The time series trends for production and yield per hectare of sugar cane for the selected countries were presented on a time series chart alongside temperature changes and rainfall for the respective countries (see Figures 1-11). From the figures presented, several key trends were observed.
For the smallest producing country outlined in the study, Guyana showed an increasing and cyclical trend for its annual temperature and a horizontal and cyclical trend for its rainfall (Figure 1). In terms of its sugar production, the country showed an overall decreasing trend in production and productivity per hectare with peak periods of production occurring from 1999 to 2013. Beyond these years, a sharp decline in productivity was observed. Though the overall trend does indicate a change in production that can be attributed to varying climatic factors, the sharp decline in its production in recent years is mostly attributed to the country’s closure of several of its sugar estates because of the unsustainable labour costs in the country (Singh & DWTI, 2021).
With respect to the largest producer, Brazil has shown an increasing trend in average annual temperatures with annual rainfall having a cyclical trend (Figure 2). The overall trend for sugar cane production and yield per hectare in Brazil however, has shown a sharp increasing trend. It is worth noting that Brazil average annual rainfall over the time period ranged from 1600mm – 1950mm which is double that of some of the other countries assessed for this study. In spite of the variability in temperature and rainfall, Brazil has managed to more than triple its production and yield per hectare of sugar cane over the time series. The boost in Brazil’s productivity can be attributed to its climate smart agronomy practices and policies and the rapid expansion in the country’s bio-fuel industry (see Maroun & La Rovere, 2014; Negra, 2014).
Similar to Brazil, Colombia had an increasing trend with its average annual temperature and a cyclical trend with its average annual rainfall as the two countries share similar climatic and ecological spaces (Figure 3). Unlike Brazil, sugar cane production and yield per hectare for Columbia shows periods across the time series with an upward trend and periods with a downward trend. Colombia’s production and yield per hectare had a sporadic downward trend from 2011 onwards. Comparing the trends between rainfall patterns and the sugar cane production variables showed that the volume of rainfall in Columbia is directly impacting production levels. Colombia’s rainfall volumes are higher than Brazil ranging from 2400mm – 3000mm annually. The data trends show that the years where rainfall levels are closer to the 3000mm upper bound, production levels would show a trough and when rainfall levels are closer to the 2400mm lower bound, production variables would show a peak. Guyana also has a similar average annual rainfall volume as Colombia and its production patterns follow a similar trajectory as well.
China is another major global producer of sugarcane. Interestingly, there is little variation in average annual temperatures and annual rainfall totals (Figure 4). Temperature variations throughout the time series (1989-2019) have only fluctuated within the 7-8 degrees Celsius range, while annual rainfall totals were constantly between 550-700mm. The overall trend for sugarcane production and yield per hectare in China has shown an increasing trend, with peak production from 2009-2015, after which, a decline was observed. From the graphs, there seems to be no correlation between rainfall and temperature on production levels. As the second largest producer of sugarcane, India has shown an increasing trend in average annual temperature and a cyclical trend for average annual rainfall (Figure 5). The graphs illustrate that sugar cane production and yield per hectare for India shows periods across the time series with an upward trend and periods with a downward trend. For the most part, comparison of the trends in temperature and rainfall patterns with sugarcane production showed that the temperature and volume of rainfall in India is directly impacting production levels. The data trends show that the years where rainfall levels are closer to the 1200mm upper bound, production levels are generally lower. Similarly, when temperature levels are closer to the lower bounds (around 24.5 oC), production levels peak, and when temperature levels are closer to the higher bounds (25.5 oC), production levels show a trough. The lower bound of temperature is also associated with higher productivity.
In South East Asia’s Oceania region, Indonesia showed an increasing trend with its average annual temperature and a cyclical trend with its average annual rainfall (Figure 6). The graphs show that sugarcane production and yield per hectare for Indonesia has a downward trajectory. The late 1980’s and early 1990’s showed periods of peak production and productivity, after which, a moderate downward trend was observed. Similar to India and Columbia, the data trends show that the years where rainfall levels are closer to the upper bound (3500 mm in this case), production showed a trough, and when rainfall levels are closer to the lower bound (2500 mm), production levels peaked. A similar observation was made for temperature levels. The data trends show that when temperature levels were closer to the lower bounds (around 25.8 oC), production and productivity levels peak, and when temperature levels were closer to the higher bounds (26.4 oC), production and productivity levels show a trough.
