Social Science & Medicine 153 (2016) 71e80

Contents lists available at ScienceDirect

Social Science & Medicine

journal homepage: www.elsevier.com/locate/socscimed

The 2008 financial crisis: Changes in social capital and its association
with psychological wellbeing in the United Kingdom e A panel study
€ m a, b, 1, Giuseppe N. Giordano a, b, *, 1
Martin Lindstro
a
Social Medicine and Health Policy, Department of Clinical Sciences, Faculty of Medicine, Lund University, Sweden
b
Centre for Economic Demography (CED), P.O. Box 7083, SE-SE-220 07, Lund, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: The global financial crisis of 2008 was described by the IMF as the worst recession since the Great
Received 26 October 2015 Depression. This historic event provided the backdrop to this United Kingdom (UK) longitudinal study of
Received in revised form changes in associations between social capital and psychological wellbeing. Past longitudinal studies
1 February 2016
have reported that the presence of social capital may buffer against adverse mental health outcomes.
Accepted 3 February 2016
Available online 15 February 2016
This study adds to existing literature by employing data from the British Household Panel Survey and
tracking the same individuals (N ¼ 11,743) pre- and immediately post-crisis (years 2007e09). With
longitudinal, multilevel logistic regression modelling, we aimed to compare the buffering effects of
Keywords:
United Kingdom
individual-level social capital (generalised trust and social participation) against worse psychological
Psychological wellbeing wellbeing (GHQ-12) during and immediately after the 2008 financial crisis. After comparing the same
Social capital individuals over time, results showed that stocks of social capital (generalised trust) were significantly
Generalised trust depleted across the UK during the crisis, from 40% trusting others in 2007 to 32% in 2008. Despite this
2008 financial crisis drop, the buffering effect of trust against worse psychological wellbeing was pronounced in 2008; those
Longitudinal not trusting had an increased risk of worse psychological wellbeing in 2008 compared with the previous
Panel data year in fully adjusted models (OR ¼ 1.49, 95% CI (1.34e1.65). Levels of active participation increased
across the timeframe of this study but were not associated with psychological health. From our empirical
evidence, decision makers should be made aware of how events such as the crisis (and the measures
taken to counter its effects) could negatively impact on a Nation's trust levels. Furthermore, past research
implies that the positive effects of trust on psychological wellbeing evident in this study may only be
short-term; therefore, decision makers should also prioritise policies that restore trust levels to improve
the psychological wellbeing of the population.
© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction major financial institutions failed to repay loans, the US Govern-

ment injected US$236 billion into the American banking system
Key adverse economic events in the United States (US), namely (Mathiason, 2008). Over the following months, worsening eco-
an unexpected rise in interest rates, the bursting of the 2005-06 nomic conditions spread internationally and developed into the
housing bubble and subsequent large-scale defaulting on mortgage 2008 global financial crisis, an event described by Alan Greenspan
payments were all precursors to the US sub-prime mortgage crisis as a “once-in-a-century credit tsunami” (Greenspan, 2008) and by
of 2007 (British Broadcasting Corporation, 2007). Negative fiscal the IMF as “… the worst recession since the Great Depres-
repercussions extended well into 2008 and by March, after three sion.”(Beale, 2009).
The 2008 crisis provided the backdrop for this United Kingdom
(UK) study of social capital and psychological wellbeing. The first
real effects of this event were felt in the UK on Monday 15th
* Corresponding author. Social Medicine and Health Policy, Department of Clin-
ical Sciences, Faculty of Medicine, Lund University, Sweden. September 2008, when Lehman Brothers declared bankruptcy and
E-mail addresses: [email protected] (M. Lindstro€ m), Giuseppe_nicola. closed their doors to business in London (British Broadcasting
[email protected] (G.N. Giordano). Corporation, 2008). By 11th October 2008, approximately £90
1
Postal address: Social Medicine and Health Policy, Department of Clinical Sci- billion were lost on the London Stock Exchange. In response to the
ences, Clinical Research Centre (CRC), Jan Waldenstro € ms gata 35, House 60, Floor
12, 205 02 Malmo €, Sweden. global financial contagion, the Bank of England announced in late

http://dx.doi.org/10.1016/j.socscimed.2016.02.008
0277-9536/© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
72 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro

