The Impacts of Social Protection Policies and Programs on Suicide: A Literature Review

Abstract

Despite vigorous interest in showing the association between economic determinants and suicide, not many studies have focused on the social protection that can moderate the detrimental impact of the economic environment on suicide. This article is the first to review the relationship between suicide and social protection. In this article, I summarize the empirical findings and theoretical approaches in published papers on the relationship between suicide and social protection, and I identify knowledge gaps for future studies. The review included all quantitative and qualitative articles published in peer-reviewed journals, regardless of study setting, language, and time period. Among 19 papers meeting the inclusion criteria, 16 studies reported at least one negative association, 2 studies failed to prove a statistical association, and 1 study showed ambiguous results. However, due to the heterogeneity of contexts, the diversity of indicators of social protection, and the paucity of theoretical mechanisms for interpreting the results, further research is required in this area.

Keywords

social protection suicide review welfare policy

Suicide can be defined as an individual’s intentional ending of his or her own life.¹ Suicide is a serious public health problem: the number of people who die by suicide per year was estimated at 800,000, and it was the 10th leading cause of death all over the world and the second leading cause of death among 15–29 year olds in 2010.² Generally, suicide is conceived of as the tragic result of individual mental problems, as mental disorders, depression, impulsive tendencies, and feelings of hopelessness are often cited as well-known risk factors of suicide.³ Nevertheless, ever since Durkheim tried to grasp suicide as a social fact in his seminal work Le Suicide, there have been some efforts to understand the sociopolitical and economic factors that affect suicide, just as with other health outcomes.

Among these sociopolitical and economic factors, the most frequently addressed determinants of suicide are the economic environment and economic activities. The associations between economic environment and activities and suicide are mainly understood in the sociological or economic literature in the following ways: (1) that changes in macroeconomic environments can have an impact on population-level suicide rates by affecting social integration and/or regulation, and (2) that shifts in individuals’ material well-being may be associated with suicide behaviors in the case that these shifts cause one’s expected lifetime utility to drop below zero.^4–6 Building on these theories, many empirical studies have investigated associations among income, unemployment, economic recession, economic deprivation, and suicide. However, despite vigorous interest in showing an association between economic determinants and suicide, few studies have focused on social protection, which can moderate the harmful impacts of the economic environment.

One of the most frequently used definitions of social protection is from the International Labor Organization (ILO), which defines social protection as

the set of public measures that a society provides for its members to protect them against economic and social distress that would be caused by the absence or a substantial reduction of income from work as a result of various contingencies.⁷

Although debates concerning the definition of social protection are ongoing and complex, the common element in the different definitions is that social protection is composed of policies for coping with conditions that may adversely affect human well-being. Various terms, such as “social security,” “welfare,” “social insurance,” “social safety nets,” and “social security,” are used interchangeably to refer to social protection.⁸ In high-income countries, social protection generally consists of social insurance, social assistance, and labor market interventions. Social insurance refers to contributory programs that provide protection against life-course and work-related contingencies, such as sickness or unemployment, while social assistance can be defined as the support provided by the government out of tax revenue to those in states of poverty and vulnerability.⁹ Labor market interventions are also regarded as a normal means for social protection because their primary purpose is to ensure basic standards and rights for people who can work.⁹

Despite the significance of the role played by social protection, no literature review has been done to date on the relationship between social protection and suicide. This article aims to fill that gap. The purpose of the review is to summarize the empirical findings and theoretical approaches of the existing literature on the relationship between suicide and social protection and to identify knowledge gaps for future studies. I will first summarize the findings of previous studies, then address some of the issues in methodology and theoretical mechanisms that arise in researching social protection and suicide, and finally identify the gaps in knowledge that should be investigated so that we may ensure that future studies are of high quality.

Method

Search Strategy

This review searched for articles published in peer-reviewed journals pertaining to social policy and mental health. The following electronic databases were employed: MEDLINE, PsychInfo, Worldwide Political Science Abstracts, International Bibliography of the Social Sciences, and Google Scholar. No limitation was placed on time and study location. Although this review included all studies regardless of language, the only studies included were those that could be searched for in English in these databases. In addition, the references of all eligible articles were reviewed, as well as any articles that cited the chosen articles. Search keywords included “social protection,” “social policy,” “social insurance,” “social security,” “income transfers,” “social transfers,” “social welfare,” “welfare,” “welfare spending,” “social spending,” “public policy,” “public spending,” “social assistance,” “social expenditure,” “labor market intervention,” “unemployment insurance,” and “employment protection.” A second search term of “suicide” was also applied.

Data Extraction

The review included studies that met the following criteria: (1) they set suicide rates at societal levels and/or suicide death at individual levels as dependent and/or output variables, (2) they addressed any form of social protection that meets the definition given by the ILO as independent and/or explanatory variables. Studies were excluded if they investigated only mental health investment. Also, studies that addressed other suicidal behaviors, such as suicidal ideation and suicide attempts, were excluded because suicide death is different in nature from other suicidal behaviors, in spite of their association.¹⁰ All studies were required to be original research (i.e., commentaries and editorials were excluded). This process is all presented in figure 1 (Figure 1).

Figure 1.

Flow chart of literature review process.

