Sage Journals: Discover world-class research

Abstract

This research examines the long-term drivers of suicides in Turkey, focusing on the effects of uncertainty and the Misery Index. In this regard, this study utilizes yearly data covering the 1980–2019 period. Additionally, it employs the augmented autoregressive distributed lag (AARDL) approach to confirm the cointegrating relationships in the proposed models and estimate the long-term effects of selected determinants on suicides. The empirical analysis produced the following long-run findings: First, cointegration exists between suicide and its determinants in Turkey. Second, uncertainty and the Misery Index significantly increase suicides. Third, tobacco consumption is positively associated with suicides. Fourth, economic growth and industrial output significantly inhibit suicides in Turkey. These findings underline that in developing open economies like Turkey, improving socio-economic factors can play a significant role in curbing suicidal behavior.

Keywords

suicide uncertainty misery index cointegration Türkiye

Introduction

Universally, one of the principal causes of death is suicide (WHO, 2021). Moreover, financial and economic difficulties and crises are among the main determinants of suicide (Baeza et al., 2022; CDC, 2022; Pridmore & Reddy, 2012; Reeves et al., 2012; Webb & Kapur, 2015). In the Turkish case, although the number of suicides is relatively low, the growth rate of suicides indicates that it may become a major healthcare problem in the future (Kartal et al., 2022). For instance, the number of suicides increased from 750 in 1980 to 3409 in 2019, corresponding to a 354.5% total increase and an average growth rate of 3.86%. Additionally, 12.62% of suicides in Turkey were caused by business failures and economic problems during the 1980–2019 period (TSI, 2014, 2022).

Turkey is a small, open economy; therefore, internal and external crises or hardships can intensify its political and economic uncertainty. Furthermore, the empirical evidence suggests that higher uncertainty is generally associated with higher suicide rates (Antonakakis & Gupta, 2017; Claveria, 2022). Although this evidence is not empirically verified for the Turkish case, uncertainties resulting from crises seem to be a leading contributor to suicides in Turkey. For example, in 2001, when a major economic crisis was experienced, the number of suicides in Turkey increased by 43.4% compared to 2000, and approximately 26% of the suicides (i.e., 673 of 2584 suicides) in 2001 were caused by economic hardships and business collapses. Similarly, in the wake of the global financial crisis, suicides caused by economic hardships and failed businesses in Turkey increased by more than 26% (i.e., from 323 to 408) in 2008 compared to 2007 (TSI, 2022). Figure 1 presents the association between uncertainty and suicides in Turkey for the 1980–2019 period. From Figure 1, it is possible to infer that a long-term association between uncertainty and suicide rates might have existed.

Figure 1.

Suicides and uncertainty in Turkey.

Besides uncertainty, unemployment and inflation are also among the contributing factors to suicidal behavior. In a meta-analysis study, Amiri (2022) found that unemployment was positively correlated with increased suicide mortality rates. Additionally, the analysis by Akyuz and Karul (2022) documented that inflation increased suicides in Turkey. The combination of inflation and unemployment is called the Misery (or Discomfort) Index (Yang & Lester, 1992); as the name suggests, this index shows a country’s citizens’ level of misery. To the best of the author’s knowledge, no research empirically investigated how the level of misery impacted suicides in Turkey. Turkey has had a chronically higher inflation rate than the OECD average since 1980. For instance, in 2019, while the average inflation rate of the OECD members was 1.7%, Turkey’s inflation rate was significantly higher (i.e., 15.2%). Additionally, Turkey’s unemployment rate has been higher than the OECD average since 2001. For example, in 2019, the unemployment rate was 5.4% in the OECD; however, in Turkey, the unemployment rate was 13.7% (The World Bank, 2022a, 2022b). Given these stylized facts, examining the Misery Index’s impact can provide fresh evidence regarding the sources of suicides in Turkey.

All in all, this study investigates how the level of discomfort and uncertainty influenced suicides in Turkey by controlling the effects of other selected factors (i.e., alcohol and tobacco abuse, income growth, and industrial output). The impacts of uncertainty and the Misery Index on suicides are under-searched in the literature. Turkey constitutes a suitable case for this subject; as a developing open economy, Turkey has been vulnerable to uncertainties resulting from external and internal shocks and has relatively high inflation and unemployment rates. Consequently, revealing the long-run effects of the uncertainty and misery indices on suicides can assist policymakers in improving the mental health of Turkish citizens and lowering suicide rates.

This research consists of the following parts: the next section provides a conceptual framework and a literature summary. Besides data and model, the third section summarizes the method. The fourth section summarizes the empirical expectations and the hypotheses to be tested. The fifth and the sixth sections, respectively, present and discuss the empirical evidence. The conclusion takes up the last section.

Conceptual Framework and Literature Summary

Besides presenting a literature survey, the first subsection initially develops a conceptual framework regarding the possible impacts of the Misery Index’s components (i.e., unemployment and inflation) and uncertainty on suicides based on the existing theoretical notions. In the second subsection, other selected variables’ (such as income growth, industrial output, alcohol, and tobacco use) association with suicides are discussed.

Uncertainty and the Misery Index as the Drivers of Suicides

Durkheim (2005) classifies suicides by considering various factors. Some of these factors include an individual’s alienation from society and, therefore, lacking social capital to function properly (egoistic suicides) and experiencing dramatic (social and economic) disturbances at individual or societal scales (anomic suicides). Additionally, Durkheim (2005) claims that unless a person’s needs are appropriately balanced with their means, their motivation to exist can diminish. In their economic theory, Hamermesh and Soss (1974) mathematically prove that if a person’s perceived discounted lifetime utility drops lower than a threshold level, this person could commit suicide.

