1. Introduction
Despite modern technologies allowing scientists to forecast the likely routes and strengths of hurricanes with rising precision, thus allowing some preparation, the devastation they scatter during their passage in the Caribbean can still be extremely costly (Strobl and Walsh 2008; Deryugina 2011; Strobl 2011).1 In the nineteenth century none of these technologies existed. Rather, hurricanes arrived practically unannounced in the region and were a unique challenge for British colonists. These storms were entirely new to them and were the most feared aspect of colonial life, destroying entire crops, plantations, and towns in a single blow and disrupting shipping and trade, which resulted in major economic losses (Mulcahy 2006). Moreover, the destruction they caused at this moment in history had particular implications for European countries, as the Caribbean islands served as the main source of a commodity increasingly in fashion for European palates: sugar (Hersh and Voth 2009). Hence, hurricanes, because they were able to reduce the supply of sugar in a matter of hours in places like Barbados, Jamaica, and Antigua, could decide the fortunes of the planters.
This paper aims to understand the impact of these exceptional climatic episodes on sugar prices in the first half of the nineteenth century in the United Kingdom. Numerous contemporary accounts, through correspondences, newspapers in the Caribbean, and in England, report the difficult situation of the Caribbean plantations after the passage of these storms.2 Sugarcane is traditionally planted from November to May. The Atlantic hurricane season occurs from June to November but can begin as early as May, and hurricanes may even occur outside of these months. The force of winds and flooding from these storms can annihilate the harvest of the year and seriously endanger the following year’s harvest.3 Additionally, the strong winds and rainfall destroyed equipment, storage facilities, and ports, thereby disrupting supplies and shipping from the colonies to England, causing a sugar shortage in the London sugar market. Based on historical accounts hurricanes caused sugarcane output to fall by 50% or more for at least 1 year, while in the case of especially violent storms the effect lasted for a second year (Mulcahy 2006). This decline in supply had important consequences for the price of sugar in England.
Planters tried to take some measures to protect their activities by harvesting sugarcanes, processing and sending sugar offshore before the brunt of the hurricane season, reinforcing the buildings, and keeping sufficient reserves of construction material and food. However, the strongest hurricanes took advantage of even the best prepared planters and were said to be more destructive than wars (Mulcahy 2004). Further, higher priced sugar did not necessarily offset losses caused by hurricanes and west Indian planters were often unable to salvage enough sugar from a storm-damaged crop to provide income and cover expenses. Moreover, the cost for plantation supplies rose along with the prices for staples since the storms destroyed not only sugar but also food and building material. Our hypothesis here is that the news of hurricanes in the Caribbean arriving at British harbors is likely to have driven up sugar prices mainly because the market anticipated that sugar supply in the Caribbean colonies would drop dramatically in the near future. Contemporaneous accounts in the newspaper tend to accredit this hypothesis.4 Mulcahy (2006, p. 75) argues that the “potential disruption from a hurricane and its effects on prices meant that correspondents in England anxiously awaited word from the colonies.” For example, he noted that the “London sugar market fluctuated widely upon news of a hurricane in the Leeward Islands or Jamaica (the largest producers of muscovado sugar), often producing a rise of between two and ten shillings per hundredweight” (Mulcahy 2006, p. 75).5 Timely information meant significant profits for merchants selling sugar to metropolitan grocers and refiners. In this paper, we look for statistical evidence of this phenomenon.
The study of hurricanes in the economic literature is fairly recent and mainly empirical, looking primarily at the impact on gross domestic product (GDP) growth (Loayza et al. 2012; Hochrainer 2009; Noy 2009; Noy and Nualsri 2007; Raddatz 2007). These studies are generally limited to beyond the 1950s, where statistical data on economic activity across countries over time are readily available. Nevertheless, there are some historical studies. For example, Mohan and Strobl (2013) show that hurricanes had a large, negative impact on sugar exports in Caribbean colonies from 1700 to 1960. Similarly, Pielke et al. (2008) calculate the total direct losses due to hurricanes for the United States for the period 1900–2005. Additionally, Boustan et al. (2012) investigate the impact of hurricanes among other natural disasters on migration in the United States in the 1920s–30s. This paper contributes to this growing literature on the economic consequences of natural disasters (Lamb 1995; Worthington and Valadkhani 2004; Ewing et al. 2006; Blau et al. 2008; Fink et al. 2010; Strobl 2011; Fink and Fink 2013), from a historical perspective.
The paper investigates the impact of hurricanes in Caribbean sugar colonies on London sugar prices in the first half of the nineteenth century using historical data and the role played by transportation technologies in the ease of information transmission through a process of price discovery, that is, the determination of price at a given time in the marketplace through the interactions of buyers and sellers (Hasbrouck 1995). While it is well documented in historical accounts that hurricanes were pivotal events in Caribbean development during the colonial era, there is hardly any empirical evidence of this. This paper provides a better understanding of the economic effects of hurricanes beyond modern times and fills an important knowledge gap in the role of these storms in influencing the economy of both the colonies and the colonists. It allows for an insight into an aspect of economic history that constituted an important part of life. Further, while economic historians have investigated the consequences of natural disasters (Bridbury 1973; Ó Gráda and O’Rourke 1997; Odell and Weidenmier 2004; Pereira 2009; Boustan et al. 2012; Slavin 2012; Bai and Kung 2014; Stone 2014), they are generally interested in the effect of a single event, whereas hurricanes are annual phenomena and are treated as such by this study.