Database searching yielded a total of 2,621 articles, of which 49 articles were retained for full-text review. Of these, 35 articles were excluded because they did not meet the inclusion criteria, such as by addressing suicidal behaviors other than suicide mortality or by using explanatory variables that cannot be regarded as social protection (e.g., only health care or mental health care are considered). In addition to the electronic database search, 5 relevant studies were identified by references from the chosen studies and their authors based on previous research knowledge of the topic. Therefore, 19 articles were finally selected for full article review, including 3 articles written in Korean and 1 article written in Japanese.

Results

All study characteristics and findings are summarized in Table 1. Nineteen articles employed quantitative methods, and one¹¹ used a qualitative method in their study design and analysis. Among the quantitative studies, one¹² was a cross-sectional study, four^13–16 were repeated cross-sectional studies, and the rest used time-series analysis. The qualitative study conducted fuzzy set/Qualitative Comparative Analysis (fs/QCA), using cross-sectional datasets.¹¹ All studies that met the inclusion criteria used ecological designs.

Table 1.

Summary of Included Studies.

Author(s)	Study design	Setting and time studied	Suicide data and data source	Social protection and data source	Other explanatory variables or confounders	Findings
Zimmerman¹²	Cross-sectional study with ecological-level exposures (multiple regression)	50 U.S. states in 1982	Average total suicide rates from National Center for Health Statistics	State per capita spending for public welfare from the U.S. Bureau of Census	Population change rates, state income distribution, state unemployment rates, racial composition of states	State public welfare expenditures did not have statistical association with suicide, but all the variables together accounted for 0.73 of the variance in suicide rates.
Zimmerman¹³	Repeated cross-sectional study with ecological-level exposures (multiple regression)	50 U.S. states in 1960, 1970, 1980, 1985, and 1990	Average total suicide rates from National Center for Health Statistics	State per capita spending for public welfare from the U.S. Bureau of Census	State divorce rates, population change rates, population density per square mile, states’ unemployment rates, racial composition and gender ratio	Welfare expenditure’s relationship with suicide grew over the 30 years in a negative direction: r = 0.02 in 1960, −0.19 in 1970, −0.36 in 1980, −0.42 in 1985, and −0.57 in 1990. In 1985 and 1990, states’ welfare spending was associated with increased variance of states’ suicide rates (beta = −0.18 and −0.21, respectively) and (R² = 0.83 and 0.84, respectively)
Zimmerman¹⁷	Time series study with ecological-level exposures (fixed effects)	50 U.S. states in 1960, 1970, 1980, 1985, 1990, and 1995	Average total suicide rates from National Center for Health Statistics	State per capita spending for public welfare from mortality U.S. Bureau of Census	States’ divorce rates, population change rates, population density per square mile, states’ unemployment rates, racial composition and gender ratio, and year/state dummy	Welfare spending remained significant and negative: for 1 dollar reduced in states’ per capita spending for welfare, suicide rates rose by 0.004 per 100,000 between 1960 and 1995.
Minoiu and Andres¹⁸	Time series study with ecological-level exposures (GMM)	47 U.S. states during period 1998–1997	Age-adjusted total, male, and female suicide rates among people aged 20–69 from the Centers for Disease Control and Prevention	Share of public health spending or welfare spending in the state budget from the Statistical Abstract of the U.S. Bureau of the Census	Gini coefficient of income inequality, share of migrants in population density, unemployment rate, mountain state dummy, divorce rates, sunshine, and year/state dummy	An increase of the share of public health expenditure and welfare expenditure by 0.01% points would lead to a reduction in total suicide rates by 1.54% and 0.74% in the following year. The effect was significant and robust for male and total suicide rates.
Flavin and Radcliff¹⁴	Repeated cross-sectional study with ecological exposures (multiple regression)	50 U.S. states from 1990 to 2000	Average total suicide rate per 100,000 from McIntosh, John L. U.S.A. SUICIDE: State and Regional Data	States’ spending per capita on transfer payments, medical benefits, family assistance, total state spending from Bureau of Economic Analysis	Social capital, state’s rate of residential mobility, divorce rate, and unemployment rate	Higher transfer payments, medical benefits, and family assistance were all related to reducing suicides (beta = −1.275, −2.579, and −22.394, respectively) and (R² = 0.80, 0.80 and 0.81, respectively), but total spending did not show statistical significance. Higher social capital index was associated with higher suicide rates.
Park et al.¹⁵	Repeated cross-sectional study with ecological exposures (multiple regression)	27 OECD countries from 1980 to 2003	Average age-adjusted male and female suicide rates and their annual rates of change from OECD health data	Share of public social expenditure in GDP from the OECD health data	GDP per capita, annual change in GDP per capita, male unemployment rates, annual change in male unemployment rates, divorce rates, fertility rates, alcohol consumption, male life expectancy, and female life expectancy	Public social expenditure did not have statistical associations with suicide rates for both males and females, but it had negative association with annual change rates in male and female suicide rates (beta = −0.227 for male and −0.276 for female).
Stuckler et al.²¹	Time-series study with ecological exposures (fixed effects)	17 European countries from 1980 to 2003	Age-adjusted and age- and gender-specific suicide rates from the WHO European Health	Spending on ALMP, unemployment cash benefits, family support, health care and housing support from the OECD health data	Changes in unemployment rates, year dummy and country-year trends from the WHO European Health	Every US$10 per person increased investment in ALMP and family support decreased the effect of unemployment on suicide rates by 0.038% and 0.023%, respectively. ALMP had the strongest effect among all spending on social protection.
Lee¹¹	Cross-sectional study with ecological exposures; qualitative study (fs/QCA)	24 OECD countries in 2005	Gender-adjusted suicide rates among the elderly aged 65–74 from OECD	Share of welfare spending for the elderly in the total welfare spending from OECD social expenditure from the WHO European Mortality Database and the Eurostat database	GDP per capita, poverty rates among the elderly, life satisfaction score among the elderly	The combination of high poverty rates, low share of welfare spending, and low level of life satisfaction were causal conditions of high elderly suicides in South Korea and Switzerland, but a combination of low poverty rates, low share of welfare spending, and high economic growth rates explained low suicide rates in Luxemburg, Norway, Canada, and Denmark.
Matsubayashi and Ueda²⁷	Time-series analysis with ecological exposures (fixed effects)	21 OECD countries from 1980 to 2004	Age-adjusted total and gender-specific suicide rates and 3 different age groups (males and females aged under 24, 25–64, and over 65) from the WHO Mortality Database	The total government social expenditures and unemployment benefit replacement rates from the OECD Stat Extracts	The national suicide prevention program, left party, Christian Democratic party, GDP growth rates, unemployment rates, inflation rates, economic inequality, GDP per capita, union density, female labor participation, divorce rates, dependent population, population size, and time/country dummy	Social expenditure was the strongest variable to account for suicide rates among most population groups (beta=20.946 among the total population). Unemployment benefit replacement did not show statistical significances among working population.
