1-s2.0-s0277953610007215-main
TRANSCRIPT
-
7/31/2019 1-s2.0-S0277953610007215-main
1/8
Economic fluctuations and suicide: A comparison of Taiwan and Hong Kong
Ying-Yeh Chen a,b,*, Paul SF Yip c,d, Carmen Lee d, Hsiang-Fang Fan a, King-Wa Fu e
a Taipei City Psychiatric Centre, Taipei City Hospital, Taipei, Taiwanb Institute of Public Health and Department of Public Health, National Yang-Ming University, Taipei, Taiwanc Department of Social Work and Social Administration, University of Hong Kong, Hong Kongd Center for Suicide Research and Prevention, University of Hong Kong, Hong KongeJournalism and Media Studies Centre, University of Hong Kong, Hong Kong
a r t i c l e i n f o
Article history:
Available online 21 October 2010
Keywords:
Taiwan
Hong Kong
Suicide
Unemployment
Economic fluctuations
Charcoal burning suicide
a b s t r a c t
This study examines the impact of unemployment on suicide rates in Taiwan and Hong Kong during the
period of rising unemployment (1997e2003) and its subsequent decline (2003e2007), with 2003 as the
turning point. During these initial years of high unemployment, suicide rates increased markedly in Hong
Kong and Taiwan; however, as employment conditions improved, suicide rates fell in Hong Kong but
continued to increase in Taiwan. ARMAX time-series models with appropriate time lags were used to
assess the impact of unemployment on suicide rates for both periods. It was found that for Taiwan, the
unemployment rate was positively related with the suicide rate for both males and females during the
period of high unemployment, whereas a negative relationship was observed as the rate of unemploy-
ment decreased. On the other hand, the reduction in suicide rates since 2003 was not statistically
significantly related to the improvement of employment conditions for Hong Kong; whereas the suicide
rate in Taiwan still remained at a high level due to the increasing number of charcoal burning suicide
deaths despite improvements in employment conditions.
In conclusion, lower unemployment was not necessarily associated with lower suicide rates. Exoge-
nous factors other than economic ones have been suggested to be important for understanding differ-
ences in suicide patterns in Hong Kong and Taiwan. The impact of employment conditions on suicideacross different countries deserves further investigation.
2010 Elsevier Ltd. All rights reserved.
Introduction
Recent studies have shown that general mortality ratesfluctuate
with economic conditions, rising in expansions and falling in
recessions (Khang, Lynch, & Kaplan, 2005; Tapia Granados, 2008;
Tapia Granados & Diez Roux, 2009). The only exception is suicide,
which tends to increase in unfavorable economic environments
(Chan, Yip, Wong, & Chen, 2007; Chang, Gunnell, Sterne, Lu, &
Cheng, 2009; Chen, Chou, Lai, & Lee, 2010; Gunnell, Peters,
Kammerling, & Brooks, 1995; Khang et al., 2005; Morrell, Taylor,Quine, & Kerr, 1993; Norstrom, 1995; Tapia Granados, 2008; Tapia
Granados & Diez Roux, 2009; Yu et al., 2007). In other words,
structural socioeconomic conditions may have a substantial impact
on suicidal behavior. The association between unemployment
conditions and the increase in suicide rates is particularly prom-
inent among working age males, still most often the breadwinners
in society (Blakely, Collings, & Atkinson, 2003; Chan et al., 2007;
Chen et al., 2010; Platt & Hawton, 2000; Tapia Granados, 2005 ). A
recent meta-analysis of the association between suicide and
socioeconomic characteristics of geographical areas revealed that
studies from Asian countries, compared to research from the West,
were more likely to observe a positive impact of adverse socio-
economic conditions on the increase in suicide rates (Rehkopf &
Buka, 2006).
In addition to the ecological correlation between societal
economic fluctuations and suicide mortality, individual level dataalso indicate that suicide is more common among the unemployed
andamongthose whoare marginalized in thelabormarket (Agerbo,
2005; Blakely et al., 2003; Platt & Hawton, 2000; Qin, Agerbo, &
Mortensen, 2003). However, there have been debates about
whether the unemploymentesuicide relationship at the individual
level is causal or is actually confounded by health selection(i.e. poorer health or mental health leading to suicide as well as
unemployment) (Agerbo, 2003; Preti, 2003). Only well-designed
longitudinal follow-up studies are able to provide convincing
evidence for the unemployment-suicide link, but conducting
such studies is extremely labor intensive and time consuming.
* Corresponding author. 309 Songde Road, XinYi District, Taipei City Psychiatric
Center, Taipei, Taiwan. Tel.: 886 2 27263141.
E-mail address: [email protected] (Y.-Y. Chen).
Contents lists available at ScienceDirect
Social Science & Medicine
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / s o c s c i m e d
0277-9536/$ e see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2010.09.043
Social Science & Medicine 71 (2010) 2083e2090
mailto:[email protected]://www.sciencedirect.com/science/journal/02779536http://www.elsevier.com/locate/socscimedhttp://dx.doi.org/10.1016/j.socscimed.2010.09.043http://dx.doi.org/10.1016/j.socscimed.2010.09.043http://dx.doi.org/10.1016/j.socscimed.2010.09.043http://dx.doi.org/10.1016/j.socscimed.2010.09.043http://dx.doi.org/10.1016/j.socscimed.2010.09.043http://dx.doi.org/10.1016/j.socscimed.2010.09.043http://www.elsevier.com/locate/socscimedhttp://www.sciencedirect.com/science/journal/02779536mailto:[email protected] -
7/31/2019 1-s2.0-S0277953610007215-main
2/8
A case-control design that takes account of mental disordersmay be
a more plausible alternative. However, the results of some case-
control studies have been inconsistent (Beautrais, Joyce, & Mulder,
1998; Jones, Forster, & Hassanyeh, 1991; Newman & Bland, 2007).