In line with the conceptual and theoretical arguments of Durkheim (2005) and Hamermesh and Soss (1974), the effect of the Misery Index on suicides is expected to be positive. The Misery Index’s first component is unemployment. When unemployed, individuals lack a constant source of income, face financial difficulties, and lose the opportunity to socialize with their co-workers and friends. As a result, they can be alienated from society by losing their social role, social mobility, capital, and integration (Vandoros et al., 2019). In line with Durkheim’s (2005) egoistic suicides definition, these negative aspects of unemployment can cause individuals to display suicidal behavior (Mattei & Pistoresi, 2019).

The Misery Index’s second component, inflation, can increase the number of suicides by simultaneously escalating the negative impact of unemployment on mental health and increasing the cost of living and investing. First, increased cost of living and investing can reduce the employment prospects for unemployed individuals. Therefore, prolonged (long-term) unemployment status can lead to suicidal behavior. Second, reducing inflation’s negative impact on the cost of living might be possible by increasing the minimum wage (or unemployment benefits) higher than inflation rates. However, this is a politically difficult task, and this difficulty can prevent policymakers from curbing suicide rates by improving minimum wages (or unemployment benefits) (Stack, 2021).

In line with Durkheim’s (2005) anomic suicides definition, uncertainty can increase suicidal behaviors. Uncertainty is generally caused by sudden (political and economic) crises/shifts, and these sudden changes can induce economic and psychological disturbances at individual and societal levels. Further, such crises could jeopardize the employment status of the employees and prolong the unemployment status of the individuals. Expectedly, uncertainties could negatively impact the mental state of individuals and contribute to suicidal behavior (de Bruin et al., 2020). Moreover, considering the economic approach of Hamermesh and Soss (1974), one can posit that individuals facing uncertainties can have lower threshold levels to attempt suicide because uncertainties could negatively impact their expected lifetime utility.

To the author’s knowledge, only the works of Yang and Lester (1992, 1999) examined the association between the Misery Index and suicides, and they revealed that a positive correlation existed between them in the USA. Nonetheless, several studies examined the drivers of suicidal behavior by considering the impacts of unemployment, uncertainty, or inflation. For the Italian case, Mattei and Pistoresi (2019) examined the impact of unemployment on suicide by using the error correction models (ECM). Their results revealed that unemployment significantly increased suicide, yet public spending on unemployment curbed this effect. Considering the 2001–2015 period, Vandoros et al. (2019) used the ordinary least squares (OLS) method to estimate the impact of unemployment and economic uncertainty on England’s and Wales’s suicide rates. Their analysis showed that economic uncertainty significantly increased suicide rates; unemployment’s effect was insignificant. Using the monthly data from February 1990 to September 2018, Botha and Nguyen (2022) investigated how unemployment and consumer sentiment affected suicides in Australia. Their nonlinear autoregressive distributed lag (NARDL) analysis disclosed that positive shocks in male unemployment intensified suicide rates, and increases (decreases) in consumer sentiment significantly decreased (increased) suicides in both sexes.

On a global level, Claveria (2022) examined the impacts of economic uncertainty, unemployment, and growth on suicides in 183 economies for the 2000–2019 period. The study’s panel data estimates proved that uncertainty, unemployment, and income growth were positively related to suicides. de Bruin et al. (2020) analyzed the drivers of suicides in 17 economies for the 1990–2015 period. Their panel data fixed effect results indicated that economic uncertainty, political instability, and substance abuse were positively associated with suicides. Also, per-person income was negatively associated with suicides, while unemployment’s impact was generally insignificant. Moreover, by using time series analysis, Men et al. (2022) revealed that unemployed people had significantly higher suicide rates than employed people in Hong Kong; besides, inflation (i.e., consumer price index) had an insignificant yet positive effect on the suicide rates of employed and unemployed citizens through the 2003–2019 period. For the case of Kazakhstan, Inoue et al. (2021) showed that unemployment was significantly and positively associated with suicide rates; however, inflation’s impact was positive yet insignificant in the 2000–2019 period.

Other Drivers of Suicides

Other economic variables considered to be associated with suicide are per capita income growth and industrial output (IO). Higher (lower) income is generally associated with lower (higher) suicide rates (Stack, 2021). Also, Okada and Samreth (2013) demonstrated that higher income level was associated with lower suicide rates in the majority of European OECD economies. For Turkey, Altinanahtar and Halicioglu (2009) and Akyuz and Karul (2022) established that income growth lowered suicide levels in the long-run. Similar to income, Akyuz and Karul (2022) showed that industrial production reduced Turkey’s suicide rates in the 1988–2018 period, confirming their expectation. However, Inoue et al. (2010) discovered that the correlation between industrial output and suicides was insignificant in South Korea.

Income growth and IO are decisive development indicators for emerging economies such as Turkey. Numerically, Turkey’s average per capita income growth rate in the 1980–2019 period was 2.78%, significantly higher than the Euro Area’s 1.5%. Similarly, in the 1991–2019 period, the average IO (in terms of manufacturing value added as a percentage of output) was 18.36% in Turkey, while the average for the Euro Area was lower (i.e., 16.19%) (The World Bank, 2022c). Given these stylized facts and the models of previous studies, both indicators are included in this research’s model.

Additional factors associated with higher suicide risk and included in this research are alcohol and tobacco consumption (Amiri & Behnezhad, 2020; Echeverria et al., 2021). Substance abuse is known to aggravate depression’s effects and induce impulsive behavior such as attempting suicide (Stack, 2021). Akyuz and Karul (2022) demonstrated that alcohol use’s effect on suicides was insignificant yet positive in Turkey. Finally, Okada and Samreth (2013) found that in only 3 out of 13 countries, alcohol use significantly increased suicides.