The paper looks at the consequences of natural disasters on globalization through international trade and the role played by technology in the transmission of information. Today, sugar and other agriculture products remain important exports from the Caribbean, since these crops were first introduced during colonization and continue to be affected by hurricanes. Moreover, hurricanes have been increasing in frequency and severity given the effects of climate change, while information transmission is now almost instantaneous with advances in information communications technology. These changes only serve to amplify price instability caused by hurricanes in the Caribbean in modern times.
Our econometric results suggest a significant rise in sugar prices in the London market due to the news of hurricanes in Caribbean colonies. Moreover, the reaction time of the market to hurricane news has decreased over our period of analysis, where this change might be through the technological innovations marking this era through price discovery. More specifically, technological progress in transport is likely to have reduced the time required for information to cross the Atlantic, making markets more reactive to the news of hurricanes.
The remainder of this paper is organized as follows. Section 2 describes the workings of the sugar market and its relationship with hurricanes. Section 3 presents our methodology and our results. Section 4 concludes.
2. Caribbean sugar in the United Kingdom and hurricanes in the first half of the nineteenth century
Because of its economic importance and its relationship with slavery at the beginning of the modern era, many scholars have written about sugar (Engerman 1984; Solow 1985). There is thus an abundance of information about the changes in its production process, the wealth it provided for a small group of planters in the Caribbean and throughout all the economy in England (Curtin 1954; Parker 2012), and the changing habits in Europe regarding its consumption (Shammas 1984; Hersh and Voth 2009). There is, however, much less written on the impact of hurricanes on sugar production and exports from the British colonies and prices.
The workings of the sugar market and the relationship between hurricanes and the transmission of information and prices may be explained using the concept of price discovery, borrowed from the financial literature. The news of a hurricane strike in Caribbean colonies could cause a price increase in England through the process of price discovery. Price discovery is the determination of price at a given time in the marketplace through the interactions of buyers and sellers. It is concerned with how information flows into the market, how the market responds to this flow of information, and how quickly this information is incorporated into prices. When new information of a hurricane strike in the Caribbean colonies arrives in England, this affects market conditions, causing changes to anticipated supply and triggering an increase in price. This transmission of information on hurricane strikes in the Caribbean is affected by the information communications technology present, which would affect timeliness, reliability, and frequency of information received, consequently impacting sugar prices in London.
a. Sugar imports into the United Kingdom
The period we focus on witnesses dramatic changes for the old production centers of sugar. For more than a century, the West Indies colonies (first Barbados, surpassed in the second half of the eighteenth century by Jamaica) dominated sugar production for the British market. Nevertheless, the region did have some competition. At the beginning of the nineteenth century, production started to expand to Mauritius and India. The tariff system that was advantaging the Caribbean was dismantled, first, in 1825, when the tariffs on Mauritius muscovado sugar were lowered to the level of the West Indies and then again in 1836 when the tariffs on the East Indies muscovado sugar were lowered (House of Commons, custom tariffs of the United Kingdom from 1800 to 1897; House of Commons 1898). Additionally, slavery ended in the British Empire on the 1 August 1834 with consequences for the cost of production of sugar. As can be seen from Fig. 1, sugar imports to the United Kingdom from the West Indies peaked just before abolition and decreased substantially afterward. The decline was compensated by the increase in production in Mauritius and east India between 1825 and the end of our period. The Caribbean colonies, however, remained the primary supplier of muscovado sugar to England over the period 1815–41.
Sugar import in the United Kingdom from 1815 to 1841.
Citation: Weather, Climate, and Society 9, 4; 10.1175/WCAS-D-17-0015.1
Within the context of these developments, our interest lies in the impact of hurricanes in Caribbean colonies on the London sugar market. More specifically, given that the West Indies was still the principal center of production over our period of analysis, we expect an increase in the price of sugar in the London market due to hurricane news.
b. Brown muscovado price in London
It is important to note that the sugarcane planters were not actually exporting the end product, that is, white sugar, consumed by Europeans, but instead were shipping brown sugar, issued from a first step transformation of sugarcane juice, to refiners in Europe. They had several reasons for doing so: white sugar could alter during the long voyage between the Americas and Europe; refining operations required large amounts of fuel, which was lacking in the sugar colonies; and sugar planters were constrained by the tariffs set out by the metropolitan governments if they tried to export the end product, where the tariffs were intended to protect the refiners located in London. The London refiners justified their advantaged position to the government by arguing that sugar transformation was a process providing employment to many Englishmen (Galloway 1989).
We compiled weekly sugar prices from the various House of Commons parliamentary papers scanned and housed at the House of Lords Library, which are based on prices reported in the gazette over the years (House of Commons 1831). The prices constitute the average weekly prices of brown muscovado, exclusive of duty, sold on the London market. All prices are expressed in shillings throughout our analysis. A gap exists in the reporting of the price of brown muscovado in the gazette between the second week of December 1826 and the last week of July 1828.
