The Impact of Sustainable Policies on Subjective Well-Being in a Developing Nation: The Case of Ecuador

Abstract

Objective: This study explores the impact of public policies that promote social equity and rights to nature on subjective well-being (SWB). The focus is on how sustainability policies, particularly environmental protection and poverty reduction, have influenced life satisfaction in Ecuador. Design/methodology/approach: The research uses ordinal logistic regression models to analyze data from Ecuador’s National Survey of Employment, Unemployment and Underemployment. Life satisfaction is the dependent variable, while independent variables include income, education, job satisfaction, concern for the environment, and regional differences. The analysis covers data from 2007 and 2014, representing periods before and after Ecuador’s 2008 constitutional reform. This highlights the study’s temporal focus on changes in subjective well-being. Findings: The results indicate that implementing public social investment policies increases life satisfaction, particularly in regions such as the Coast. After the policy was implemented, income, higher education, job satisfaction, and concern for the environment were associated with higher levels of subjective well-being. However, disparities persist, with lower life satisfaction among Indigenous groups, women, the elderly, and those living in poverty. Originality/value: This study provides valuable insights into the relationship between sustainable policies and subjective well-being, highlighting Ecuador’s unique constitutional framework. It offers lessons for other developing nations seeking to balance social equity, economic growth, and environmental sustainability to improve overall quality of life.

1. Introduction

Studies establishing a relationship between sustainability and subjective well-being have become increasingly relevant in studying social policies and how they impact people’s quality of life. Ecuador presents a unique case with its 2008 Constitution, which enacts social equity and poverty reduction but also recognizes the rights of nature, an unprecedented concept in this era. This study aims to explore how these disruptive constitutional principles have influenced the life satisfaction of Ecuadorians, using data from before and after the implementation of the analyzed Constitution.

The 2008 Ecuadorian Constitution marked an unprecedented shift towards a more sustainable development model; this mandate is characterized by environmental protection and equitable redistribution of resources. The orientation of this new mandate was to promote benefits mainly for the most vulnerable. Through this study, we want to examine life satisfaction from the perspective of sustainability. Additionally, we intend to identify whether these policies have improved the quality of life in different demographic groups and regions of Ecuador.

To achieve this, we apply ordinal logistic regression models to data from 2007 and 2014, representing the period before and after the promulgation of the Constitution. The study employs several factors that affect life satisfaction, among them income, education, job satisfaction, and, finally, concern for the environment. At the same time, it considers the regional disparities between each of the country’s regions (Coast, Center, and East).

This comparative analysis establishes the Constitution’s effects on subjective well-being and provides insights into considerations for designing sustainable policies that seek to improve life satisfaction in other developing nations. The Ecuadorian experience provides valuable lessons for countries seeking to balance economic growth, social equity, and environmental sustainability in their quest for a better quality of life for all citizens.

Subjective well-being (SWB) has emerged as an important field of study in the field of psychology, and this, in turn, to other fields of knowledge, the economic sciences being no exception. To process these indicators, the starting point is how people evaluate their own lives. This evaluation is based on cognitive judgments but also on how they define their affective relationships, either as happiness or sadness [1]. The concept of SWB is considered multidimensional as it usually includes components such as life satisfaction and positive and negative effects. Research such as Diener [2] highlights that SWB includes reflective cognitive appraisals and affective reactions to life events, making it a comprehensive measure of individuals’ overall quality of life.

Within the literature, SWB can be influenced by socioeconomic status, social networks, and personal competence. For example, Pinquart and Sorensen [3] established that the aforementioned elements significantly influence subjective well-being in adults. On the other hand, George K. and Stokes et al. [4,5] established that the role of social support is preponderant; that is, people who receive more emotional support tend to report higher levels of life satisfaction. This suggests that social and economic contexts play a vital role in shaping individuals’ perceptions of their well-being [6].

Additionally, establishing a relationship between income and well-being has been much debated. The “wealth–happiness paradox” argues that, beyond a certain point, an increase in income does not necessarily lead to a corresponding increase in happiness. Although indicators such as GDP are often a good proxy for SWB, income inequality has been shown to reduce average SWB across societies. These results suggest that economic factors are important but not the sole determinants of well-being in its full context [7,8,9].

It is not only individual factors that can affect subjective well-being. It is social and political implications that can affect their well-being. That is why states are considering this field of subjective well-being more and incorporating it into their agendas when formulating policies. Several programs, such as the “National Wellbeing Measurement Program” in the United Kingdom, employ SWB indicators in structuring social policies [10]. This integration of well-being signals the importance and relevance it gains within the political discourse.

Cross-national research has generated relevant information on how social contexts affect social welfare. Among the factors established in such studies, economic development, democratization, and even social tolerance are considered significant predictors of happiness [11]. However, it is necessary to deepen the study with analyses that consider specific ages, based on older adults, to understand better how these factors interact as a function of age [4]. However, it is also necessary to establish their relationship in cultural contexts because, according to Jun [12], cultural beliefs and values significantly affect people’s perception of well-being.

It is necessary to conceptualize happy life expectancy (HLE), which measures the number of years people can expect to live happily. Studies in the United States and the Netherlands show that HLE has increased over time, suggesting improvements in both life expectancy and the quality of those years [4].