The literature survey above highlights several gaps in the empirical domain. First, including the works of Altinanahtar and Halicioglu (2009) and Akyuz and Karul (2022), the number empirical works on Turkey is rather limited. Second, while some studies considered the effects of unemployment and/or inflation, their combined effects (i.e., the Misery Index) are not econometrically examined. Considering one variable instead of two improves the degrees of freedom in the analysis. Also, the expected effects of inflation and unemployment on suicides are positive. Since Turkey has been facing relatively high unemployment and inflation rates, decomposing their effects would not significantly improve our political implications. Third, uncertainty’s effect on suicides is overlooked in the works examining Turkey. Turkey is a small, open economy that is vulnerable to uncertainties caused by shocks in political and economic spheres. Quantifying uncertainty’s effect can provide Turkish policymakers in the healthcare division to devise more inclusive approaches in preventing suicide attempts.

Data, Model, and Method

This analysis utilizes the yearly time series spanning the 1980–2019 period on Turkey. The employed econometric model is based on the previous literature. The econometric analysis consists of three stages in this study. In the analysis’ first stage, the integration orders of the time series are determined. To do so, this research applies two unit-root tests, namely the Phillips and Perron (1988) (PP) and the Kwiatkowski et al. (1992) (KPSS) tests. In the second stage, the existence of the long-run (LR) association among the time series is determined by three bounds tests based on the augmented ARDL (AARDL) method developed by Pesaran and Shin (1995) and Pesaran et al. (PSS) (2001), and improved by Sam et al. (SMG) (2019). This method is selected because it is applicable even if the sample size is limited. Besides, the AARDL method allows the employed time series to have different integration orders. In other words, the variables can be integrated at the level or first difference; nonetheless, the variables should not be integrated at the second difference. The econometric model of this research can be presented in its typical ARDL form in the following (unconditional) error-correction equation:

\begin{array}{l} {Δ L S U I}_{t} = ψ_{0} + ψ_{1} {L S U I}_{t - 1} + ψ_{2} {L A L C}_{t - 1} + ψ_{3} L {T O B C}_{t - 1} + ψ_{4} {(U N I, U N I S)}_{t - 1} + ψ_{5} {M I S I}_{t - 1} + ψ_{6} {I N D P}_{t - 1} + ψ_{7} {G D P G}_{t - 1} + \\ \sum_{h = 1}^{q} ψ_{8 h} ∆ {L S U I}_{t - h} + \sum_{h = 0}^{p - 1} ψ_{9 h} ∆ {L A L C}_{t - h} + \sum_{h = 0}^{m - 1} ψ_{10 h} {∆ L T O B C}_{t - h} + \sum_{h = 0}^{n - 1} ψ_{11 h} {∆ (U N I, U N I S)}_{t - h} + \sum_{h = 0}^{j - 1} ψ_{12 h} {∆ M I S I}_{t - h} + \\ \sum_{h = 0}^{k - 1} ψ_{13 h} {∆ I N D P}_{t - h} + \sum_{h = 0}^{l - 1} ψ_{14 h} {∆ G D P G}_{t - h} + ε_{t} \end{array}

(1)

UNI and UNIS in equation (1) are Turkey’s normalized and smoothed uncertainty indices, respectively. Based on these uncertainty measures, two models are employed. Model (A) corresponds to UNI, and Model (B) uses UNIS. These indices are constructed by Ahir et al. (2018, 2022a) and converted from quarterly to yearly frequency by averaging. Additionally, higher index values mean higher uncertainty. LSUI is the number of suicides caused by business collapses and economic complications. LALC and LTOBC are alcohol and tobacco consumption per person, respectively. MISI is the Misery (Discomfort) Index constructed by summing inflation and unemployment rates. INDP and GDPG are economic conditions and indicate industrialization as a percentage of output and per capita economic growth, respectively. L shows the logarithmic transformation. Time series presented in percentages or as indices are not logarithmically transformed. The Appendix provides the variables’ sources, contents, and descriptive statistics.

Two of the three abovementioned bounds tests (i.e., t test and F-test) are proposed by PSS (2001) and consider the lagged variables’ coefficients (i.e., ψ₁ through ψ₇). The t test tests the null hypothesis of ψ₁ = 0 against the alternative hypothesis ψ₁ < 0. Moreover, for the F-test, the null of ψ₁ = … = ψ₇ = 0 is tested against the alternative of ψ₁≠…≠ψ₇≠0. Rejection of the null hypotheses in both tests requires that both test statistics are significant compared to their upper-bound critical values compiled by PSS (2001). According to the ARDL approach of PSS (2001), the significance of the F-test and t test is sufficient for confirming the cointegrating relationship given that the response variable’s (LSUI) integration order is one (i.e., I(1)).

However, this may not always be the case since stationarity tests can generate conflicting results. For instance, if the response variable is possibly I(0) (integrated at level), then the significance of the aforementioned bounds tests cannot be reliable because of the “degenerate case of lagged independent regressors” (Goh et al., 2017; McNown et al., 2018). To fix this degenerate case and confirm cointegration, McNown et al. (2018) and SMG (2019) augment the ARDL approach (also referred to as the augmented ARDL or AARDL method) by proposing an additional F-test on the regressors (i.e., F_IV-test). In F_IV-test, the null of ψ₂ = … = ψ₇ = 0 is tested against its alternative ψ₂ ≠ … ≠ ψ₇ ≠ 0. Akin to the F- and t tests, rejecting the null requires that F_IV-test statistic is significant compared to its (upper-bound) critical values computed by SMG (2019). In short, a cointegrating relationship requires the significance of the F-, t, and F_IV-tests.

