The consumption of sugar-sweetened beverages in Jamaica and its association with household expenditure allocation

Background . Overweight and obesity are important drivers for the increasing healthcare and other social costs. Apart from present social costs, consumption of SSB may have an effect on future outcomes, from family future wellbeing to national economic growth. If expenditures on SSB decrease expenditures on other goods and services, such as education and healthcare, this “crowding-out” may have a lasting effect. The objective of this article is to estimate the statistical association between the decision of spending in SSB and budget allocation in Jamaica Methods . Using the Jamaican Household Expenditure Survey 2004-2005 a generalized ordered probit model was estimated to examine the association between socioeconomic variables and the decision to purchase SSB. Seemingly Unrelated Regression Equations (SURE) of all the expenditure groups (except the SSB group) were used to estimate the association between the decision of buying SSB and budget allocation on other goods and services. Results . Purchases of SSB are negatively affected by the size of the household and the area of residence (rural households purchase more SSB than urban ones), while having a larger proportion of children (15 or younger) and having a larger total budget is associated to more expenditures on SSB. Households with positive expenditure on SSB allocate significantly less budget to “Health” and “Education”, when compared to those who did not buy SSB. Conclusions . SSB purchases may crowd-out expenditures in essential goods and services, which implies that decreasing the amount spent on SSB may have important present and future consequences on poorer households’ welfare.


Background
According to a 2018 World Health Organization Global report, obesity has nearly tripled between 1975 and 2016 and, globally, more than 1.9 billion people aged 18 and older were overweight, with more than 650 million considered obese.(1) Among children the overweight/obesity epidemics is also growing: in 2016, 340 million children and adolescents aged 5-19 years, and around 40 million children under the age of 5 years, were considered overweight or obese.(1) The region of the Americas has the highest prevalence of adult obesity in the world, with 28.6%, which is more than double the global prevalence of 13.1%.(1) In the case of Jamaica, the prevalence of adults with overweight doubled from 27.4% in 1975 to 55.5% in 2016, while the prevalence of obesity among adults increased from 6.9% in 1975 to 24.7% in 2016, nearly a 4-fold increase. (2) For childhood obesity the situation is even worse. Its prevalence increased from 1% in 1975 to 13% in 2016, thirteen times greater, with an average annual increase of 6.3%. (2) Increases in overweight and obesity have been associated with a number of health conditions, including the most burdensome ones such as cardiovascular diseases, different types of cancer, and diabetes.(3) Prominent experts in public health have signalled the consumption of sugar-sweetened beverages (SSB) as "the single largest driver of the obesity epidemic", calling for extensive taxation and regulation of such products. (4) In addition to its impact on overweight/obesity, the scientific evidence relating the consumption of SSB to negative health outcomes is vast and has been accumulating over the last decade. A 2007 systematic review found that SSB consumption was associated with an increase in caloric intake, beyond the levels contributed by the said beverages and an increase in body weight. (5) In addition, it found negative associations (moderate but significant) between the consumption of SSB 5 and certain nutritious foods like milk and essential nutrients like calcium. There is also evidence of a positive relationship between SSB consumption and cardiovascular diseases and type 2 diabetes mellitus. (6)(7)(8)(9) These associations points toward a significant increase in future health system costs associated with overweight/obesity and noncommunicable diseases, in relation to SSB consumption. (10,11) A recent study measured global, regional and national consumptions of SSB and milk. (12) The study found that the intake of SSB in Jamaica is higher than milk intake for women aged 20-49 years old and for males aged 20-69 years old. It has also demonstrated that the average SSB intake in Jamaica is 3.29 and 3.58 servings/day (serving = 8oz = 237 ml) for female and male (aged 20-30 years), respectively. These figures are more than three times higher than the global average intake for female and male population with similar age range (20-39 years): 0.94 and 1.04 servings/day, respectively. Another, related study, found that, in 2010, more than 380 people (over 20 years old) died in Jamaica from diseases directly attributed to SSB consumption. The vast majority of these deaths were related to diabetes. (13) As with tobacco and alcohol, taxation has been proposed as an effective tool to decrease SSB consumption.(14) As recent as last year, a high-profile Task Force on Health Taxes, chaired by a former US Secretary of Treasure, concluded that "raising taxes on sugary beverages is prudent because taxes can incentivize healthier diets and address the growing burden of disease from obesity and diabetes". (15) Overweight and obesity are also important drivers for the increasing healthcare costs. Direct healthcare costs increase because of the many conditions that are caused by overweight/obesity, but other indirect, usually much higher costs are also present, such as loss of human capital, job absenteeism and presenteeism, disability pension, loss of qualitylife years, and premature deaths. (16) Family costs, such as the cost of suffering and loss 6 earnings for caregivers, can also be significant and greatly increase total social costs, though they are often difficult to measure. Studies measuring such costs for low-and middle-income countries are scarce, though a recent study conducted in Chile, Ecuador and Mexico found that direct and indirect costs for overweight/obesity may range from 0.2% of GDP (Chile) to 1.7% of GDP (Ecuador) with important increases expected in the near future.(17) In the case of Jamaica and other Caribbean countries, although there are no reasons to expect they would be different from the abovementioned three countries, there are no studies quantifying such costs.
Apart from present social costs, consumption of SSB may have an effect on future outcomes, from family future wellbeing to national economic growth. If expenditures on SSB decrease expenditures on other goods and services, such as education and healthcare, this "crowdingout" may have a lasting effect. Families in such a situation may end up with lower human capital accumulated, which could imply lower earnings and higher healthcare costs in the future. (18) At the aggregate level, lower human capital is related to permanent lower economic growth. (19) Although a number of studies have analyzed the effect that consumption of unhealthy products (e.g. tobacco) has on the allocation of households expenditures,(20, 21) such analysis has not been conducted for SSB. The objective of this article is to estimate the statistical association between the decision of spending in SSB and budget allocation, in the case of Jamaica. To the best of our knowledge, this is first work to conduct such an analysis for SSB. Two types of analyses were conducted. The first one seeks to shed light on what socio-demographic variables are related to the purchase of SSB according to households' levels of SSB expenditures. The second analysis seeks to establish the statistical association between expenditure allocation on SSB and on other goods and services. Given their fixed budget, households have to decide on how to allocate it and therefore in which goods and services they would spend it on.