Figure 2 demonstrates the wide variation of the price of brown muscovado over our time period, reaching a maximum of 78.4 shillings in the first week of 1815 and a low point of 21.8 shillings in the third week of November 1831. One may also want to note that during this period the muscovado price shows a downward trend. However, the decline in price occurs mainly during the first 9 years after the end of the Napoleonic Wars. After this the price seems to fluctuate more around its period average price, 39.9 shillings, with a standard deviation of 9.64. From the end of 1831 to the end of our period, the price of brown muscovado is on an upward trend. In this regard, the 1898 report on the customs tariffs of the United Kingdom from 1800 to 1897, presented to the Houses of Parliament, attributed the sharp fall in price occurring after 1819 to the competition of foreign sugar from the continental European market, and the increase in the price of sugar at the end of our period is argued to be due to the decrease in supply from the West Indies (House of Commons, custom tariffs of the United Kingdom from 1800 to 1897; House of Commons 1898). It has to be noticed that a consequent part of the raw sugar imported to the United Kingdom after the Napoleonic War was refined and reexported to continental Europe.
Brown muscovado price in London.
Citation: Weather, Climate, and Society 9, 4; 10.1175/WCAS-D-17-0015.1
c. Hurricanes in the British Caribbean
A key element in our study is the information about the number and strength of hurricanes occurring in the Caribbean region in the first half of the nineteenth century. In this regard, we identify the tropical cyclone events affecting Caribbean colonies/countries for the period 1815–41 using information from Chenoweth (2006). More specifically, Chenoweth (2006) meticulously went through Poey’s (1855) well-known and widely used list of known historical tropical cyclones in the region, verifying and validating these with other sources, to then construct an updated and corrected chronology of 383 unique tropical storms striking the Caribbean over the period 1700 until 1855. In the 26 years we cover, 37 hurricanes affecting the English Caribbean colonies6 are recorded. Importantly for the purposes here, Chenoweth (2006) identifies for each storm the localities of colonies/countries affected and estimates according to their description of whether the cyclone was likely to have been of tropical storm or hurricane strength.7 We thus use this list to identify the number of hurricane strength tropical storms affecting colonies in each week for the period 1815–41. One may note that the quality of this list, as well as all accepted tropical storms in the Chenoweth data, in terms of providing an accurate picture of all potentially damaging cyclones affecting localities in the Caribbean region will ultimately depend on the quality and availability of the primary sources on which they are based.
Our hurricane index, weighted according to the value of the previous year’s exports of those islands affected, has a mean, standard deviation, maximum, and minimum of 0.004, 0.040, 0.479, and 0 for all weeks, respectively, while for hurricane-affected weeks the equivalent values are 0.184, 0.170, 0.479, and 0.0002, respectively. The index has a simple intuitive interpretation. For instance, the expected share of sugar exports affected by hurricanes was 0.4% of the Caribbean export in any week. The weeks when a hurricane did strike, on average 18.4% of Caribbean export was affected, while the most destructive event was a week in which nearly 50% of all sugar export in the Caribbean was potentially affected by a hurricane.
3. Results
Our main task is to investigate how the price of sugar in London reacted to hurricane news from the British Caribbean colonies. Hurricane news would act as a type of market shock signaling a disruption in supply and thereby affecting prices through price discovery. How the information flows to the market would determine how quickly the news is incorporated into prices and how the market responds. To this end we use our price and hurricane index data in an autoregressive distributed lag model, where details underlying our statistical model choice are provided in the appendix. To identify the nature of any price effect of hurricane news, we experiment with different lags on hurricanes given the likely delayed response between a hurricane strike and sugar price. More specifically, information flows into the market during this era had to travel by sea either through mail packet ships or by merchantmen. In terms of how quickly information flows, we check reports for each hurricane in our database in English newspapers for information on when news on a hurricane strike would reach London (Fig. 3).
News of hurricanes in the West Indies in London.
Citation: Weather, Climate, and Society 9, 4; 10.1175/WCAS-D-17-0015.1
During our period of study, 1815 to 1841, the West Indies colonies were the principal center of production of sugar imported in England. This plays a significant role in the extent to which sugar prices in London respond to hurricane news from the British Caribbean colonies, since a shock to supply would have a large anticipated impact. Only from the 1840s onward (after our period of study) did sugar imported into England from East Asia and Mauritius and other foreign supply begin to pick up and present some level of competition to the Caribbean British colonies (as seen in Fig. 1). Nonetheless, while other territories may have been exporting sugar to the United Kingdom, these historical data are difficult, if not impossible, to obtain. Moreover, based on historical accounts the West Indies colonies were the principal exporters of sugar to England during our period of study.
Table 1 provides our estimation results, where the first column uses the whole sample of observations. As can be seen, both the fourth and tenth lag (L4 and L10) have a significant and positive impact on price. To further investigate this result, we divide our sample into two subperiods using the year 1827. Our choice of this year as the cutoff point was because 26 April 1827 was the date of the first voyage of a steam paddle ship from Europe to South America via the Caribbean. More specifically, the steamship, the Curaçao, made the voyage between Rotterdam to Paramaribo in Suriname in 28 days, operating mainly via steam (Bradlee 1925). This changed how quickly news could travel. Steamship lines started to cross the North Atlantic more regularly in 1838 (Kludas 1999). For instance, in 1841 the Royal Mail Steam Packet Company Line was inaugurated to travel between England and the Caribbean. Between 1827 and 1841, we also know of a steamship packet, the Spitfire, being employed on intra-Caribbean voyages between the Caribbean and Plymouth, although it was reported damaged after a hurricane in the Caribbean.9 At the same time of the gradual introduction of steamship lines on Caribbean–United Kingdom voyages, there was also an increase in the organization of northern transatlantic trade into shipping. For instance, at the end of 1825 a weekly connection existed between New York and Liverpool. Importantly, most of the ships on these lines when traveling east were crossing on average in 24 days (Laakso 2006). While greater strength in the connection between the United States and England does not necessarily imply a quicker exchange of information between the Caribbean and England, we found evidence in newspaper reports that at least on two occurrences information about hurricanes in the Caribbean arrived first in the United Kingdom, passing by the United States.10 Arguably these changes, that is, the competition and complementarity between sail and steam—the steam ships being used more on short distance journeys and the sail vessels still widely used on the transatlantic journey—and the better organization of shipping lines through packet lines, might have acted to decrease the number days required for information about hurricanes in the Caribbean reaching London.