In the last (third) stage, the cointegrating model should be estimated through equation (1) and its conditional version to determine the long-run (LR) and short-run (SR) impacts of selected determinants on the number of suicides (LSUI). Consistently, the sign and magnitude of the error correction term's (ect) coefficient are retrieved in the same estimation procedure.

Hypotheses to be Tested

This section outlines the hypotheses that are tested in this research. Besides common sense, these hypotheses are based on the theoretical framework summarized in the second section. Furthermore, even though the SR impacts of the selected variables are also reported, the LR effects (i.e., coefficients) are considered to test the proposed hypotheses.

The intuition behind considering the LR effects lies in the previously mentioned Hamermesh and Soss' (1974) economic theory. They claim that a person considers their expected lifetime utility to attempt suicide. As the term “lifetime” connotes a long-term process, this research contemplates that considering the LR impacts is more appropriate. Besides, this study posits that considering the selected drivers’ cumulative (i.e., long-term) effects is more proper. For instance, it would be rather marginal to state that a person can attempt suicide right after becoming unemployed or starting to use harmful substances such as alcohol or tobacco.

The first hypothesis (H1) considers the effect of the Misery Index. Since the expected LR impacts of the Misery Index’s components (i.e., inflation and unemployment) on suicides are positive, this empirical work hypothesizes that the Misery Index intensifies the suicide rates in the LR. It is also possible to maintain that H1 corroborates with the Turkish case, as Turkey’s inflation and unemployment rates are relatively higher. To accept H1, the calculated LR coefficient of MISI should be significantly positive (i.e., $- ψ_{5} / ψ_{1} > 0$ ). The second hypothesis (H2) tests the LR effect of uncertainty and posits that uncertainty intensifies suicide rates. In H2, it is asserted that a significant positive LR association between uncertainty and the number of suicides should exist, confirming the Durkheim’s (2005) previously mentioned “anomic suicides” definition. H2 can be validated if the uncertainty’s coefficient is significant and positive (i.e., $- ψ_{4} / ψ_{1} > 0$ ).

The third hypothesis (H3) indicates that better long-term economic conditions, in terms of increased income growth and higher IO, lower suicides. To validate H3, the signs of the GDPG’s and INDP’s LR coefficients should be negative and statistically significant (i.e., $- ψ_{6} / ψ_{1} < 0$ and $- ψ_{7} / ψ_{1} < 0$ ). The last (and fourth) hypothesis (H4) postulates that a long-term positive association exists between substance abuse (alcohol and tobacco use) and suicides. To confirm H4, the signs of the LR coefficients for LALC and LTOBC should be positive and significant (i.e., $- ψ_{2} / ψ_{1} > 0$ and $- ψ_{3} / ψ_{1} > 0$ ).

Empirical Results

Before testing for the LR association between the response variable and its regressors, the stationarity (unit-root) properties should be decided. The findings from the unit-root tests are in Table 1.

Table 1.

Unit-Root Tests.

	PP			KPSS
variables	level	1^st Diff.	I(.)	level	1 ^st Diff	I(.)
LSUI	−4.7536^c	-	I(0)	0.2026^b	0.3329	I(1)
LALC	−4.2057^b	-	I(0)	0.0477	-	I(0)
LTOBC	−1.4551	−6.6237^c	I(1)	0.1486^b	0.2356	I(1)
MISI	−2.2341	−7.3469^c	I(1)	0.1016	-	I(0)
UNI	−4.8366^c	-	I(0)	0.0552	-	I(0)
UNIS	−4.4856^c	-	I(0)	0.0551	-	I(0)
INDP	−2.5994	−6.0263^c	I(1)	0.0986	-	I(0)
GDPG	−6.7322	-	I(0)	0.0373	-	I(0)

Notes. Test equations have intercept and trend in levels and have intercept in first differences. I(.) shows stationarity level. Bandwidth selection is based on the Andrews approach. b and c indicate 5% and 1% significance levels.

The findings show that none of the time series becomes stationary at the second difference (i.e., they are all either I(0) or I(1)). Therefore, the stationarity properties of the variables are appropriate for using the AARDL method. Further, according to the PP test, the response variable (i.e., LSUI) is stationary at level (i.e., its integration order is zero or I(0)). This result requires using the AARDL bounds testing procedure for the proposed suicide models to avoid the degenerate cointegration case explained previously. With these findings at hand, the cointegrating properties of the proposed models can be tested. Table 2 reports the cointegration tests and diagnostics for each model.

Table 2.

Bounds Tests and Diagnostics.

	Model (A)	Model (B)
B ounds T ests	test statistics
F-test	16.4697^c	16.5221^c
F_IV-test	10.7384^c	10.7781^c
t test	−8.6992^c	−8.7226^c

Diagnostic T ests
	test statistic	p-value	test statistic	p-value
serial correlation	0.4431	0.5056	0.2269	0.6338
normality	3.7245	0.1553	2.0149	0.3651
functional-form	0.7641	0.3895	0.5149	0.4791
homoscedasticity	0.0087	0.9254	0.0069	0.9335

Notes. Maximum lag length is set to 1. For lag selection, the Akaike criterion is chosen. Selected lags are (1, 0, 0, 1, 0, 1, 0) in both models. c indicates 1% significance level. Significance is based on the upper-bound critical values obtained from the PSS (2001) and SMG (2019) studies. Graphical representations of the stability diagnostics (i.e., the CUSUM and CUSUMSQ tests) are provided in the Appendix.