On the other side of the globe, in North America, Mexico also showed an increasing trend with its average annual temperature and a cyclical trend with its average annual rainfall (Figure 7). Unlike Indonesia however, the data trends for Mexico shows a slow but constant increase in production levels (with no major declines) throughout the time series. The graphs show a 50% increase in sugarcane production for Mexico from the early 1990’s to 2019. Comparing yield and temperature trends from graph 7(c) showed a similarity in trend movements between temperature and sugarcane productivity indicating a distinct relationship. The data trends show that when temperature levels were closer to the lower bounds (around 20.7 oC), productivity levels peak, and when temperature levels were closer to the higher bounds (22.2 oC), productivity levels show a trough.
As observed in the data trends for majority of the countries within this study, Pakistan shows a similar trend with respect to annual average temperature and annual rainfall totals (Figure 8). An increasing trend with its average annual temperature and a cyclical trend with its average annual rainfall can be observed. Unlike the other countries previously discussed, Pakistan’s average annual temperature has increased by almost 2oC over the time series, which represents a much sharper increase in mean annual temperature. Despite the sharp increase in temperature and highly variable rainfall patterns, an overall increase in production and productivity can still be observed over the time series. It is worth noting that when rainfall levels are below 1200mm, sugar cane production are usually well irrigated. As Pakistan has a low annual rainfall level, the irrigation structure in the country can be attributed to the increasing trend of productivity.
For the Philippines (Figure 9), a slight increase in annual mean temperature and cyclical rainfall patterns can be observed from the data trends. Interestingly, production levels also seem to be following the cyclical patterns of rainfall. There are periods of rapid decline and increase of sugarcane production over the time series. There is no clear trend as it relates to an overall increase or decline of production levels. However, sugarcane productivity has shown to be on a constant decline.
Similar to the Philippines, Thailand has also shown a slight overall increase in annual mean temperature and cyclical rainfall patterns can be observed from the data trends. The difference however, is that large temperature and rainfall fluctuations has been seen over the time series. Irrespective of these highly variable conditions, production and productivity levels has increased significantly. It is worth noting that Thailand has increased its sugarcane production from forty million tons in 1989 to one hundred and thirty million tons in 2019. This represents an increase of more than 200% over the time series. It can be seen in the graphs that the higher bounds of the temperature levels (27.5 oC) is associated with lower productivity and vice versa. The same trend is also evident for rainfall on production levels.
Similar to China, the USA has shown little variation in average annual temperatures and annual rainfall totals (Figure 11). Temperature variations throughout the time series have only fluctuated within the 9-10 oC range, while annual rainfall totals were constantly between 680-820mm. There is no distinct positive or negative trend as it relates to sugarcane production or productivity for the USA. There are periods increasing and decreasing production and productivity levels over the time series, with peak production and productivity occurring from 1999-2003, after which, a decline was observed. From the graphs, there seems to be an inverse relationship between rainfall and temperature on production parameters. As temperature and rainfall levels increase, production and productivity of sugarcane drops and vice versa for the USA.
The data trends in the time series observed in the eleven (11) countries used in this study, showed that majority (55% of the sample) of the countries had an increasing trend in sugarcane production, with some of the countries doubling (India, Pakistan, China) and even tripling (Brazil, Thailand) production levels over the thirty (30) year time series. These countries were able to utilize the climatic parameters through climate smart techniques to enhance production and productivity (Kumar et al, 2018). A vast majority of the countries showed increasing trends for average annual temperature and cyclical trends for average annual rainfall with relatively consistent annual rainfall totals. For two (2) countries (Philippines and the USA), no distinct trajectory was observed as it relates to an overall increase or decrease in production levels over the time period. However, for the Philippines, a decline in productivity levels was evident. Countries like Guyana and Columbia showed a decline in production levels during the latter portion of the time series. This is attributable to various factors. For Guyana, the sharp decline in production and productivity was due to the closure of several sugar estates. This can be viewed as a socio-economic issue that led to the decline. In the case of Colombia however, the decline in production levels was mainly due to the temperature and rainfall variations. The data trends show that when rainfall totals are at its upper bound (3000 mm) production drops significantly, and peaks when rainfall totals are at the lower bound (2400 mm). This was also observed for temperature fluctuations. The decline in production levels in Colombia started after a sharp steady increase in annual mean temperature within the last decade. The trends observed for the major sugarcane producing countries and Guyana follows the pattern outlined in recent research which shows that climate variability has both positive and negative impacts on sugarcane production (Arora, 2019; Taskinsoy, 2019).