Methods
Data for the analyses come from the Jamaican Household Expenditure Every category, except the first four ones, follows the Classification of Individual Consumption According to Purpose (COICOP) classification. (23) The first group of COICOP classification was further divided into four subgroups "Food", "Tea, coffee and cocoa", "SSB" (juices, carbonated beverages, nectars, etc.) and "Water" to perform the 8 analysis. The COICOP groups of "Miscellaneous goods and services", "Taxes" and "Donations" were grouped in the category specified as "Others".
Two different models are estimated to characterize households' decisions related to SSB purchases. First, a generalized ordered probit model (GOPM) was estimated to examine the association between socioeconomic determinants and the decision to purchase SSB. The dependent variable is ordinal and takes four possible values: 0 if the household does not purchase SSB; 1 if household spend a "low amount" on SSB; 2 if they spend a "medium amount" on SSB; and 3 if they spend a "high amount" on SSB. Categories of expenditures on SSB are ad-hoc and constructed using tertiles (33% of the distribution) of the total household expenditure on SSB. GOPM are more parsimonious than probit models when data is ordered, (24) as it is this case. In addition, GOPM do not have to satisfy the parallel lines assumption that ordered probit models (OPM) have to satisfy.(24) A likelihood-ratio test, testing the parallel lines assumption, is conducted to choose between them. (25) The functional form for these models has been described in detail elsewhere. (25) In our case, the independent variables include the area of residence of the household (urban or rural); the sex, age and age squared of the head of the household; the natural logarithm of the household size; the proportion of women in the household; the proportion of children in the household  for category i means that the purchase of SSB is associated to an increase (decrease) in the share of total expenditures devoted to that particular category of goods/services. In the case the coefficient is negative, this indicates that purchasing SSB is related to a decision to spend less in that category of goods/services. Because, as mentioned before, the decision by the household is made simultaneously, this system of equations may result in errors that are correlated ( ℎ ).
The estimation of both models takes into account the structural information of sample design and sampling weights. Models are estimated using Stata 15.1/MP. Place holder for Table 1. Table 2 shows budget shares of each of the 16 groups in which the universe of goods and services purchased by Jamaican households was divided. The category of goods or services that concentrate, on average, most of the total expenditure is "Foods" (36%), followed by "Transport" (12%) and "Housing, water, electricity, gas and other fuels" (12%). These three groups are the ones that concentrate the greatest proportion of the total expenditures in all the SSB consumption groups.

Results
Place holder for Table 2.
The likelihood-ratio test to test the parallel lines assumption is rejected at a significance of 1% (results not shown but available form authors). Hence, the GOPM is preferred over the OPM. The marginal effects of the GOPM are presented in Table 3. It shows that living in urban areas, significantly increases the probability of not purchasing SSB: rural households have a 2 percentage points higher probability of purchasing SSB than urban ones, keeping everything else constant. Larger households also increase such a probability: a 10% increase in the household size increases the probability of not purchasing SSB by 4%. If the proportion of children increases by 10%, the probability of not purchasing SSB decreases by 8% and the proportion of having high SSB purchases increases by 6%. Finally, increasing by 10% the total household expenditures (a proxy for household income) decreases the probability of not purchasing SSB by 17% and increases the probability of having high SSB purchases by 21%.
Place holder for Table 3. Place holder for Table 4.