Table 2 shows that bounds tests are uniformly significant at 1% significance level for both models. In other words, suicide and its determinants have a cointegrating relationship in Turkey. Also, diagnostic tests indicate that the proposed models do not contain misspecification, non-normality, autocorrelation, or heteroskedasticity problems. Last, the cumulative sum tests’ plots in the Appendix indicate that estimated parameters are stable. Table 3 reports the estimation findings.

Table 3.

Estimation Results.

	Model (A)		Model (B)
Long-run (LR) E stimates
regressors	coefficients	t-statistics	coefficients	t-statistics
LALC	−0.4225	−0.6967	−0.4031	−0.6725
LTOBC	1.1947^c	4.5369	1.2082^c	4.6464
MISI	0.0113^b	2.4814	0.0111^b	2.4684
UNI	1.1319^c	3.6738	-	-
UNIS	-	-	3.5625^c	3.7298
INDP	−0.2143^c	−5.2219	−0.2121^c	−5.2221
GDPG	−0.0344^b	−2.7558	−0.0349^c	−2.8342
Short-run (SR) E stimates
variables	coefficients	t-statistics	coefficients	t-statistics
constant	0.2633^c	8.6373	0.1695^c	6.3784
∆MISI_t	0.00195	0.9998	0.00196	1.0057
∆INDP_t	−0.0334	−1.4277	−0.0317	−1.3581
ect	−0.6441^c	−11.7958	−0.6514^c	−11.8145

Notes. c and b respectively show 1% and 5% significance levels. ect is the error correction term.

The LR findings in Table 3 indicate that the signs and magnitudes of the coefficients of the determinants are similar in Models (A) and (B). Alcohol use (LALC) has a negative but insignificant LR effect on suicide in Turkey. Nevertheless, consuming tobacco has an intensifying LR effect on suicides in Turkey. The results from both models reveal that, all else the same, a 1% increase (decrease) in tobacco use correlates with approximately 1.2% higher (lower) suicides in Turkey.

The LR results additionally show that the Misery Index (MISI) positively correlates with suicides in Turkey. Ceteris paribus, one percentage-point higher discomfort index results in approximately 1% higher suicide rates in the LR. Similarly, the evidence reveals that uncertainty (UNI, UNIS) increases suicides in Turkey in the long-term. However, expectedly, the magnitudes of the coefficients are different in both models. In Model (A), a 0.01-point increase in the index of uncertainty (UNI) escalates suicides by more than 1.1%. Further, a 0.01-point increase in the uncertainty index (UNIS) in Model (B) intensifies suicides in Turkey by more than 3.5%.

The evidence in Table 3 also documents that economic growth (GDPG) in both models lowers suicides in Turkey in the LR. Numerically, in the LR, a one-percentage point growth in per-person GDP lowers suicides by more than 3.4% in Turkey. Similarly, industrial output (INDP) curbs suicides in Turkey. For instance, a 0.1 percentage-point increase in manufacturing value added curtails suicides in Turkey by more than 2% in the LR. Finally, Table 3 also includes the SR findings. While none of the SR coefficients is significant in both models, the ect terms are significantly negative, and their magnitudes are expectedly between zero and −1.

Discussion

The empirical results have significant LR implications for Turkey. First, the LR increasing effect of the Misery Index on suicides confirms H1 and generally aligns with Yang and Lester’s (1992, 1999) findings. The impact of the misery level of Turkish citizens on their mental health is usually overlooked and under-searched. This finding shows that prolonged higher discomfort levels can lead to increased suicide attempts in the LR. Second, the result that higher uncertainty leads to increased suicide levels in Turkey validates H2 and is in line with the results in Antonakakis and Gupta (2017), Claveria (2022), and de Bruin et al. (2020). Uncertainty’s long-term impact on mental health is also ignored in Turkey. Since economic and/or political instability is generally associated with higher uncertainty, this study’s finding shows that commitment to long-term economic and political stability can lead to better mental health (i.e., lower suicide levels) in Turkey by reducing uncertainty. Another possible implication of this finding is that ignoring long-term economic and political commitments for short-term (political and economic) gains can lead to higher uncertainty and eventually increased suicide levels.

Furthermore, the findings regarding the LR impacts of income growth and IO indicate that improving the economic welfare of Turkish citizens can lower suicide numbers. The significant negative effect of income growth on suicides agrees with the findings of Altinanahtar and Halicioglu (2009) and Akyuz and Karul (2022). Additionally, the curbing effect of IO on suicide levels confirms the result in Akyuz and Karul (2022) regarding the Turkish case. These LR empirical outcomes validate H3, underscoring that promoting long-term economic conditions can improve a person’s perceived lifetime utility, thus preventing suicidal behavior. Another implication of achieving better economic conditions is that the Turkish economy can become more robust to uncertainty caused by political and economic instabilities. In other words, robustness resulting from better economic conditions can also help curb the LR intensifying effect of uncertainty on suicide rates.

Finally, alcohol consumption’s insignificant influence on suicide levels is in accordance with the result of Akyuz and Karul (2022). This result is not untoward because although Turkey is a secular state, its majority religion is Islam. Further, because alcohol abuse is forbidden in Islam, alcohol consumption is relatively limited in Turkey. Numerically, annual alcohol use per capita in Turkey varied between 1.2 to 1.8 L in the 1980–2019 period OECD (2022). However, tobacco consumption is a significant LR predictor of suicides in Turkey. This result confirms the proposal of Echeverria et al. (2021). Likewise, this finding is unsurprising given that the percentage of daily smokers among the population aged 15 and over was relatively high (27.3%) in 2019, ranking Turkey second among the European Union economies (Eurostat, 2022). In summary, the LR findings regarding the effects of substance abuse partially confirm H4 and indicate that limiting tobacco use can contribute to curbing LR suicide levels in Turkey.