The aim of this study is to provide empirical evidence to
demonstrate how changing economic environments impact suicide
rates in Hong Kong and Taiwan, which have experienced dramatic
economic decline and expansion for the period 1997e2007. Taiwan
(with a population of 23 million) and Hong Kong (7 million), two
major economic centers in Asia, have both witnessed dramatic
changes in suicide rates over the past decade (1997e2007). The
suicide rate for Hong Kong increased from 12.6 per 100,000 in 1997
to a historical high of 18.6 in 2003, returning to about 14.5 in 2007,
a 50% increase up to 2003 and 30% decrease thereafter, whereas
Taiwans suicide rate increased from 10.0 per 100,000 in 1997 to
14.2 in 2003 and to 17.2 in 2007 e a 70% monotonic increase in
a decade. The suicide rate in Taiwan had leveled off a bit in
2007e2009. One important contributing factor in this sharp rise is
related, at least in part, to the emergence of a newsuicide methode
burning barbecue charcoal in a closed space to induce carbon
monoxide poisoning (Kuo et al., 2008; Liu, Beautrais, Caine, Chan,
Chao, Conwell et al., 2007; Yip & Lee, 2007). The rapid rise in the
use of this new method is believed to be related to economicdownturns for the period (1997e2003) (Chan, Yip, Au, & Lee, 2005;
Chang et al., 2009). Furthermore, after the media portrayed carbon
monoxide poisoning as a painless and peaceful way of eliminating
financial burdens; charcoal burning suicides dramatically increased
by 30% in Hong Kong and 20% in Taiwan during high rates of
unemployment during 1997e2003 (Liu et al., 2007). The unem-
ployment rate in 2003 was 7.9% and 5.0% for Hong Kong and
Taiwan, respectively e historically high figures in both places.
Research has indicated that this novel method of suicide has drawn
a new cohort of people to engage in suicidal behaviors who
otherwise would not have tried to kill themselves (Chan et al.,
2005; Chen, Lee, Chang, & Liao, 2009; Kuo et al., 2008; Liu et al.,
2007; Yip & Lee, 2007). The marked increases in the suicide rate
in both Taiwan and Hong Kong for the period 1997e
2003 weredriven by a complex amalgamation offinancial disruptions, fear of
losing jobs, and adjustment to a rapidly changing external envi-
ronment, as well as by the rapid rise in charcoal burning suicide
(Chan et al., 2005; Liu et al., 2007; Yip & Lee, 2007 ).
The Asian financial crisis, which took off in late 1997, reached its
greatest economic slump in 2003 due to the SARS epidemic. The
economies and economic conditions of Hong Kong and Taiwan
graduallyimprovedfromthe secondhalfof 2003during 2003e2007
(Directorate-General of Budget Accounting and Statistics Executive
Yuan Taiwan, 2009; International Monetary Fund, 2009), but the
aftermath of severe financial crises is usually deep and protracted,
andit wasestimatedthat thebanking crisisand consequentdeclines
in output and employment lasted in Hong Kong for about six years
(from 1997/1998e
2003) following the Asian financial crisis(Reinhart & Rogoff, 2009). A prominent economic decline was also
observed in Taiwan (Chen, 2008). Although there was a slight
economicexpansion in Taiwan after2001,it was short-lived and was
quickly overturned by the increase in oil prices in 2002, followed by
the SARS epidemic in 2003 (Chen, 2008). Economic conditions
started to improve, however, during 2003e2007, with both Taiwan
andHongKongentering a stageof economic expansion (Chen,2008;
Reinhart& Rogoff, 2009). Using unemployment rates as an example,
in Taiwan, the unemployment rate before 2000 was generally
around 3%; the yearly average unemployment rate reached a
historical peak of about 5.0% in 2002 and 2003; the rate then grad-
ually decreased to 3.7% in 2007, reflecting a steady recovery of the
economy during 2003e2007. Similarly, in Hong Kong, the unem-
ployment rate reached a high of 7.3% and 7.9% in 2002 and 2003
respectively; the rate steadily decreased thereafter to 4.0% in 2007.
Overall, the year 2003 can be viewed as a turning point for
employment conditions in Taiwan and Hong Kong, with both
economic entities somewhat recovering during the period 2003e
2007. The fluctuation in unemployment rates in Taiwan and Hong
Kong from 1997to 2007provides an opportunity to examine suicide
rates e particularly charcoal burning suicides, which have not
previously beenstudiede in relation to employmentconditions. It is
of interestto comparethe similaritiesand differences in suicide rates
in generaleand in charcoal burning suicide in particulare inTaiwan
and Hong Kong during the course of this economic transformation,
since not only are the methods of suicide used in these two places
similar, but they also have a common cultural inheritance and
a similar ethnic composition.
Methods
Data Collection
Dependent variables
Official monthly suicide mortality data from Taiwan and Hong
Kong were available from the Department of Health of the Execu-
tive Yuan of Taiwan and the Coroners Court of Hong Kongrespectively for the period 1997e2007. Suicide is coded as
E950e959 in the International Classification of Diseases, Ninth
Revision (ICD-9) (World Health Organization, 1977) and X60-84 in
ICD-10 (World Health Organization, 1992). Prior studies have
shown no marked differences in the coding of suicide deaths
between the two ICD revisions (Brock & Griffiths, 2003).
There is no specific coding for charcoal burning suicide under
the ICD classification system. It is usually coded under E952 and
X67: intentional self-poisoning by other gases and vapors. In both
Taiwan and Hong Kong, E952/X67 contributed to less than 2% of
suicide deaths before 1998, revealing a stark difference from the
current figure of approximately 30% in both areas. Coronial docu-
ments reveal that 95.6% and approximately 88% of the E952/X67
cases in Hong Kong and Taiwan respectively were charcoal burningsuicides (Lin, Chen, Huang, & Lu, 2008; Liu et al., 2007). Overall, the
majority of E952/X67 cases in both Taiwan and Hong Kong today
are charcoal burning suicides. In the current paper, we refer to
suicide methods covered by these two codes as charcoal or other
gas poisoning.