Considering more factors in these analyses provides a broader understanding of SWB, where longevity and happiness converge. To this end, various scales and methodological tools have been established to measure and evaluate vital well-being. Prominent among them are, for example, the Satisfaction with Life Scale (SWLS) and the Positive and Negative Affect Schedule (PANAS) [13,14]. However, it is necessary to consider cultural and gender contexts in order to ensure its universal applicability [14]. This is crucial when making comparisons between countries and diverse populations.

Subjective well-being is a multidimensional concept influenced by various factors, such as social support, economic conditions and even cultural contexts. Although significant progress has been made in understanding these influences, there still needs to be more knowledge about the interaction of the different variables that shape people’s perceptions of well-being. Considerations for measuring the fullness of life must be informed by the applicability of the tools and their measurement in different cultural contexts. This study will explore global studies on subjective well-being, such as Salameh et al. [8], focusing on methods like ordinal logistic regression and their findings on socioeconomic determinants while providing comparative insights from diverse contexts, including developed and developing nations.

This paper is organized as follows: Section 2 reviews the literature; Section 3 describes the data and methodology; Section 4 presents the findings and discussion; Section 5 provides policy recommendations; and Section 6 concludes with limitations and future research directions.

2. Structure of the Data

In Ecuador, there is a higher concentration of poverty among Indigenous people in the Eastern areas. With the econometric models, we will test whether Eastern people are more satisfied than other regions. Thus, this aspect of the study explores the regional disparities further. Consequently, we also want to determine through the econometric models which region has more satisfied people. This is crucial to understanding the regional disparities and identifying which region has higher levels of life satisfaction.

Dummy variables and categorical variables were created for the regions (Eastern, Coast, and Central regions and urban/rural areas), age, gender, ethnicity, education, job satisfaction, income, poverty, extreme poverty, marital status, and environmental concern (see

Table 1. Variables in the models.

Type of Variable		Variables	Name of the Variable in the Regression Models
Dependent variable		Life satisfaction	LifeSatisf
Independent variables	Economic variables	Income/Net income (USD)	IncomeCAT/Net income
	Demographic variables	Age	AgeCAT
		Tertiary education	Edu_Dumm
		Gender	Female
		Ethnicity	Indigenous
		Poverty	Poor_Dumm
		Extreme poverty	EXPO
		Religion	Religion
		Marital status	Married_Lawunion
		Regions	RegionsCAT
		Urban/rural areas	Rural_Dumm
	Other variables	Environmental concern (information about environmental concerns is available only in ECV 2014)	Enviro_Dumm
		Job satisfaction	JobSat_Dumm

Source: authors’ summary.

). Job satisfaction is a subjective variable. It is not part of the 14 “life satisfaction” questions but is included in the ENEMDU. Income is measured differently each year. Income in 2007 is related to monthly wages, and income in 2014 is related to net earnings by year. Even though income data in 2014 is limited and only includes net year earnings, it cannot be omitted since the literature review showed that income is highly correlated with well-being. The information about poverty and extreme poverty is related to the survey question about the household heads’ living conditions. The question about poverty is related to being poor or non-poor, and extreme poverty is related to being extremely poor (indigent) or not.

We used the 2014 mean as a reference to classify life satisfaction into three categories. High levels of life satisfaction were classified as more significant than the mean of life satisfaction plus one standard deviation (SD) (9–10 points on the Likert scale), moderate life satisfaction was classified as a value falling within one SD above and below the mean of life satisfaction (6–8), and low life satisfaction was classified a value more than one SD below the mean (5). (The mean of overall life satisfaction was 7.64, and the standard deviation was 1.69 in 2014). Thus, the three classifications of satisfaction are unsatisfied, satisfied, and very satisfied. The sample includes 18,933 responses in 2007 and 28,970 responses in 2014, offering a comprehensive view of regional and demographic differences. Specifying these numbers ensures data transparency and robustness.

Income data were collected in the surveys for both years in association with the economic variables. The way these data were collected is different for 2007 compared to 2014. For example, in 2007, household head income was measured as monthly wages, whereas in 2014, it was defined as the net annual earnings of the household head from the previous year. (Net earnings refer to total income minus total expenses over a one-year period for the household head).

Thus, the income variable is not able to be compared across both periods. However, we analyzed how income affects life satisfaction in each analysis year. Variations in data collection methods, such as income reporting, create challenges in comparability between 2007 and 2014. These limitations are acknowledged, and interpretations are made with caution.

Related to the sample by region, the ENEMDU 2007 and ECV 2014 surveys have different sample sizes but have the same questions in the perception of life section. For example, the ECV survey collected more information from the population in the Eastern area in 2014 (15.37%) than ENEMDU in 2007 (4.5%). The ECV 2014 survey increased the sample in comparison to ENEMDU 2007. Also, the ECV 2014 survey was created to be more representative at the national, urban, and rural levels, with four natural regions, 24 provinces, 9 planning zones, and 4 self-represented cities (Quito, Guayaquil, Cuenca, and Machala) included [15]. Additionally, the Ecuadorian National Institute of Statistics and Censuses (INEC) included topics in the ECV 2014 that include psychosocial well-being, perception of the standard of living, social capital, citizen security, use of time, and good environmental practices. (INEC is the institution that develops and conducts all national surveys in Ecuador, including ENEMDU and the ECV). For instance, regarding regional characteristics, the ENEMDU survey gathered more information about the condition of people in the Eastern region, where there is a greater concentration of poverty [16]. The increase in the sample size of the Eastern region from 2007 to 2014 was due to the change in the collection methodology to cover more geographic areas. However, as noted previously, the questions in both surveys were the same and were addressed to the same groups of people (heads of households). Most heads of households in Ecuador are males; the ENEMDU survey had 78.1% of male heads of households in 2007, and the ECV survey had 75.51 in 2014.