Weekly results, 1815–41. Notes that 1) standard errors are given in parentheses, and time and week dummies are included but not reported.
We present our results for the period 1815–27 in column 2 of Table 1 and in column 3 for the period 1827–41. As can be seen, there is evidence that hurricanes affected sugar prices in London for both subperiods. In the earlier period (Model 2) L10 is the only significant effect, whereas L4 is the only significant effect in the later period (Model 3). This suggests that while in the earlier period, 1815–27, hurricanes caused an increase in prices 10 weeks after their occurrence, during the later period, 1827–41, price increases were felt only 1 month after a storm struck the Caribbean. More specifically, as outlined above, after 1827 more advanced ships were used in commercial transatlantic transport, while at the same time shipping lines became better organized so that a hurricane tended to have a positive effect on the sugar price earlier, in considerably less time after its occurrence. The market incorporated news of a hurricane strike in British colonies into sugar prices in London more quickly.11, 12
To ensure that our results for the two periods are not driven by single events, we also dropped individual events and reestimated the specification. The resultant coefficients and 90% confidence bands for each period, where events were dropped in order of their strengths, are shown for 1815–27 and 1827–41 in Figs. B1 and B2 in appendix B, respectively. As can be seen, generally the coefficients do not change much quantitatively and remain significant, no matter which event is dropped.
One can also consider the economic significance of our results. For example, the coefficient in column 2 suggests that a hurricane affecting 18% of Caribbean sugar export (equivalent to a one standard deviation of the hurricane index between 1815 and 1827) increased sugar prices by 5.8 percentage points. After 1827, a hurricane affecting 9.4% of Caribbean sugar export (equivalent to a one standard deviation of the hurricane index between 1827 and 1841) increased sugar prices by 9.8%. On average, hurricanes would have been expected to increase prices by 9.34 and 10.21 percentage points, respectively, while the largest observed price hikes due to these storms were 16.21 and 50, respectively. Considering that the absolute average weekly price change over our sample period was 2.26% (2.40% in the first period and 2.14% in the second), the impact of the news of hurricanes in the Caribbean is arguably not negligible. On average, hurricanes would increase the price of 4.91 shillings in our first period and 3.96 shillings in our second period. These numbers are well in the range of the 2 to 10 shillings suggested by Mulcahy (2006). It is noticeable that in the second period, the impact of the hurricanes seems stronger. In this period of expansion of sugar production in east India, the uncertainty of the growth of this production and the expectation that production in the West Indies Caribbean would drop with the end of slavery might have made the market more sensitive to the news of hurricanes coming from the Caribbean. At the same time, during this last period, the hurricane index was on average less important than in the preceding one, meaning that fewer production centers were strongly affected by the hurricanes.
The welfare significance of the price increase of sugar in London caused by news of a hurricane strike in the British West Indies colonies may be highlighted when compared to relative prices of other commodities consumed during this time period. According to Allen (2001), during the nineteenth century sugar was the most expensive source of calories in the typical British basket of goods. Measured in grams of silver per 1000 calories, the following prices hold: bread (0.283), beans (0.424), meat (0.885), butter (0.476), cheese (0.758), eggs (1.260), olive oil (0.926), beer (1.104), wine (1.135), and sugar (2.317). Additionally, Hersh and Voth (2009) show that in the early 1800s, per capita, sugar was the most important commodity imported into England, where 23 lb of sugar was imported compared to 2 lb of tea, 1 lb of tobacco, and 0.1 lb of coffee. Furthermore, sugar was consumed by all strata of society, including the lowest income households. Sugar also constituted a large part of the household expenditure, where the average Englishman spent 15% or more of his income on sugar and accompanying tea (Hersh and Voth 2009). Therefore, the slightest price increase caused by news of a hurricane in the Caribbean negatively affected all in London, especially the poor during our period of study.
While intuitively plausible, our results nonetheless should be taken with some caution. More specifically, our newspaper data indicate that the information about hurricanes in our first time period arrived on average after 54 days, that is, 8 weeks following a hurricane. Our statistical finding of 10 weeks might be driven by the hurricanes for which the information took more time to arrive. Indeed the slightly longer average effect that our econometric analysis implies might be due to the possibility that some hurricanes, particularly the stronger ones, may have also delayed ships traveling to the United Kingdom.