Independent variables
Main predictor: unemployment rate. The unemployment rate is one
of the most commonly used socioeconomic indicators of the
economic well-being of a geographic area and is most often used by
researchers to indicate economic fluctuations (Watkins, 1985). The
unemployment rate data for Taiwan and Hong Kong were derived
from the monthly manpower survey report, 1997e2007 (Center for
Suicide Research and Prevention, 2009; Directorate-General ofBudget Accounting and Statistics Executive Yuan Taiwan, 2009).
Other control covariates. The divorce rate, the Gross Domestic Pro-
duct (GDP) percapita, andseasonality within a year were controlled
in the model as these factors were also possible determinants of
suicide rates according to previous studies (Gunnell, Middleton,
Whitley, Dorling, & Frankel, 2003; Ho, Chao, & Yip, 1997).
Data analyses
Following the steps stated in Cryer and Chan (2008), time series
analyses with prewhitening procedure were conducted in order to
examine the relationship between monthly suicide rate, St (depen-
dent variable) and monthly unemployment rate, Ut (independent
Y.-Y. Chen et al. / Social Science & Medicine 71 (2010) 2083e20902084
-
7/31/2019 1-s2.0-S0277953610007215-main
3/8
variable) at time t. First, we need to determine the correct time lag
between the two time series variables. We differenced (if the series
were found to be non-stationary) and log-transformed (stabilizing
the datavariance)the monthly suicide rateand unemployment rate.
To avoid obtaining spurious correlation between two strongly
autocorrelated time series, a prewhitening procedure was therefore
usedto remove the temporalcorrelationof independent variable,i.e.
making Ut a white noise. Then, we calculated the cross-correlation
function between the prewhitened time series to investigate
whether the suicide rate and unemployment rate were really cross-
correlated and, if so,to determine the time lag lu between them. The
specification of the noise process Zt was obtained by examining the
residuals from an ordinary least squares fit of suicide rate on the
time-lagged unemployed rate. Finally, ARMAX (an autoregressive
moving average model (ARMA) with an exogenous inputs model
(X)) models were fitted to find the estimated coefficients of unem-
ployment rates b and their standard errors.
St b$Utlu Zt
where Stis suicide rateat time t; Utlu is unemployment rate at time
(tlu); and Zt is an autoregressive moving average (ARMA) model
with p autoregressive terms and q moving average terms, usuallydenoted by ARMA p; q c 3t Pp
i 14i$Zti
Pqi 1
qi$3ti;
where c is a constant; 3t is a white noise; and b, 4i and qi are the
coefficients of the model (i 1..p or q)
Results are reported in Table 1. By repeating the above proce-
dures, the monthly divorce rate Dt (with time lag ld) and GDP Gt(with time lag lg) were added into the models as covariates to test
whether there were changes in the adjusted coefficients of the
unemployment rate b.
St b$Utlu c$Dtld d$Gtlg Zt
where c and d are the coefficients of the model.
The relationships between the overall suicide rate, the charcoal
or other gas poisoning suicide rate, and the unemployment rate
from 1997 to 2007 were estimated and stratified by sex. In order to
assess the impact of age on the charcoal burning suicide rate, an
age-gender analysis was carried out. To capture the impact of thechange in employment conditions on the changein suicide rates for
different age groups for the charcoal burning suicide deaths, we
further grouped our analysis into two different periods: 1997e2003
(the period of an increase in the unemployment rate) and
2003e2007 (the period of a decrease in the unemployment rate) to
highlight the intriguing relationship between employment condi-
tions and suicide rates. All analyses were performed using R version
2.10.1 with the TSA package.
Results
Suicide rates, particularly male suicide rates, rose as unem-
ployment rose (1997e2003) in both Taiwan and Hong Kong (Fig. 1a
and b). However, when unemployment decreased in both placesafter 2003, suicide rates in Taiwan continued to rise (Fig. 1a).
Conversely, in Hong Kong, suicide rates declined along with
unemployment rates during the period of economic recovery and
improvements in employment conditions (2003e2007) (Fig. 1b).
Charcoal burning suicide, once an unknown method, has been
a popular method of suicide in both Taiwan and Hong Kong since
1998 (Fig. 2a and b). Charcoal or other gas poisoning was the
second leading cause of suicide deaths in 2003 in Hong Kong
Table 1
ARMAX regression analyses of therelationships betweenunemployment rateand suicide ratetrend,charcoal or other gas poisoning(CB) suicide ratetrendin Taiwanand Hong
Kong, 1997e2007.
Taiwan Hong Kong
1997e2003a 2003e2007b 1997e2003a 2003e2007b
Crude Adjusted# Crude Adjusted# Crude Adjusted# Crude Adjusted#
Overall suicide rate
Unemployment 0.4477** 0.4468* 1.7030*** 1.4226** ns ns ns ns
(se) (0.1724) [8] (0.1875) [8] (0.4116) [5] (0.4136) [5]
Male suicide rate
Unemployment 0.5429** 0.5249** 1.6195** 1.6967** ns ns ns ns
(se) (0.1805) [8] (0.1886) [8] (0.5294) [4] (0.5064) [4]
Female suicide rate
Unemployment 0.5633* 0.4705* 1.0013** 0.7258ns 0.1787* 0.2221* ns ns
(se) (0.2550) [8] (0.2234) [8] (0.3285) [5] (0.3819) [5] (0.0778) [15] (0.0809) [15]
CB suicide rate
Unemployment ns ns 0.2157* 1.0539ns ns ns ns ns
(se) (0.1090) [7] (0.8070) [7]
Male CB suicide rate
Unemployment ns ns 2.4735* 1.2933ns ns ns ns ns
(se) (1.1420) (0.9290) [7]
Female CB suicide rate
Unemployment (se) 0.0578*** 0.0581*** ns ns ns ns 0.1084* 0.5974ns
(0.0123) [8] (0.0125) [8] (0.0494) [1] (0.3597) [1]
Note: *p < .05, **p < .01, ***p < .0001.
Crude: regression analyses with unemployment rate as the single predictor and suicide rate as outcome variable.