October 2008 that it would lower its base-rate of interest and the (2000), could have positive effects on psychological wellbeing
UK Government promised to prop up failing financial institutions through increased social ties and community integration (Kawachi
with tax-payers money (Kingsley, 2012). and Berkman, 2001).
For the purposes of this study, the 2008 financial crisis was In relation to the buffering effects of social capital against
considered a natural experiment of sorts, i.e. an historical (period) worsening mental health, to date only a handful of longitudinal
event, simultaneously experienced across the UK. As theorised by empirical research papers have been identified. A recent on-line
Pildes (1996), a negative event such as this had the potential to search within PUBMED, Web of Science, SCOPUS and Google
deplete stocks of social capital, a resource loosely defined as ‘social Scholar revealed just nine articles to investigate social capital and
networks, norms of reciprocation and trust’ (Putnam, 2004). mental health outcomes over time (Bertotti et al., 2013; Ding et al.,
Interestingly, previous empirical evidence from British House- 2015; Frank et al., 2014; Giordano and Lindstro € m, 2011; Hall et al.,
hold Panel Survey data (BHPS) provides some support for this 2014; Han, 2015; Murayama et al., 2013; Tsuboya et al., 2015;
theory, with Giordano et al. (2012) revealing that 45% of individuals Verduin et al., 2014). All demonstrated positive associations be-
reported changes in generalised trust levels over a seven-year tween social capital and better individual mental health outcomes.
period (2000e07). Other national panel datasets to include trust Of these, just one had tested the hypothesis that social capital
measurements show similar short-term fluctuations (for examples buffered against economic stress and worsening mental health
see: Glanville et al., 2013; van Ingen and Bekkers, 2015). Such (Frank et al., 2014). In their paper, Frank et al. (2014) concluded that
findings contradict the belief that generalised trust is determined in the presence of social capital in rural Canada moderated the effects
early life and is resistant to change, irrespective of later-life expe- that financial strain had on perceived stress and depressive
riences (Putnam, 2000; Uslaner, 2002). Instead, such fluctuations symptoms (NT ¼ 317). Conversely, in their cross-sectional study of
reflect the economic viewpoint that trust is a summary measure of the effects of post-crisis austerity on mental health outcomes in
individual experiences, good and bad (Glaeser et al., 2000). Greece, Economou et al. (2014) concluded that the buffering effects
Social capital is considered a resource, which has been theorised of social capital were fully attenuated if individuals were under
both at the contextual- (Berkman and Kawachi, 2000; Putnam, high economic stress (NT ¼ 2256). From the above, there are a
2000) and individual-level (Bourdieu, 1986; Coleman, 1988; limited number of longitudinal empirical social capital and mental
Portes, 1998). Attempting to resolve such conceptual differences health studies and, regarding the buffering effects of social capital
empirically, multilevel research suggests that the effects of social against mental health outcomes when considering financial strain,
capital on health appear strongest at the individual-level, with results are not in agreement.
typically only 0e4% of all variation in individual health being To address this, and to investigate events surrounding the 2008
attributed to community-level aggregates (e.g. Giordano et al., crisis in the UK, we prepared panel data to compare the same in-
2011; Islam et al., 2006; Murayama et al., 2012; Poortinga, 2006; dividuals (NT ¼ 11,743) pre- and post-crisis (2007e09). In this
Subramanian et al., 2002; Suzuki et al., 2010; Waverijn et al., 2014). fashion, there was opportunity to investigate: i) if the advent of the
To date, there is a large body of empirical evidence suggesting crisis coincided with depletion in social capital, and ii) if the
that the presence of social capital is positively associated with a presence of individual-level social capital protected against worse
variety of individual general health (Islam et al., 2006; Murayama psychological wellbeing during the 2008 financial crisis. The aim of
et al., 2012) and mental health outcomes (Bassett and Moore, this study, therefore, was to investigate if stocks of social capital in
2013; Cao et al., 2015; Fujiwara and Kawachi, 2008; Giordano and 2008 continued to buffer against worse psychological wellbeing
Lindstro€m, 2011; see also: Ehsan and De Silva, 2015; McPherson during times of potential financial turmoil. We hypothesise that
et al., 2014; Whitley and McKenzie, 2005 for reviews). From social capital stocks were depleted during the crisis and for those
these, the proxy ‘generalised trust’ provided the most consistent individuals who retained high levels of social capital (trust and
(positive) associations with health outcomes (Kim et al., 2008). participation), its presence would buffer against worsening psy-
However, there remains strong critique surrounding the research chological wellbeing. To our knowledge, this is the first longitudinal
field of social capital and health (Muntaner, 2004; Pearce and Davey research paper to investigate the buffering effects of social capital
Smith, 2003), which is further compounded by the lack of a uni- on psychological wellbeing in the UK, against the backdrop of this
versally accepted social capital definition, or consistent measure- historic event.
ments to quantify its effects on health (Macinko and Starfield,
2001). 2. Methods
Despite this critique, several mechanisms have been postulated
as to how social capital could influence health by: i) deterring so- 2.1. Data collection
cially ‘deviant’ behaviours (e.g. excess alcohol consumption,
smoking and crime); ii) increasing the dissemination of positive The British Household Panel Survey (BHPS) is a longitudinal
health messages and health behaviour ‘norms’; iii) increasing ac- survey of randomly selected private households, which has been
cess to resources, i.e. greater availability and use of prevention conducted annually by the UK's Economic and Social Research
services, and; iv) providing a buffer against psychosocial stress Centre since 1991. The first cohort sample was randomly selected
(Kawachi et al., 1999). Several of these mechanisms seem relevant by using a two-stage cluster design, and a total of 8166 private
to the outcome of this study (psychological wellbeing); for postal addresses around the UK were originally selected. In-
example, high crime levels, associated lower levels of generalised dividuals aged 16 years or older were invited to participate, with a
trust and increased psychosocial stresses are well known pre- total of 10,264 individual face-to-face interviews being completed
cursors to worse psychological wellbeing (Aneshensel and Sucoff, in the first BHPS wave (participation rate 95%). Since then, in-
1996). Chronic stress can lead to raised levels of blood cortisol via dividuals from this nationally representative sample of selected
the hypothalamic-pituitary-adrenal axis (Shively et al., 2009), households have been interviewed annually with a view to iden-
which in turn, have been linked to several deleterious health out- tifying social and economic changes within the British population.
comes, such as type II diabetes, cardio-vascular disease and poor Data derived from individual-level responses from Waves 17 (2007)
mental health (Hemingway and Marmot, 1999; Watson and and 18 (2008) were used for this study, with a total of 13,826 in-
Mackin, 2006). Furthermore, active social participation, consid- dividuals consenting to participate in 2007 and 13,710 in 2008. All
ered by Putnam as the cornerstone of social capital generation data were weighted after collection by the Research Centre to
 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro 73