The main characteristics of the sample in terms of age groups are that the younger group (13/14–34 years old) fell more often into the “very satisfied” category than other age groups in 2007, and the older group (65 years and above) was more unsatisfied than the other groups. These results are tested within the econometric models.

We created a dummy variable for tertiary education to identify those who have a post-school study, such as university or superior institute studies (non-university post-secondary school), and those who do not have such a qualification. According to the literature review, people who have studied for more years are more satisfied with life than others. High education was shown to have a positive impact on Ecuadorian well-being. However, a high level of education has been found to have a stronger association with well-being in low-income countries such as Ecuador [17].

In addition, subjective well-being within the marital status categories was tested through the following categories: married or in a de facto union; divorced, separated, or widowed; and single. According to Nicola, Bravo, and Sarmiento [18], in Ecuador, people who are married or in a de facto union have greater well-being than single people or those who are separated or divorced. These results are also supported by European, American, Asian, and Latin American studies [19].

Thus, a dummy variable of environmental concern was created, where one (1) represents an excessive, significant, or moderate concern and zero (0) represents low or no concerns.

Table 2. Characteristics of the sample.

YEAR	2007	2014	2007	2014	2007		2014	2007	2014
Characteristics of the Sample					Mean	Median	Mean	Range
Variable	N		%		(SD)		(SD)
Total data	18,933	28,970
Economic variables					322.5 (747.5)
Monthly wage 2007	15,715		100			198		0	50,050
0–322 USD (base)	11,497		73.2		143.5	140
323–747 USD	2983		19.0		476.9	450
748–1070 USD	584		3.7		882.8	880
1071 + USD	651		4.1		2272.4	1580
							1089 (3519)
Net earnings 2014 (Net earnings represent the total income earned by the household head in 2014, subtracting all expenses they incurred throughout the year).		11,731		100		327		0	160,638
0–1089 USD (base)		9308		79.4		210	294
1090–4608 USD		1912		16.3		1183	2132
4609 + USD		511		4.4		7630	11,675
Demographic variables
Regions
Coast + GP	7827	10,259	41.3	35.4
Central	10,186	14,259	53.8	49.2
Eastern (base)	855	4452	4.5	15.4
Area
Urban (base)	10,684	13,908	56.4	48.0
Rural	8249	15,062	43.6	52.0
Age					49.9 (16.2)		47.80 (16.5)	13–99	14–98
Age group 13–34 Y (Base)	3605	7125	19.1	24.6
Age group 35–44 Y	4192	6665	23.0	24.2
Age group 45–54 Y	4115	5639	22.6	20.5
Age group 55–64 Y	3108	4350	17.1	15.8
Age group 65 Y +	3904	5191	21.4	18.9
Gender
Male (base)	14,789	21,874	78.1	75.5
Female	4144	7096	21.9	24.5
Education level					4.86 (2.2)		6.68 (2.2)	1–10	1–11
Non-tertiary education (base)	16,329	17,116	86.2	86.7
Tertiary education (university and superior institute)	2604	11,854	13.8	13.3
Ethnic group
Indigenous	1485	4052	7.8	14.0
Non-Indigenous (base)	17,448	24,918	92.2	86.0
Marital status
Married or de facto union (base)	13,300	19,911	70.3	68.7
Divorced, separated, or widowed	4119	6490	21.8	22.4
Single	1514	2569	8.0	8.9
Other variables
Environmental concern	N/A	16,212	N/A	56.0
Satisfied with job	16,053	N/A	62.6	N/A
Overall life satisfaction
Unsatisfied	8204	3237	43.5	11.2
Satisfied	8008	17,254	42.5	59.6
Very satisfied	2630	8479	14.0	29.3

Source: authors’ summary STATA; Related to environmental concern, individuals with medium, high, or very high levels of concern are coded as 1, while those who show little or no concern for the environment are coded as 0.

shows the characteristics of the variables used in 2007 and 2014 to create the models.

3. The Economic Model

The economic theory on which this research is based hypothesizes that the new Constitution increased life satisfaction for Ecuadorians, considering all other variables. Thus, one expects these explanatory variables to positively affect life satisfaction (used as a measure of subjective well-being) (see Table 2).

The econometric model shows the association between subjective well-being in Ecuador and the implementation of the 2008 Constitution. Ordinal logistic regression effectively captures the ordered nature of life satisfaction categories. The model highlights meaningful relationships among predictors and subjective well-being despite low R-squared values. Future analyses could explore complementary models.

