What is Happening with Quality of Life Among the Oldest People in Southern European Countries? An Empirical Approach Based on the SHARE Data

Population aging in developed countries has created new challenges to improve the well-being of individuals at different age cohorts. This issue is especially significant for Southern European countries, were aging societies have worse health and less socio-economic resources. The aim of this study is to contribute to this body of literature and to estimate the effect of aging on quality of life of oldest people. This paper uses the latest available data (6th wave) from the Survey on Health, Ageing and Retirement in Europe (SHARE). Specifically, robust ordinary least squares and multilevel regressions are employed to analyse the effects of socioeconomic, health, and community factors on quality of life among the oldest population for Southern European countries. Our findings confirm the significance of several factors on life satisfaction among the oldest population in this group of countries. Moreover, we show that the determinants which are correlated with quality of life include predisposing, health, geographic area and social isolation factors.


Introduction
Population aging is a long-term trend that began several decades ago in Europe. The changing demographic structure of European Union, which is manifested in the increase in the proportion of people aged 65 or over (old dependency ratio), 1 is creating new challenges regarding broader welfare policies and improving the well-being of individuals. In any case, to understand the causes and assess the consequences of this progression, it must also bear in mind that the aging factor goes together with the rise of longevity understood as life expectancy at birth 2 and also, life expectancy in good health status, 3 has grown very significantly over the years. Thus, population aging is a problem that affects several dimensions and it is expected to have widespread direct and indirect economic impacts. Besides, there exist relevant effects of several illnesses that need long time care. These chronic diseases or disability conditions are very usual among the old people and deeply affect the quality of life from many aspects so they may require ongoing care and support (Wiles and Jayasinha 2013).
Over the past decades, aging has a significant impact on society due to a higher use of health care services and social resources. Precisely, Europe is one continent where a significant number of countries face population ageing in the near future, as the full generations born in the Baby Boom years would joined the upper stratum of the population pyramid. Aging has a direct influence on the labour market, since improvements in life expectancy affect individual behaviour in deciding to remain in work longer.
Moreover, it is well known that health and economic status are the most significant factors of quality of life. Consequently, there is a sizeable literature on the main factors of subjective well-being (Dolan et al. 2008). Data in early studies under represent the oldestold (Ng et al. 2017), however recent ones (Mojon-Azzi and Sousa-Poza 2011; Niedzwiedz et al. 2014Niedzwiedz et al. , 2015Stolz 2015;Conde-Sala et al. 2017;Gibney et al. 2017) take the advance of the availability of micro data for the elderly population which represent a great opportunity for Europe comparing with other continents (Jagodzinski 2010).
The main source of data used in this study is the latest wave of the Survey on Health, Ageing and Retirement in Europe (SHARE). We contribute to the existing literature regarding the well-being of the oldest people in order to disentangle the effect of aging and health on the quality of life. Specifically, robust ordinary least squares (OLS) and multilevel regressions are designed to analyse the effects of socioeconomic, health, and community factors on quality of life among the oldest population for Southern European countries.
In spite of the fact that aging population is a generalized trend, it must be taken into account that there is a huge different incidence between countries, and there are unequal repercussions due to the dynamics of fertility and migration (Nicholson 2012). The knowledge gaps that we want to cover are to disentangle the principal specificities and differences of the Southern European countries in relation to other European welfare systems because the economic crisis and austerity policies have greatly increased the level of dissatisfaction with health and social care provision (Lou 2010;Diener et al. 2013;Layte et al. 2013;Tavares and Zanto 2017). For this reason, we present results for the full sample and the selected group of Southern European countries.
Our analysis confirms the significance of several factors affecting life satisfaction among the oldest in Southern European countries. Thus, we show that determinants which are correlated with quality of life are in accordance with previous literature and include predisposing, health, geographic and social isolation characteristics. Population aging is a direct result of the decrease in the fertility rate and the increase in life expectancy and it is useful to understand the growth of life satisfaction in developed countries, so more information is needed for coordinated public policies to enhance health promotion and disease prevention among the various sectors involved. 4 The structure of the paper is as follows. Next, we describe the data, based on the latest wave of the SHARE longitudinal survey (Wave 6), and the econometric model. Empirical findings and discussion are presented in the following section and final section concludes.

