates increase, so does productivity, even if slight. This means that more goods for export would be produced and therefore, that trade levels would rise.
X4: Global Employment Rates
Global Employment rates have been taken from ‘Labour Force and Employment, 1800 – 1960' (Stanley Lebergott, 1966). This data is provided by country. As this does not provide complete coverage for a global model, other sources were also used, such as ‘The International Comparison of Unemployment rates’ (Gallenson and Zellner, 1957). A positive coefficient is predicted for the level of employment rates as the more workers employed, the more crops and goods produced and therefore the higher the trade levels for any given period of time.
X5: Trade Costs
Data for global transport costs during the later 19th century have been taken from ‘Trade Costs, 1870 -2000’. There are multiple factors that affect transport costs, including the presence of tariffs, the cost of shipping or transporting the goods (by ship, airplane or train) and the cost of implementing the infrastructure needed for transport. A negative coefficient is expected as the higher the price of trade, the less likely countries are to trade.
X6: Democracy
Data for the number of democratic countries was taken from ‘Introducing the Historical Varieties of Democracy Dataset: Political Institutions in the Long 19th Century’ (Knutsen et al, 2018). A positive coefficient is anticipated, as democratic countries normally have a ‘positive link’ with trade levels in comparison with other less democratic countries, according to Abeliansky and Krenz (2015).
5. Approach
In order to address the research question that this paper puts forward, a panel regression model will be used for the period 1820 – 1939. The model is grounded in the work of Skidmore and Toya (2002), but with the fundamental addition of a variable that captures the occurrence of adverse weather conditions during this period in addition to natural disasters. This is a key variable of interest in this research paper.
Regression Model:
〖Trade〗_it= β_0+ β_1 〖wage rates〗_it+β_2 〖 trade costs〗_it+ β_3 〖employment rates〗_it+ β_(4 ) 〖Democracy〗_it+ β_5 〖Natural disasters〗_it+ β_6 〖Climatic conditions〗_it+ μ_i+U_it
This is a static regression model where subscript i donates continents and t donates time period, 1800,…,1939. The error term is given by the sum of an individual error term that does not change across time and an idiosyncratic error term. μ_i captures unobserved individual characteristics that do not change across time.
However, there are three different ways of assessing panel data; pooled OLS, fixed effect and random effect. This study looks to identify the model most suitable for this specific data. The pooled ordinary least squares estimation is the OLS of the above regression model using the pooled sample and ignoring that the error term is a composite error term. The fixed effect estimation is equivalent to the OLS estimation of the model rewritten transforming each of the variables, i.e. in deviations from the individual mean. The random effect model can be estimated using the Generalized Least Squares.
Basically what this means is that there is a need to identify weaknesses in the data and to then choose the model that will address those weaknesses. As there may be a specific error term in the regression, it is expected that a fixed effects panel regression would give the most efficient estimators. However, it is impossible to tell without running tests to figure this out.
To choose between pooled OLS and random or fixed effect estimations, it is possible to use Breusch-Pagan Lagrange Multiplier. To choose between random or fixed effect estimation it is possible to use a Hausman test. Regardless, once pooled OLS has been ruled out, both (random and fixed) are consistent estimators, meaning that they have insignificant variations as the sample size grows larger. However, it is expected that fixed effects would likely be preferable as there could be qualities unique and integral to each continent that need consideration. These could be difficult to pin-down, definitely difficult to measure, and therefore impossible to control for. Using the fixed effects model would remove that.
6. Conclusion
In conclusion, this research proposal aims to offer a way of tackling an issue that has long been overlooked by economists, due in part perhaps, to a lack of data. This paper seeks to expand our understanding of what impacted global trade levels during the nineteenth century by investigating whether or not the occurrence of natural disasters and adverse climatic conditions was a significant determinant of the trade level.
References
Abeliansky, A. and Krenz, A. (2015). Democracy and International Trade: Differential Effects from a Panel Quantile Regression Framework. SSRN Electronic Journal.
Da Silva, J. and Cernat, L. (2012). Coping with loss: The impact of natural disasters on developing countries' trade flows. European Commission.
Data.nodc.noaa.gov. (2018). World Weather Records. [online] Available at: https://data.nodc.noaa.gov/cgi-bin/iso?id=gov.noaa.ncdc:C00160 [Accessed 25 Nov. 2018].
De Zwart, P., van Leeuwen, B. and van Leeuwen-Li, J. (2014). Real wages since 1820.
Emdat.be. (2018). EM-DAT | The International Disasters Database. [online] Available at: https://www.emdat.be/emdat_db/ [Accessed 25 Nov. 2018].
Emdat.be. (2018). Explanatory Notes | EM-DAT. [online] Available at: https://www.emdat.be/explanatory-notes [Accessed 28 Nov. 2018].
European Commission (2012). Trade, growth and development. Tailoring trade and investment policy for those countries most in need. Brussels.
Galenson, W. and Zellner, A. (1957). International Comparison of Unemployment Rates. National Bureau of Economic Research, pp.p. 439 – 584.
Hynes, W., Jacks, D. and O'rourke, K. (2012). Commodity market disintegration in the interwar period. European Review of Economic History, 16(2), pp.119-143.
IPCC (2014). Climate Change 2014 Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.
Jacks, D., Meissner, C. and Novy, D. (2008). Trade Costs, 1870–2000. American Economic Review, 98(2), pp.529-534.
Knutsen, C., Teorell, J., Cornell, A., Gerring, J., Gjerllw, H., Skaaning, S., Wig, T., Ziblatt, D., Marquardt, K., Pemstein, D. and Seim, B. (2018). Introducing the Historical Varieties of Democracy Dataset: Political Institutions in the Long 19th Century. SSRN Electronic Journal, 65.
Magerman, G., Studnicka, Z. and Van Hove, J. (2016). Distance and Border Effects in International Trade: A Comparison of Estimation Methods. Economics: The Open-Access, Open-Assessment E-Journal, 10, pp.1-31.
Narayan, P. and Smyth, R. (2009). The effect of inflation and real wages on productivity: new evidence from a panel of G7 countries. Applied Economics, 41(10), pp.1285-1291.
Oh, C. and Reuveny, R. (2010). Climatic natural disasters, political risk, and international trade. Global Environmental Change, 20(2), pp.243-254.
Roser, M. (2018). Democracy. [online] Our World in Data. Available at: https://ourworldindata.org/democracy [Accessed 30 Nov. 2018].
Skidmore, M. and Toya, H. (2002). DO NATURAL DISASTERS PROMOTE LONG-RUN GROWTH?. Economic Inquiry, 40(4), pp.664-687.
Appendix
Appendix A
Appendix B
Appendix C