adjust for non-response in the standard fashion (Elliot, 1991); 2.3.3. Socio-economic variables
further weighting was not recommended (Taylor et al., 2010, p.184). Annual household income was weighted according to size by
For each Wave, face-to-face interviews took place between summing the total income of all household members (net of
September and May, with the date, month and year for all re- taxation) and dividing this sum by the square root of the household
spondents being recorded (i.e. all interviews for Wave 1 were size (Burkhauser et al., 1996). The log of this income measure was
conducted between 1st Sept 1991e30th April 1992). In all eighteen kept as a continuous variable (per £1000 increase) for all analyses.
Waves of the BHPS, approximately half of all anticipated interviews Social class was determined by respondents' most recent occupa-
for that year were completed by the end of October. Full details of tion, derived from the Registrar General's Social Classification of
the selection process, weighting and participation rates can be occupations. The usual six categories (see appendix) were main-
found on-line (Taylor et al., 2010). tained for all analyses (reference (0) being Class I), with those yet to
The raw data for this panel study come from the BHPS individ- be employed labelled ‘never worked’. To compliment this, a further
ual-level responses in Wave 17 (Sept 2007eApril 2008) and Wave current employment status variable was categorized as ‘Employed’
18 (Sept 2008eApril 2009). Unique cross-wave identifiers meant (0), ‘Retired’ (1), ‘Fulltime student’ (2) or ‘Unemployed’ (3).
that individuals, who responded to all considered variables in this Highest achieved education was categorised as ‘Undergraduate
study, could be followed (N ¼ 11,743). Full interview participation or higher’ (0), ‘Year 13’ (1), ‘Year 11’ (2) or ‘No formal qualifications’
rates for year 2008 (as compared to year 2007) were 95.0%, with (3). Finally, perceived financial stress was assessed by asking re-
44.5% being from the original 1991 cohort sample. spondents about their current financial situation compared to the
The Research Centre fully adopted the Ethical Guidelines of the previous year. Possible responses were ‘About the same’ (0), ‘Now
Social Research Association; informed consent was obtained from better off’ (1) or ‘Now worse off’ (2). These three categories were
all participants and strict confidentiality protocols were adhered to maintained.
throughout data collection, processing procedures and subsequent
analyses for this study by the authors. 2.3.4. Confounders
Self-rated health (SRH) was assessed by the question:
‘Compared to people your own age, would you say that your health
2.2. Dependent variable has on the whole been: excellent, good, fair, poor or very poor?’ As
standard, this five-point scale was recoded into the dichotomous
The dichotomous outcome in this study was psychological variable ‘good’ (0) (excellent, good) and ‘poor’ (1) (fair, poor, very
wellbeing, obtained using the 12-item General Health Question- poor) health (Manor et al., 2000).
naire (GHQ-12) (Golderberg and Williams, 1988). Depending on the Age and gender (men (0), women (1)) were also considered
answers obtained, if three or more of the twelve items deemed to confounders in this study, with age being stratified into quintiles
reflect poor psychological wellbeing were selected, then re- for descriptive purposes (see Table 1) and employed as a contin-
spondents were categorised as having ‘worse’ psychological well- uous variable (per 10 year increase) in all analyses.
being (1); all others were labelled ‘better’ psychological wellbeing Values for all variables were obtained from the same individuals
(0) (Golderberg and Williams, 1988) - see appendix for more detail. (N ¼ 11,743) in years 2007 and 2008 (BHPS Waves 17 and 18). The
The GHQ-12 item has been shown to have an overall sensitivity of two data panels were subsequently merged to form a single
83.4% and a specificity of 76.3% when compared with the GHQ-28 multilevel, longitudinal dataset (occasions (time - level 1) clustered
item in a study of psychological disorders in general health care, on individuals (level 2)).
suggesting that the GHQ-12 tool is comparable to other more
complex assessment instruments (Goldberg et al., 1997). 2.3.5. Statistical analysis
Analysis 1 tested if changes in social capital between 2007 and
2.3. Independent variables 2008 (Table 1b) were statistically significant. To this effect, we
performed logistic regression pairwise tests using the multilevel,
2.3.1. Social capital proxies longitudinal dataset previously described, for full year-on-year
Generalised trust was assessed by asking people: ‘Would you comparison (N ¼ 11,743). Trust (can trust e (0); cannot trust e
say that most people can be trusted, or that you can't be too care- (1)) was the dichotomous outcome, with time (2007 e (0) and 2008
ful?’ Possible answers were ‘Most people can be trusted’, ‘You can't (1)) as the sole covariate. Odds ratios derived here described the
be too careful’ and ‘It depends’. This variable was dichotomised, same individuals' risk of not trusting in 2008 compared to 2007,
with only those respondents stating that most people could be alongside a 95% confidence interval (See Table 1c (i)).
trusted being labelled ‘Can trust’ (0); all negative responses In analysis 2, we further investigated associations between
(including ‘it depends’) were labelled ‘Can't trust’ (1) (Uslaner, changes in social capital over time as the financial crisis unfolded,
2002). with all data from Analysis 1 now being stratified by date and
Social participation was measured by asking respondents month (Table 1c (ii)). As dictated by data collection and events
questions about being active members of voluntary community described in the introduction, timeframe cut-offs in 2008e09 were:
groups or any sports, hobby or leisure group activity found locally. 1st e 14th Sept; 15th e 30th Sept; 1st e 31st Oct; 1st e 30th Nov;
Only those who answered positively to any of these were judged to and 1st Dec e 30th April. Identical tests were repeated with social
participate (0), with all others being labelled ‘No participation’ (1). participation as the outcome (participates (0); no participation (1))
Note that the question was posed slightly differently in 2008 against time (Table 1c).
compared to 2007 - see appendix for more detail. Analysis 3 investigated associations between worse psycholog-
ical wellbeing in 2008 and all considered covariates (including
time) in a multilevel, longitudinal logistic regression model (time
2.3.2. Marital status clustered on individuals). The model allowed a random intercept
Respondents were asked if they were ‘married, separated, for each individual and we obtained standard errors that were
divorced, widowed or never married’. Marital status was recoded adjusted for the temporal correlation of GHQ-12 scores within the
into ‘married’ (0) and ‘unmarried’ (1) (separated, divorced, wid- same individual across the timeframe of this study, whilst allowing
owed, never married) (Afifi et al., 2006). a between-Wave comparison. The outcome of interest was ‘worse
74 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro

Table 1a
Baseline (yr 2007) frequencies of all considered variables expressed as integers and percentages in italics (%) of NT (11,743) stratified by psychological wellbeing (GHQ-12).

Psychological wellbeing

Better Worse Total (NT)

a
Age (years) 16e34 2555 861 3416
21.8% 7.3% 29.1%
35e44 1723 594 2317
14.7% 5.1% 19.7%
45e54 1472 512 1984
12.5% 4.4% 16.9%
55e64 1376 408 1784
11.7% 3.5% 15.2%
65þ 1712 529 2241
14.6% 4.5% 19.1%

Total 8838 2904 11,742

75.3% 24.7% 100.0%
Gender Male 4219 1055 5274
35.9% 9.0% 44.6%
Female 4620 1849 6469
39.3% 15.7% 55.1%

Total 8839 2904 11,743

75.3% 24.7% 100.0%
Generalised trust Trusts others 3732 959 4691
31.8% 8.2% 39.9%
Cannot trust 5107 1945 7052
43.5% 16.6% 60.1%

Total 8839 2904 11,743

75.3% 24.7% 100.0%
Active participationb Participates 3936 1189 5125
33.5% 10.1% 43.7%
No participation 4901 1715 6616
41.7% 14.6% 56.3%

Total 8837 2904 11,741

75.3% 24.7% 100.0%
Self - rated healtha Good health 6877 1448 8325
58.6% 12.3% 70.9%
Poor health 1962 1455 3417
16.7% 12.4% 29.1%

Total 8839 2903 11,742

75.3% 24.7% 100.0%
Marital statusa Married 4956 1422 6378
42.2% 12.1% 54.3%
Not married 3882 1482 5364
33.1% 12.6% 45.7%