Ross et al.¹⁹	Time-series analysis and cross-sectional study with ecological exposures (GMM)	51 U.S. states from 1997 to 2005	Age-adjusted male and female suicide rates (all age groups and those aged 25–64) from Centers for Disease Control WISQARS injury mortality reports	State per capita welfare spending on mental health and public health, states’ total welfare spending from U.S. Bureau of Census and National Association of State Mental Health Directors Research Institute	Per capita personal income, unemployment rates, population density per mile, net migration, divorce rates, proportion of non-white, mountain state dummy, number of sunny days in the year, and state/year dummy	Comparing the results of fixed effects and generalized method moments (GMM) showed that states’ per capita welfare spending on mental health and total welfare spending have no significant effects on both male and female suicide rates. Public health spending per capita negatively associated with suicide rates among males (beta = −0.038).
Cheong et al.¹⁶	Repeated cross-sectional study with ecological exposures (multiple regression)	248 cities in South Korea from 2006 to 2008	Age-adjusted male and female suicide rates (2 age groups: the elderly and non-elderly) from Statistics Korea	Share of the social welfare budget in regional general accounting	Single household, female householder, divorce or separated, poor house environments, housing in non-apartment overcrowding, lower education level, lower social class, and no car	The statistical relationship between regional suicide rates among both the elderly and non-elderly had no statistical significance.
Yur’yev et al.²²	Time-series analysis with an ecological exposure (correlation)	17 European countries (country-specific analysis) from 1980 to 2005	Age-adjusted gender-specific suicide rates from the WHO European Mortality Database	Per capita social expenditure in GDP from the OECD health data	None	There were negative correlations between social expenditure and suicide trends and the correlations were stronger among males. There were no correlations in Spain, Poland, and Ireland among males and in Spain among females.
Cylus et al.²⁰	Time-series analysis with ecological exposures (fixed effects)	51 U.S. states from 1968 to 2008	Age-adjusted total suicide rates from the Centers for Disease Control and Prevention	Maximum unemployment insurance benefits from the Employment and Training Administration of the U.S. Department of Labor	States’ unemployment rates, state-specific linear trends, age cohort, sex cohort, population size, percentage of the population that had graduated from college, average real state wages and salaries, and year/state dummy	The negative interaction between benefits and unemployment rates (beta = −0.57) on suicides suggested that higher unemployment benefits offset the effects of unemployment rates on suicide rates, but the main impact of maximum unemployment benefit was not significant. Generosity of unemployment benefit did not have direct effects on suicide rates.
Baumbach and Gulis²³	Time-series analysis with ecological exposures (correlation)	8 EU countries from 200 to 2010	Age-adjusted total suicide rates from Eurostat	Share of social spending in GDP	Unemployment rates, GDP growth rates	The correlation between unemployment rates and suicide mortality was higher with a lower social spending (r = −0.83).
Lee and Kang²⁹	Time-series analysis with ecological exposures (random effects)	256 districts in South Korea from 2005 to 2011	Age-adjusted total suicide rates from Statistics Korea	Share of welfare spending in regional general accounting at community level	Population size, proportion of the elderly of population, GDP per capita at community level, employment rates, divorce rates, numbers of doctors per 1,000, and trend dummy	Increased share of welfare spending was associated with decreased suicide rates at district levels (beta = −0.11).
Shibata²⁸	Time-series analysis with ecological exposures (GMM)	26 OECD countries from 1980 to 2007	Age-adjusted total suicide rates from OECD Health data	Share of ALMP in GDP from the OECD social expenditure database	Real GDP per capita, change rates in unemployment, average working hours per person a year, unionization rates, women’s labor market participation rates, proportion of the elderly of population, marriage rates, fertility rates, divorce rates, alcohol consumption per person, share of revenue from tax and social insurance in GDP, share of other expenditure except for social spending in GDP, share of expenditure for the elderly in GDP, share of health expenditure in GDP, share of government expenditure on disability in GDP, year/state dummy, Japan dummy, and trend dummy	ALMP had an impact on decreasing suicide rates (beta = −0.955) and the impact was much stronger in Japan (beta = −21.56), controlling for all the variables. Other social programs did not have any impacts on suicides.
Norström and Grönqvist²⁴	Time-series analysis with ecological exposures (fixed-effects)	30 countries (European and some English-speaking countries) from 1960 to 2012	Gender-specific mortality data for the age groups 20–64 years and 65+ from the WHO Mortality Data Base	Five welfare state regimes (a proxy of welfare generosity) Eastern European Southern European Anglo-Saxon Bismarckian Scandinavian classified by unemployment replacement rates	Unemployment rates and country dummy	Unemployment rates increased most in Eastern and Southern European countries with low unemployment protection and the increased unemployment in these countries had the strongest impacts on increased suicide rates. There was a clear gradient among the different type of welfare regimes in the effects.
Antonakakis and Collins²⁵	Time-series analysis with ecological exposures (GMM)	Greece, Ireland, Italy, Portugal, and Spain from 1968 to 2012	Age-adjusted and age-specific male and female suicide rates among those aged 10–89 from the WHO Mortality Database and the Eurostat.	Total government expenditure, Employment Protection Legislation Indicator (version EPRC_V1) and unemployment benefit (the gross replacement rate) from the OECD labor market statistics database	Unemployment rates, share of government expenditure in GDP (a proxy of fiscal austerity), growth rate of per capita GDP, fertility rates, per capita alcohol consumption, divorce rates, interaction between EPL/unemployment benefit and share of government expenditure in GDP, and time/country dummy	Unemployment benefits had negative impacts on suicide rates among both males and females aged 25–44 years old and the elderly among males. EPL had negative impacts on suicide rates among females across all the age groups, except for those aged 10–24 years old.
Reeves et al.²⁶	Time-series analysis with ecological exposures (fixed effects)	24 European countries from 1981 to 2011	Age-adjusted and age-specific male suicide rates among those aged 25–64 years and those aged 65+ from the WHO Human Mortality Database	Moderators (unemployment Ells, spending on ALMP, and total social protection spending) from the OECD social Expenditure Database	Unemployment, indebtedness, unaffordable housing, moderators (social capital and antidepressant daily dosage per 1,000), year dummy, year, interaction between changes in unemployment and the moderators, and year and country-specific trend dummy	Every US$10 increase in investment on ALMP decreased the effect of a percentage increase in male unemployment on the male suicide rate by −0.026%. Unemployment benefits and social protection spending turned out to be not statistically significant.