Conclusion

While the literature on the determinants of suicide considered many variables as possible contributing factors, the impacts of the uncertainty and misery indices are under-searched. This study seeks to fill this gap in the literature for a small open economy, namely, Turkey. Accordingly, by utilizing the AARDL method and covering the 1980–2019 period, this study examines the determinants of suicides. The empirical investigation showed that suicide and its selected determinants had a cointegrating relationship. Furthermore, estimation results produced the following LR evidence: first, uncertainty and the Misery Index significantly increased suicides in Turkey. Second, tobacco consumption was positively associated with suicides in Turkey, whereas alcohol use had no statistically significant impact. Last, income and industrial growth benefited Turkey in the LR by lowering suicide levels.

Some policy suggestions can be put forward in line with these findings and their implications. First, policymakers should improve Turkey’s long-term political and economic stability by committing to Western democratic values and implementing structural reforms. Following such a policy path in the LR can benefit Turkey by making it more robust to uncertainty caused by internal and external instabilities. Eventually, improved economic robustness can lead to lower suicide rates induced by economic problems and financial hardships. In addition, improving Turkey’s political and economic stability may also positively influence its economic and industrial growth. Since income growth and industrial output hinder Turkey’s suicide rates in the LR, improving stability may produce more positive outcomes than expected. Second, lowering the discomfort levels of Turkish citizens by keeping unemployment and inflation under check through necessary economic policies would not only improve the welfare of these citizens but also curb suicide rates caused by economic and financial hardships.

Third, because unemployment is a component of the Misery Index, improving the unemployed individuals’ well-being can contribute to lowering suicides in Turkey. Accordingly, Turkey’s public unemployment insurance in terms of the “unemployment benefit” and the “job loss compensation” is in place; however, besides their strict requirements, their duration is rather limited. For example, the maximum number of days an individual can receive the unemployment benefit is 300. Moreover, the received benefit is well below Turkey’s official minimum wage (for more details, see ISKUR (2023)). Especially during political and economic crises, the duration and amount paid by the unemployment benefit should be improved to ensure unemployed individuals’ mental well-being. Such an approach would not only inhibit the Misery Index’s aggravating effect on suicides but also curb uncertainty’s impact on suicides. With these possibilities in mind, necessary changes in the social care system should be implemented. Last, tobacco use should be curbed by following necessary healthcare policies and informing the public about its negative consequences on mental health. The long-term commitment to such policies would hopefully lower tobacco use and inhibit suicides in Turkey.

Limitations and Further Research Directions

One limitation of this research is its limited spatial coverage. Future research can focus on other emerging open economies to determine if the uncertainty and discomfort levels have similar effects on suicides. This study’s other limitation is the use of aggregated suicide data. Since the selected drivers (such as uncertainty and the Misery Index) can have different mental health effects in different age and gender groups, further empirical research can test whether these drivers have varying impacts in different age/gender groups. Finally, this study used linear methods in line with the presented theoretical framework. Nevertheless, future works can use nonlinear methods (such as the NARDL approach) to capture if a nonlinear relationship exists between suicide rates and their drivers. Therefore, these future studies can also introduce the necessary theoretical concepts, justifying the use of nonlinear methods.

Footnotes

Author’s Note

This manuscript has not been previously published in any language anywhere, and it is not under simultaneous consideration or in press by another journal.

Acknowledgments

The author thanks the anonymous referee(s) for the constructive and insightful comments and suggestions. The author is responsible for any remaining errors. As a result of the agreement between Sage Publishing and ANKOS, the open-access article processing charge for this research is significantly reduced. The author is grateful to both parties for this agreement.

Declaration of Conflicting Interests

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

Funding

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

ORCID iD

Alper Karasoy

Appendix

Table A1.

Information About the Variables.

Variable		Definition	Source
LSUI		Total number of suicides (caused by business failures and economic issues)	TSI, 2014, TSI, 2022
LALCO		Alcohol use (litres per capita (15+))	OECD (2022)
LTOBU		Tobacco use (grams per capita (15+))	OECD (2022)
Uncertainty	UNI	Uncertainty index of Turkey	Ahir et al. (2022b)
Uncertainty	UNIS	Uncertainty index of Turkey (smoothed version)	Ahir et al. (2022b)
MISI (Misery index)		Inflation (consumer prices, annual %)	The World Bank (2022c)
MISI (Misery index)		Unemployment (% of total labor force)	IMF (2022)
INDP		Industrial output (manufacturing value added, % of GDP)	The World Bank (2022c)
GDPG		GDP per capita growth (yearly %)	The World Bank (2022c)

Notes. Uncertainty indices are converted from quarterly to annual frequency by averaging. Higher index values correspond to higher uncertainty and misery. TSI: Turkish Statistical Institute.

Table A2.

Descriptive Statistics.

	LSUI	LALC	LTOBC	MISI	UNI	UNIS	INDP	GDPG
Mean	5.4322	0.3886	7.2183	48.4129	0.2337	0.0778	18.9109	2.7771
Median	5.6094	0.4054	7.3313	43.5704	0.2171	0.0767	18.2679	3.5611
Maximum	6.5117	0.5877	7.5503	113.2281	0.7221	0.2287	23.1219	9.5099
Minimum	3.0445	0.1823	6.7855	15.4548	0.0000	0.0000	15.0539	−7.1478
Std. Dev.	0.6194	0.1024	0.2356	29.3168	0.1546	0.0495	2.4704	4.2061
Skewness	−1.4256	−0.0427	−0.3248	0.4769	0.8634	0.7583	0.2791	−0.8029
Kurtosis	6.6075	2.3794	1.5693	1.9589	3.9365	3.6952	1.6554	2.9172

Figure A1.