Total 8838 2904 11,742

75.3% 24.7% 100.0%
Social class: based on most recent (RGSC) occupation I 385 113 498
3.3% 1.0% 4.2%
II 2633 813 2446
22.4% 6.9% 29.3%
IIIa 1970 731 2701
16.8% 6.2% 23.0%
IIIb 1508 405 1913
12.8% 3.4% 16.3%
IV 1413 466 1879
12.0% 4.0% 16.0%
V 432 171 603
3.7% 1.5% 5.1%
Never worked 498 205 703
4.2% 1.7% 6.0%

Total 8839 2904 11,743

75.3% 24.7% 100.0%
c
Education achieved Undergraduate or higher 1976 625 2601
17.0% 5.4% 22.4%
Year 13 1774 601 2375
15.3% 5.2% 20.5%
Year 11 2766 850 3616
23.8% 7.3% 31.2%
No formal qualifications 2219 790 3009
19.1% 6.8% 25.9%

Total 8735 2866 11,601

75.3% 24.7% 100.0%
 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro 75

Table 1a (continued )

Psychological wellbeing

Better Worse Total (NT)

Employment status Employed 5412 1482 6894

46.1% 12.6% 58.7%
Full-time student 518 177 695
4.4% 1.5% 5.9%
Retired 1916 597 2513
16.3% 5.1% 21.4%
Unemployed 993 648 1641
8.5% 5.5% 14.0%

Total 8839 2904 11,743

75.3% 24.7% 100.0%
Current financial situation Still the same 4280 1165 5445
36.4% 9.9% 46.4%
situation compared with last year Worse off 2905 1196 4101
24.7% 10.2% 34.9%
Better off 1654 543 2197
14.1% 4.6% 18.7%

Total 8839 2904 11,743

75.3% 24.7% 100.0%
Household income, (annual) - size weighted < £6552 2179 748 2927
18.6% 6.4% 24.9%
£6553-£12,623 2113 811 2924
18.0% 6.9% 24.9%
£12,624-£21,343 2218 725 2943
18.9% 6.2% 25.1%
£21,344 þ 2329 620 2949
19.8% 5.3% 25.1%

Total 8839 2904 11,743

75.3% 24.7% 100.0%

Missing aN ¼ 1.
Missing bN ¼ 2.
Missing cN ¼ 142.
Source: BHPS Wave 17, 2007.

Table 1b being stratified by date and month as described in Analysis 2

Frequencies of social capital levels (as measured by generalised trust and social (Table 3).
participation) from the same respondents in years 2007 and 2008, expressed as
integers and percentages in italics (%) N ¼ 11,743.
All analyses were conducted using GLLAMM version 2.3.15
(Rabe-Hesketh et al., 2005), within the statistical software package
Social capital BHPS Waves STATA 11.2 (StataCorp, 2009).
Wave 17 (2007) Wave 18 (2008)

Generalised trust Trusts others 4691 3775 3. Results

40.0% 32.1%
Cannot trust 7052 7968
60.0% 67.9% Table 1a shows frequencies and total percentages (N ¼ 11,743) of
all considered explanatory variables, stratified by baseline psy-
Total 11,743 11,743
100.0% 100.0%
chological wellbeing (2007).
Social participation Active participation 5125 6149 Table 1b compares the levels of generalised trust and social
43.6% 52.4% participation from the same respondents in 2007 and 2008
No participation 6616 5594 (N ¼ 11,743), and showed a decrease in generalised trust from 40%
56.4% 47.6%
in 2007 to 32% in 2008. Conversely, there was an increase in
Total 11,741a 11,743 participation, from 43.6% in 2007 to 52.4% in 2008.
100.0% 100.0% Table 1c presents pairwise odds ratios (OR) with 95% confidence
Missing a ¼ 2 intervals (95%CI) for trust and participation over time. Interpreta-
Source: BHPS, Waves 17 & 18. tion of the year-on-year comparison (i) (OR ¼ 1.96, (95%CI)
1.81e2.11) implies that for each individual (N ¼ 11,743), the odds
for not trusting was 1.96 times higher in 2008 than in 2007, i.e.
psychological wellbeing’ in 2008 (1) whilst considering the same
reduced trust levels seen in 2008 compared to 2007 (Table 1b) are
individuals' GHQ-12 score in 2007; odd ratios derived from Anal-
statistically significant and not due to chance. Conversely, the odds
ysis 3 (the fixed effects from the mixed model) reflected an in-
for no participation in 2008 are 0.70 lower in 2008 than in 2007.
dividuals' risk of worse GHQ-12 in 2008 compared with 2007
These patterns appear consistent when the same data were
according to changes in considered covariates from 2007 to 2008,
stratified by crisis timeframe cut-offs (Table 1c (ii)), with the
having allowed for ‘within’ individual correlation of GHQ-12 scores
greatest risk of not trusting occurring during the two weeks
across the two Waves (Table 2).
immediately after the global collapse of Lehman Brothers
In Analysis 4, as with the pairwise testing, associations between
(OR ¼ 2.20, (95%CI) 1.90e2.56).
worse psychological wellbeing and all considered variables were
The likelihood of not trusting in 2008 (compared to 2007) ex-
further investigated as the financial crisis unfolded, with all data
tends across all timeframes from 1st September through to 30th
 76 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro

Table 1c
Pairwise odds ratios (ORs) with 95% confidence intervals (95% CI) describing the likelihood of the same individuals having reduced social capital (as measured by trust and participation) in
2008 compared to 2007, derived from multilevel, longitudinal analysis: i) full year-on-year comparison; ii) stratified by crisis timeframe cut-offs (N ¼ 11,743).

Social capital

i) Year-on-year comparison (NT ¼ 11,743) Trusts others 1.0

Cannot trust in 2008 1.96 (1.81e2.11)***
Participates 1.0
No participation 0.70 (0.66e0.74)***
ii) Timeframe (cut-offs)
September 1ste14th (NT ¼ 2320) Trusts others 1.0
Cannot trust in 2008 1.96 (1.66e2.31)***
Participates 1.0
No participation 0.80 (0.71e0.90)***
September 15the30th (NT ¼ 3434) Trusts others 1.0
Cannot trust in 2008 2.20 (1.90e2.56)***
Participates 1.0
No participation 0.67 (0.61e0.71)***
October 1ste31st (NT ¼ 3776) Trusts others 1.0
Cannot trust in 2008 1.82 (1.59e2.08)***
Participates 1.0
No participation 0.67 (0.61e0.74)***
November 1ste30th (NT ¼ 1580) Trusts others 1.0
Cannot trust in 2008 1.92 (1.56e2.36)***
Participates 1.0
No participation 0.65 (0.56e0.76)***
December 1st - April 30th (NT ¼ 633) Trusts others 1.0
Cannot trust in 2008 1.74 (1.25e2.41)***
Participates 1.0
No participation 0.77 (0.61e0.96)**

Reference group ¼ 1.0.