Abbreviations: ALMP, active labor market program; EPL, employment protection legislation; EU; European Union; GDP, gross domestic product; OECD, Organisation for Economic Co-operation and Development; WHO, World Health Organization.

Study Setting and Times

Seven studies^{12–14,17–20} targeted the United States, 6 studies^21–26 addressed European countries, 4 studies^11,15,27,28 investigated several Organization for Economic Cooperation and Development (OECD) countries, and 2 studies^16,29 examined cases in South Korea. Aside from three^12,13,17 pioneering studies, 16 of the articles were published after 2008, probably due to increased attention to the relationship between economic recession and its detrimental impacts on health outcomes. There may exist studies written in other languages, but they were not available in peer-reviewed international journals.

Measurement of Suicide

All suicide deaths were measured by number of suicides per 100,000 at the population level. Eight studies used only age-adjusted suicide rates.^{12–14,17,20,23,28,29} Ten studies^{15,16,18,19,21,22,24–27} conducted gender-specific analyses, and 7 studies^{11,19,21,25–27} conducted age-specific analyses. A single study¹⁵ set the rate of change in the suicide rate as a dependent variable.

Results of Studies and Synthesis of Findings

Table 2 summarizes the associations between different indicators of social protection and suicide in the chosen articles. Among the 19 articles reviewed, 16 reported a negative association between social protection indicators and suicide mortality, while three^11,12,16 showed a null association or ambiguous results. Since most studies set multiple indicators as explanatory factors, studies that showed at least one negative association are regarded as showing a negative association. The findings are listed according to 7 different social protection indicators. Afterwards, theoretical mechanisms for the link between social protection and suicide are suggested. It should be noted in advance that results from the same indicator were often inconsistent due to heterogeneity in study setting, timespan, and methods.

Table 2.

Positive, Negative, Null, and Ambiguous Associations Between Social Protection Indicators and Suicide.

	Negative	Null	Ambiguous
Total social protection spending	Zimmerman¹³, Zimmerman¹⁷, and Yur’yev et al.²²	Zimmerman¹², Ross et al.¹⁹, and Reeves et al.²⁶
Share of GDP spent on social protection	Minoiu and Andres¹⁸, Lee and Kang²⁹, Park et al.¹⁵, Matsubayashi and Ueda,²⁷ and Baumbach and Gulis²³	Cheong et al.¹⁶	Lee (2010)¹¹
Unemployment insurance benefits	Antonakakis and Collins,²⁵ Cylus et al.²⁰	Stuckler et al.²¹ and Reeves et al.²⁶	Matsubayashi and Ueda²⁷
Spending on ALMP	Stuckler et al.²¹ and Reeves et al.²⁶
Share of GDP spent on ALMP	Shibata²⁸
Spending on family support	Flavin and Radcliff¹⁴, Stuckler et al.²¹
Spending on public health	Ross et al.¹⁹
Spending on mental health		Ross et al.¹⁹
Spending on medical benefits	Flavin and Radcliff¹⁴
Spending on health care		Stuckler et al.²¹
Share of GDP spent on public health	Minoiu and Andres¹⁸	Shibata²⁸
Share of GDP spent on disability		Shibata²⁸
Total government spending	Antonakakis and Collins²⁵	Flavin and Radcliff¹⁴
Spending on housing support		Stuckler et al.²¹
Spending on transfer payments	Flavin and Radcliff¹⁴
EPL	Antonakakis and Collins²⁵
Types of welfare regime	Norström and Grönqvist²⁴

Abbreviations: ALMP, active labor market program; EPL, employment protection legislation; GDP, gross domestic product.