To produce more meaningful results, the ordinal logistic model in STATA uses “unsatisfied” as the base to compare with “satisfied” and “very satisfied” responses, ordered from 0 to 2.

The economic model is expressed in this way:

Equation (1), economic model

$$LS=f(Economic\_variables;Demographic\_variables;Other\_variables)$$

(1)

Information about job satisfaction is only available in ENEMDU 2007, and information about religion and environmental concerns is available only in ECV 2014.

The explanatory variables—regions, area, age, gender, job, and ethnicity—and their correlations are related to life satisfaction. This specification allows the model to test whether life satisfaction is associated with the 2008 Constitution. Ordinal logistic regression assumes proportional odds, which may not always hold. Diagnostics verify this assumption, and alternative models like multinomial logistic regression may be explored to confirm results.

3.1. Correlations Between the Variables

Before conducting a regression, it is important to understand the underlying data. While the distribution of the variables was already examined, examining their correlations is the first indicator of whether multicollinearity exists in the regression [20,21]. This section shows the correlation matrix between all model variables for 2007 and 2014.

3.1.1. Correlation Between the Variables in 2007

When we examine a correlation matrix of the 2007 data, the relevant results are that living in the Coast and the Central regions, income (above the mean), job satisfaction, and education positively correlate with life satisfaction. Living in the Eastern region, living in rural areas, age (above 55 years old), poverty and extreme poverty, being female, being Indigenous, and being married each have a negative correlation with life satisfaction. There are apparent disparities in gender, ethnicity, and income in Ecuador. Also, in Latin America, more broadly, poverty and inequality among females and Indigenous people are higher than for males and non-Indigenous people [22].

Viewing the correlations from high to low, female and married have the highest correlation (0.49); then, income and poverty (−0.44) have the next highest (negative) correlation, followed by tertiary education and income (0.43), and poverty and living in rural areas (0.33).

The other variables have weak correlations among each other, suggesting multicollinearity is not a problem in the model. Although it is difficult to define a high, moderate, or weak correlation between two variables, Hayes [23] suggests that correlations less than 0.30 are weak, those between 0.31 and 0.69 are moderate, and those greater than 0.70 are high. None of the correlations identified in

Table 3. Correlation table of the variables for 2007.

	Life Satisfaction	IncomeCAT	AgeCAT	EDU-Sup-Dumm	Female	Indigenous	POV	EXPOV	Married-Law	RegionsCAT	Rural-Dumm	JobSat-Dumm
Life Satisfaction	1.0000
IncomeCAT	0.1828	1.0000
AgeCAT	−0.0647	−0.0024	1.0000
EDU-Sup-Dumm	0.1737	0.4458	−0.0639	1.0000
Female	−0.0615	−0.0708	0.0653	0.0158	1.0000
Indigenous	−0.0761	−0.1065	0.0040	−0.0831	0.0002	1.0000
POV	−0.1440	−0.3822	−0.0303	−0.2585	−0.0105	0.1536	1.0000
EXPOV	−0.0978	−0.2271	0.0043	−0.1622	0.0077	0.1583	0.5818	1.0000
Married-Law	−0.0482	−0.0595	0.0168	0.0171	0.4903	−0.0293	−0.0770	−0.0368	1.0000
RegionsCAT	0.0210	−0.0539	0.0078	−0.0570	−0.0361	−0.2344	0.0153	−0.0375	−0.0101	1.0000
Rural-Dumm	−0.1545	−0.2722	0.0890	−0.2944	−0.0827	0.2070	0.3344	0.2719	−0.0512	−0.0304	1.0000
JobSat-Dumm	0.1241	0.1837	0.0592	0.1166	0.0071	0.0073	−0.1395	−0.0931	0.0176	−0.0327	−0.0570	1.0000

Source: authors’ summary STATA.

are sufficiently high nor are there enough to cause problems with multicollinearity.

3.1.2. Correlation Between the Variables in 2014

In the 2014 data, as shown in

Table 4. Correlation table of the variables for 2014.

	LS	Eastern Dumm	Rural	Mingas	Caring	Disabi	Net Income	Age	Ter EDU	Female	Indigen	Po- very	Envi- Rome	Re- ligion	Ma- rried
LS	1.00
Eastern D	−0.03	1.00
Rural	−0.08	0.20	1.00
Mingas	−0.03	0.05	0.19	1.00
Caring	0.01	0.06	−0.03	0.01	1.00
Disabi	−0.02	−0.01	0.00	−0.02	0.02	1.00
Net income	0.07	−0.03	0.09	0.01	−0.03	−0.02	1.00
Age	−0.06	−0.13	0.11	−0.04	−0.32	0.03	0.01	1.00
Ter EDU	0.09	−0.05	−0.23	−0.06	0.04	−0.01	0.06	−0.09	1.00
Female	−0.09	−0.06	−0.06	−0.05	0.10	0.05	−0.15	0.03	−0.05	1.00
Indigen	−0.11	0.29	0.32	0.29	0.04	−0.01	−0.06	−0.06	−0.11	−0.04	1.00
Poverty	−0.15	0.09	0.22	0.03	0.00	0.04	−0.13	0.09	−0.18	0.09	0.14	1.00
Environ	0.06	−0.01	−0.15	−0.03	0.05	0.00	0.04	−0.10	0.14	−0.03	−0.12	−0.13	1.00
Religion	−0.02	−0.02	0.03	−0.06	0.03	0.01	0.02	−0.04	−0.01	0.01	−0.06	0.02	−0.01	1.00
Married	−0.08	−0.05	−0.07	−0.06	0.00	0.04	−0.14	0.07	−0.01	0.61	−0.08	0.08	−0.03	0.03	1.00

Source: authors’ summary STATA.

