Opposite nonlinear effects of unemployment and sentiment on male and female suicide rates: Evidence from Australia

https://doi.org/10.1016/j.socscimed.2021.114536Get rights and content

Highlights

  • We study gender differences in the effects of unemployment and sentiment on suicide.

  • Higher unemployment and lower sentiment increase suicide rates.

  • Higher unemployment increases male suicide rates.

  • Lower sentiment increases female suicide rates.

  • Economic expectations are strong predictors of suicide rates.

Abstract

We investigate gender differences in the effects of unemployment and sentiment on suicide rates. Using monthly Australian data, we find a positive relationship between the unemployment rate and the suicide rate, and a negative relationship between consumer sentiment and the suicide rate. However, there is strong evidence of nonlinearity in the effects of both unemployment and sentiment on suicide rates, with substantial gender differences. For men, an increase in the unemployment rate increases the suicide rate, but an unemployment decrease has no effect; we find the opposite for women. For men, an increase in sentiment has stronger effects on the suicide rate than a decrease in sentiment. Again, we observe the opposite effect for women. Among sentiment components, forward-looking expectations are stronger predictors of suicide rates than present conditions. Sentiment has a much stronger effect on male suicide rates than on female suicide rates.

Introduction

Globally, suicide is one of the leading causes of non-natural death, with the World Health Organization (WHO) reporting roughly 703,000 suicide deaths each year (WHO, 2021). Although 77% of global suicides occurred in low- and middle-income countries (WHO, 2021), according to the 2017 Global Burden of Disease Study suicide rates in high-income and high Social-Demographic Index (SDI) countries are considerably higher than the world average (IHME, 2018). While the high suicide rates in high-income and high-SDI countries pose a significant public health concern in these countries, it also raises an important question of the influences of economic factors on suicide rates.

Among high-income and high SDI countries, Australian male and female suicide rates, at 17.4 and 5.0 deaths per 100,000 population respectively, are around the averages (IHME, 2018). Among OECD countries, Australian male and female suicide rates are slightly above the OECD averages (OECD, 2019). Moreover, among OECD nations with suicide rates around the OECD averages, Australia is the largest economy. These factors make Australia an interesting and relevant case to study the impacts of macroeconomic conditions on suicide rates. Since the mid-1980s, at least 2000 suicides (on average 12.8 deaths per 100,000 population) were reported annually in Australia, and from 2017 to 2019 there were between 3100 and 3300 suicides each year (equating to about 13 deaths per 100,000). Importantly, there is a substantial gender difference with men accounting for just more than three quarters of reported suicides (AIHW, 2019; ABS, 2020a; WHO, 2021).

The substantial gender difference in suicide rates is, importantly, a phenomenon observed across the world (IHME, 2018). There are likely a multitude of psychological and socio-economic factors that cause this gender difference. However, abstracting from social and psychological factors, as individuals within an economy, both men and women are subject to the same economic conditions and macroeconomic events such as financial crises or recessions. This raises the question of whether there are major differences in the ways that men and women respond to changes in economic conditions. Research into the factors that influence suicide rates and the mechanisms through which such factors influence suicide rates can provide valuable information for well-targeted policy initiatives and interventions.

In this paper, we focus on the roles of economic factors in explaining suicide rates. We use two popular macroeconomic variables, namely the unemployment rate and a consumer sentiment index, as proxies for economic factors indicative of current and future economic conditions, respectively. The unemployment rate is often used as one of the key economic variables to construct the coincident index of economic activity, to monitor current economic conditions (Stock and Watson, 1991, 2011; Lim and Nguyen, 2015). The consumer sentiment index is mainly predictive of future economic conditions, particularly household consumption (Carroll et al., 1994; Bryant and Macri, 2005; Chua and Tsiaplias, 2009).

We estimate linear and nonlinear Auto-Regressive Distributed Lag (ARDL) models to allow for both symmetric and asymmetric effects of economic variables on suicide rates, on monthly Australian data from February 1990 to September 2018. We also split the sample by gender to study possible different mechanisms through which economic factors influence male and female suicide rates. In addition to the reasons provided earlier, our focus on Australian data is also partly due to advantages of the Westpac-Melbourne Institute Consumer Attitudes, Sentiments, and Expectations (CASiE) survey which is used to construct sentiment measures. CASiE is modelled on the University of Michigan Survey of Consumers (MSC), but it consists of 1200 consumers per month as compared to around 500 consumers in the MSC. Importantly, CASiE is stratified by gender, age and location so it is representative of the Australian population.