For our second time period, our econometric results point to a delay of 4 weeks, whereas our news data indicate an average time of 45 days. Even if the results are possibly mostly driven by the news of hurricanes arriving the most promptly from the Caribbean, this difference might indicate that actors on the market may have received private information before it was diffused to the general public. To try to test for this, we also used the date when a hurricane was reported in the United Kingdom newspaper instead of the relative date of when a hurricane actually occurred. However, we found no significant result (even if the sign of the coefficient was in accordance with a positive impact in the week of the news reception). This might be due to the fewer number of observations that we have regarding the reports in the news. To further investigate this issue we also, in Fig. 4, present descriptive statistics of the evolution of the price difference between the current and preceding weeks around the time the news of a hurricane is received in England (±3 weeks).
Price evolution around the date of arrival of the news of a hurricane in England.
Citation: Weather, Climate, and Society 9, 4; 10.1175/WCAS-D-17-0015.1
As we lack price data for the year 1827 and some hurricanes occurring in short sequences in the Caribbean were reported in the same week, we present the results for 12 individual, isolated events. Accordingly, we have at the beginning of our period (before 1827) a peak in price corresponding either to the week the news is received or shortly afterward. After 1827, the pattern is less clear, where only in the case of the 1834 hurricane does a peak price seem to be synchronizing with the reception of the news of a hurricane. For the cases of the hurricanes of 1830, 1835, and 1838, a peak in price seems to occur 1 week before we record the news in the newspaper. See the appendix for additional robustness tests and goodness of fit model specification tests.
4. Conclusions
Hurricanes were a unique challenge for British colonists in the Caribbean. These storms were entirely new to them and were the most feared aspect of colonial life, destroying entire crops, plantations, and towns in a single blow and disrupting shipping and trade, which resulted in major economic losses. We investigated whether hurricanes striking the English Caribbean affected sugar prices in the London market. We find that before 1827, when the shipping lines were still in their infancy and more advanced steamships had not yet been widely used in commercial transatlantic transportation, it took on average 10 weeks before news of a hurricane in the Caribbean affected London sugar prices. After 1827, the 10 weeks effect was reduced to 4 weeks. One possible explanation for this change could be the introduction of more rapid nautical vehicles during that time, which greatly reduced the sugar-concerned information from the Caribbean islands to Great Britain.
This study contributes to filling an important gap in the literature. The study of the economic impact of hurricanes is fairly recent and limited mainly to GDP growth, while the impact on other economic outcomes such as commodity prices is generally ignored. The historical study of hurricanes has also been overlooked. Also, the study of hurricanes during colonization in the Caribbean provides an insight into an aspect of history that played an important part of life both in the region and the “mother land.” More generally, this paper provides new insights into the historical impact of natural disasters on sugar prices through price discovery and the consequences of globalization, particularly international trade and the role of technology for the transmission of information. This has important lessons for today, where modern changes only serve to amplify the negative consequences of natural disasters, especially in small island developing countries.
APPENDIX A
Price and Hurricane Time Series
We begin our analysis with the inspection of the time plots of sugar price and the weighted hurricane destruction index for stationarity and cointegration. Figure A1 at least visually suggests that the weekly prices are nonstationary, while our hurricane index is likely to be stationary. Further, price and hurricane destruction do not seem to move together, indicating no cointegration. Moreover, at least sugar prices are likely to be skewed.
Plot of annual British sugar price and weighted hurricane destruction index, 1815–41.
Citation: Weather, Climate, and Society 9, 4; 10.1175/WCAS-D-17-0015.1
Next, we formally test for skewness and stationarity. The skewness–kurtosis test, which is based on the Jarque–Bera statistic, indicates that the price data are skewed (see Table A1). We thus log prices to take account of this.
Stationarity and skewness tests: annual data. Notes the 90% level of significance with no intercept and no trend; the critical values of the upper and lower bound test are calculated by Pesaran et al. (1996). The term I( ) indicates the order of integration.
In testing for stationarity we implement the Phillips–Perron and Dickey–Fuller tests, although the Phillips–Perron is preferred as it is robust to general forms of heteroscedasticity in the error term. Both tests indicate what we earlier presumed, that price is nonstationary and contains 1 unit root, while hurricane is stationary (shown in Table A1). Since our variables are integrated for different orders, we use the autoregressive distributed lag (ARDL) bounds testing approach to cointegration pioneered by Pesaran and Shin (1995), and Pesaran et al. (1996) to determine whether a long-run relationship exists. The model allows for the estimation of a cointegrating vector, which is equivalent to the error correction model (ECM). Our resultant test statistic was 1.73, indicating that sugar price and hurricanes are not cointegrated; that is, they are not moving together in the long run. We thus cannot obtain long-run coefficients, and the ECM does not exist. We thus use change in log prices throughout our analysis, which, as can be seen from Table A1, transforms our variable into a stationary series.
APPENDIX B
Robustness Tests
Additional robustness tests are provided in Table B1 and Figs. B1 and B2, which show that our results remain unchanged. First, we add more lags to see whether the significant effect of hurricane strikes on sugar prices fades out (Table B1, columns 1 to 6). Second, we add a “placebo test” to check whether there is a null effect prior to the month in which the news of a hurricane strike could have arrived. In columns 7 and 8, we implement a placebo effect simulating an arrival of the news of a hurricane a month before its supposed arrival on the London market. In columns 9 and 10, we implement a placebo effect simulating an arrival of the news of a hurricane a week before its supposed arrival on the London market. Third, we add an additional lag a month later (columns 11 and 12) and an additional lag a week later (columns 13 and 14). Fourthly, we use the actual first reporting day and leads and lags (+2/−2) rather than assuming that the news arrived at the same time during both periods of time (Table B2). Finally, Figs. B1 and B2 act as a robustness check by excluding one hurricane event at a time to see whether the results hold up and indeed they do.