Total social protection spending

Six studies^{12,13,17,19,22,26} examined the role of total spending on social protection policies. Three studies reported a negative association between total spending on social protection policies and suicides: Zimmerman investigated the association between welfare spending per capita and suicide rates in the 50 U.S. states with different statistical models in 1987,¹² 1995,¹³ and 2002.¹⁷ The 2 studies^13,17 conducted with more rigorous designs reported significant negative associations. Another study showing a negative association conducted only correlation analysis for the European countries.²² Ross and colleagues examined the association between welfare spending and suicide in the United States and demonstrated that, although the results lacked statistical significance, welfare expenditure showed more reliable results than other variables despite multiple robustness checks.¹⁹ Reeves and colleagues investigated social protection spending as a potential moderator of the detrimental effects of the economic recession in Europe but found no effects.²⁶

Share of gross domestic product (GDP) spent on social protection

Of 7 studies, five^{15,18,23,27,29} reported a negative association between share of GDP spent on social protection, while 2 studies showed a null¹⁶ or ambiguous¹¹ association. Minoiu and Andres reported a consistent negative association for both lagged and non-lagged suicide rates in the United States in a model with and without outliers.¹⁸ Park and colleagues examined the negative association between changes in suicide rates and share of social expenditure in GDP among the OECD countries, using pooled OLS.¹⁵ Another study that used the same indicator and the same setting as Park and colleagues showed similar results using a fixed effects model.²⁷ Baumbach’s study reported that the impact of unemployment on suicide varied by level of social spending.²³

Meanwhile, 2 studies of regions in South Korea showed contradictory results. Lee and Kang found a negative association between suicides and the percentage of welfare spending in the GDP of local government using a random effects model.²⁹ Cheong and colleagues reported a null association using a model with pooled OLS.¹⁶ As opposed to other studies using share of social spending in government expenditure, Lee conducted an fs/QCA analysis using share of social expenditure for the elderly within total social expenditure.¹¹ The findings of the study were contradictory with respect to social protection and poverty: in South Korea and Switzerland, low levels of social protection, combined with high levels of poverty and low levels of life satisfaction, were associated with increased suicides among the elderly. By contrast, in Canada and Luxembourg, high levels of social protection in combination with low levels of poverty and high levels of GDP per capita were associated with increased suicides among the elderly.

Unemployment benefits and replacement rates

Five studies were concerned with the impact of unemployment insurance benefits. Two studies^20,25 showed negative associations, one²⁷ showed ambiguous results, and 2 other studies^21,26 showed a null association. A study²⁰ using data from the United States confirmed the statistical impact of the maximum amount of benefits in each state on suicide rates using a fixed effects model and multiple robustness checks. The other study²⁵ showing a negative association employed the gross replacement rates of unemployment benefits for both males and females in the working population in 5 European countries going through a severe economic recession. Two studies that failed to find a negative association used similar designs in the same contexts during similar periods of time, testing unemployment cash benefits as an interaction term with unemployment rates.^21,26 In a study showing equivocal results, the effect of unemployment insurance benefits was statistically significant for those aged 24 and below among males and 65 and above among females.²⁷ The validity of these results is dubious because those age groups are not likely to be beneficiaries of unemployment insurance.

Spending on active labor market programs (ALMP) and its share in the government budget

Two studies^21,26 confirmed the impact of spending on ALMP as a moderator on suicide during the economic recession. These 2 studies were similarly designed, except for the included confounding variables, and showed that ALMP moderated the detrimental impacts of economic recession on suicide, but the size of the impact was small in both cases. On the other hand, a study, which focused on the share of ALMP in GDP, also provided evidence of a negative association, and the impact of a higher proportion of ALMP turned out to be much stronger in Japan.²⁸

Family support

Two studies that tested the impacts of spending on family support all showed a negative association. One article,¹⁴ which investigated state spending per capita on transfer payments, medical benefits, family assistance, and total spending in the United Sates, demonstrated that, of the 4 types of spending, it was support for low-income families that had the strongest negative association with decreased suicide rates (see Table 1, coefficient of low-income family support = −22.394). The other study reported that, for each US$10 spent on universal family support, suicide rates were lowered by 0.023% during the recession in Europe.²¹

Spending on public health, mental health, medical benefits, health care, and its share in GDP

Three^14,18,19 of 6 studies showed a negative association. Ross and colleagues,¹⁹ examining data from the United States, found that, while spending on public health had a consistent negative association with state suicide rates under different estimation models, spending on mental health showed an inconsistent association under the same models. One article confirmed the impact of spending on medical benefits such as Medicare, Medicaid, and public assistance medical care on decreased suicides in the United States.¹⁴ Another study using multiple models also showed a negative association between public health expenditure and suicides in the United States.¹⁸ In contrast to those studies, which targeted the U.S. population, other studies^21,28 examining spending on health care and share of public health spending in European countries and OECD countries failed to show a statistical association.