Existing studies on determinants of suicide rates mostly focus on the effects of objective economic factors. The unemployment rate is most frequently used as a proxy for objective economic factors and generally has a significant effect on suicide rates across a range of countries. Specifically, an increase in unemployment rates leads to an increase in suicide rates, and vice versa (Gerdtham and Johannesson, 2003; Barstad, 2008; Altinanahtar and Halicioglu, 2009; Andrés and Halicioglu, 2010, 2011; Andrés et al., 2011; Ceccherini-Nelli and Priebe, 2011; Antonakakis and Collins, 2014; Ruhm, 2015; Agrrawal et al., 2017; Mattei and Pistoresi, 2019; Abdou et al., 2020; Atalay et al., 2020, Meda et al., 2021). Real economic growth is the other popular proxy in the literature though its effect on suicide rates is less pronounced than that of the unemployment rate (Ceccherini-Nelli and Priebe, 2011; Gonzalez and Quast, 2011; Okada and Samreth, 2013; dos Santos et al., 2016).

The limited focus on consumer sentiment in explaining suicide is surprising. Suicide is ultimately an individual outcome and might therefore also be dependent on personal circumstances or perceptions as opposed to current, objectively measured, economic factors. Moreover, there is growing evidence that negative future economic expectations and uncertainty can adversely impact reported stress levels and individual mental health (Rohde et al., 2016; van Giesen and Pieters, 2019; Avdic et al., 2020; Johnston et al., 2020), and suicide (Korhonen et al., 2016; Christian et al., 2019; Vandoros et al., 2019; de Bruin et al., 2020). The studies of Berk et al. (2006) and Collins et al. (2021) begin to shed some light on the influence of economic indicators on suicide rates using a consumer sentiment index (Nowzohour and Stracca, 2020). As a self-reported measure, a consumer sentiment index can reflect the emotional responses of individuals to their own perceived current or expected economic circumstances and “capture the underlying feeling of economic pressures on an individual that are unlikely to be reflected in measures of unemployment” (Collins et al., 2021:6).

Berk et al. (2006) investigate the effects of a range of current economic indicators as well as a consumer sentiment index on suicide rates using Australian data from January 1968 to August 2002. Estimating a linear regression of suicide rates on a single variable at a time, they find a positive relationship between the unemployment rate and the suicide rate for men, but a negative relationship for women. They also find a negative association between the consumer sentiment index and the male suicide rate, but no association with the female suicide rate, though the correlations with the consumer sentiment index and its components are generally very small. Collins et al. (2021) estimate several dynamic panel models using US state-level data and find that an increase in the consumer sentiment index reduces state suicide rates, and that this effect is larger for females than for males. The authors also find that the unemployment rate is mostly insignificant in explaining suicide rates. Neither Berk et al. (2006) nor Collins et al. (2021) investigates the possibility of nonlinearity in the effects of current and forward-looking economic indicators on suicide rates.

With growing evidence that consumers respond asymmetrically to economic information (Soroka, 2006; Nguyen and Claus, 2013), and that economic stressors play more important roles in suicide among men than among women (Qin et al., 2000; Wunderlich et al., 2001), recent studies also investigate possible asymmetry in the relationship between suicide rates and current economic conditions. Wu and Cheng (2010) find that economic expansion in the US, measured by positive deviations of the unemployment rate from its trend, significantly reduces fatalities from suicide while recession has no significant effect. This asymmetry only holds for men and working-age groups. Chang and Chen (2017) examine the responses of suicide rates to unemployment rates across age groups in the US using annual data between 1928 and 2013. The authors find evidence that economic expansion (a decrease in the unemployment rate) has a greater effect on the suicide rate of those aged over 45 than economic recession does. Lin and Chen (2018) also find evidence of asymmetry in the effects of economic cycles on the suicide rates of different age groups. None of these studies, however, considers possible asymmetries in the effects of economic indicators such as a consumer sentiment index on suicide rates.

To our knowledge, this paper is the first to study the asymmetric effects of both unemployment and sentiment on suicide rates in a unified framework. Several studies also use other social and economic factors such as the divorce rate, fertility rate, migration, population density, alcohol and cigarette consumption, income, expenditures on mental health, public health and public welfare as controls in their analyses (Wu and Cheng, 2010; Chang and Chen, 2017; Collins et al., 2021). However, because most of these variables are often insignificant in explaining changes in suicide rates (as is also the case in our study), we focus only on two of the most important factors in the literature, namely the unemployment rate and consumer sentiment, in favor of more parsimonious models.