# Adjusted: regression analyses assessing the relationship between unemployment rate and suicide rate adjusting for GDP and divorce rate.
[ ]: time lag lua period of increase of unemployment rate.b
period of decrease of unemployment rate.
Y.-Y. Chen et al. / Social Science & Medicine 71 (2010) 2083e2090 2085
-
7/31/2019 1-s2.0-S0277953610007215-main
4/8
(second to jumping) and Taiwan (second to hanging). After 2003,
the reduction in suicide rates can be attributed to the decrease in
suicides from charcoal or other gas poisoning in Hong Kong
during the period of lower unemployment rates (2003e2007)
(Fig. 2). Nonetheless, in Taiwan, charcoal or other gas poisoning as
a means of suicide, which emerged during the period of wors-
ening employment conditions, continued to display an upward
trend during the period of improving employment condition,
2003e2007 (Fig. 2a).
Table 1 presents the results from an ARMAX regression proce-
dure analyzing the relationship between detrended unemployment
rates and suicide rates on monthly basis for Taiwan and Hong Kong
for the two periods 1997e
2003 and 2003e
2007, which aremodeled separately. For Taiwan, a positive relationship between
suicide and unemployment rates for both sexes during 1997e2003,
with a time lag of eight months, was detected. In other words, the
unemployment rate eight months earlier was shown to be posi-
tively related to the suicide rate eight months later. Also, for the
period 1997e2003 with a high unemployment rate, a significant
positive relationship was found among Taiwanese females, but not
among males, between the rates of suicide resulting from charcoal
or other gas poisoning and an increase in unemployment (see Table
1). When compared to the rate for their male counterparts, the rate
of suicide by charcoal or other gas poisoning for females was more
closely linked to a higher rate of unemployment. During the period
of improving employment condition (2003e2007), the opposite
relationship (negative) was observed between the unemployment
rate and the overall suicide rate for both sexes with a shorter time
lag of 4e5 months, whereas a positive relationship still remained
for charcoal or other gas poisoning suicide deaths for males.
However, after adjustment of the other covariates, GDP and the
divorce rate, the rates of suicide by charcoal or other gas poisoning
werefound not to be significantly related to unemployment rate for
both sexes during periods of economic recovery. Apparently there
were other factors which were possibly more related to the
increase in suicide during the period. The overall suicide rates were
negatively correlated to unemployment rates in Taiwan during the
periods of economic recovery (2003e2007), indicating a discon-
nection between unemployment and suicide rates in times of
improving employment conditions in comparing to 1997e
2003.For Hong Kong, only female suicide rates were found to be
positively related to unemployment rates for the period 1997e
2003 with a long time lag of 15 months. However, a negative
relationship was observed between the female suicide rates for
charcoal/other gas poisoning suicide and unemployment rates
before adjusting of other covariates for the period 2003e2007.
After adjustment of other covariates, the significant relationship
between female suicide rates and unemployment rates during
economic recovery became insignificant as well.
Age-gender specific analysis was performed to examine the
impact of unemployment rates on charcoal or other gas poisoning
suicide rates. For Hong Kong, suicide by charcoal or other gas
poisoning among the three age groups (15e24, 25e59, and 60 or
over) were not related to unemployment rates. For Taiwan, during
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Suicidedeathsper100,0
00
0.00
1.00
2.00
3.00
4.00
5.00
6.00
Unemploymen
rrate(%)
Increase of unemployment rate Decrease of unemployment rate
Female suicide rate
Unemployment rate
Male suciide rate
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Suicidedeathsper100,00
0
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
Unemploymentrate(%)
Increase of unemployment rate Decrease of unemployment rate
Female suicide rate
Male suicide rate
Unemployment rate
a
b
Fig. 1. (a) Overall and sex-specific suicide rate trends in relation to economic fluctuations in Taiwan, 1997e2007. (b) Overall and sex-specific suicide rate trends in relation to
economicfl
uctuations in Hong Kong, 1997e
2007.
Y.-Y. Chen et al. / Social Science & Medicine 71 (2010) 2083e20902086
-
7/31/2019 1-s2.0-S0277953610007215-main
5/8
the period of 1997e2003, it was found that the charcoal or other
gas poisoning suicide rate for each of the age groups were not
related to the increase in unemployment. It is important to note
that the proportion of dying from charcoal burning poisoning
among all suicide deaths in Taiwan was still relatively small before
2003 (less than 10%). For the period of improvement of employ-
ment condition (2003e2007), number of charcoal burning suicide
has increased significantly in Taiwan since 2003. The suicide rate
for the age group of 25e59 was shown to be positively related to
unemployment rate with a time lag of 7 months despite a negative
relationship was observed between unemployment rate andoverall suicide rate. Apparently, the impact of unemployment on
the charcoal or other gas poisoning suicide for the two different
age groups 15e24 and 25e59 have cancelled out one another (see
Table 1 and Table 2).
Discussion
We found an upward trend in suicide rates in Taiwan and Hong
Kong in times of increased unemployment. This increase in
unemployment appeared to trigger charcoal burning suicides in
both areas, as suggested by Chang et al. (2009). However, the
improvement in employment conditions for the period 2003e2007
revealed different suicide patterns in Taiwan and Hong Kong. While
we observed a declining trend in overall suicide rates as well as the
rates for suicide by charcoal or other gas poisoning in Hong Kong,
the effect of unemployment rates was shown to not be statistically
significantly related to the decline. The picture was different in
Taiwan, where a divergence in suicide and unemployment rates
was observed during 2003e2007, indicating that suicide rates
continued to rise even after employment conditions improved. Our
results indicate that while suicide rates generally rise in times of
increased unemployment, improving employment conditions is
neither a sufficient nor a necessary condition for suicide rates to
fall. It is important to understand and appreciate the disconnection
between the employment condition and suicide rate at thecommunity level.
Despite the many similarities between Taiwan and Hong Kong in
cultural, ethnic, and economic development, there is a clear
divergence in suicide rates in relation to employment conditions.