Other social protection indicators

Share of GDP spent on disability support did not show an impact on suicide in one study that examined OECD countries.²⁸ Two studies^14,25 that examined total government spending reported contradictory results. A study investigating the moderating effects of housing supports also failed to prove an association.²¹ In the case of one study examining the impacts of transfer payments, including retirement and disability insurance benefits, medical benefits, income maintenance benefits, unemployment insurance compensation, veterans’ benefits, and education and training assistance, a negative association was shown.¹⁴ Another article showed that indicators of employment protection legislation (EPL), which measure the strictness and flexibility of employment protection legislation, have a negative association with suicide in 5 European countries. Finally, one study showed an association between 5 types of welfare state regimes with different levels of unemployment protection and suicide.²⁵

Theoretical Approaches to the Link Between Social Protection and Suicide

The chosen papers present 4 different theoretical rationales for the existence of a link between social protection and suicide: (1) the social integration approach, (2) the social regulation approach, (3) the psychosocial approach, and (4) the materialist approach. The most frequently presented theoretical backgrounds for the link in the selected papers are from social integration theory, mainly as developed by Durkheim.^{11–13,16,17,29} This theory defines social integration as the strength of a person’s ties to the macro society. These ties can be measured by governments’ welfare policies, because welfare policy implies the will of political institutions to make decisions to meet the needs of individuals.^12,13,17 Under this theory, the association between social protection and suicide has less to do with the impact of material factors per se than with the norms of mutual aid and support.^12,13,17 In addition, social integration theory proposes that non-economic roles (e.g., cognitive and informative aids of labor market intervention) of social protection protect the unemployed from social isolation, which results in decreased suicide behaviors.

The social regulation approach is also deeply affected by Durkheim’s anomie theory. According to this approach, a society’s normative system for regulating individuals’ behaviors is loosened by the instability of the market economy, which encourages a cynical, instrumental, and narrowly self-interested view of interpersonal relationships.³⁰ These collapsed social norms are claimed to result in the failure to regulate deviant behaviors, including suicidal behaviors.^14,30 The role of social protection is to act as a counterforce to the anomie-encouraging nature of the market economy.

The psychosocial approach can be understood in terms of the meaning of social protection, just as with the social integration approach. From this perspective, devotion to social protection policies is a sign of the government’s degree of preference for redistributive policies to offset disadvantaged social status.¹⁸ Thus, the will of the government for social fairness is likely to affect people’s levels of life satisfaction because people subjectively perceive that social protection plays a role in decreasing income inequality and that increased life satisfaction may lead to decreased suicide mortality.¹⁸

The most common approach presented in the selected articles is the materialist approach. Under a materialist approach, it is argued that social protection can provide a strong safety net to individuals.^{14,20,22,25,28} Three pathways from the financial support of social protection to the prevention of suicide are proposed. To begin with, social protection is claimed to act as a moderator for chronic and acute stressors deriving from decreased material well-being by improving the financial conditions among those in poverty and economic desperation owing to their vulnerable status in society.^14,20,22 Moreover, the existence of a safety net can also be helpful in improving the mental health status of those who are still employed yet face anxiety over potential unemployment due to economic uncertainty.^14,20 Lastly, financial support from social protection is argued to have a direct impact on the quality of marriages and other types of intimate personal relationships, which can be closely related to suicide.^14,18

Discussion

This is the first review to examine the role of social protection on suicide. The present study reviewed 19 published articles on social protection and suicide that employed both quantitative and qualitative methods. The results of this review indicate a negative association between social protection and suicide rates, although the studies varied with respect to setting, observation period, variables represented as social protection indicators, and other explanatory variables. Because of the heterogeneity and the limited number of studies, it is impossible to draw wholly consistent and clear conclusions. Several methodological and theoretical issues must be addressed when understanding the results and planning further research. These are (1) the lack of studies in low- and middle-income countries (LMIC), (2) the use of different social protection indicators, (3) methodological challenges, and (4) the paucity of theory.

Lack of Studies in LMIC

All of the selected studies aimed at high-income countries, and not a single paper exploring LMIC was identified in peer-reviewed journals published in English. This is in part attributable to the less-developed welfare state present in LMIC, affording scarce social protection.⁹ However, the effect of social protection on suicide must still be explored in LMIC. This is not only because LMIC bear a greater burden of suicide than high-income countries³¹ but also because social protection may have a stronger prevention impact in LMIC than richer countries. A systematic review that investigated the association between suicidal ideation, suicidal behavior, and economic poverty in LMIC reported that poverty and economic adversity are consistently associated with suicidal ideation and suicidal behavior at the individual level.³² This finding suggests that social policies implemented to moderate the detrimental impacts of poverty and unemployment in LMIC may prevent suicidal behavior and that the size of the impact is likely to be even bigger in LMIC than in high-income countries, considering Esterlin’s paradox, which states that the same amount of resources brings more life satisfaction and happiness to poor persons or poor countries than to rich persons or rich countries.³³

Uses of Different Social Protection Indicators

With respect to studies that investigated the impact of labor market intervention, the following variables were tested: generosity of unemployment insurance, investment on ALMP, and EPL indicators. There are several limitations in using these variables. First, most studies investigating the roles of total spending on unemployment benefits and ALMP only considered the total amount of money, instead of the per capita spending, which makes it possible for the results to be biased when the total number claiming unemployment benefits increased during the economic recession. Second, there are aspects that should be considered in measuring programs other than the quantity of money spent. For instance, unemployment generosity can be measured by replacement rates, duration of unemployment insurance, eligibility criteria, waiting time, and so on. Third, there are still many other labor market intervention policies that have not yet been addressed in examining their moderating role on suicide. Moreover, there are important policy areas which remain poorly investigated: changes to legislation over the laws regulating employment conditions such as minimum wages, precariousness and safe working conditions, despite the fact that one study examined EPL.