*p < 0.05.
**p < 0.01.
***p < 0.001.
Source: BHPS, Waves 17 & 18.

April 2009. Similar patterns for participation are also seen in date- participation) across the UK and quantify its buffering effects on
stratified data, with the odds for no participation being least in worse psychological wellbeing against the backdrop of the 2008
November 2008 (OR ¼ 0.65, (95%CI) 0.56e0.76). global financial crisis. Results revealed that there was a large, sig-
Table 2 presents the risk of having poor psychological wellbeing nificant drop in generalised trust levels (from 40% to 32%) across
in 2008 after adjusting for changes in all considered covariates from the UK in 2008 compared to 2007. Conversely, social participation
2007 to 2008. Individuals with low levels of trust in 2008 had an seemed to increase in 2008. However, care must be taken when
increased risk of worse psychological wellbeing in 2008 compared drawing inference from this social capital proxy, as the question
to 2007 (OR ¼ 1.49, (95%CI) 1.34e1.65), even after considering in- posed to determine participation rates differed between the two
dividual perceptions of financial strain. Women (OR ¼ 1.69, (95%CI) Waves (see appendix). Regarding any buffering effects of social
1.51e1.89), those not married (OR ¼ 1.41, (95%CI) 1.34e1.65), those capital, a lack of trust coincided with a greater overall negative
with poor SRH (OR ¼ 4.88, (95%CI) 4.38e5.43), the unemployed effect on psychological wellbeing during and immediately after the
(OR ¼ 1.98, (95%CI) 1.71e2.30) and those whose financial situation 2008 crisis. After stratification by month (see Table 3(ii)), the
was perceived as ‘now worse off’ (OR ¼ 2.23, (95%CI) 2.01e2.48) all negative effect of the crisis on psychological wellbeing extended up
had an increased risk of worse psychological wellbeing in 2008. to December 2008 for those individuals who could not trust, with
Covariates that protected against worse psychological wellbeing the greatest risk for worse GHQ-12 scores occurring immediately
in 2008 compared with 2007 included being ‘better off’ financially after the global collapse of Lehman Brothers (OR ¼ 1.73, (95%CI)
and those with lower levels of education (Year 11 or no qualifica- (1.42e2.12). No significant association between participation and
tions). Every ten-year increase in age also protected against worse GHQ-12 scores was observed.
psychological wellbeing. There was no significant association be- This study further adds to the existing body of longitudinal so-
tween psychological wellbeing and active participation. cial capital and mental health research by allowing pre- and post-
Table 3(ii) presents associations between poor psychological crisis comparisons of the same individuals across the UK. By treat-
wellbeing in 2008 and social capital only, still having adjusted for ing the crisis as a natural experiment of sorts, we were able to
changes in all considered covariates (2007e08) but now stratified investigate the buffering role of social capital across consecutive,
by crisis timeframe cut-offs. The association between worse GHQ- yet starkly contrasting years. Our results mirror other longitudinal
12 and lack of participation consistently lacks statistical signifi- research to demonstrate the protective effects of social capital
cance; conversely, those who cannot trust maintained their against poor mental health outcomes (Bertotti et al., 2013; Ding
increased risk for worse psychological wellbeing from 1st et al., 2015; Frank et al., 2014; Giordano and Lindstro €m, 2011;
September until 30th November 2008. Hall et al., 2014; Han, 2015; Murayama et al., 2013; Tsuboya et al.,
2015; Verduin et al., 2014). However, our results are in contrast to
those of Economou et al. (2014), who concluded that the buffer
4. Discussion provided by social capital on psychological wellbeing was fully
attenuated by the effects of financial strain. In our study, lack of
The aim of this longitudinal UK panel study was to investigate trust remained strongly associated with worse psychological
changes in social capital levels (generalised trust and social
 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro 77

Table 2
Odd ratios (ORs) with 95% confidence intervals (95% CI) of the same individuals having worse psychological health in 2008 compared to 2007 according to multilevel, lon-
gitudinal analysis of all considered explanatory variables and potential confounders (Nt ¼ 11, 743).

Explanatory variables Worse psychological wellbeing in 2008 (N ¼ 11,743)

ORs (95% CI)

Time 0.91 (0.83e0.99)*

Age Per 10 year increase 0.91 (0.86e0.96)***
Gender Male 1.0
Female 1.69 (1.51e1.89)***
(Log) Household income - size weighted Continuous 1.04 (0.97e1.13)
Social class: derived from most recent occupation-based RGSC schema I 1.0
II 1.01 (0.77e1.33)
IIIa 0.97 (0.73e1.30)
IIIb 0.78 (0.59e1.06)
IV 0.79 (0.58e1.07)
V 0.83 (0.57e1.19)
Never worked 0.83 (0.61e1.12)
Generalised trust Trusts others 1.0
Cannot trust 1.49 (1.34e1.65)***
Social participation Active participation 1.0
No participation 1.03 (0.94e1.13)
Marital status Married 1.0
Not married 1.41 (1.26e1.58)***
Self-rated health Good health 1.0
Poor health 4.88 (4.38e5.43)***
Education Undergraduate þ 1.0
Year 13 0.86 (0.73e1.02)
Year 11 0.73 (0.63e0.86)***
No Qualifications 0.77 (0.65e0.93)**
Employment status Employed 1.0
Unemployed 1.98 (1.71e2.30)***
Retired 1.13 (0.93e1.38)
FT student 1.09 (0.85e1.38)
Financial situation Still the same 1.0
Now better off 0.85 (0.75e0.96)*
Now worse off 2.23 (2.01e2.48)***

Reference group ¼ 1.0.