This divergence requires an in-depth exploration. It is possible that
factors other than employment conditions account for the contin-
uous rise in suicide rates during periods of economic recovery in
Taiwan. One potential factor could be the influence of exaggerated
media reporting of suicide events. Unlike Hong Kong, where tabloid
journalism has long existed, tabloid journalism did not appear in
Taiwan until the launch of one Hong Kong-based newspaper Apple
Daily in May, 2003 (Pai, 2004). Gaining its readership through
sensationalism, exaggerated headlines, and flashy graphic images,
Apple Daily changed the norms of suicide reporting and news
0.0
0.5
1.0
1.5
2.0
2.5
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Suicidedeathsper100,0
00
0.00
1.00
2.00
3.00
4.00
5.00
6.00
Unemployment
rate(%)
Increase of unemployment rate Decrease of unemployment rate
CB suicide
non-CB suicide
Unemployment rate
0.0
0.5
1.0
1.5
2.0
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Suicidedeathsper100,00
0
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
Unemploymentrate(%)
Increase of unemployment rate Decrease of unemployment rate
CB suicide
non-CB suicide
Unemployment rate
a
b
Fig. 2. (a) Charcoal or other gas poisoning suicide (CB suicide) and all other suicides not in the charcoal or other gas poisoning category (non-CB suicide) in relation to economic
fluctuations in Taiwan,1997e
2007. (b) Charcoal or other gas poisoning suicide (CB suicide) and all other suicides not in the charcoal or other gas poisoning category (non-CB suicide)in relation to economic fluctuations in Hong Kong, 1997e2007.
Y.-Y. Chen et al. / Social Science & Medicine 71 (2010) 2083e2090 2087
-
7/31/2019 1-s2.0-S0277953610007215-main
6/8
reporting in general in Taiwan soon after its entrance into the
market (Audit Bureau of Circulations, 2009). Recently, a series of
studies from Taiwan have pointed to copycat suicides provoked by
extensive media reporting of celebrity suicides (Chen et al., 2010;
Cheng, Hawton, Chen, Yen, Chang et al., 2007; Cheng, Hawton,Chen, Yen, Chen et al., 2007; Cheng, Hawton, Lee, & Chen, 2007),
indicating the increasing influence of the media on suicide-related
deaths in Taiwan.
Political instability may be another factor responsible for the
upswing in suicide rates during periods of improved employment
conditions in Taiwan. Political issues have been prominent sources
of social turmoil in Taiwan since the transfer of political powerfrom
the KMT (Kuomintang, the Nationalist Party) to the DPP (Demo-
cratic Progress Party) between 2000 and 2008 (Wu, 2007). The
potential impact of political turbulence on suicide rates in Taiwan
has been suggested (Chen et al., 2010), but not rigorously investi-
gated. Designing a reliable and valid measure of political stability to
disentangle the relationships between political turbulence and
suicide may help illuminate suicide patterns in Taiwan. Suiciderates in Hong Kong tend to respond to unemployment rates;
however, they have not been shown to be directly proportional to
improvement in unemployment conditions. It seems we still need
to search for other factors to explain thefluctuation of suicide rates.
An additional explanation for the disconnection between
suicide and unemployment rates in Taiwan is the uneven economic
benefits gained in times of economic prosperity. In other words, an
improvement in employment conditions may not benefit every
individual equally. A group of working poor has emerged in Hong
Kong and Taiwan due to relatively low wages and rising living costs
(Center for Suicide Research and Prevention, 2009; Lo, 2007). Prior
research has indicated that financial problems accounted for a high
proportion of suicide deaths in Hong Kong and the emergence of
a signifi
cant number of working poor in the community (Center for
Suicide Research and Prevention, 2009). More than 20% (about
200,000) of the working population earn less than US$750 a month
in Hong Kong, and the medium income of the Hong Kong working
population is US$1250. The most disadvantaged among the
employed may still be at great risk even though unemploymentrates have improved. Also, more employed people chose charcoal
burning suicide over other suicide methods (Chan et al., 2005).
It is noteworthy that unemployment rates in Taiwan have
recovered at a much slower pace than those in Hong Kong. In fact,
even though unemployment rates decreased after 2003 in Taiwan,
they did not return to pre-crisis levels, and protracted economic
stagnation in Taiwan may have resulted in the continued rise in
suicide rates.
Despite many similarities in social-economic conditions Taiwan
and Hong Kong have adopted different measures to lower suicide
ratesin recent years. Recognizing that depression is a key risk factor
in suicide, suicide prevention strategies adopted by Taiwans
Suicide Prevention Center have mainly focused on the early
recognition and treatment of depression (Taiwans Suicide
Prevention Center, 2009). In Hong Kong, however, a more gener-
alized public health approach to lower the suicide rate has been
adopted (Center for Suicide Research and Prevention, 2009).
Determining to what extent suicide rates in Taiwan and Hong Kong
have been affected by these differences in approach deserves
further exploration.
In Taiwan, we did not observe prominent gender differences in
suicide rates in response to changes in employment conditions
during the study period. Among both males and females in Taiwan,
unemployment rates were positively related to suicide rates during
economic recessions and negatively related to suicide rates during
economic prosperity; however, adjusted models indicate stronger
relationships among males, a finding compatible with prior
research results that suggest male suicide rates are more affected
Table 2
Age-specified ARMAX regression analyses of the relationships between unemployment rate and charcoal or other gas poisoning (CB) suicide rate trend in Taiwan and Hong
Kong, 1997e2007.