Robustness tests.
Coefficient on the tenth lag (L10) dropping events according rank of strength: 1815–1927.
Citation: Weather, Climate, and Society 9, 4; 10.1175/WCAS-D-17-0015.1
Coefficient on the tenth lag (L10) dropping events according rank of strength: 1827–41.
Citation: Weather, Climate, and Society 9, 4; 10.1175/WCAS-D-17-0015.1
Results for reporting day. Note that standard errors are given in parentheses.
APPENDIX C
Goodness of Fit Model Specification Tests
In Table C1, we run additional goodness of fit model specification tests. We implement the reset test of Ramsey for our baseline specification. At the 5% significance level, we fail to reject the null hypothesis of correct specification. This indicates that the functional form is correct. Nonetheless, if the result of the reset test is indicative that we may have omitted an important variable, we investigate the problem further. In Table C1, we account for major political/economic events of the time: the financial crisis of 1825,13 the abolition of slavery in 1833,14 and the end of slavery (emancipation) in 1838.15
The impact of single events. Note that standard errors are given in parentheses.
In columns 1, 2, and 3 of Table C1, we control for the 1825 financial crisis. In column 1, we use a dummy for the 1825 calendar year (January to January) that we interact with the variable hurricane. Neither the coefficient for the year 1825 nor the interaction term is significant. In column 2, we create a dummy for the month of November 1825. This month precedes the wave of bankruptcies characteristic of the crisis of 1825. We find a negative and significant relationship of this month with the price of sugar. The significance of our hurricane variable remains unchanged. It has to be noted that no hurricane occurred in this month. Finally, we control for the year following the start of the crisis (July 1825 to July 1826), and we interact this variable with our hurricane variable. Neither the coefficient for the year 1825–26 nor the interaction term is significant
In columns 4, 5, 6, and 7, we control for the end of slavery, more precisely in columns 4 and 5 for the legal end of slavery. We create a dummy abolition, for the month of August 1833 when the king ratified the law passed in Parliament at the end of July 1833. We find a positive and significant relationship with the price of sugar. This is not unexpected; at the confirmation of the end of slavery sugar traders anticipating an increase in the cost of production of sugar may have tried to buy more sugar. We also control for the year following the abolition of slavery (from August 1833 to August 1834), and we interact this variable with our hurricane variable. Neither the coefficient for the year 1833–34 nor the interaction term is significant. In columns 6 and 7, we control for the effective end of slavery. We create a dummy emancipation for the month of August 1838 when the period of apprenticeship for the slaves started to end. We find no effect for the month of August 1838. Finally, we control for the year following the emancipation of slaves (from August 1838 to August 1839), and we interact this variable with our hurricane variable. When we do so, the variable hurricane loses its significance and the interaction term becomes significant. This suggests that the impact of the hurricane is exacerbated by the emancipation of slaves at the end of our period.
REFERENCES
Allen, R. C., 2001: The great divergence in European wages and prices from the Middle Ages to the First World War. Explor. Econ. Hist., 38, 411–447, doi:10.1006/exeh.2001.0775.
Bai, Y., and J. K. Kung, 2014: The shaping of an institutional choice: Weather shocks, the Great Leap Famine, and agricultural decollectivization in China. Explor. Econ. Hist., 54, 1–26, doi:10.1016/j.eeh.2014.06.001.
Blau, B., R. Van Ness, and C. Wade, 2008: Capitalizing on catastrophe: Short selling insurance stocks around Hurricanes Katrina and Rita. J. Risk Insur., 75, 967–996, doi:10.1111/j.1539-6975.2008.00293.x.
Boustan, L. P., M. E. Kahn, and P. W. Rhode, 2012: Moving to higher ground: Migration response to natural disasters in the early twentieth century. Amer. Econ. Rev., 102, 238–244, doi:10.1257/aer.102.3.238.
Bradlee, F. B. C., 1925: The First Steamer to Cross the Atlantic. Salem Institute, 10 pp.
Bridbury, A. R., 1973: The black death. Econ. Hist. Rev., 26, 577–592, doi:10.2307/2593699.
Caledonian Mercury, 1818: South America, 2 November.
Carlisle Journal, 1817: The Late Tremendous Hurricane Morning Post, 19 December.
Chenoweth, M., 2006: A reassessment of historical Atlantic basin tropical cyclone activity, 1700–1855. Climatic Change, 76, 169–240, doi:10.1007/s10584-005-9005-2.
Cornwall Royal Gazette, 1835: Packet News, 30 October.
Curtin, P. D., 1954: The British sugar duties and west Indian prosperity. J. Econ. Hist., 14, 157–164, doi:10.1017/S002205070006544X.
Deerr, N., 1949: The History of Sugar. Chapman and Hall, 636 pp.
Deryugina, T., 2011: The dynamic effects of hurricanes in the US: The role of non-disaster transfer payments. Massachusetts Institute of Technology MIT Working Paper CEEPR WP 2011-007, 51 pp.
