Economic Development and Marriage Stability: Evidence for a Concave Relationship Between per Capita Income and Divorce Rate

Abstract

When the economy grows, people become wealthier and more autonomous from their intimate partners. This autonomy potentially has a negative impact on relationship stability: As per capita income increases, so does the divorce rate. Nevertheless, there is evidence that, after a certain income level, the divorce rate starts to decline, suggesting that the relationship between the two factors is not linear. The purpose of the current research is to examine the relationship between per capita income and divorce rate by analyzing historical data from the UK (obtained from the UK Office of National Statistics) and USA (obtained from the CDC and the Federal Reserve), as well as contemporary data encompassing a sample of 107 societies (obtained from the World Population Review and the World Bank). Our analysis finds a concave relationship between the two variables: an increase in per capita income corresponds to a rise in the divorce rate. However, beyond a certain threshold, the increase in per capita income is associated with a decrease in the divorce rate.

1. Introduction

Rising divorce rates and a growing prevalence of singlehood indicate that marriages in modern post-industrial societies are becoming less stable (Apostolou et al. 2023a; Klinenberg 2012; Ortiz-Ospina 2019; Raley and Sweeney 2020). The complexities behind this heightened fragility are multifaceted, with numerous influencing factors. It has been argued that one such factor is the level of economic development (Cooke et al. 2013; Greenstein and Davis 2006). In particular, increased economic development is associated with less dependence on one’s intimate partner, resulting in higher fragility of marriage. Conversely, there exists some evidence suggesting a reversal of this trend in certain societies, where marriages exhibit greater stability (Cohen 2019). The current research aims to examine whether the relationship between per capita income and divorce rate is indeed non-linear using historical and contemporary data. We will first review the literature on how macro-level factors such as economic development are likely to influence divorce rate.

The Current Literature

The existing literature predominantly focuses on divorce rates and their correlation with macro-level factors. These inquiries link escalating divorce rates to structural shifts such as urbanization and increased incomes, which offer alternatives to traditional family roles (Cole and Powers 1973; Goode 1993). Moreover, the rising status, education, and career opportunities for women enable greater independence from marriage (Clark 1990; Cooke et al. 2013; Greenstein and Davis 2006; Schwartz and Gonalons-Pons 2016). Economic development exerts influence on societal values, catalyzing shifts in family behaviors (Jeng and McKenry 1999). Scholars highlight factors like individual freedom, post-materialist values, gender equality, secularization of marital behaviors, and individualism as negatively affecting marriage stability (Lesthaeghe and Surkyn 1988; Inglehart and Norris 2003; Beck and Beck-Gernsheim 2002).

Schneider and Hastings (2015) observed a negative impact on many families in the United States due to the economic recession, attributing this as a driver for increased divorce rates. Conversely, Hill (2015) primarily attributed the reasons to the foundation upon which marriages are established, noting that marriages formed during challenging economic times were more likely to endure compared to those facing sudden economic turbulence. Wang and Schofer (2018) argued that global institutions legitimize cultural principles such as individualism, human rights, and gender equality, thereby reshaping modern interpretations of marriage and family relations. Their panel regression models, analyzing 84 countries between 1970 and 2008, demonstrated a robust association between global cultural influence and divorce rates. Alola et al. (2020) examined the association between divorce rates, per capita income, and unemployment rates using panel data from 33 OECD countries spanning 1995–2016. Their model suggested a significant negative impact of per capita income on the long-term divorce rate. Additionally, they found that increased female unemployment levels correlated with a decreased divorce rate, while increased male unemployment predicted an increase in divorce rates in the long run.