Taiwan Hong Kong
1997e2003a 2003e2007b 1997e2003a 2003e2007b
Crude Adjusted# Crude Adjusted# Crude Adjusted# Crude Adjusted#
CB suicide rate
Unemployment (se)
15e24 ns ns 0.2690*
(0.1223) [0]
4.9144 ns
(2.8600) [0]
ns ns ns ns
25e59 ns ns 0.3441*
(0.1547) [7]
0.9105 ns
(0.6976) [7]
ns ns ns ns
60 ns ns ns ns ns ns ns ns
Male CB suicide rate
Unemployment (se)
15e24 ns ns 0.4438*
(0.1967) [0]
1.6722*
(0.8421) [0]
ns ns ns ns
25e9 ns ns 0.7516**
(0.2622) [7]
2.4575*
(0.9953) [7]
ns ns ns ns
60 ns ns ns ns ns ns ns ns
Female CB suicide rate
Unemployment (se)
15e24 ns ns ns ns ns ns ns ns
25e
59 ns ns ns ns ns ns 0.6217 ns
(0.3461) [1]0.5409 ns
(0.3465) [1]
60 ns ns ns ns ns ns ns ns
Note: *p < .05, **p < .01, ***p < .0001.
Crude: regression analyses with unemployment rate as the single predictor and suicide rate as outcome variable.
# Adjusted: regression analyses assessing the relationship between unemployment rate and suicide rate adjusting for GDP and divorce rate.
[ ]: time lag lua period of increase of unemployment rate.b period of decrease of unemployment rate.
Y.-Y. Chen et al. / Social Science & Medicine 71 (2010) 2083e20902088
-
7/31/2019 1-s2.0-S0277953610007215-main
7/8
by economic conditions (Blakely et al., 2003; Chan et al., 2007;
Chen et al., 2010; Platt & Hawton, 2000; Tapia Granados, 2005 ).
In the case of suicide by charcoal or other gas poisoning, we
found that female rates were more affected by the increase in
unemployment rates in Taiwan during 1997e2003. Although this
finding is counter-intuitive, we have parallel findings in Hong Kong
in that unemployment rates were significantly related to female
suicide rates (but were not related to male suicide rates) when GDP,
divorce rates, and time trend effects were simultaneously consid-
ered. In Hong Kong, we also found that in times of economic
recovery, female rates for suicide by charcoal or other gas poisoning
did not follow the decrease in unemployment rates. Overall, it
seems that female suicide rates were particularly affected by
unemployment conditions in Hong Kong, and that when the
economic situation improved, female rates for suicide by charcoal
or other gas poisoning failed to decrease accordingly (although this
was not statistically significant in the adjusted model). This could
be related to the increase in womens labor force participation in
recent decades in both places; as a result of this, women may be as
affected as men, if not more so, by deteriorating employment
conditions. In addition, working women, often not afforded the
same opportunities as men, are more likely to be given low-skilled,
low-wage jobs with very limited possibility of promotion (Lee,2007; Sung, 2003). This type offemale labor is usually the most
expendable in times of economic recession. It is possible that
a disadvantaged employment situation concomitant with the
socio-cultural treatment of suicide as a protest against social
oppression in a Chinese cultural context (Ji, Kleinman, & Becker,
2001) may have contributed to the gender paradox (i.e. female
suicides were more sensitive to economic conditions) observed in
Hong Kong.
The age and gender stratified analysis for charcoal or other gas
poisoning suicide rates revealed that this specific method of suicide
in middle aged males were positively related to employment
conditions, even though a disconnection between unemployment
and overall suicide rates were found in Taiwan during 2003e2007.
Thefindingindicatesthat although miscellaneous factors other thaneconomic conditions drove up suicide rates in Taiwan during
2003e2007; economic factorwas still an importantdriving forcefor
the increase in charcoal burning suicide rates among middle aged
males. The finding provides further evidence for the observation
that charcoal burning suicide is a method particularly favored by
middle aged males under economic stress (Chan et al, 2005).
There are several limitations that need to be addressed when
interpreting the current results. First, the ecological associations
between population-level unemployment rates and suicide rates
may not be generalized to the individual level. However, prior
psychological autopsy studies using individual-level data in both
Taiwan and Hong Kong have demonstrated the adverse effect of
unemployment on risk of suicide (Chen et al., 2006; Cheng, Chen,
Chen, & Jenkins, 2000). In addition, although we controlled forseveral potential confounders, there are still other factors that may
confound the relationship betweenunemploymentand suiciderates.
Tracing the vicissitude of suicide trends in the course of
economic transitions in Taiwan and Hong Kong, our research has
revealed a complex picture of interactions between employment
conditions, suicide rates, gender, methods of suicide, and country-
specific reactions to contextual socioeconomic conditions. The
continuous rise of the charcoal burning suicide rate in Taiwan,
despite improvements in economic conditions, suggests that we
should consider limiting access to charcoal as a way of preventing
suicide. Preliminary results in Hong Kong have shown that a slow
access strategy of removing charcoal from open shelves in
convenience stores is effective in reducing the rates of charcoal
burning suicide in the community (Yip, 2009; Yip, Law, Fu, Law,
Wong, & Xu, 2010). In Hong Kong, the fluctuation of female
suicide rates with economic downturns has directed the prevention
focus toward improving social policies in relation to labor market
conditions for women. In the face of recent global economic
uncertainty, the troubling trends of suicide in Taiwan and Hong
Kong call for early attention and active intervention.
Acknowledgement
This study was partly supported by a grant from Department of
Health, Taipei City Government (Grant No: 99001-62-016) and
Hong Kong Research Grant Council (784210).
References
Agerbo, E. (2003). Unemployment and suicide. Journal Epidemiology CommunityHealth, 57(8), 560e561.
Agerbo, E. (2005). Effect of psychiatric illness and labour market status on suicide:a healthy worker effect? Journal of Epidemiology & Community Health, 59(7),598e602.
Audit bureau of circulations. (2009). http://www.abc.org.tw/. Accessed 25 06 2010.Beautrais, A. L., Joyce, P. R., & Mulder, R. T. (1998). Unemployment and serious
suicide attempts. Psychological Medicine, 28(1), 209e218.Blakely, T. A., Collings, S. C., & Atkinson, J. (2003). Unemployment and suicide:
evidence for a causal association? Journal of Epidemiology & Community Health,57(8), 594e600.
Brock, A., & Griffiths, C. (2003). Trends in suicide method in England and Wales,1979e2001. Health Statistics Quarterly, 20, 7e17.