*p < 0.05.
**p < 0.01.
***p < 0.001.
Source: BHPS, Waves 17 & 18.

wellbeing, even after considering individuals' perceptions of revealed in these (and other) panel data add strength to the theory
financial strain (Tables 2 and 3). However, it is important to note that generalised trust is indeed a summary measure of individual
that this study investigated associations during and immediately experiences. From all BHPS data Waves when trust was measured
after the onset of the 2008 crisis in the UK; the study by Economou (years 1998, 2000, 2003 and 2005), analyses showed that trust
et al. (2014) was conducted in Greece after several years of harsh levels were relatively stable across the UK, between 38 and 44%.
Governmental austerity. One notable exception, however, was the year 2005. Trust levels
There were no effects on psychological wellbeing from the social decreased from 44% in 2003 to 33% in 2005, the year when the city
capital proxy ‘active social participation’ in this study. However, as of London was subjected to the July 7th terror attacks (British
the question employed to determine participation differed some- Broadcasting Corporation, 2005). The 25% drop in trust (in real
what between 2007 and 2008 (see appendix), we remained terms) from BHPS data seen in 2005 adds weight to the economic
cautious in drawing inferences. viewpoint of fluctuating generalised trust levels, and further sup-
Year-on-year comparisons of social capital levels also provided ports the plausibility that trust could also drop because of events
insight into the possible impact of the financial crisis on UK stocks surrounding the 2008 financial crisis. Of interest, other survey data
of social capital (Tables 1b & 1c). Pre- and post-crisis analyses (the World Values Survey) also reported similarly low levels of UK
revealed a substantial and statistically significant decrease in trust in 2005 (30% can trust, N ¼ 1041); unfortunately, no 2008 data
generalised trust across the UK, from 40% in 2007 to 32% in 2008 were available for the UK from this source for comparison (World
(Tables 1b & 1c). As data for each Wave were collected from Values Survey, 2005).
September onwards (and considering the timeframe of the 2008 Associations between worse psychological wellbeing in 2008
crisis), such empirical evidence provides strong support for the (compared with 2007) and changes in all other considered variables
theory offered by Pildes (1996), that this global event and/or the from 2007 to 2008 were presented in Table 2. Notably, unemployed
subsequent Governmental response could be responsible for the individuals had an increased risk of worse psychological wellbeing
drop in levels of generalised trust. Conversely, there seemed to be a in 2008 (OR ¼ 1.98, (95%CI) 1.71e2.30), which may reflect feelings
significant increase in participation levels from 2007 to 2008; of uncertainty for the future. Poor self-rated health was also asso-
however, as stated previously, the question employed to determine ciated with worse psychological wellbeing (OR ¼ 4.88, (95%CI)
participation differed between 2007 and 2008, so we are unsure if 4.38e5.43). This was expected (Tessler and Mechanic, 1978) and
the reported increase is an artefact. lends further weight to our other results.
As touched upon in the Introduction, fluctuations in trust Unexpectedly, social class (as determined by most recent
78 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro

Table 3
Odd ratios (ORs) with 95% confidence intervals (95% CI) of the same individuals having worse psychological health in 2008 compared to 2007 according to
multilevel, longitudinal analysis of all considered explanatory variables and potential confounders a (Nt ¼ 11,743) i) Full year, and ii) Crisis timeframe cut-
offs. Note: only social capital results shown.

Date/month Social capital Worse psychological health in 2008

ORs (95% CI)

a
i) Full year (2008) (NT ¼ 11,743) Trusts others 1.0
Cannot trust 1.49 (1.34e1.65)***
Active participation 1.0
No participation 1.03 (0.94e1.13)
ii) Timeframe a (cut-offs)
September 1ste14th (N ¼ 2406) Trusts others 1.0
Cannot trust 1.40 (1.09e1.79)**
Active participation 1.0
No participation 1.03 (0.83e1.28)
September 15the30th (N ¼ 3384) Trusts others 1.0
Cannot trust 1.73 (1.42e2.12)***
Active participation 1.0
No participation 0.96 (0.81e1.14)
October 1ste31st (N ¼ 3780) Trusts others 1.0
Cannot trust 1.46 (1.22e1.76)***
Active participation 1.0
No participation 0.96 (0.82e1.13)
November 1ste30th (N ¼ 1549) Trusts others 1.0
Cannot trust 1.36 (1.06e1.75)**
Active participation 1.0
No participation 1.17 (0.93e1.48)
December 1st - April 30th (N ¼ 624) Trusts others 1.0
Cannot trust 1.19 (0.78e1.81)
Active participation 1.0
No participation 1.38 (0.94e2.03)

*p < 0.05.
**p < 0.01.
***p < 0.001.
Reference group ¼ 1.0.
a
Model adjusted for Gender, Age, Marital status, Social class, Employment status, Highest achieved education, Self-rated health, Financial situation and
Household income.
Source: BHPS Waves 17 and 18.

employment) showed no significant association with worse psy- comparing the same individuals pre- and post-crisis, with a high
chological wellbeing (Marmot et al., 1997). This could be a product number of individual respondents (N ¼ 11,743). That these panel
of the inclusion of other well-known mental health variables, such data span the advent of the 2008 financial crisis allowed us to
as gender and employment status. As Wave 18 (year 2008e9) was consider this event as a natural experiment of sorts and to draw
the last Wave of the BHPS, it was not possible to investigate if inference from any changes in associations over the timeframe of
patterns of association with social class changed over subsequent this study. This is, however, not a typical example of a natural
years, post-crisis. Future research may consider investigating the experiment, as there is no similar ‘control group’ to compare
longer-term effects of austerity on psychological wellbeing within against in our study population. To our knowledge, this is the first
the different social class groups in the UK. Furthermore, the suc- longitudinal research paper to investigate the buffering effects of
cessor to the BHPS, ‘Understanding society’ which rolled out in social capital on psychological wellbeing against the backdrop of
2010, no longer asked respondents about generalised trust atti- this historic event. The data were obtained via interview rather
tudes. It is, therefore, not possible to investigate the longer-term than relying on postal questionnaires, which contributed to the
buffering effects of trust on psychological wellbeing in the UK very high participation rate of around 95%, year on year (Taylor
from these data sources. et al., 2010). Although there are more complex GHQ instruments
Interestingly, low/no qualifications seemed to protect against (28- or 60-item) to measure psychological wellbeing, there seems
worse psychological wellbeing in 2008 (Table 2). As the timeframe little difference in validity between these and the GHQ-12 item
of this study was just two years, this result could represent those used in this study (Goldberg et al., 1997). Unfortunately, there is no
who remained within the school system during the timeframe of ‘gold standard’ with which to validate generalised trust; however, it
this study. However, cross-tabulations revealed that this assump- has been considered an acceptable proxy of social capital for two
tion is unlikely, as the 16e24 year old age groups represented just decades (Islam et al., 2006; Putnam, 2001). To reduce the risk of
12.2% of those with Year 11 or less qualifications, compared with potential confounding, we further included numerous well-known
those aged 25 years or older in 2008 (87.8% - results not published). mental health determinants in our full model analyses.
That lower levels of education protected individuals against worse One major limitation is that the BHPS sample was originally
psychological wellbeing may rather reflect perceptions regarding selected to reflect the UK population as a whole, so deliberately
the short-term ramifications that the financial crisis may have on avoided oversampling of smaller-sized communities. Values ob-
this specific subgroup (Malhotra and Margalit, 2010). tained for GHQ-12, SRH and perceived financial strain were relative,
i.e. responses given were dependent on respondents' usual levels
5. Strengths and weaknesses (of GHQ-12, SRH, financial situation); as such, some self-report bias
may have been introduced, though the validity of our outcome
A major strength of this panel study is that it is longitudinal, measure (GHQ-12) is high (see Methods). By year 2007, only 44.5%
 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro 79