, life satisfaction has a positive relationship with the Coast and Central regions relative to the Eastern region, as well as net income (above the mean), education, and environmental concern. Also, there is a positive relationship between Indigenous and rural areas. It was noted that poverty is higher in Ecuadorian rural areas where there is the highest concentration of Indigenous people and that Indigenous women are much more vulnerable due to the lack of household income and access to jobs. In the Eastern region, there is also a higher proportion of the population living in rural areas than in other regions.

Living in the Eastern region, being older, living in a rural area, being a female, being part of an Indigenous group, being poor, and being married have a negative correlation with life satisfaction.

As in the 2007 data, all the variables mentioned above have a weak correlation with the 2014 model, again suggesting multicollinearity is not a problem for this year.

All the analyses presented so far have been bivariate, and the other variables may be affecting any of these correlations. A rigorous analysis uses regression to identify the impact of life satisfaction, controlling for all other variables.

3.2. Using Ordered Logit Regression

Ordered logit models are used to estimate the relationships between an ordinal dependent variable and a set of independent variables [7,24,25]. An ordinal variable is categorical and ordered, such as “unsatisfied”, “satisfied”, and “very satisfied”, which could indicate the current life perception of a head of household. Very satisfied has a higher order than satisfied, and satisfied has a higher order than unsatisfied.

The ordered logit regression fits the model since we are comparing the three ordered categories of life satisfaction. It also represents the results more clearly. Ordinal logit coefficients represent changes in the log odds for a one-unit increase in the independent variables. If the x variable is a dummy variable, we can exponentiate its coefficient β to obtain an adjusted odds ratio.

3.3. R-Square and Pseudo-R-Square in Regression Models

The use of R-square has recently been debated in the literature on political science. Hagquist and Stenbeck [26] discussed that R-square is only a measure of the degree of agreement between the effect of the model and the result. According to Hu, Shao, and Palta [27], “there is no clear interpretation of the pseudo-R²s in terms of variance of the outcome in logistic regression... the pseudo-R²s for a given data set are point estimators for the limiting values that are unknown”. Also, the McFadden statistic in logit models points to a low degree of explanation of the control variables [28]; for example, the pseudo-R² has a limited interpretation in ordinal or categorial logit. “In linear regression, the standard R² converges almost surely to the variability ratio due to the covariates over the total variability as the sample size increases to infinity” [27]. Thus, logistic regression does not have an equivalent to the R-squared in OLS regression [9].

After fitting the logistic regression model, we want to know how well we can predict the dependent variable: life satisfaction. A model can fit the data well but does a poor job of predicting, or a model can predict the outcome well but needs to fit better (low pseudo-R-square). Low pseudo-R-squared values reflect the multifaceted nature of subjective well-being and its determinants. Including additional predictors or exploring mixed methods could improve variance explanation.

4. The Models

Identifying the determinants of life satisfaction after the implementation of the new Constitution is the priority in this research to compare life satisfaction in 2007 and 2014. The models include a set of predictor variables that, according to the literature and this study’s descriptive findings, are the determinants of life satisfaction.

We used statistical software (STATA 18) to correct multicollinearity and heteroskedasticity problems in all the econometric models. Thus, we obtained a better estimate of the results (the goodness of fit).

The 2007 Model

The complete model adds the regional dummy variables to the intermediate model. This allows us to understand the influence of the Coast, Central, and Eastern regions on life satisfaction. In this case, the model is expressed by the following relationship:

Equation (2), full model

$$LS={\beta }_0+{\beta }_{1}Economic+{\beta }_{2}Demographic+{\beta }_{3}Other\_variables+u$$

(2)

Through this equation or model, we show the likelihood of being more satisfied rather than less satisfied because of a change in all the explanatory variables.

5. The Econometric Model with the 2007 Dataset: Life Satisfaction, Regions, and Control Variables

The full model adds all variables, including regions. The significant variables and the odds ratios are presented below (see

Table 5. Controlled and demographic variables with regions in 2007 (odds ratios).