The current literature identifies macro-level factors, notably economic development, as having a negative impact on relationship stability, evident in the increase in divorce rate. However, in some instances a decline in divorce rate has been observed despite increasing economic growth (Cohen 2019). For instance, according to the American Community Survey data from the Census Bureau, in 2019, for every 1000 marriages in the last year, only 14.9 ended in divorce—the lowest rate observed in the last 50 years (United States Sensus Bureau). The observed decline in divorce suggests that its relationship with economic development is not linear. There have been some attempts to account for this decline in divorce rate, linking it also to macro-level factors. More specifically, Hankins et al. (2011) argued that couples experience improved relationship quality following a sudden positive income shock, leading to greater marital stability as family income improves. González-Val and Marcén (2017) found that the rate of unemployment was negatively related to divorce rates. In particular, a one-percentage-point increase in the unemployment rate involves almost 0.02 fewer divorces per thousand inhabitants. Killewald et al. (2023) argued that wealth may stabilize marriage by easing financial stress. They showed that having $40,000 in wealth rather than $0 is associated with as big a decline in average predicted divorce risk as having $400,000 rather than $40,000. Additionally, some recent studies have linked a decline in divorce rates with the COVID-19 epidemic (e.g., Fallesen 2021). Yet, the data from the USA discussed above indicate that a decline in the divorce rate was observed prior to the epidemic.

Overall, the current literature has identified economic development as an important predictor of divorce rate; however, the relationship between the two factors has not been adequately explored. That is, most studies attribute or produce evidence supporting an increase in divorce rate to economic development, hinting at a linear relationship between the two. Yet, this is not the full story, as there is evidence that divorce rate may start declining as an economy keeps growing, hinting at a non-linear relationship between the two factors. Consistent with this argument, Elmi and Mohamadi (2021), using panel data from Iran, found evidence for a U shape effect of per capita income on the probability of divorce: the probability of divorce in low- and high-income groups was higher than the middle-income group. The current research aims to contribute to the existing literature by providing further evidence that the relationship between economic development proxied by the per capita income and divorce rate is indeed non-linear.

2. Materials and Methods

2.1. Data Sources

We employed two measures of divorce: the Crude Divorce Rate (CDR) and the refined divorce rate (RDR). The CDR is determined by dividing the number of divorces in a given year by the total population of the country or region in that year. Conversely, the RDR is calculated by dividing the number of divorces by the number of marriages. The latter provides a more accurate depiction of divorce likelihood by accounting for the actual number of marriages. For economic development, we used per capita income, computed by dividing the Gross Domestic Product (GDP) by the population. GDP represents the total market value of all final goods and services produced within a country within a specified period, while population denotes the number of individuals residing in a country at a particular time.

In our study, we used historical data from the UK and USA, alongside current data from 107 countries. For the UK, data on the refined divorce rate from 1960 to 2021 were obtained from the UK Office of National Statistics (ONS). Concurrently, GDP per capita data for the same period were sourced from the Federal Reserve. The GDP per capita for historical data was represented in real US dollars, adjusted for inflation to reflect the dollar’s purchasing power over time.

Regarding the USA, divorce rate data were sourced from the Centers for Disease Control and prevention (CDC). As the CDC reported crude divorce rates, data on the number of marriages and divorces from 1960 to 2021 were obtained, and the refined divorce rate was calculated by dividing the latter by the former. GDP per capita in real US dollars for the same period was sourced from the Federal Reserve. A summary of the data for the UK and the USA is provided in Appendix A.

Additionally, we acquired the latest available crude divorce rates for 107 countries from the World Population Review. GDP per capita data for these countries were obtained from the World Bank in current US dollars, which were then converted to 2020 US dollars. Due to varying years of reported divorce rates, each divorce rate was matched with the respective year’s per capita income. For instance, if China’s latest available divorce rate was from 2018, the per capita income employed was for China in 2018. A summary of this data is provided in Appendix B.

2.2. Data Analysis

For data analysis, we utilized regression analysis with the divorce rate as the dependent variable and both per capita income and per capita income squared as the independent variables. In essence, we sought to fit a quadratic model represented as Y = a + bX + cX². This analysis was conducted separately for the UK, USA, and the dataset comprising 107 countries. Furthermore, regression outcomes can be influenced by outliers (Salini et al. 2015). To address this, we employed the Box-plot analysis to identify and subsequently eliminate any extreme values from the dataset.

3. Results

Beginning with the UK, Figure 1 indicates a non-linear relationship between per capita income and divorce rate. Subsequently, there is a downward trend, with divorce rate decreasing as the income level increases. With respect to regression analysis, Box-plots did not reveal any extreme values, so no data points were removed. The results are presented in

Table 1. Regression results for historical UK, USA, and the contemporary cross-cultural sample of societies.