Chang and Chen (2017) and Collins et al. (2021) are the two studies closest to our analysis. However, in explaining the suicide rates of different demographic groups, such as males and females, both Chang and Chen (2017) and Collins et al. (2021) use aggregate measures of the economic indicators rather than demographic-specific measures. Although aggregate measures, such as economy-wide unemployment rate and sentiment, do influence, say, the male suicide rate, using the male-specific unemployment rate and sentiment is likely to be more relevant in explaining male suicide rates. Furthermore, although men and women are exposed to the same macroeconomic conditions, gender-specific unemployment rates and consumer sentiment indices, to a certain extent, capture possible differences in the impacts of common macroeconomic conditions on male and female economic outcomes, and hence different suicide rates. We therefore use gender-specific unemployment rates and consumer sentiment and match these to gender-specific suicide rates to more accurately capture the potential differential influences of economic circumstances on male and female suicide rates.

Our empirical analyses reveal four key findings. First, the results confirm a long-run level relationship between suicide rates and the unemployment rate and the consumer sentiment index. As expected, the suicide rate is positively associated with the unemployment rate and negatively associated with sentiment. Second, among the consumer sentiment index's components, expectations of family finances and expectations of short-term and longer-term economic conditions significantly influence suicide rates, but components relating to present conditions are less important. This suggests that influencing individuals' economic expectations could be a key channel for policy interventions.

Third, we find that unemployment and consumer sentiment have very different asymmetric effects on male and female suicide rates. An increase in the unemployment rate significantly increases the suicide rate for men, but a decrease has no effect on male suicide rates. The opposite is true for women. An increase in sentiment tends to have a stronger effect on the suicide rate in men than a decrease. Again, the opposite is true for women. This suggests that a policy intervention targeting both men and women, through influencing the unemployment rate and/or sentiment, may not be as effective as policy interventions tailored specifically at men and at women.

Finally, we find that sentiment has a much stronger effect on the suicide rate among men than among women. This contrasts with Collins et al. (2021) finding, for which there are two possible explanations: (i) Collins et al. (2021) do not consider the possibility of asymmetric effects of sentiment on suicide rates. If the association between sentiment and suicide rates is indeed nonlinear, a linear framework could obscure the true relationship and the true magnitude of their association; (ii) in explaining male and female suicide rates, Collins et al. (2021) use aggregate factors, including an aggregate consumer sentiment index, rather than gender-specific measures. As mentioned above, gender-specific economic variables are likely to be more relevant in explaining gender-specific suicide rates. Our findings suggest that policy intervention targeting sentiment will likely have a stronger impact on male suicide rates than on female suicide rates.

Section snippets

Data

We use monthly data on Australian suicide rates, unemployment rates, and the consumer sentiment index and its components for the period February 1990 to September 2018. The choice of our sample period is due to two factors. First, unit records for CASiE are only available from February 1990. These are required to construct gender-specific consumer sentiment indices and their components. Second, detailed suicide rates up to September 2018 were provided in a customised report (see Section 2.1).

Econometric methods

Abstracting from social and psychological factors, we assume that the suicide rate, st, is determined by economic factors indicative of current and future economic conditions, denoted Ct and Ft, respectively. This is written as:st=f(Ct,Ft)

To proxy for economic factors informative about current conditions, we use the unemployment rate, ut, and to proxy for mainly forward-looking economic conditions, we use the consumer sentiment index, csit. We assume that f(.) is a linear function, and hence

Results

In this section, we report and discuss the estimation results. Specifically, we estimate equation (4) (ARDL) to investigate the impacts of economic conditions on the suicide rate in a linear framework and estimate equation (7) (NARDL) to investigate if these factors have asymmetric impacts on the suicide rate. For each equation, we estimate three sets of results: one for all respondents (aggregate), one for male respondents (males) and one for female respondents (females). For ease of

Robustness checks

In our preliminary analyses, we also considered a broader set of economic variables to proxy for different dimensions of economic conditions. Specifically, we explored controlling for the monetary policy environment via the central bank policy rate (the Reserve Bank of Australia's target cash rate) and also used the ASX/S&P200 index to control for activity in the Australian share market. However, these variables were never significant in both the short-run and long-run analyses, and hence made

Conclusion

In this paper we combined monthly suicide rates in Australia with both unemployment and consumer sentiment to test how these affect suicide rates over time. We use linear and nonlinear ARDL models, proxying current economic circumstances with the unemployment rate, and future economic conditions with the consumer sentiment index and its components. Our study is the first to conduct a detailed investigation into how consumer sentiment—that captures perceptions and expectations of both personal

Author statement

Ferdi Botha: Conceptualization, Methodology, Formal analysis, Data curation, Writing – original draft, Writing – review & editing, Project administration, Funding acquisition. Viet Nguyen: Conceptualization, Methodology, Software, Formal analysis, Writing – original draft, Writing – review & editing, Visualization.

References (66)

  • J.P. dos Santos et al.