Center for suicide research and prevention. (2009). http://csrp.hku.hk/WEB/eng/pageHandler.asp?id 1. Accessed 25.06.2010.
Chan, K. P., Yip, P. S. F., Au, J., & Lee, D. T. (2005). Charcoal-burning suicide in post-transition Hong Kong. British Journal of Psychiatry, 186, 67e73.
Chan, W. S., Yip, P. S. F., Wong, P. W., & Chen, E. Y. (2007). Suicide and unemploy-ment: what are the missing links? Archives of Suicide Research, 11(4), 327e335.
Chang, S. S., Gunnell, D., Sterne, J. A., Lu, T. H., & Cheng, A. T. (2009). Was theeconomic crisis 1997e1998 responsible for rising suicide rates in East/South-east Asia? A time-trend analysis for Japan, Hong Kong, South Korea, Taiwan,Singapore and Thailand. Social Science & Medicine, 68(7), 1322e1331.
Chen, E. Y., Chan, W. S., Wong, P. W., Chan, S. S., Chan, C. L., Law, Y. W., et al. (2006).Suicide in Hong Kong: a case-control psychological autopsy study. PsychologicalMedicine, 36(6), 815e825.
Chen, H. W. (2008). Determining the reference date for the 11th business cycle peakin Taiwan. Economic Research, 9, 1e26.
Chen, V. C., Chou, J. Y., Lai, T. J., & Lee, C. T. (2010). Suicide and unemployment rate inTaiwan: a population-based study, 1978e2006. Social Psychiatry and PsychiatricEpidemiology, 45(4), 447e452.
Chen, Y. Y., Lee, M. B., Chang, C. M., & Liao, S. C. (2009). Methods of suicide indifferent psychiatric diagnostic groups. Journal of Affective Disorders, 118(1e3),196e200.
Cheng, A. T. A., Chen, T. H., Chen, C. C., & Jenkins, R. (2000). Psychosocial andpsychiatric risk factors for suicide: case-control psychological autopsy study.British Journal of Psychiatry, 177, 360e365.
Cheng, A. T. A., Hawton, K., Chen, T. H., Yen, A. M., Chang, J. C., Chong, M. Y., et al.(2007). The influence of media reporting of a celebrity suicide on suicidalbehavior in patients with a history of depressive disorder. Journal of AffectiveDisorders, 103(1e3), 69e75.
Cheng, A. T. A., Hawton, K., Chen, T. H., Yen, A. M., Chen, C. Y., Chen, L. C., et al.(2007). The influence of media coverage of a celebrity suicide on subsequentsuicide attempts. The Journal of Clinical Psychiatry, 68(6), 862e866.
Cheng, A. T. A., Hawton, K., Lee, C. T., & Chen, T. H. (2007). The influence of media
reporting of the suicide of a celebrity on suicide rates: a population-basedstudy. International Journal of Epidemiology, 36(6), 1229e1234.
Cryer, J. D., & Chan, K. S. (2008). Time series regression models. In Time seriesanalysis with applications in R (pp. 249e276). New York: Springer.
Directorate-General of Budget Accounting and Statistics Executive Yuan Taiwan.(2009). http://www.dgbas.gov.tw/ct.asp?xItem17144&ctNode3246. Accessed25.06.2010.
Gunnell, D., Middleton, N., Whitley, E., Dorling, D., & Frankel, S. (2003). Why aresuicide rates rising in young men but falling in the elderly? e a time-seriesanalysis of trends in England and Wales 1950e1998. Social Science & Medicine,57(4), 595e611.
Gunnell, D., Peters, T. J., Kammerling, R. M., & Brooks, J. (1995). Relation betweenparasuicide, suicide, psychiatric admissions, and socioeconomic deprivation.British Medical Journal, 311(6999), 226e230.
Ho, T. P., Chao, A., & Yip, P. (1997). Seasonal variation in suicides re-examined: nosex difference in Hong Kong and Taiwan. Acta Psychiatrica Scandinavica, 95(1),26e31.
International monetary fund. (2009). http://www.imf.org/external/pubs/ft/weo/
2009/01/weodata/download.aspx. Accessed 25.06.2010.
Y.-Y. Chen et al. / Social Science & Medicine 71 (2010) 2083e2090 2089
http://www.abc.org.tw/http://csrp.hku.hk/WEB/eng/pageHandler.asp?id=1http://csrp.hku.hk/WEB/eng/pageHandler.asp?id=1http://csrp.hku.hk/WEB/eng/pageHandler.asp?id=1http://csrp.hku.hk/WEB/eng/pageHandler.asp?id=1http://www.dgbas.gov.tw/ct.asp?xItem=17144&ctNode=3246http://www.dgbas.gov.tw/ct.asp?xItem=17144&ctNode=3246http://www.dgbas.gov.tw/ct.asp?xItem=17144&ctNode=3246http://www.dgbas.gov.tw/ct.asp?xItem=17144&ctNode=3246http://www.imf.org/external/pubs/ft/weo/2009/01/weodata/download.aspxhttp://www.imf.org/external/pubs/ft/weo/2009/01/weodata/download.aspxhttp://www.imf.org/external/pubs/ft/weo/2009/01/weodata/download.aspxhttp://www.imf.org/external/pubs/ft/weo/2009/01/weodata/download.aspxhttp://www.dgbas.gov.tw/ct.asp?xItem=17144&ctNode=3246http://www.dgbas.gov.tw/ct.asp?xItem=17144&ctNode=3246http://www.dgbas.gov.tw/ct.asp?xItem=17144&ctNode=3246http://csrp.hku.hk/WEB/eng/pageHandler.asp?id=1http://csrp.hku.hk/WEB/eng/pageHandler.asp?id=1http://csrp.hku.hk/WEB/eng/pageHandler.asp?id=1http://www.abc.org.tw/ -
7/31/2019 1-s2.0-S0277953610007215-main
8/8
Ji, J., Kleinman, A., & Becker, A. E. (2001). Suicide in contemporary China: a review ofChinas distinctive suicide demographics in their sociocultural context. HarvardReview of Psychiatry, 9(1), 1e12.