of the original (1991) cohort members were able to answer the References
questions posed (Taylor et al., 2010). This would have introduced
further selection bias into this study. As the BHPS was in its final Afifi, T., Cox, B., Enns, M., 2006. Mental health profiles among married, never-
married, and separated/divorced mothers in a nationally representative sam-
year in 2008, we were only able to investigate the shorter-term ple. Soc. Psychiatry Psychiatric Epidemiol. 41, 122e129.
effects of the financial crisis on social capital and psychological Aneshensel, C., Sucoff, C., 1996. The neighborhood context of adolescent mental
wellbeing. Another limitation is that there were differences in the health. J. Health Soc. Behav. 37, 293e310.
Bassett, E., Moore, S., 2013. Social capital and depressive symptoms: the association
question regarding active participation in years 2007 and 2008 (see of psychosocial and network dimensions of social capital with depressive
appendix). Care, therefore, must be taken when interpreting the symptoms in Montreal, Canada. Soc. Sci. & Med. 86, 96e102.
effects of this social capital proxy. Marital status was reduced to the Beale, L.M., 2009. In the Wake of Financial Crisis. International Monetary Fund. Last
accessed on 1st Feb 2016 at. http://www.imf.org/external/np/seminars/eng/
dichotomous ‘married’ and ‘not married’ (separated, divorced, 2009/finstability/pdf/lbeale.pdf.
widowed, never married). Though this method of reduction has Berkman, L., Kawachi, I. (Eds.), 2000. Social Epidemiology. Oxford University Press,
been previously validated (Afifi et al., 2006), it may hide more New York.
Bertotti, M., Watts, P., Netuveli, G., Yu, G., Schmidt, E., Tobi, P., et al., 2013. Types of
complex pathways regarding cohabitation common in society
social capital and mental disorder in deprived urban areas: a multilevel study of
today. 40 disadvantaged London neighbourhoods. PLoS One 8.
Bourdieu, P. (Ed.), 1986. The Forms of Capital. Greenwood Press, New York.
British Broadcasting Corporation, 2005. 7 July Bombings. BBC, London. Last
6. Conclusion accessed on 1st Feb 2016 at. http://news.bbc.co.uk/2/shared/spl/hi/uk/05/
london_blasts/what_happened/html/.
British Broadcasting Corporation, 2007. The Downturn in Facts and Figures. Last
The longitudinal, multilevel design of this study allowed for the accessed on 1st Feb 2016 at. http://news.bbc.co.uk/2/hi/business/7073131.stm.
comparison of the same individuals over time, whilst treating the British Broadcasting Corporation, 2008. Lehman Bros Files for Bankruptcy. Last
accessed on 1st Feb 2016 at. http://news.bbc.co.uk/2/hi/business/7615931.stm.
2008 financial crisis as a natural experiment of sorts. From this, we Burkhauser, R., Smeeding, T., Merz, J., 1996. Relative inequality and poverty in
can report two major findings from this study: i) the crisis itself Germany and the United States using alternative equivalence scales. Rev. In-
coincided with a significant drop in generalised trust levels across come & Wealth 42, 381e401.
Cao, W., Li, L., Zhou, X., Zhou, C., 2015. Social capital and depression: evidence from
the UK, from 40% in 2007 to 32% in 2008 and, ii) despite this
urban elderly in China. Aging & Ment. Health 19, 418e429.
reduction, the buffering effects of trust remained fully evident, with Coleman, J., 1988. Social capital in the creation of human-capital. Am. J. Sociol. 94,
those not trusting being at risk of worse psychological wellbeing in S95eS120.
Ding, N., Berry, H.L., O'Brien, L.V., 2015. One-year reciprocal relationship between
2008 compared to the previous year (OR ¼ 1.49, (95%CI) 1.34e1.65).
community participation and mental wellbeing in Australia: a panel analysis.
From this, we can conclude that social capital (generalised trust) Soc. Sci. & Med. 128, 246e254.
seems a relatively changeable resource, whose levels are suscepti- Economou, M., Madianos, M., Peppou, L.E., Souliotis, K., Patelakis, A., Stefanis, C.,
ble to individuals' current perceptions and experiences. Despite its 2014. Cognitive social capital and mental illness during economic crisis: a
nationwide population-based study in Greece. Soc. Sci. & Med. 100, 141e147.
apparent volatility, however, those individuals who managed to Ehsan, A.M., De Silva, M.J., 2015. Social capital and common mental disorder: a
retain trust during times of uncertainty were protected against systematic review. J. Epidemiol. Community Health 69, 1021e1028.
worsening psychological wellbeing during the 2008 financial crisis. Elliot, D., 1991. Weighting for Non-response: A Survey Researchers Guide. Office of
Population Census and Surveys:Social Survey Division, London, UK.
Decision makers should be made aware of how events such as the Frank, C., Davis, C.G., Elgar, F.J., 2014. Financial strain, social capital, and perceived
crisis (and the measures taken to counter its effects) could nega- health during economic recession: a longitudinal survey in rural Canada.
tively impact on a Nation's levels of trust. The positive (short-term) Anxiety Stress Coping 27, 422e438.
Fujiwara, T., Kawachi, I., 2008. A prospective study of individual-level social capital
effects of trust on psychological wellbeing evident in this study and major depression in the United States. J. Epidemiol. Community Health 62,
implies that decision makers should also make it a priority to 627e633.
ensure that social capital stocks can be restored to pre-crisis levels; Giordano, G.N., Bjo €rk, J., Lindstro
€m, M., 2012. Social capital and self-rated health - a
study of temporal (causal) relationships. Soc. Sci. & Med. 75, 340e348.
as Economou et al. (2014) imply, reserves of social capital and its €m, M., 2011. Social capital and change in psychological
Giordano, G.N., Lindstro
positive effects on psychological wellbeing may become fully health over time e a panel study. Soc. Sci. & Med. 72, 1219e1227.
attenuated if allowed to dwindle over years of austerity and Giordano, G.N., Ohlsson, H., Lindstro € m, M., 2011. Social capital and health - purely a
question of context? Health & Place 17, 946e953.
financial strain.
Glaeser, E., Laibson, D., Scheinkman, J., Soutter, C., 2000. Measuring trust. Q. J. Econ.
65, 811e846.
Glanville, J.L., Andersson, M.A., Paxton, P., 2013. Do social connections create trust?
Acknowledgements An examination using new longitudinal data. Soc. Forces 92, 545e562.
Goldberg, D., Gater, R., Sartorius, N., Ustun, T., Piccinelli, M., Gureje, O., et al., 1997.
The validity of two versions of the GHQ in the WHO study of mental illness in
The data used in this study were made available through a third general health care. Psychol. Med. 27, 191e197.
party, the UK Data Archive. These data were originally collected by Golderberg, D., Williams, P., 1988. A User's Guide to the General Health Question-
the ESRC Research Centre on Micro-social Change at the University naire. NFER-Nelson, Windsor, UK.
Greenspan, A., 2008. Testimony of Alan Greenspan. The Financial Crisis and the Role
of Essex (now incorporated within the Institute for Social and
of Federal Regulators. Hearing before the House Committee on Oversight and
Economic Research). Neither the original collectors of the data nor Government Reform, 110th Cong. Washington, U.S. Last accessed 1st Feb 2016
the Archive bear any responsibility for the analyses or in- at. http://vorige.nrc.nl/multimedia/archive/00183/greenspan_statement_
183123a.pdf.
terpretations presented here. This study was supported by Swedish
Hall, B.J., Tol, W.A., Jordans, M.J.D., Bass, J., de Jong, J.T.V.M., 2014. Understanding
Research Council Linnaeus Centre for Economic Demography (VR resilience in armed conflict: social resources and mental health of children in
79), the Swedish Research Council (K2014-69X-22427-01-4), Burundi. Soc. Sci. & Med. 114, 121e128.
Swedish ALF Government Grant (Dnr M M 2014/354), Riksbankens Han, S., 2015. Longitudinal association between social capital and self-esteem: a
matter of context. Psychiatry Res. 226, 340e346.
Jubileumsfonds: The Swedish Foundation for Humanities and So- Hemingway, H., Marmot, M., 1999. Evidence based cardiology - psychosocial factors
cial Sciences (NHS14-2035:1), and the Research Funds of the Uni- in the aetiology and prognosis of coronary heart disease: systematic review of
versity Hospital in southern Sweden. prospective cohort studies. Br. Med. J. 318, 1460e1467.
Islam, M., Merlo, J., Kawachi, I., Lindstro€m, M., Gerdtham, U., 2006. Social capital and
health: does egalitarianism matter? A literature review. Int. J. Equity Health 5.
Kawachi, I., Berkman, L.F., 2001. Social ties and mental health. J. Urban Health 78,
Appendix A. Supplementary data 458e467.
Kawachi, I., Kennedy, B.P., Glass, R., 1999. Social capital and self-rated health: a
contextual analysis. Am. J. Public Health 89, 1187e1193.
Supplementary data related to this article can be found at http://
Kim, D., Subramanian, S., Kawachi, I., 2008. Social capital and physical health: a
dx.doi.org/10.1016/j.socscimed.2016.02.008.
80 €m, G.N. Giordano / Social Science & Medicine 153 (2016) 71e80
M. Lindstro