    More than just numbers: suicide rates and the economic cycle in Portugal (1910-2013)

    SSM – Population Health

    (2016)
  • U.-G. Gerdtham et al.

    A note on the effect of unemployment on mortality

    J. Health Econ.

    (2003)
  • D. Kwiatkowski et al.

    Testing the null hypothesis of stationarity against the alternative of a unit root

    J. Econom.

    (1992)
  • V.H. Nguyen et al.

    Good news, bad news, consumer sentiment and consumption behavior

    J. Econ. Psychol.

    (2013)
  • K. Okada et al.

    A study on the socio-economic determinants of suicide: evidence from 13 European OECD countries

    J. Soc. Econ.

    (2013)
  • S. Payne et al.

    The social construction of gender and its influence on suicide: a review of the literature

    J. Mens Health

    (2008)
  • N. Rohde et al.

    The effect of economic insecurity on mental health: recent evidence from Australian panel data

    Soc. Sci. Med.

    (2016)
  • C.J. Ruhm

    Recessions, healthy no more?

    J. Health Econ.

    (2015)
  • S. Vandoros et al.

    The association between economic uncertainty and suicide in the short-run

    Soc. Sci. Med.

    (2019)
  • R.I. van Giesen et al.

    Climbing out of an economic crisis: a cycle of consumer sentiment and personal stress

    J. Econ. Psychol.

    (2019)
  • W.-C. Wu et al.

    Symmetric mortality and asymmetric suicide cycles

    Soc. Sci. Med.

    (2010)
  • P. Agrrawal et al.

    Suicides as a response to adverse market sentiment (1980-2016)

    PLoS One

    (2017)
  • L. Artazcoz et al.

    Unemployment and mental health: understanding the interactions among gender, family roles, and social class

    Am. J. Publ. Health

    (2004)
  • K. Atalay et al.

    Lives Saved during Economic Downturns: Evidence from Australia

    (2020)
  • Customised Report

    (2019)
  • Causes of Death, Australia, 2019. ABS Cat. No. 3303.0

    (2020)
  • Labour Force, Australia. ABS Cat. No. 6202.0

    (2020)
  • Australian Demographic Statistics, June 2020. ABS Cat 3101.0

    (2021)
  • Suicide in Australia: Trends and Analysis, 1964 to 2018. Injury Research and Statistics Series No. 132. Cat. No. INJCAT 212

    (2020)
  • D. Avdic et al.

    Economic Downturns and Mental Wellbeing

    (2020)
  • A. Barstad

    Explaining changing suicide rates in Norway, 1948-2004: the role of social integration

    Soc. Indicat. Res.

    (2008)
  • M. Berk et al.

    The effect of macroeconomic variables on suicide

    Psychol. Med.

    (2006)
  • W.A. Bryant et al.

    Does sentiment explain consumption?

    J. Econ. Finance

    (2005)
  • Cited by (11)

    • The impact of the COVID-19 pandemic on suicides: A population study

      2022, Psychiatry Research
      Citation Excerpt :

      Considering that the association was also negative in the cross-sectional analysis with the latest data of 2005 in the above-mentioned study in India, it is likely that the effect of economic pressure on suicide may become small in developed economies. Consumer sentiment index was reported to be negatively associated with suicide rates (Berk et al., 2006; Botha and Nguyen, 2021; Collins et al., 2021). The absence of their association in this study suggests also that the economic impact on suicide was not like as had been expected.

    • Revisiting the relationship between economic uncertainty and suicide: An alternative approach

      2022, Social Science and Medicine
      Citation Excerpt :

      However, recently, some research has challenged this assumption. Botha and Nguyen (2022), Chang and Chen (2017) and Wu and Cheng (2010) examine the asymmetric response of suicide rates to changes in the unemployment rate. Their studies provide empirical evidence supporting the asymmetry in the unemployment-suicide relationship.

    • Global economic uncertainty and suicide: Worldwide evidence

      2022, Social Science and Medicine
      Citation Excerpt :

      The estimated coefficient suggests that a one-point increase in the unemployment rate is associated with an increase in the mean suicide rate of 0.138. Botha and Nguyen (2022) and Phillips and Nugent (2014) also found a positive relationship between unemployment and suicide in Australia and the US, respectively. The relationship between unemployment and economic uncertainty and suicide can often be channelled by intermediate variables that have an impact on mental health.

    View all citing articles on Scopus

    This research was funded by the Australian Government through the Australian Research Council's Centre of Excellence for Children and Families over the Life Course (Project ID CE140100027 and CE200100025). We are grateful to the Mortality Data Centre at the Australian Bureau of Statistics for providing the suicide data, as well as to Steve Zubrick and three anonymous referees for comments and suggestions.

    View full text