Data, Measurements and Methods
Data for this study are taken from SHARE, which is a longitudinal survey from individuals aged fifty and over in a host of different European countries. It was established specifically to address multidisciplinary areas of aging and the cross-sectional analysis utilises contemporaneous data from easySHARE release 6.0.0 (Wave 6 for 2014/15).
Countries here included in the full sample represent Nordic (Sweden and Denmark), Continental (Austria, Germany, France, Switzerland, Belgium and Luxembourg), Southern (Spain, Italy, Greece, Portugal, Slovenia and Croatia), Eastern (Czech Republic, Poland and Estonia) European countries, and Israel although SHARE questionnaires gathered information on households, demography, education, labour, income, health status, and other indicators for the social network.
Our analytical sample is restricted to those aged more than 50 (n = 62,715 individuals). This is mainly due to the analysis of the quality of life (QoL) measure (CASP-12) in SHARE data. That is, the CASP-12v Quality of life and well-being index which ranges between 12 and 48. It is interpreted as follows: scores QoL below 35 = low QoL; 35-37 = moderate QoL; 37-39 = high QoL; ≥ 39 very high QoL. Additionally, the QoL is a frequently used measure for well-being and it is usually stable across countries and time.
Using SHARE database, Fig. 1 presents QoL distinguishing between the full sample and Southern European countries. It is observed similar findings regarding moderate and high responses, but the inverse one regarding low and very high perceptions. Therefore, worse results are obtained for the Southern European countries sample being Slovenia, Spain and Croatia the countries with lower perceptions scores. Italy, Portugal and Greece have the highest levels of low perceptions.
Additionally, Fig. 2 shows huge differences by age cohort among the full sample and the subsample of Southern European Countries. Better outcomes in terms of QoL are obtained for the full sample, where also percentages are almost stables by age cohort. The worst consequences are for the oldest-old in Southern European countries and the proportion of individuals with good perceptions with their life falls after 70 and 79 years because for people between 50-59 and 60-69 years is quite similar. Moreover, Fig. 3 shows the differences in   according to the samples considered by age cohort-years (full sample and Southern European countries). Source: Authors' elaboration based on easy-SHARE release 6.0.0 (Wave 6) QoL for the Southern European countries by age cohort. Precisely, Spain, Italy and Croatia obtain the most similar results with our findings obtained in Fig. 2 whereas Slovenia and Greece show more discrepancies. Indeed, the 50-59 and 60-69 age cohorts display very high levels in terms of QoL for Slovenia. Far from these levels, we can point out the case of an ageing society as Spain where the QoL is focused again on the youngest cohorts.
Summing up all these findings and in order to investigate the relationship between aging and health on the QoL, we have taken into account four sets of socio-economic determinants as exogenous variables following the results of previous studies which have shown that they all affect the QoL (Clark 2011;Angelini et al. 2012;Niedzwiedz et al. 2014Niedzwiedz et al. , 2015Bourassa et al. 2015;Wahrendorf and Blane 2015;Tomini et al. 2016;Ng et al. 2017) and their availability in the SHARE database. Firstly, we have included as predisposing factors (P) a group of variables related with individual characteristics like sex, age, education, marital and labour status. Secondly, health factors (H) are considered attending several determinants like Body Mass Index, suffering any chronic condition, having limitations in daily activities and depression, hospital utilization, and lifestyles. Thirdly, with geographic characteristics (G) we have measured if the person lives in a rural area or not, and if the country that we consider is a Southern European one or not. Finally, we show with social isolation factors (SI) if the person lives alone or not because we have hypothesized that is related with QoL.  The list of variables and description are presented in Table 1. It covers the whole relevant aspects, such as sex (1 if female), age (years), 5 educational level (middle and high), current marital status (married or with a registered partner), employment status (retired, employed, or permanent sick or disabled), Body Mass Index (overweight, obesity), limited (if respondent reports any difficulties), chronic (if respondent declares any chronic disease), depression (if respondent has depression), hospital admission (if respondent has stayed overnight in hospital during the last 12 months), ever smoke (whether respondent has ever smoked daily), if the person lives in a rural area or not (1 if rural), macro region (defined as 1 if it is a Southern European country; 0 otherwise) and alone (if the respondent live alone or with more people in the household). Thus, the choice of these particular variables is made so that they are correlated with the QoL (dependent variable). Besides, Table 2 contains the descriptive statistics for QoL variables. It shows that the mean score of QoL was 37.06 for the full sample and 35.09 for Southern European Countries. These levels are moderate because ranges are between 12 and 48 and it shows a division between these two groups of countries.
Our dependent variable is QoL and robust OLS and multilevel regressions are used to study the impact of socioeconomic, health, and community factors on QoL and health

Results
In this section we present the empirical results for the model described above based on demographics, education, marital status, employment status, health, lifestyles, geographical area and social isolation factors. To fully understand these effects, Tables 3 and 4 report the main estimates of our two models: OLS regression estimates and multilevel regression. Thus, Column 1 presents the variables and Column 2 describes the detailed findings for the full sample, whereas the following one does it for each of the Southern European countries. As we can see, coefficients are statistically significant and have the expected signs according to the a priori economic criteria. Besides, the robustness of our estimates is checked as findings are consistent between the methods here applied (robust OLS and multilevel regressions, respectively). Overall, and in line with other papers (Angelini et al. 2012), our study shows that older age, disabled, poorer health status (overweight, obesity, limitations in daily activities, chronic diseases, depression, hospital admissions and smoking) are associated with lower QoL. Surprisedly, the proxy for social isolation does not report negative effects probably due to the special consideration of these variables in countries like the Southern European ones. Gender effect is no significant for Southern European countries. Moreover, middle and high education, being retired or employed, and living in rural areas are vinculated with better QoL. The analysis also indicates that Southern European countries are related with lower QoL. Besides, the unique diference regarding the effects is concentrated on the gender variable that is not significant when it is considered the subsample of the Southern European countries, but it is for the full sample of the European countries in the SHARE database where it is appreciated that being a male is associated with worse QoL.
We also have run a sensitivity analysis including the same specification than in the models that we estimate but using other econometric approaches. Again, our empirical results seem to be suitable in terms of statistical significance and signs of the relationship between the variables of interest. All results for additional sensitivity analyses are available from the authors based on request.
Moreover, different tests have been run in order to determine possible multicollinearity. Precisely, the "Appendix" contains the Variance Inflation Factors for our OLS estimates (see Tables 5 and 6). Furthermore, correlation matrix has been observed. So, it its verified the inclusion of all these factors in our model.