As for social insurance, the following indicators were used: unemployment insurance and spending on health care, public health, and mental health. The present literature review will skip discussion of unemployment insurance because it was addressed in the previous paragraph. It is noteworthy that the impacts of public health and health services were conflicting in the United States and other high-income countries. The absence of universal health care in the United States is a provisional explanation for the exceptional statistical impact of health care on suicide in the United States. This is because universal health care makes it possible for anyone easily to seek needed medical care during a suicide crisis. It also plays a role in redistributing income, thereby preventing suicide arising from desperation over catastrophic medical expenditures and severe disease. Many important areas in social insurance still remain insufficiently examined. The roles of social pensions for the retired and the elderly, injury compensation, and sickness benefits on buffering suicide must be studied further.

Different social assistance indicators were examined, including spending on medical benefits, spending on family supports, spending on transfer payments for low-income families, spending on housing support, and spending on disability. Due to the heterogeneous results of these indicators, it is impossible to draw a uniform conclusion from each variable. Rather, it is worth emphasizing that, since the roles of social assistance vary by context with different models of the welfare state, it is necessary to understand the structure of social policies and historical development of welfare states before investigating social assistance.

Lastly, studies that used total government spending, social welfare spending, and welfare regime typologies have strengths in clarifying the impacts of all social policies, including the net impact of the individual policy and the interaction with other social policies implemented simultaneously from multiple sectors. However, these approaches provide little insight into which individual policies mitigate suicide among which populations.

Methodological Challenges

The vast majority of the time-series studies examined in this review adopted fixed effects, random effects, and generalized method of moments estimation to control for unobserved time and place effects. Due to the nature of suicide, which is strongly affected by cultural characteristics,^6,10 the results of cross-national studies may be biased, unless these omitted variables inherent in each culture are cancelled out.

All the included studies used ecological designs. Although ecological designs have strengths in examining the impact of policy at the population level,³⁴ the simple association between aggregate policy indicators and suicide rates at the population level cannot distinguish between the effects of the program on the recipients per se and the contextual impacts of prevention in whole populations. In addition, studies with individual-level data make it possible to differentiate the materialistic mechanisms in which social protection operates at an individual level from the psychosocial mechanisms acting as contextual effects at a societal level. The shortage of studies with these designs might be due to the difficulty of collecting the appropriate data. Cohort studies employing available register-based data targeting the entire population and case-control studies with well-designed matching methods would be great alternatives.

Only one qualitative study is included in this review paper. This study applied fs/QCA (sometimes called a “mixed method”) rather than a true qualitative method.¹¹ Qualitative studies should be conducted more often, so that we can understand the qualitative aspects and details of social protection and the specific mechanisms by which certain policies and programs work well on certain populations. Information from populations at high risk of suicide, from families and close friends of people who died by suicide, from social workers in the field of suicide prevention, and from suicide prevention experts would be especially informative in clarifying detailed program mechanisms.

Paucity of Theory

Some studies presented theoretical grounds for the association between social protection and suicide, while others did not demonstrate theoretical rationales at all. Even studies that provided multiple theoretical rationales did not include variables that could properly serve as appropriate proxies for theoretical concepts. In addition, few studies have tried to interpret the conclusions of their analysis under relevant theories, in part due to methodological limitations that make it impossible to differentiate between mechanisms at an individual level and the societal level. In other words, many of the studies selected here did not attempt to demonstrate the theoretical reasons by which policies or programs worked or did not work for certain populations. Future research should be able to reveal the mechanisms of social protection through rigorous multilevel designs and setting variables that fit the formulated theory.

This review has some limitations. First, it is possible that this paper missed papers relevant to this topic, not only because the terms and concepts used for the policies and programs are heterogeneous but also because our searches were run only in English, excluding papers that are not available in databases in English. Second, publication bias may have occurred because many researchers are not likely to publish research protocols with null or opposite associations. Nevertheless, the present literature review has achieved its purpose, namely to provide a summary of relevant concepts, theories, and methodologies and to clarify the knowledge gap.

In conclusion, the findings suggest that social protection may have a preventing impact on suicide mortality, although the studies conducted thus far have been only partial and poor quality. Future studies should include more theory-oriented analysis to examine causal mechanisms and inform policy makers. More rigorous quantitative studies, employing individual longitudinal datasets and qualitative studies for intensive behavioral mechanisms, are required to conduct a theory-oriented analysis. Additionally, more studies of LMIC will have to be carried out, using indicators for social protection that reflect country-specific contexts.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the authorship and/or publication of this article.

Funding

The author disclosed no receipt of financial support for the research, authorship, and/or publication of this article.

Author Biography

Chungah Kim is a PhD student in the Department of Health Policy at McMaster University. She earned her MPH in the Department of Public Health at Seoul National University in 2014 and worked as a researcher at Health Insurance Review and Assessment in South Korea. Her research focuses on the political economy perspective on population health. She published a paper titled “Working conditions and leisure-time physical activity among waged workers in South Korea: a cross-sectional study.”