Engerman, S. L., 1984: Economic change and contract labor in the British Caribbean: The end of slavery and the adjustment to emancipation. Explor. Econ. Hist., 21, 133–150, doi:10.1016/0014-4983(84)90021-4.
Ewing, B. T., S. E. Hein, and J. B. Kruse, 2006: Insurer stock price responses to Hurricane Floyd: An event study analysis using storm characteristics. Wea. Forecasting, 21, 395–407, doi:10.1175/WAF917.1.
Fink, J. D., and K. E. Fink, 2013: Hurricane forecast revisions and petroleum refiner equity returns. Energy Econ., 38, 1–11, doi:10.1016/j.eneco.2013.02.005.
Galloway, J. H., 1989: The Sugar Cane Industry: An Historical Geography from its Origin to 1914. Cambridge University Press, 266 pp.
Hasbrouck, J., 1995: One security, many markets: Determining the contributions to price discovery. J. Finance, 50, 1175–1199, doi:10.1111/j.1540-6261.1995.tb04054.x.
Hersh, J., and H. J. Voth, 2009: Sweet diversity: Colonial goods and the rise of European living standards after 1492. CEPR Discussion Paper DP7386, 45 pp.
Hochrainer, S., 2009: Assessing the macroeconomic impacts of natural disasters: Are there any? World Bank Policy Research Working Paper WPS4968, 43 pp.
House of Commons, 1831: An account of the quantities of unrefined sugar retained for home consumption in the United Kingdom, 17 October 1831.
House of Commons, 1898: Custom tariffs of the United Kingdom from 1800 to 1897, H. C. Sessional Papers, Vol. 85, 890ff.
Jamaica Papers, 1815: Exeter Flying Post, 28 December 1815.
Kludas, A., 1999: Record Breakers of the North Atlantic: Blue Riband Liners 1838–1953. Chatham, 160 pp.
Laakso, S. R., 2006: Across the oceans: Development of overseas business information transmission, 1815-1875. Ph.D. dissertation, University of Helsinki, 361 pp.
Lamb, R., 1995: An exposure-based analysis of property-liability insurer stock values around Hurricane Andrew. J. Risk Insur., 62, 111–123, doi:10.2307/253695.
Loayza, N., E. Olaberría, J. Rigolini, and L. Christiansen, 2012: Natural disasters and growth-going beyond the averages. World Dev., 40, 1317–1336, doi:10.1016/j.worlddev.2012.03.002.
London Colonial Markets, 1834: British Plantation Sugars, 15 November.
London Standard, 1827: Copy of a letter received this morning from the agent to the North and South American Coffee-house, 15 October.
Martin, R. M., 1843: The British Colonies; History, Extent, Condition, and Resources. London Printing and Publishing Company, 788 pp.
Mohan, P., and E. Strobl, 2013: The economic impact of hurricanes in history: Evidence from sugar exports in the Caribbean from 1700 to 1960. Wea. Climate Soc., 5, 5–13, doi:10.1175/WCAS-D-12-00029.1.
Mulcahy, M., 2004: Weathering the storms: Hurricanes and risk in the British Greater Caribbean. Bus. Hist. Rev., 78, 635–663, doi:10.2307/25096952.
Mulcahy, M., 2006: Hurricanes and Society in the British Greater Caribbean, 1624–1783. Johns Hopkins University Press, 257 pp.
Noy, I., 2009: The macroeconomic consequences of disasters. J. Dev. Econ., 88, 221–231, doi:10.1016/j.jdeveco.2008.02.005.
Noy, I., and A. Nualsri, 2007: What do exogenous shocks tell us about growth theories? University of Hawaii Working Paper 07-28, 31 pp.
Odell, K. A., and M. D. Weidenmier, 2004: Real shock, monetary aftershock: The 1906 San Francisco earthquake and the panic of 1907. J. Econ. Hist., 64, 1002–1027, doi:10.1017/S0022050704043062.
Ó Gráda, C., and K. H. O’Rourke, 1997: Migration as disaster relief: Lessons from the Great Irish Famine. Eur. Rev. Econ. Hist., 1, 3–25, doi:10.1017/S1361491697000026.
Parker, M., 2012: The Sugar Barons: Family, Corruption, Empire and War. Windmill Books, 480 pp.
Pereira, A. S., 2009: The opportunity of a disaster: The economic impact of the 1755 Lisbon earthquake. J. Econ. Hist., 69, 466–499, doi:10.1017/S0022050709000850.
Pesaran, M. H., and Y. Shin, 1995: An autoregressive distributed lag modelling approach to cointegration analysis. Faculty of Economics, University of Cambridge, Cambridge Working Papers in Economics 9514, 33 pp.
Pesaran, M. H., Y. Shin, and R. J. Smith, 1996: Testing for the ‘existence of a long-run relationship.’ Faculty of Economics, University of Cambridge, Cambridge Working Papers in Economics 9622, 37 pp.
Pielke, R., Jr., J. Gratz, C. Landsea, D. Collins, M. Saunders, and R. Musulin, 2008: Normalized hurricane damage in the United States: 1900–2005. Nat. Hazards Rev., 9, 29–42, doi:10.1061/(ASCE)1527-6988(2008)9:1(29).