Ordered logistic regression			Number of obs = 15,292	Wald chi2(17) = 1263.50	Prob > chi2 = 0.0000	Pseudo R2 = 0.0398
Log pseudolikelihood = −14,742.109
Robust
LifeSatisf	Odds ratio	std. err.	z	P > z [95% conf. interval]
IncomeCAT
1	1.3414	0.0593	6.6400	0.0000	1.2300	1.4629
2	1.5124	0.1154	5.4200	0.0000	1.3023	1.7563
3	1.6118	0.1244	6.1800	0.0000	1.3855	1.8751
AgeCAT
35–44	0.8908	0.0409	−2.5200	0.0120	0.8140	0.9747
45–54	0.8754	0.0406	−2.8700	0.0040	0.7992	0.9588
55–64	0.8136	0.0421	−3.9900	0.0000	0.7352	0.9004
65+	0.6791	0.0387	−6.7800	0.0000	0.6073	0.7595
JobSat_Dumm	1.5177	0.0502	12.6200	0.0000	1.4225	1.6194
POV	0.8015	0.0315	−5.6400	0.0000	0.7421	0.8656
Female	0.8266	0.0460	−3.4200	0.0010	0.7413	0.9218
Indigenous	0.7466	0.0486	−4.4900	0.0000	0.6572	0.8482
EDU_Sup_Dumm	1.5818	0.0720	10.0700	0.0000	1.4467	1.7295
Married_Lawunion
1	0.8311	0.0450	−3.4200	0.0010	0.7474	0.9242
3	0.7919	0.0513	−3.6000	0.0000	0.6974	0.8991
RegionsCAT
Central	1.0618	0.0837	0.7600	0.4470	0.9098	1.2394
Coast	1.1736	0.0946	1.9900	0.0470	1.0022	1.3743
Rural_Dumm	0.7439	0.0263	−8.3600	0.0000	0.6940	0.7974
/cut1			−0.4690	0.1050	−0.6747	−0.2632
/cut2			1.7792	0.1068	1.5699	1.9885

Source: authors’ summary STATA. The pseudo-R² in the full model is also low: 0.039. When regions are added to the model, only the Coast region and living in rural areas are significant. The impact of age groups is significant compared with the youngest group. All income categories, job satisfaction, poverty, gender, ethnicity, tertiary education, and marital status are also significant.

).

A test for multicollinearity confirmed a variance inflation factor (VIF) of 1.46 (mean). Thus, no multicollinearity problems were found in the full model for 2007.

5.1. Interpretation of the Variables 2007 After Running the Ordinal Regression

The odds ratio of 1.17 for the Coast region means that living on the Coast is associated with 17 percent greater odds of being in a higher rather than a lower category of LS. However, the odds ratio of 0.743 for rural/urban areas means that living in rural areas is associated with 25.7 percent lower odds of being in a higher than a lower category of LS.

The odds ratio of 0.89 for those aged between 35 and 44 years means this age group is associated with 11 percent lower odds of being in a higher than a lower category of LS. The odds ratio of 0.813 for the age group of 55–64 years means it is associated with 18.7 percent lower odds of being in a higher rather than a lower category of LS, and 0.679 for the 65+ age group means it is associated with 32.1 percent lower odds of LS. Younger people between 13 and 34 are the basis of the age range analysis.

Life satisfaction decreased in 2007 as age increased for both the “satisfied” and “very satisfied” ranges. Ecuador, a developing country and a part of Latin America, shows a progressive decrease in well-being with age, in contrast with developing countries, where lower levels of well-being are evident in the age groups below 65 years [19].

The odds ratio of 1.341 for income in the USD 323–747 range means that this income category is associated with 34.1 percent greater odds of being in a higher rather than a lower category of LS. The odds ratio of 1.512 for income in the USD 748–1070 range means that this category is associated with 51.2 percent greater odds of being in a higher rather than a lower category. The odds ratio of 1.611 for income of more than USD 1070 is associated with 61.1 percent greater odds of being in a higher rather than a lower category of LS relative to the income base (lower than USD 322).

The odds ratio of 1.517 for job satisfaction means that being satisfied with a job is associated with 51.7 percent greater odds of being in a higher rather than a lower category of LS. The odds ratio of 1.581 for tertiary education indicates that each additional year of education (after 12 years of formal education) is associated with 58.1 percent greater odds of being in a higher rather than a lower category of LS.

However, the odds ratio of 0.801 for poverty means that being poor is associated with 19.9 percent lower odds of being in a higher rather than a lower category of LS, and the odds ratio of 0.826 for gender means that being a female is associated with 17.4 percent lower odds of being in a higher rather than a lower category of LS. For ethnicity, being Indigenous is associated with 25.4 percent lower odds of being in a higher rather than a lower category of LS.

The odds ratio of 0.831 for being divorced or separated means that this marital status is associated with 16.9 percent lower odds of being in a higher than a lower category of LS relative to being married or in a de facto union. Also, an odds ratio of 0.791 for being single means that single status is associated with 20.9 percent lower odds of being in a higher than a lower category of LS relative to being married or in a de facto union.

5.2. The 2014 Model: Determinants of Life Satisfaction in 2014 (Regions and Control Variables)

The complete model includes all controlled variables; specifically, regional variables were included in the model.

The results from this model are shown in

Table 6. The 2014 model: determinants of life satisfaction in 2014 (regions and controlled variables).