References

O’Connor

Nock

MK.

The psychology of suicidal behaviour. Lancet Psychiatry. 2014; 1(1):73–85.[TQ1]

World Health Organization. http://www.who.int/mediacentre/factsheets/fs398/en/. Published 2010. Accessed October 1, 2016.

Turecki

Brent

DA.

Suicide and suicidal behaviour.

Lancet. 2016; 387(10024):1227–1239.

Hamermesh

Soss

NM.

An economic theory of suicide.

J Polit Econ. 1974; 82(1):83–98.

Chen

Choi

Mori

Sawada

Sugano

Socio‐economic studies on suicide: a survey. J Econ Surv. 2012; 26(2):271–306.

Stack

Suicide: a 15‐year review of the sociological literature part I: cultural and economic factors. Suicide Life Threat Behav. 2000; 30(2):145–162.

International Labour Organization. http://www.ilo.org/public/english/protection/download/lifecycl/lifecycle.pdf. Published 2003. Accessed September 15, 2016.

Drolet

JL.

Social Protection and Social Development: International Initiatives. Dordrecht, The Netherlands: Springer; 2014.

Barrientos

Introducing basic social protection in low-income countries: lessons from existing programmes. In: Townsend

, ed. Building Decent Societies. 1st ed. Berlin, Germany: Springer; 2009:253–273.

10.

O’Connor

Platt

Gordon

International Handbook of Suicide Prevention: Research, Policy and Practice. Hoboken, NJ: John Wiley and Sons; 2011.

11.

Lee

Socio-economic factors of elderly suicide. Korean Soc Secur Stud. 2010; 26(4):1–19.

12.

Zimmerman

SL.

States’ public welfare expenditures as predictors of state suicide rates.

Suicide Life Threat Behav. 1987; 17(4):271–287.

13.

Zimmerman

SL.

States’ spending for public welfare and their suicide rates.

J Nerv Ment Dis. 1995; 183(7):425–434.

14.

Flavin

Radcliff

Public policies and suicide rates in the American states.

Soc Indic Res. 2009; 90(2):195–209.

15.

Park

Kim

Kwon

Shin

The association between public social expenditure and suicides: evidence from OECD countries.

J Prev Med Public Health. 2009; 42(2):123–129.

16.

Cheong

K-S

Choi

M-H

Cho

B-M

et al . Suicide rate differences by sex, age, and urbanicity, and related regional factors in Korea. J Prev Med Public Health. 2012; 45(2):70–77.

17.

Zimmerman

SL.

States’ spending for public welfare and their suicide rates, 1960 to 1995: what is the problem?

J Nerv Ment Dis. 2002; 190(6):349–360.

18.

Minoiu

Andres

AR.

The effect of public spending on suicide: evidence from US state data.

J Socio-Econ. 2008; 37(1):237–261.

19.

Ross

Yakovlev

Carson

Does state spending on mental health lower suicide rates?

J Socio-Econ. 2012; 41(4):408–417.

20.

Cylus

Glymour

Avendano

Do generous unemployment benefit programs reduce suicide rates? A state fixed-effect analysis covering 1968–2008. Am J Epidemiol. 2014; 180(1):45–52.

21.

Stuckler

Basu

Suhrcke

Coutts

McKee

The public health effect of economic crises and alternative policy responses in Europe: an empirical analysis.

Lancet. 2009; 374(9686):315–323.

22.

Yur’yev

Värnik

Sisask

Leppik

Role of social welfare in European suicide prevention.

Int J Soc Welf. 2012; 21(1):26–33.

23.

Baumbach

Gulis

Impact of financial crisis on selected health outcomes in Europe.

Eur J Public Health. 2014; 24(3):399–403.

24.

Norström

Grönqvist

The Great Recession, unemployment and suicide.

J Epidemiol Community Health. 2015; 69(2):110–116.

25.

Antonakakis

Collins

The impact of fiscal austerity on suicide mortality: evidence across the ‘Eurozone periphery’.

Soc Sci Med. 2015; 145:63–78.

26.

Reeves

McKee

Gunnell

et al . Economic shocks, resilience, and male suicides in the Great Recession: cross-national analysis of 20 EU countries. Eur J Public Health. 2015; 25(3):404–409.

27.

Matsubayashi

Ueda

The effect of national suicide prevention programs on suicide rates in 21 OECD nations.

Soc Sci Med. 2011; 73(9):1395–1400.

28.

Shibata

The effect of active labor market policies on suicide rates: a panel data analysis for 26 OECD countries, 1980–2007.

Jpn Sociol Rev. 2014; 65(1):116–133.

29.

Lee

Kang

Changes in suicide rates and social integration in Korea: a community-level panel data analysis. Korea J Popul Stud. 2014; 37(2):1–19.

30.

Durkheim

Suicide: A Study in Sociology [1897]. New York, NY: The Free Press; 1951.

31.

World Health Organization. http://www.who.int/mental_health/prevention/suicide/suicideprevent/en/. Published 2010. Accessed October 1, 2016.

32.

Iemmi

Bantjes

Coast

et al . Suicide and poverty in low-income and middle-income countries: a systematic review. Lancet Psychiatry. 2016; 3(8):774–783.

33.

Easterlin

McVay

Switek

The happiness-income paradox revisited. Proc Natl Acad Sci. 2010; 107(52):22463–22468.

34.

Rose

Sick individuals and sick populations.

Int J Epidemiol. 2001; 30(3):427–432.