Discussion
This study has reaffirmed the relevance of the impact of aging and health on the QoL of the oldest-old who lived in the Southern European Countries. Precisely, this is the area inside the Europe Union with the highest proportion of people aged 65 or over and where the proportion of oldest people is rising (OECD Health 2017). In this regard, for example, our findings show that QoL among the oldest people is significantly influenced by health status. Thus, despite of major exposure to important health problems like strokes, diabetes, arthritis or other diseases, QoL among the oldest old does not always depend on their objective health level (Gwozdz and Sousa-Poza 2010;Singh et al. 2017). Hence, our findings suggest again that age-related decline in functional level of the individual and it can be mitigated by QoL (Palgi et al. 2015).
In any case, it might not be forgotten that the Southern European Countries are usually characterised by National Health Services and decentralized tax-funded models. Then, in spite of guaranteeing minimum levels of health care everywhere, our results mean that these decentralized tax-funded systems do not reduce the quality and quantity of health services.
Indeed, one important role of applying research on the well-being of the oldest-people is to disentangle their perceived QoL (Angelini et al. 2012). Not surprisingly, results can change inside Europe and these effects are more accused for all the variables with no relevant differences by gender (Stolz 2015).
In order to overcome this issue, we have used data from the SHARE, that it is one of the best multidisciplinary dataset containing a huge amount of information on both the economic and non-economic characteristics of oldest people (individuals aged 50 and over). Nevertheless, possible biases in self-evaluation of QoL are partly explained by controlling several predisposing, health, geographic and social isolation factors. For example, the main problem in other studies focused on the validity of the individual response is that, in the case of smoking behaviour, individuals misreport their current status due to social restrictions (Bago d'Uva et al. 2011).
Besides, despite modeling an OLS and multilevel model of QoL, the special relationship between aging and health and QoL can be two faces of the same coin. Thus, while older adults with chronic limitations can have low levels of QoL, the opposite is also a possible scenario: higher QoL may also enhance their health status (Steptoe et al. 2015) although the elderly people's dissatisfaction will be vanished along with the growth by age of this population group (Jin et al. 2017). This conclusion is also supported by our results but according to the multilevel regression estimates, Southern European countries should be also more considered as possible recipients of new social funds in order to mitigate the economic effect of aging populations.
It is generally understood that demographic changes (fertility, mortality, aging, migration, and social mobility) are closely linked to welfare policies, both as a cause and as a result. For this reason, social demography and welfare policies share a compelling way to know how these factors are reshaping the main characteristics of aging populations. Hence, some demographic trends that influence welfare policies will be highlighted as changes in the size and ages of the population that demands social protection and trends pertaining to household structure and income distribution that demand a better financial sustainability and more equal access to public services of Welfare States.

Conclusions
The aging process that has been taking place for decades will have to face a policy priority in the next years. Precisely, in this study the attention is put on the needs of the elderly population in order to enhance well-being because Europe is the area with the highest percentage of people aged 65 or over and the percentage of older individuals is heavily rising (Eurostat 2017). Hence, different factors have been analysed: predisposing, health, geographic and social isolation.
Our main results can be summarized as follows. Southern European countries, traditionally characterized by poorer socioeconomic conditions, highlighted lower QoL coefficients that the full sample (all the European countries included in the SHARE database). However, the rest of factors almost behave similarly to the full sample but gender is not statistically significant for the Southern European countries subsample.
Therefore, population aging issues should be considered as an irreversible global trend in our developed societies due to fertility declines and rising life expectancy. Moreover, the ability of each country to assimilate demographic changes and their enormous capacity to take better advantage of opportunities that come with the newest demographic structure, would determine the success of the European Welfare States. Besides, we really think that the needs and capabilities of future older generations which speed of aging is projected to be different to those of elders now. For example, the way in which welfare policies are organized in developed countries can have a deep effect on their cost effectiveness and resource constraints. Thus, the problem in this group of countries is to sustain existing health resources and social services but not to include older people into social policies. As a result, a shift in the priorities of meeting the needs of elders will also be required in order to play an important role as a solution for old people with health problems or disabilities.
All in all, limitations and extensions of this study should be indicated. The most important one is that we are working with self-reported data in most of the socio economic variables that we use. That is, despite SHARE information allows incorporating individualspecific characteristics, data drawback is the lack of objectivity. Besides, here we present the empirical results considering only one wave of the SHARE (the latest available, Wave 6). Further studies require to explore the evolution on individual data and to consider more plausible differences by countries in order to gain a better understanding for coordinated social policies.