Ordered logistic regression Number of obs = 11,725
Wald chi2(17) = 604.39
Prob > chi2 = 0.0000
Log pseudolikelihood = −10,363.268 Pseudo R2 = 0.0290
Robust
LifeSatisf |	Odds ratio	std. err.	z	P > |z|	[95% conf. interval]
Regions
Coast	1.27514	0.0787258	3.94	0	1.129811	1.439164
Central	0.8655154	0.0465154	−2.69	0.007	0.7789842	0.9616587
Rural	0.9660288	0.0421933	−0.79	0.429	0.8867723	1.052369
NetIncomeCAT
1090–4608 USD	1.240964	0.0646756	4.14	0	1.120462	1.374426
4609 USD +	1.2841	0.1183424	2.71	0.007	1.071895	1.538315
AgeCAT
35–44 years	0.8691385	0.0554016	−2.2	0.028	0.7670626	0.9847981
45–54 years	0.8642123	0.0554279	−2.28	0.023	0.7621264	0.9799726
55–64 years	0.8441846	0.0550386	−2.6	0.009	0.7429189	0.9592536
65+ years	0.8249512	0.0545984	−2.91	0.004	0.7245903	0.9392128
Edu_Dumm	1.437972	0.0991581	5.27	0	1.256187	1.646064
Female	0.8229604	0.0511633	−3.13	0.002	0.7285508	0.9296041
Indigenous	0.7076548	0.0380359	−6.43	0	0.6368982	0.7862722
Poor_Dumm	0.616715	0.0250357	−11.91	0	0.5695472	0.6677891
Enviro_Dumm	1.093773	0.041892	2.34	0.019	1.014672	1.17904
Religion	0.8339482	0.0456503	−3.32	0.001	0.7491078	0.9283972
Married_Lawunion
Divorce_Separated	0.8518621	0.053213	−2.57	0.01	0.7536983	0.962811
Single	0.7938266	0.0645279	−2.84	0.005	0.6769147	0.9309307
/cut1	−3.049822	0.1621875	−3.367703	−2.732
/cut2	0.0659302	0.1580761	−0.2438932	0.3758

Source: authors’ summary STATA. First, the pseudo-R² (0.029) is low. Table 6 shows that net income is significant for both categories (USD 1090–4608 and USD 4069+) relative to income less than USD 1089. The impact of all age ranges between 35 and 64 years old compared to the youngest group (14–34) is also significant. Finally, tertiary education, gender, ethnicity, poverty, environmental concern, religion, and marital status are statistically significant, similar to the intermediate model.

.

A test for multicollinearity confirmed a variance inflation factor VIF of 1.76 (mean). Thus, no multicollinearity problems were found in the full model for 2014.

5.3. Interpretation of the Variables 2014 After Running the Ordinal Regression

In terms of odds ratios, when regions are added into the model, we observe an odds ratio of 1.275 for the Coast region, meaning that living on the Coast is associated with 27.5 percent greater odds of being in a higher rather than a lower category of LS relative to the Eastern region. For the Central region, the odds ratio of 0.865 means that living in the Central region is associated with 13.5 percent lower odds of being in a higher than lower category of LS relative to the Eastern region.

The odds ratio of 1.24 for the first category of net income per year (USD 1090–4608) means that this income category is associated with 24 percent greater odds of being in a higher rather than a lower category of LS. The odds ratio of 1.284 for the second category of income (USD 4609+) means that this category is associated with 28.4 percent greater odds of being in a higher rather than a lower category of LS. The net income category of less than USD 1089 is the basis of the net income range analysis.

The odds ratio of 0.833 for the 35–44 age group means that this category is associated with 16.7 percent lower odds of being in a higher rather than a lower category of LS. The odds ratio of 0.869 for the 45–54 age group tells us that this category is associated with 13.1 percent being in a lower than a higher category of LS. The odds ratio of 0.844 for the 55–64 age group indicates that this category is associated with 15.6 percent lower odds of being in a higher than a lower category of LS. Finally, the odds ratio of 0.824 for the 65+ age group tells us that this category is associated with 17.6 percent lower odds of being in a higher than a lower category of LS. The youngest age group of 14–34 years old is the basis of the age range analysis.

The odds ratio of 1.437 for tertiary education means that having higher education (after 12 years of formal education) is associated with 43.7 percent greater odds of being in a higher rather than a lower category.

For the environmental concern variable, the odds ratio of 1.093 means that people who are worried about the environment have 9.3 percent greater odds of being in a higher rather than a lower category of LS.

In contrast, the odds ratio of 0.822 for gender means that being a female is associated with 17.8 percent lower odds of being in a higher than a lower category of LS. For ethnicity, the odds ratio of 0.707 means that being Indigenous is associated with 30.3 percent lower odds of being in a higher rather than a lower category of LS. For poverty, the 0.616 odds ratio means that being poor is associated with 38.4 percent lower odds of being in a higher rather than a lower category of LS. Finally, for religion, the 0.833 odds ratio means that being religious is associated with 16.7 percent lower odds of being in a higher rather than a lower category of LS.

The odds ratio of 0.851 for being divorced, separated, or widowed means that this marital status is associated with 14.9 percent lower odds of being in a higher than a lower category of LS relative to being married or in a de facto union. Also, the odds ratio of 0.793 for being single indicates that this type of marital status is associated with 20.7 percent lower odds of being in a higher than a lower category of LS relative to being married or in a de facto union.

Ecuadorians who live in the Coast and Eastern regions seem to have higher levels of life satisfaction after seven years with the new Constitution. The Coast region had a higher odds ratio for increased LS in 2014 than in 2007. The Eastern region increased the odds ratio for higher LS in 2014 (the Eastern region was not significantly related to higher LS in 2007). The coefficients of control variables in both years of analysis are similar in terms of their signs and significance.