Plots for the CUSUM and CUSUMSQ tests.

Author Biography

Alper Karasoy is currently working as an assistant professor (Ph.D.) in the Department of Economics at Afyon Kocatepe University's Faculty of Economics and Administrative Sciences. He obtained his Ph.D. in Economics from Afyon Kocatepe University's Institute of Social Sciences. His research interests include, but are not limited to, applied econometrics, time series analysis, environmental economics, and energy economics. His research has been published in Technology in Society, Natural Resources Forum, International Migration, Environmental Science and Pollution Research, Management of Environmental Quality: An International Journal, International Journal of Energy Sector Management, Review of Economic Perspectives, and OPEC Energy Review.

References

Ahir

Bloom

Furceri

(2018). The World uncertainty index. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3275033

Ahir

Bloom

Furceri

(2022a). The World uncertainty index. National Bureau of Economic Research (NBER) Working Paper No. 29763. https://doi.org/10.3386/w29763

Ahir

Bloom

Furceri

(2022b). World uncertainty index database. Retrieved December 19, 2022, from. https://worlduncertaintyindex.com/research/

Akyuz

Karul

(2022). The effect of economic factors on suicide: An analysis of a developing country. International Journal of Human Rights in Healthcare, 16(5), 473–482. https://doi.org/10.1108/IJHRH-03-2022-0023

Altinanahtar

Halicioglu

(2009). A dynamic econometric model of suicides in Turkey. The Journal of Socio-Economics, 38(6), 903–907. https://doi.org/10.1016/j.socec.2009.05.008

Amiri

(2022). Unemployment and suicide mortality, suicide attempts, and suicide ideation: A meta-analysis. International Journal of Mental Health, 51(4), 294–318. https://doi.org/10.1080/00207411.2020.1859347

Amiri

Behnezhad

(2020). Alcohol use and risk of suicide: A systematic review and meta-analysis. Journal of Addictive Diseases, 38(2), 200–213. https://doi.org/10.1080/10550887.2020.1736757

Antonakakis

Gupta

(2017). Is economic policy uncertainty related to suicide rates? Evidence from the United States. Social Indicators Research, 133(2), 543–560. https://doi.org/10.1007/s11205-016-1384-4

Baeza

González

Benmarhnia

Vives Vergara

(2022). Effects of the Great Recession on suicide mortality in Chile and contributing factors. SSM - Mental Health, 2, 100104. https://doi.org/10.1016/j.ssmmh.2022.100104

10.

Botha

Nguyen

V. H.

(2022). Opposite nonlinear effects of unemployment and sentiment on male and female suicide rates: Evidence from Australia. Social Science and Medicine, 292, 114536. https://doi.org/10.1016/j.socscimed.2021.114536

11.

CDC . (2022). Risk and protective factors | suicide. Retrieved December 30, 2022, from. https://www.cdc.gov/suicide/factors/index.html

12.

Claveria

(2022). Global economic uncertainty and suicide: Worldwide evidence. Social Science and Medicine, 305, 115041. https://doi.org/10.1016/j.socscimed.2022.115041

13.

de Bruin

Agyemang

Chowdhury

M. I. H.

(2020). New insights on suicide: Uncertainty and political conditions. Applied Economics Letters, 27(17), 1424–1429. https://doi.org/10.1080/13504851.2019.1686453

14.

Durkheim

(2005). Suicide: A study in sociology. Routledge.

15.

Echeverria

Cotaina

Jovani

Mora

Haro

Benito

(2021). Proposal for the inclusion of tobacco use in suicide risk scales: Results of a meta-analysis. International Journal of Environmental Research and Public Health, 18(11), 6103. https://doi.org/10.3390/ijerph18116103

16.

Eurostat . (2022). Tobacco consumption statistics - statistics explained. Retrieved January 18, 2024, from. https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Tobacco_consumption_statistics#Level_of_cigarette_consumption

17.

Goh

S. K.

Sam

C. Y.

McNown

(2017). Re-examining foreign direct investment, exports, and economic growth in Asian economies using a bootstrap ARDL test for cointegration. Journal of Asian Economics, 51, 12–22. https://doi.org/10.1016/j.asieco.2017.06.001

18.

Hamermesh

D. S.

Soss

N. M.

(1974). An economic theory of suicide. Journal of Political Economy, 82(1), 83–98. https://doi.org/10.1086/260171

19.

IMF . (2022). World economic outlook database (october 2021). Retrieved December 19, 2022, from. https://www.imf.org/en/Publications/WEO/weo-database/2021/October

20.

Inoue

Nishimura

Nishida

Fukunaga

Masaki

Fujita

Nata

Okazaki

Ono

(2010). Relationships between suicide and three economic factors in South Korea. Legal Medicine, 12(2), 100–101. https://doi.org/10.1016/j.legalmed.2009.11.002

21.

Inoue

Seksenbayev

Chaizhunusova

Moldagaliyev

Ospanova

Tokesheva

Zhunussov

Y. T.

Takeichi

Noso

Hoshi

Kawano

(2021). An exploration of the labor, financial, and economic factors related to suicide in the republic of Kazakhstan. International Journal of Environmental Research and Public Health, 18(13), 6992. https://doi.org/10.3390/ijerph18136992

22.

ISKUR . (2023). Unemployment benefit. Retrieved July 6, 2023, from. https://www.iskur.gov.tr/en/job-seeker/unemployment-insurance/unemployment-benefit/

23.

