PhD thesis entitled “Essays on the economics of obesity” defended on January 27, 2019 at the Department of Economics, University of Barcelona, directed by Joan Gil Trasfi (University of Barcelona), which earned the Cum Laude International PhD rating. Academic examiners: Joan Costa Font (London School of Economics), Toni Mora (Universitat Internacional de Catalunya) and Apostolos Davillas (University of Essex).
Obesity is one of the greatest contemporary challenges to the health of the world population and is determined by a combination of factors. Recent statistics indicate that obesity rates have continued to rise during the last decade, resulting in severe health implications as well as in substantial costs. My thesis deals with three different yet connected critical aspects of the analysis of obesity in Spain and is aimed to provide a holistic picture of its key determinants as well as to elicit valuable information with policy relevance on the specific matter. It comprises three main articles that focus on an in-depth analysis of obesity and the Body Mass Index (BMI) and their association with various risk factors.
In the first article (co-authored with Joan Gil Trasfi), we focus on the way income and obesity are associated to one another to describe the health status (proxied here by obesity) of the poor in comparison to the rich. As the health risks associated with being obese are increasing even more at the top of the BMI distribution, we do not evaluate income inequality only in obesity status but also in its depth and severity. Specifically, we make use of overweight indices expressing the depth and severity of the problem, as the excess BMI above the overweight threshold (depth) and the squared excess (severity), respectively. The next basic step of the analysis is the measurement of inequality in the distribution of the above three obesity measures using the concentration index (CI). This index relates the concentration of a health variable with the cumulative rank of the income distribution. The CI ranges between -1 and +1, which makes the comparison of inequality between years, groups as well as populations, possible. In the case where the health variable of interest corresponds to a measure of ill-health (i.e., obesity), negative values of the CI indicate that this variable is concentrated among the worse-off, which means that inequalities favour those individuals with high socioeconomic status (pro-rich). A zero value of the index represents an equal distribution of obesity across income. We examine changes in obesity rates as well as in income-related inequality in obesity over time using data from the Spanish National Health Survey (ENSE) and the European Health Interview Survey (EHIS), spanning the last two decades from 1997 to 2017. In order to observe the main factors explaining income-related inequality in obesity, we decompose the overall inequality in obesity status, depth and severity to its main contributing factors, comparing two points in time (2017 with 1997) and splitting by gender. Our results show a declining trend of income-related inequalities in all three measures of obesity, mostly for women. This could be explained by changes in the social standards, as nowadays women might be more affected by the societal attitude towards obesity than men, in comparison to the past. Furthermore, we witness that income and education are the most important contributors to the overall inequality, as well as that the income-related disparities in depth and severity of obesity tend to be noticeably larger than the ones in obesity status. This is in line with similar findings arguing that highest levels of BMI are often observed among those in disadvantaged socioeconomic circumstances. Overall, data show that obesity is concentrated among the poor in Spain, but the degree of inequality has a declining trend (mainly among women), even though obesity prevalence, as well as depth and severity, increase over the years. These findings may imply a switch in the basic determinants of obesity. That is, BMI status might not be linked only to individual attributes, but environmental influences that similarly affect all income groups may be important as well.
The previous findings set up the main motivation for the second article (Raftopoulou, 2017), which seeks to explain the basic determinants of individual weight status and obesity risk by accounting for the interplay between individual and regional attributes in a multilevel (ML) framework. Utilising data from the National Health Survey of Spain for 2011-2012, two econometric models are fitted: a linear multilevel model for analysing individual BMI and a non-linear logit multilevel model for analysing obesity. The estimates confirm the expected individual-level and regional effects on BMI and obesity. More specifically, net income is statistically significant and negatively associated both with BMI and obesity. That is, individuals with higher net income tend to exhibit lower levels of BMI and lower probability of being obese. The same applies for high-educated individuals who appear to have lower expected BMI and obesity levels compared to the low-educated. Smoking seems to be a protective factor, as daily smokers have lower levels of BMI and obesity than non-smokers. There is significant evidence that the proxies of the social environment (criminality and green spaces) have a positive and statistically significant effect on female BMI and the prevalence of obesity. Data also confirm that the average self-reported BMI at the provincial level is higher in the southern part of Spain than in the northern part of the country, while income distribution presents the opposite pattern. This article stresses the fact that some health determinants can only be observed while investigating interactions between variables at different levels of analysis. These findings are important for public health authorities, since they witness the importance of environmental and regional characteristics in shaping individual BMI and obesity. This means that local governments can play an important role in implementing specific policies, such as promoting environments that encourage and support healthy lifestyles.
The third article (Di Paolo et. al, 2018) is inextricably linked and mainly driven by the spatial pattern of BMI that has been observed in the above-mentioned paper. We contribute to the North-to-South health divide in Spain, using decomposition techniques to analyse the main contributors of the BMI gap between the North and the South, based on the Spanish data of the European Health Survey of 2014. First, we analyse the relationship between BMI and several potential conditioning factors and examine whether their conditional correlation with BMI is different between the two groups of regions. Second, we decompose the observed average gap into the part attributed to differences in observable determinants of BMI and the part that is left unexplained (return to observables), using the classical Oaxaca-Blinder (OB) decomposition. Moreover, as long as important differences in BMI occur away from the average, we proceed with a distributional analysis by applying the Recentered Influence Function (RIF) regression and the corresponding decomposition. This way, we can observe what happens at every part of the distribution and subsequently draw conclusions for the upper tail (obesity, severe obesity) where relationships might vary. The analysis is carried out separately by gender, as the underlying mechanisms that affect BMI appear to be different for women and men. Our findings indicate that the South to North gap in BMI is mostly driven by women, whereas it is lower and insignificant for men. Around 73% of the cross-regional gap in BMI among women is accounted by differences in observable characteristics, as women residing in the South have lower education and income levels. The distributional analysis reveals that the South to North gap in BMI for Spanish women tends to increase over its unconditional distribution, with observable factors (especially schooling) making a growing contribution in explaining the differential across the quantiles of BMI. Given that the education gradient in obesity seems to be much stronger in women than in men, efforts aimed at improving years of schooling for women in the South would substantially mitigate differences in overweight and obesity between the two groups of regions. Such a policy intervention would additionally reduce differences in obesity-related diseases and/or improve health in general, inasmuch as obesity constitutes a key risk factor for many chronic conditions and health complications. However, it must also be stressed that even equalising the female endowments across the two groups of regions, there would still be a certain differential in BMI that penalises southern Spanish regions in terms of the prevalence of overweight and obesity problems.
Despite intense public health focus to curtail growing levels of obesity worldwide, obesity remains a grave public health threat. In the current doctoral thesis, obesity is examined through three different – though interconnected – standpoints, so that relevant and useful information for policymakers is provided. The strong and robust relationship between education and obesity, as well as the important effect of the surrounding environment on individual obesity risk epitomizes the basic outgrowth of this work. The causal effect of education on BMI and other health-related variables as well as the mechanism through which inequalities in educational attainment translate to health inequalities should be further investigated in future research. The exploration of the effect of the so called “obesogenic environment” also calls for further analysis, focusing in unmasking the interplay of different forces contributing to excess weight gain.