6. Results and Discussion

The models presented in this study used an ordinal logit regression and cross-section model to estimate the proportional change in three categories of life satisfaction, from unsatisfied to satisfied to very satisfied, related to several explanatory variables, for 2007 and 2014. Odds ratios, p-values at the 1% and 5% levels, and other tests were considered to interpret the findings.

The pseudo-R² values are low in all models. The McFadden statistic (pseudo-R²) in logit models points to a low degree of explanation of the variables. All models have a low pseudo-R², but in the full model with all variables for both 2007 and 2014, the pseudo-R² increased.

The regression model in 2007 showed that most of the controlled variables mentioned in the literature review were significant. The odds ratios for living on the Coast, having a higher income, having tertiary education, experiencing job satisfaction, and being married are associated with greater odds of being in a higher rather than a lower category of life satisfaction. However, the odds ratios for being older, living in rural areas, being female, being Indigenous, and being in poverty are associated with lower odds of being in a higher rather than a lower category of LS. The findings about tertiary education align with those of Nicola, Bravo, and Sarmiento [18], who reported higher levels of well-being in Ecuador with increasing years of education.

In 2007, the variables with the most significant odds ratio values and the most influential factors were tertiary education, job satisfaction, and income (the highest category). As the literature mentions, the higher people’s satisfaction with their job, marital, educational, or financial situation, the higher their level of well-being.

In the 2014 model, people living in the Coast region rather than the Eastern region, having a higher income, having tertiary education, being married or in a de facto union, being young, being male, experiencing job satisfaction, and being environmentally concerned (heads of households who are aware that caring for the environment has a positive relationship with well-being) have a greater probability of increased life satisfaction.

Having a higher income (2007) and a higher net income (2014) was associated with a greater likelihood of being in a higher rather than a lower category of life satisfaction in all models. Job satisfaction (available only for 2007), tertiary education, and being married or in a de facto relationship were associated with a greater likelihood of being in a higher rather than a lower category of LS relative to not having a tertiary education and being single, separated, or widowed.

These findings resonate with much of the existing research, which consistently shows that factors like income, education, and job satisfaction play a vital role in shaping how people perceive their well-being [8]. As seen in numerous studies, women, Indigenous communities, and those living in poverty often experience lower levels of life satisfaction, mainly due to systemic inequalities and limited opportunities. This trend is observed in both developing and developed countries, underscoring how deeply rooted structural barriers continue to impact vulnerable groups. However, Ecuador offers a unique perspective. The country’s 2008 constitutional reforms appear to have influenced well-being in meaningful ways, mainly by amplifying the impact of these factors in underserved regions such as the Eastern area.

One shocking discovery in this study is that older adults aged 65 and over in Ecuador report lower levels of life satisfaction than younger generations. This finding is striking because it contrasts with patterns in many Western countries, where older adults often enjoy higher well-being [5]. Japan, for example, despite having one of the world’s oldest populations, older adults there report much higher satisfaction levels than their Latin American counterparts [29]. This is owing largely to strong social safety nets, universal healthcare, and cultural traditions emphasizing respect and care for elders, providing economic stability and a sense of belonging. Unfortunately, this is not true for many older Ecuadorians, who often face limited pensions, healthcare inequalities, and weaker support networks. These challenges illuminate how institutional and cultural factors shape well-being differently across regions. Ultimately, this study underscores the urgent need for tailored policies in Latin America to address these gaps and improve the quality of life for older adults.

Limitations include low explanatory power and data comparability issues, which restrict definitive conclusions. Future research could use longitudinal data and richer datasets to explore causal relationships.

7. Conclusions

Implementing the 2008 reform could be associated with increased life satisfaction in 2014 in Ecuador. The data presented align with the regression model, showing that people were more satisfied in 2014 than in 2007, and that the youngest age group was more satisfied in both periods than the other age groups. Living in the Coast region was associated with greater odds of being in a higher rather than a lower category of LS. Living in the Central region was associated with greater odds of being in a higher rather than a lower category of LS rather than the Eastern region, but only in 2014. In contrast, being in poverty or extreme poverty, being Indigenous, and being female were associated with a lower likelihood of being in a higher rather than a lower category of LS. The age group of 65+ years was associated with the lowest likelihood of being in a higher rather than a lower category of LS relative to the youngest group. Finally, a better distribution of incomes, redistributive social policies, and better care for the environment could have helped Ecuador to improve its social welfare in 2014 and reduce poverty and income inequality in the Eastern region (where there is a higher concentration of poverty and inequality) [30]. This study concludes that there is an association between sustainable policies from the 2008 Constitution and the increase in subjective well-being for 2014.

The study suggests targeted policy interventions to improve life satisfaction, particularly for vulnerable groups such as older adults and marginalized communities. Strengthening social support systems, including pensions and universal healthcare, would address critical gaps faced by Ecuador’s elderly population. Additionally, policies should focus on reducing regional disparities by promoting localized economic development and ensuring equal access to education and job opportunities. Environmental concerns must also be prioritized through sustainable practices that align with well-being goals. These measures collectively aim to enhance life satisfaction and reduce inequalities.