Sample	GDP per Capita		GDP per Capita Squared
	Coefficient (Standardized)	p-Value	Coefficient (Standardized)	p-Value	Adjusted R-Squared
UK	6.99	<0.001	−6.53	<0.001	0.92
USA	6.76	<0.001	−6.63	<0.001	0.75
Full sample (extreme values included)	0.53	0.024	−0.34	0.147	0.05
Sample of 99 societies	0.94	0.003	−0.70	0.028	0.11

, where we can see that the quadratic model was significant. In addition, the GDP per capita squared argument of the model had a negative coefficient indicating a concave relationship between income and divorce rate. The R-squared value was over 90%, indicating that the model fit the data well. Based on these results, divorce rates are predicted to be highest when per capita income is roughly 37,000 USD.

A similar relationship was observed for the US data. More specifically, as we can see from Figure 2, initial increases in per capita income were associated with a sharp increase in divorce rate. In terms of statistical analysis, no extreme values were detected, so no data points were removed. As we can see from Table 1, the quadratic model was significant. As in the case of the UK, the relationship was concave. Furthermore, the R-squared value was over 75%, indicating that the model fit the dataset well. Based on these results, divorce rates are predicted to be highest when per capita income is roughly 46,800 USD.

Turning to the cross-cultural sample, the Box-plot analysis identified eight extreme cases (Belarus, Bermuda, Kazakhstan, Liechtenstein, Luxembourg, Maldives, Monaco, and Russia), which were consequently removed from the dataset. The remaining 99 observations are depicted in Figure 3, where we can observe a concave relationship between income and divorce rate, which is not however, as pronounced as in the cases of the UK and the USA. As income increases, so does the divorce rate, with a maximum reached at around $40,000. After this level of income, the divorce rate starts declining. From Table 1, we can see that the quadratic model was significant, indicating a concave relationship between per capita income and divorce rate. Yet, the R-squared value was lower than that observed in the models fitted for the UK and the USA. Based on these results, divorce rates are predicted to be highest when per capita income is roughly 35,930 USD.

4. Discussion

Our analysis identified a concave relationship between divorce rate and per capita income. In particular, our analysis revealed a significant quadratic relationship between per capita income and divorce rates, utilizing historical data from the UK and USA, as well as contemporary data from a sample of 107 societies. Initially, an increase in per capita income corresponded to higher divorce rates, but beyond a certain threshold, further income increases were linked to a decline in divorce rates.

Our findings suggest that increases in economic development, as reflected in per capita income, have a strong negative impact on relationship stability, as indicated by the divorce rate. Specifically, as income increases, there is a sharp rise in the divorce rate. As discussed in the introduction, increased financial independence can potentially account for this finding. In particular, marriage constitutes an alliance between two parties, aiming to enable them to effectively face life’s challenges, including the cost of raising children (Coontz 2006; Fisher 2017). However, as individuals become wealthier, they also become less dependent on their spouses, weakening the marital bond (Cooke et al. 2013; Greenstein and Davis 2006). For example, during most of the previous century, women were primarily devoted to maintaining the household and heavily relied on their husband’s income. This financial dependence reduced women’s willingness to seek divorce, even if they were dissatisfied with the relationship. The increasing participation of women in the labor force over the last few decades has enhanced their financial autonomy. Consequently, they are now more willing to seek divorce if they are dissatisfied with their marriage.

While higher financial independence could potentially explain the positive association between per capita income and divorce rate, it does not account for why, after a certain income level, the divorce rate starts to decline. There are likely other factors at play. We propose that one such factor is an increasing dependence on the spouse for emotional support and social interaction. An intimate partner serves not only as a source of financial support but also as a crucial source of emotional well-being. For instance, a study involving Greek-speaking participants assessed several potential benefits of being in an intimate relationship (Apostolou 2022; Apostolou et al. 2023b). It was found that having someone to provide support and engage in shared activities were considered among the most important benefits of an intimate relationship. Additionally, emotional support and social interaction are also derived from friendships (Apostolou et al. 2021). Nonetheless, as societies become wealthier, they tend to exhibit more individualistic tendencies, resulting in people having fewer close friends, and loneliness becoming a significant issue in post-industrial societies (Cacioppo et al. 2014). It is plausible that as economic development increases and social relationships deteriorate, individuals become increasingly reliant on their intimate partners for emotional support and social interaction. In effect, such dependence strengthens the bond within the relationship and reduces the willingness of the parties involved to seek divorce.