Discussion
The consumption of SSB is associated to the prevalence of obesity, diabetes, cardiovascular diseases, cancers, and other several health conditions.(6-10) The social and economic costs of SSB consumption include not only direct medical costs, associated with the treatment and care for those illnesses, but crucially, the value of productive and life-quality years lost to diseases and incapacities. (4,15) Hence, these costs are borne not only by those consuming these goods, but by the society. When this happens (i.e. the consumption of products implies social costs that are larger than the private ones), it is optimal to tax these products in order to decrease their consumption. (15) In the case of Jamaica, SSB are not taxed, apart from the general consumption tax on all goods and services (16.5% of the value of goods after duties). No other fiscal measure (e.g. 13 subsidies on water) is in place to disincentivize the consumption of SSB by altering market relative prices. There is no information on the evolution of affordability (i.e. how many hours/days of work-salary are needed to buy one liter of SSB), although it is known that in the region, prices of SSB have suffered a strong decrease when compared to nominal wages. (28) If this is the case for Jamaica (and there are no reasons to believe it is not), it is quite possible that the per capita consumption of SSB has grown strongly over the past years.
Therefore, it is highly advisable then to impose taxes that could curb such a consumption.
One of the main arguments usually given against SSB taxation is that taxing SSB may be regressive, as relatively poorer households may spend a higher budget share on these goods.(29) At least two counter arguments can be put forward. First, SSB are non-essential goods, with widely available healthy substitutes (e.g. potable water) and with own-price demand elasticities plausibly above or around 1 (in absolute terms): an increase in SSB prices, would lead to a decrease in consumption more than proportionally. (30)(31)(32) This would also decrease the budget share allocated to these products, freeing them to be spent in more essential goods and services (as shown in Table 4). In the case of Jamaica, the results displayed in Table 3 show that purchases of SSB are strongly and directly related to household total expenditures. In other words, richer households have a significantly higher probability of purchasing these products than poorer ones. This would show that, if a tax on SSB is imposed, those mostly affected in absolute monetary terms would be richer households. In addition, the fact that SSB purchases may displace purchases in essential goods and services (eg, education and healthcare) reflects the fact that decreasing the amount spent on SSB may have important future consequences on poorer households' welfare, as they may be unable to increase their investment on human capital.
Second, the consumption of SSB has negative health consequences that may have a greater effect on poorer households, as they may put them in situations of financial distress, considering they usually have less access to healthcare. Hence, lower consumption of these products by poorer households (as a consequence of, for instance, higher prices) could imply less negative health consequences and, as a result, a lower probability of incurring out-ofpocket expenditures on health. This is consistent with what has been shown for other nonhealthy products, such as tobacco or alcohol. (18,33) The second common argument against SSB taxation is related to job losses, as it is argued that taxes destroy jobs in taxed sectors. However, there is strong evidence that this is not the case with taxes on SSB. Money not spent on the taxed SSB will not be destroyed or lost; it will be spent on substitutes (non-taxed beverages) and other products and services. While

Conclusions
The consumption of SSB is associated with overweight/obesity and several diseases, including diabetes, cancers, and cardiovascular diseases. Such diseases increase the costs of medical treatment, loss of human capital and a waste of resources in the form of premature mortality and disability. Such a burden may be excessive, especially for developing countries, undermining their capacity to increase human capital stock and limiting the resources needed to face present and future challenges regarding healthcare.
As demonstrated in this study, SSB purchases displace household expenditures in housing, transport, healthcare and education, which may have a long-run effect on households' welfare (by decreasing future incomes and/or increasing future healthcare costs) and on society (by diminishing the accumulation of human capital and/or increasing healthcare costs).
Taxing SSB is a strategy that is being increasingly followed by both developing and developed countries around the world. Numerous studies have concluded that, because of the existence of healthy, nearly-free substitutes (plain water), taxing SSB does not constitute a financial burden on poorer households, as they usually are elastic goods. In addition, there is evidence that such taxes do not lead to an economic burden in terms of job losses, as the reallocation of expenditures (away from taxed SSB) imply the creation of jobs in other sectors of the economy.
Taxation can also bring fiscal revenues that can be used to increase resources devoted to healthcare and/or to promote the consumption of healthy substitutes (plain water). A program to, for instance, provide water dispensers or even unsweetened milk at schools, in both rural and urban areas, could be financed with these revenues.

Ethics approval and consent to participate
As the study uses publicly available secondary data, no ethics approval or consent to participate is needed.

Consent for publication
Not applicable

Availability of data and materials
Data is available from the Statistical Institute of Jamaica (https://statinja.gov.jm/)

Competing interests
The authors declare that they have no competing interests