The study did not take production at the community level across countries into consideration, which may have shown different results as it relates to localized impacts caused by climate change. The increasing trends in sugarcane production and the productivity per hectare observed in the time series can be attributed to several factors such as improved agronomic practices and incorporation of climate smart technology. The overall trends does tell a story of variation existing in sugar cane production that can be attributed to variation in climatic factors. As a means of further expounding on this, the panel regression modeling was done to give some empirical merit to the trend analysis observations.
4.1 Panel Regression Model
The weighted least squares panel regression model was used to establish an empirical relationship between temperature and rainfall changes and the production volumes of sugar cane. A log-log model was implemented therefore the coefficients could have been interpreted as elasticities. The panel regression model results are presented in Table 1 and shows that a 1% increase in rainfall will increase productivity by 13.5% and 1% increase in temperature will decrease productivity by 17.6%. All the variables in the regression model were statistically significant at 1%. Globally, the model shows that continued temperature increases will eventually decrease sugarcane production and rainfall decreases impacts sugarcane production positively which is a direct effect of climate change and variability. Sugarcane production is increasing at current levels of temperature and rainfall with mixed variabilities across countries. This corroborates with the literature, for instance, Marin et al. (2012) outlined that Brazilian sugarcane producers were concerned about productivity as a result of inconsistent rainfall and temperature. Despite this, Brazil has an increase in productivity primarily due to adaptive agronomic practices and technology implemented in the country. Additionally, a study done by de Medeiros Silva et al. (2019) who also used a panel regression model to empirically show the relationship between sugar cane production and climate variability in Brazil showed a similar result where temperature changes negatively impact sugar cane production and rainfall changes positively impacts production.
The model implemented by de Medeiros Silva et al., (2019) was a pooled panel model as this was outlined by the author as the most robust. The model for this study builds upon the robustness of the de Medeiros Silva et al. (2019) model by adopting a weighted model that corrects for heteroscedasticity, multicollinearity and autocorrelation and presents a new interpretive viewpoint by having a log functional form. In terms of the model’s explanatory power, the R-squared and adjusted R-squared values were over 99% R squared which showed that the model explained majority of the variation in the dependent variable. The F-value was also statistically significant at 1% which indicates well defined explanatory variables. The Cross-Sectional Dependence (CD) test accounts for the spatial effects, one country climate variability affecting another country of sugarcane production. The Pesaran CD had a p-value of 0.535 which indicates no cross-sectional dependence was present in the model.
Table 1. Weighted Least Squares Panel Regression Model on Global Sugarcane Production
Variable
|
Coefficient
|
Std. Error
|
Constant
|
3.44583***
|
0.109320
|
Ln Rainfall
|
0.135119***
|
0.0141433
|
Ln Temperature
|
-0.175539***
|
0.0217710
|
Ln Area
|
1.02469***
|
0.00475821
|
F-value
|
17193.14***
|
|
Note: ***, **,* indicates statistical significance at 1%, 5% and 10% respectively. The model has an F-value significant at 1%. The R-Squared and Adjusted R-Squared were both 0.993 respectively. The Pesaran CD test has a p-value of 0.535.
Though the findings of this study has been well established theoretically, with small scale models done at the community and national level to empirically validate it. This study presented a full-scale empirical model to further validate the theory. Additionally, there are several policy implications that can stem from the models of this study. The panel regression modeling and trend analysis show distinct patterns in rainfall variation that policy makers can use to implement adaptive strategies. For instance, the water demand for sugar cane and depending on the country, the reliance of rain fed agriculture, the model results can therefore be used to determine the potential investments needed in irrigation systems based on rainfall volume patterns (as outlined in Parkes et al., 2019). Countries like Pakistan for instance which shows low rainfall volumes annually and sporadic production volumes can look at more irrigation infrastructure such as drip irrigation in sugar cane production to compensate for rainfall shortages. On the other spectrum, countries like Colombia with high volumes of rainfall can shift investments from irrigation towards more effective drainage and run-off catchment systems.
The main policy implication from the models presented which differs from majority of the climate related studies on sugar cane production is the totality of the impacts measured to the global sugar cane market. The trend analysis shows relative stability in global sugar cane production, the panel model shows that a business as usual approach to climate change will eventually result in a significant decline in sugar cane production and a potential collapse, especially in the smaller sugar cane producing countries. The continued increase in temperatures and the subsequent impacts based on the model of this study therefore shows the need for more collective adaptive policy actions. In order to maintain sugarcane supplies in a globe with an increasing average temperature, new agronomy practices need to be researched in the areas of precision agriculture, permaculture and controlled agro-ecological zones. Most importantly, more collective policy actions are needed such as vertical and horizontal technology transfer, research sharing networks and funding support for developing countries to integrate new adaptive technologies to agronomic systems.