Kartal

Demir

Hekimoglu

Keskin

Asirdizer

(2022). Suicides in Turkey: 25-year trend (1995–2019). Journal of Forensic Sciences, 67(5), 1858–1866. https://doi.org/10.1111/1556-4029.15086

24.

Kwiatkowski

Phillips

P. C. B.

Schmidt

Shin

(1992). Testing the null hypothesis of stationarity against the alternative of a unit root. Journal of Econometrics, 54(1–3), 159–178. https://doi.org/10.1016/0304-4076(92)90104-Y

25.

Mattei

Pistoresi

(2019). Unemployment and suicide in Italy: Evidence of a long-run association mitigated by public unemployment spending. European Journal of Health Economics, 20(4), 569–577. https://doi.org/10.1007/s10198-018-1018-7

26.

McNown

Sam

C. Y.

Goh

S. K.

(2018). Bootstrapping the autoregressive distributed lag test for cointegration. Applied Economics, 50(13), 1509–1521. https://doi.org/10.1080/00036846.2017.1366643

27.

Men

Y. V.

Yeung

C. Y.

Yip

P. S. F.

(2022). The association between unemployment and suicide among employed and unemployed people in Hong Kong: A time-series analysis. Journal of Affective Disorders, 305, 240–243. https://doi.org/10.1016/j.jad.2022.03.013

28.

OECD . (2022). OECD statistics. Retrieved December 19, 2022, from. https://stats.oecd.org/

29.

Okada

Samreth

(2013). A study on the socio-economic determinants of suicide: Evidence from 13 European OECD countries. The Journal of Socio-Economics, 45, 78–85. https://doi.org/10.1016/j.socec.2013.04.009

30.

Pesaran

M. H.

Shin

(1995). An autoregressive distributed-lag modelling approach to cointegration analysis. In Strom

(Ed.), Econometrics and economic theory in the 20th century (pp. 371–413). Cambridge University Press. https://doi.org/10.1017/CCOL521633230.011

31.

Pesaran

M. H.

Shin

Smith

R. J.

(2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326. https://doi.org/10.1002/jae.616

32.

Phillips

P. C. B.

Perron

(1988). Testing for a unit root in time series regression. Biometrika, 75(2), 335–346. https://doi.org/10.1093/biomet/75.2.335

33.

Pridmore

Reddy

(2012). Financial loss and suicide. Malaysian Journal of Medical Sciences, 19(2), 74–76. Retrieved from. https://www.mjms.usm.my75

34.

Reeves

Stuckler

McKee

Gunnell

Chang

S.-S.

Basu

(2012). Increase in state suicide rates in the USA during economic recession. The Lancet, 380(9856), 1813–1814. https://doi.org/10.1016/S0140-6736(12)61910-2

35.

Sam

C. Y.

McNown

Goh

S. K.

(2019). An augmented autoregressive distributed lag bounds test for cointegration. Economic Modelling, 80, 130–141. https://doi.org/10.1016/j.econmod.2018.11.001

36.

Stack

(2021). Contributing factors to suicide: Political, social, cultural and economic. Preventive Medicine, 152(Pt 1), 106498. https://doi.org/10.1016/j.ypmed.2021.106498

37.

The World Bank . (2022a). Inflation, consumer prices (annual %) - Turkiye, OECD members | Data. Retrieved December 19, 2022, from. https://data.worldbank.org/indicator/FP.CPI.TOTL.ZG?end=2021&locations=TR-OE&start=1980

38.

The World Bank . (2022b). Unemployment, total (% of total labor force) (modeled ILO estimate) - Turkiye, OECD members | Data. Retrieved December 19, 2022, from. https://data.worldbank.org/indicator/SL.UEM.TOTL.ZS?locations=TR-OE

39.

The World Bank . (2022c). World Bank open data (WBOD). Retrieved December 19, 2022, from. https://data.worldbank.org/

40.

TSI . (2014). Turkish statistical Institute - statistical indicators 1923-2013. TSI.

41.

TSI . (2022). Turkish statistical Institute - data portal. Retrieved December 19, 2022, from. https://data.tuik.gov.tr/

42.

Vandoros

Avendano

Kawachi

(2019). The association between economic uncertainty and suicide in the short-run. Social Science and Medicine, 220, 403–410. https://doi.org/10.1016/j.socscimed.2018.11.035

43.

Webb

R. T.

Kapur

(2015). Suicide, unemployment, and the effect of economic recession. The Lancet Psychiatry, 2(3), 196–197. https://doi.org/10.1016/S2215-0366(14)00129-1

44.

WHO . (2021). Suicide worldwide in 2019: Global health estimates. World Health Organization. Licence: CC BY-NC-SA 3.0 IGO. Retrieved from. https://apps.who.int/iris/rest/bitstreams/1350975/retrieve

45.

Yang

Lester

(1992). The misery index and an index of misery. Atlantic Economic Journal, 20(3), 98–98. https://doi.org/10.1007/BF02300179

46.

Yang

Lester

(1999). The misery index and suicide. Psychological Reports, 84(3 Pt 2), 1086–1086. https://doi.org/10.2466/pr0.1999.84.3c.1086

The Drivers of Suicides in Turkey: An Econometric Approach

Abstract

Keywords

Introduction

Conceptual Framework and Literature Summary

Uncertainty and the Misery Index as the Drivers of Suicides

Other Drivers of Suicides

Data, Model, and Method

Hypotheses to be Tested

Empirical Results

Discussion

Conclusion

Limitations and Further Research Directions

Footnotes

Author’s Note

Acknowledgments

Declaration of Conflicting Interests

Funding

ORCID iD

Appendix

Author Biography

References