Jones, S. C., Forster, D. P., & Hassanyeh, F. (1991). The role of unemployment inparasuicide. Psychological Medicine, 21(1), 169e176.
Khang, Y. H., Lynch, J. W., & Kaplan, G. A. (2005). Impact of economic crisis on cause-specific mortality in South Korea. International Journal of Epidemiology, 34(6),1291e1301.
Kuo, C. J., Conwell, Y., Yu, Q., Chiu, C. H., Chen, Y. Y., Tsai, S. Y., et al. (2008).Suicide by charcoal burning in Taiwan: implications for means substitution
by a case-linkage study. Social Psychiatry and Psychiatric Epidemiology, 43(4),286e290.
Lee, K. F. (2007). Empowerment and grassroots women in Hong Kong: A case study ofHong Kong women workers association. Hong Kong: City University of HongKong.
Lin, J. J., Chen, L. H., Huang, S. M., & Lu, T. H. (2008). Problems in estimating thenumber of suicides by charcoal burning in Taiwan. Journal of EpidemiologyCommunity Health, 62(6), 566.
Liu, K. Y., Beautrais, A., Caine, E., Chan, K., Chao, A., Conwell, Y., et al. (2007). Charcoalburning suicides in Hong Kong and urban Taiwan: an illustration of the impactof a novel suicide method on overall regional rates. Journal of Epidemiology andCommunity Health, 61(3), 248e253.
Lo, W. Y. (2007). A profile of the working poor in Taiwan. Taipei: Soochow University.Morrell, S., Taylor, R., Quine, S., & Kerr, C. (1993). Suicide and unemployment in
Australia 1907e1990. Social Science & Medicine, 36(6), 749e756.Newman, S. C., & Bland, R. C. (2007). Case-control study of unemployment and
parasuicide. Comprehensive Psychiatry, 48(6), 511e515.Norstrom, T. (1995). The impact of alcohol, divorce, and unemployment on suicide:
a multilevel analysis. Social Forces, 374, 293e314.Pai, S. Y. (2004). Research of market competition and media diversity
eAn empirical
study for front page of China Times, United Daily News and Liberty Times beforeand after Appledaily Newspaper published. Kaohsiung: National Chung-ShanUniversity.
Platt, S., & Hawton, K. (2000). Suicide behaviour and the labour market. InK. Hawton, & K. van Heeringen (Eds.), The international handbook of suicide andattempted suicide (pp. 309e384). Chichester: John Wiley & Sons.
Preti, A. (2003). Unemployment and suicide. Journal of Epidemiology CommunityHealth, 57(8), 557e558.
Qin, P., Agerbo, E., & Mortensen, P. B. (2003). Suicide risk in relation to socioeco-nomic, demographic, psychiatric, and familial factors: a national register-based
study of all suicides in Denmark, 1981e1997. The American Journal of Psychiatry,160(4), 765e772.
Rehkopf, D. H., & Buka, S. L. (2006). The association between suicide and the socio-economic characteristics of geographical areas: a systematic review. Psycho-logical Medicine, 36(2), 145e157.
Reinhart, C., & Rogoff, K. (2009). The aftermath of financial crises. AmericanEconomic Review, 99(2), 466e472.
Sung, S. (2003). Women reconciling paid and unpaid work in a Confucian welfarestate: the case of South Korea. Social Policy & Administration, 37(4), 342e360.
Taiwans suicide prevention center. (2009). http://www.tspc.doh.gov.tw/tspc/
portal/center/index.jsp . Accessed 25.06.2010.Tapia Granados, J. A. (2005). Recession and mortality in Spain, 1980e1997. European
Journal of Population, 21(4), 393e422.Tapia Granados, J. A. (2008). Macroeconomic fluctuations and mortality in postwar
Japan. Demography, 45(2), 323e343.Tapia Granados, J. A., & Diez Roux, A. V. (2009). Life and death during the Great
Depression. Proceedings of the National Academy of Sciences of the United Statesof America, 106(41), 17290e17295.
Watkins, S. (1985). Recession and health: a literature review. In C. Westcott,P. G. Svensson, & H. F. K. Zollner (Eds.), Health policy implications of unemploy-ment(pp. 27e56). Copenhagen: World Health Organization, Regional Office forEurope.
World Health Organization. (1977). Manual of international statistical classification ofdiseases, injuries, and causes of death. 9th revision (ICD-9). Geneva: World HealthOrganization.
World Health Organization. (1992). International statistical classification of diseasesand related health problems. 10th (ICD-10). Geneva: World Health Organization.
Wu, Y. S. (2007). Taiwans developmental state: after the economic and politicalturmoil. Asian Survey, 47, 971e1001.
Yip, P. S. F. (2009). Prevention of suicide due to charcoal burning. In D. Wasserman(Ed.), Oxford textbook of suicidology and suicide prevention (pp. 595e597).Oxford: Oxford University Press.
Yip, P. S. F., Law, C. K., Fu, K. W., Law, Y. W., Wong, P. W., & Xu, Y. (2010). Restrictingthe means of suicide by charcoal burning. British Journal of Psychiatry, 196,241e242.
Yip, P. S. F., & Lee, D. T. S. (2007). Charcoal-burning suicides and strategies forprevention. Crisis, 28(suppl 1), 21e27.
Yu, H. S., Chen, Y. Y., Chen, C. C., Chiu, C. H., Kuo, C. J., Chang, H., et al. (2007).Socioeconomic factors and suicide mortality trend in Taipei City, 1991e2004.Taiwan Journal of Public Health, 26(1), 66e74.
Y.-Y. Chen et al. / Social Science & Medicine 71 (2010) 2083e20902090
http://www.tspc.doh.gov.tw/tspc/portal/center/index.jsphttp://www.tspc.doh.gov.tw/tspc/portal/center/index.jsphttp://www.tspc.doh.gov.tw/tspc/portal/center/index.jsphttp://www.tspc.doh.gov.tw/tspc/portal/center/index.jsp