Becker (1981) argued that a marriage ends when the expected utility of staying together falls below that of divorce; the reverse keeps couples married (see also Huber and Spitze 1980). Our reasoning fits within this framework. As incomes rise and partners gain financial independence, the perceived value of marriage may drop for some individuals, eventually falling beneath the utility of divorce and pushing divorce rates upward. Beyond a certain level of economic development, however, societies become more individualistic, and people rely more on partners for emotional support and social interaction. At this stage, the utility of marriage increases, the appeal of divorce diminishes, and divorce rates decline.

The hypothesis that, after a given level of economic growth, the intimate relationship bond strengthens as people come to rely on their partners for social interaction and emotional support, could be empirically tested using psychological research methods. For instance, future studies might measure personal income and explore its correlation with the importance individuals assign to the material resources provided by their partners. Additionally, evaluating individuals’ satisfaction with social input and emotional support from friends and family, along with assessing the significance they attribute to the social input from intimate partners, could enrich this investigation. Furthermore, this hypothesis pertains to relationship stability in general, beyond marital relationships. While divorce rates serve as a reasonable proxy for relationship stability, additional data on stability in non-marital intimate relationships would be beneficial. Future studies might explore individuals’ willingness to sustain their current relationships and examine potential associations with the significance attributed to both material and non-material contributions from their partners.

Over the past few decades, marriage rates have declined and divorce rates have risen in contemporary post-industrial societies (Raley and Sweeney 2020), making singlehood increasingly common (Apostolou et al. 2023a). Assuming continued growth of the world economy, our data indicate that these patterns are likely to reverse in several societies. In particular, as societies become wealthier and more individualistic, single people may find it difficult to maintain close friendships and may increasingly experience loneliness and sadness (Cacioppo et al. 2014). This experience, in turn, could make the prospect of a long-term committed relationship or marriage—where one has a supportive intimate partner—more appealing. People may therefore become more willing to invest the effort and make the compromises needed to attract and retain intimate partners in order to avoid being alone in an individualistic world.

Our data reveal a concave relationship between divorce rates and per capita income. Yet, with few exceptions, each passing year is associated with increases in economic growth and, consequently, per capita income. It follows that per capita income is strongly linked to the yearly timeline. However, the yearly timeline may affect the divorce rate independently of economic development. For instance, as years go by, societies may become more liberal, more sensitive to individual rights, and may develop more sophisticated welfare systems. These societal changes can impact the divorce rate. Accordingly, the observed relationship between divorce rate and per capita income may partially reflect an association with the yearly timeline. Yet, the concave relationship was also observed in a cross-cultural sample where per capita income was not directly associated with the yearly timeline, as current data points were used. Therefore, the patterns we observed cannot be solely attributed to societal and cultural changes that occur over time, as per capita income indeed has a real effect on divorce rates.

Limitations

Our study identified a significant quadratic relationship between divorce rates and per capita income, hinting at causation. Yet, due to the correlational nature of our data, establishing a causal relationship remains tentative. In addition, the divorce rate and effects on the divorce rate can often represent a complex number of factors and factor interactions that can vary country by country, which the present research has not investigated. Moreover, using per capita income as a measure of economic development overlooks income distribution’s role. Societies with identical per capita income might differ vastly in wealth distribution. In one, a minority may hold significant wealth, making them less reliant on partner contributions, while in another, wealth equality might prompt similar expectations from partners. To refine our understanding, future studies should incorporate indices of income inequality. Additionally, our study examined historical data from only two societies, warranting further investigation across diverse cultural contexts to verify the proposed relationship between divorce rates and per capita income.

5. Conclusions

Divorce rate and per capita income are related, and in the current research, we have offered evidence that this relationship is non-linear. More specifically, as people become wealthier, the divorce rate increases, but after a certain point of wealth, the divorce rate declines. We have argued that one possible reason for the observed pattern is that as economic development increases, social relationships tend to deteriorate, with individuals becoming increasingly reliant on their intimate partners for emotional support and social interaction, and thus, less likely to end the relationship. Yet, there are different ways through which per capita income may affect the divorce rate, giving rise to the observed patterns, and future theoretical and empirical work is necessary to identify these pathways.