Poey, A., 1855: A chronological table, comprising 400 cyclonic hurricanes which have occurred in the West Indies in the North Atlantic within 362 years, from 1493 to 1855. J. Roy. Geogr. Soc., 25, 291–328, doi:10.2307/1798124.
Raddatz, C., 2007: Are external shocks responsible for the instability of output in low income countries? J. Dev. Econ., 84, 155–187, doi:10.1016/j.jdeveco.2006.11.001.
Shammas, C., 1984: The eighteenth-century English diet and economic change. Explor. Econ. Hist., 21, 254–269, doi:10.1016/0014-4983(84)90008-1.
Slavin, P., 2012: The Great Bovine Pestilence and its economic and environmental consequences in England and Wales, 1318–50. Econ. Hist. Rev., 65, 1239–1266, doi:10.1111/j.1468-0289.2011.00625.x.
Solow, B. L., 1985: Caribbean slavery and British growth: The Eric Williams hypothesis. J. Dev. Econ., 17, 99–115, doi:10.1016/0304-3878(85)90024-0.
Stone, D., 2014: The impact of drought in early fourteenth-century England. Econ. Hist. Rev., 67, 435–462, doi:10.1111/1468-0289.12035.
Strobl, E., 2011: The economic growth impact of hurricanes: Evidence from U.S. coastal counties. Rev. Econ. Stat., 93, 575–589, doi:10.1162/REST_a_00082.
Strobl, E., and F. Walsh, 2008: The re-building effect of hurricanes: Evidence from employment in the US construction industry. IZA Discussion Paper 3544, 15 pp.
Worthington, A., and A. Valadkhani, 2004: Measuring the impact of natural disasters on capital markets: An empirical application using intervention analysis. Appl. Econ., 36, 2177–2186, doi:10.1080/0003684042000282489.
A hurricane is a tropical storm occurring in the North Atlantic. It is characterized by sustained winds of at least 119 km h−1. Destruction will be maximum from 16 to 80 km surrounding the eye of the storm with wind of more than 252 km h−1 for the strongest hurricane.
The Exeter Flying Post on 28 December 1815 states the following: “disastrous accounts are given of the late hurricanes, both as regards to the devastation of property on the island, and shipping. Of the 20 parishes in the Island of Jamaica, 11 or more of them, have suffered irreparable injuries, in houses, plantations and various others kinds of property. But the most to be regretted is the severe loss of lives, particularly amongst the negroes” (Jamaica Papers 1815).
Sugarcane cultivation relied on ratooning, where the root of the cane is left in the soil after harvest. The harvest from the seed matures in 12 to 17 months, and the harvest from the ratoons matures in 8 to 12 months. When a hurricane damages the crop in 1 year it hinders growth in the following year since the root of the cane is damaged.
Carlisle Journal on 15 November 1834 in Dominica states the following: “Sugar and Coffee may be quoted from 6d to 1s per cwt higher than last week” [Carlisle Journal 1817; where d (pennies, 240 pennies equal to 1 pound); s (shillings, 20 shillings equal to 1 pound); and cwt (hundredweight equal to 112 pounds or 50.8 kg)] because of hurricane destruction. Moreover, the manipulation of information in order to prop up the price of sugar on the London market is evocated in the Morning Post on 19 December 1817: “accounts received from Boston of hurricanes in our West India Islands may be, if not altogether incorrect, at least greatly exaggerated…The motive for circulating such exaggerated accounts is to produce a rise in the price of sugars but the news was brought by an American vessel to an American port, and then inserted in American newspapers” (London Colonial Markets 1834).
Muscovado sugar during the late eighteenth to nineteenth century refers to sugar of the lowest quality. This sugar was less refined since it was poorly drained of its molasses and had to be sent to British refineries before it became suitable for final consumption.
These include Antigua, Bahamas, Barbados, Berbice, Bermuda, Demerara, Dominica, Grenada, Jamaica, Montserrat, Nevis, St. Christopher, St. Lucia, St. Vincent, Tobago, and Trinidad.
A tropical storm has a maximum sustained surface wind range from 63 to 118 km h−1, just below the wind classification of hurricanes.
“On Wednesday the Scamander frigate, Captain W. Elliot, C.B. arrived at Portsmouth from the Leeward Islands, after a passage of twenty three days from Barbados…” (Caledonian Mercury 1818, p. 2).
“We are glad to learn that H.M.S Spitfire having been driven to sea by the late hurricane at Barbados reached Grenada in safety, with damage to her machinery, and loss of mainmast” (Cornwall Royal Gazette 1835).
See footnote 5 and London Standard of 15 October 1827: “The American coast, and nearly all the West India islands, were visited during last month by violent hurricanes, and the destruction of persons and property has been very great” (London Standard 1827).
The reported effects are one unit changes measured by the formula ΔY/ΔX, which is our regression coefficient.
The faster transmission of information on hurricane strikes possibly increased the relative magnitude of a reaction in sugar prices. However, this may be just one possible explanation.
The financial crisis started in July 1825; it culminated on 10 December 1825 with a run on the banks in the United Kingdom.
In July 1833, the United Kingdom adopted an act of parliament declaring slavery illegal, giving 4 years to slaver owners to emancipate their slaves and allowing financial compensation for the slave owners. The act was signed by the king in August 1833.
In August 1838, slavery came to an end in the British Empire at the exception of the territories in direct possession of the East India Company (e.g., Ceylan).