systematic review of the literature. In: Kawachi, I., Subramanian, S., Kim, D. Policy Res. 2, 41e51.
(Eds.), Social Capital and Health. Springer, New York, pp. 139e190. Putnam, R.D., 2004. Commentary: health by association: some comments. Int. J.
Kingsley, P., 2012. Financial Crisis: Timeline. The Guardian, London. Last accessed on Epidemiol. 33, 667e671.
1st Feb 2016 at. http://www.theguardian.com/business/2012/aug/07/credit- Rabe-Hesketh, S., Skrondal, A., Pickles, A., 2005. Maximum likelihood estimation of
crunch-boom-bust-timeline. limited and discrete dependent variable models with nested random effects.
Macinko, J., Starfield, B., 2001. The utility of social capital in research on health J. Econ. 128, 301e323.
determinants. Milbank Q. 79, 387e427. Shively, C., Musselman, D., Willard, S., 2009. Stress, depression, and coronary artery
Malhotra, N., Margalit, Y., 2010. Short-term communication effects or longstanding disease: modeling comorbidity in female primates. Neurosci. Biobehav. Rev. 33,
dispositions? The public's response to the financial crisis of 2008. J. Polit. 72, 133e144.
852e867. StataCorp, 2009. Stata Statistical Software. StataCorp LP, College Station, TX.
Manor, O., Matthews, S., Power, C., 2000. Dichotomous or categorical response? Subramanian, S.V., Kim, D.J., Kawachi, I., 2002. Social trust and self-rated health in
Analysing self-rated health and lifetime social class. Int. J. Epidemiol. 29, US communities: a multilevel analysis. J. Urban Health-Bulletin N. Y. Acad. Med.
149e157. 79, S21eS34.
Marmot, M., Ryff, C.D., Bumpass, L.L., Shipley, M., Marks, N.F., 1997. Social in- Suzuki, E., Takao, S., Subramanian, S., Komatsu, H., Doi, H., Kawachi, I., 2010. Does
equalities in health: next questions and converging evidence. Soc. Sci. & Med. low workplace social capital have detrimental effect on workers' health? Soc.
44, 901e910. Sci. & Med. 70, 1367e1372.
Mathiason, N., 2008. Three Weeks that Changed the World the Guardian. Last Taylor, M., Brice, J., Buck, N., Prentice-Lane, E., 2010. British Household Panel Survey
accessed on 1st Feb 2016 at. http://www.theguardian.com/business/2008/dec/ User Manual Volume A: Introduction, Technical Report and Appendices. Uni-
28/markets-credit-crunch-banking-2008. versity of Essex, Colchester, England. Last accessed 1st Feb 2016 at. https://
McPherson, K.E., Kerr, S., McGee, E., Morgan, A., Cheater, F.M., McLean, J., et al., 2014. www.iser.essex.ac.uk/bhps/documentation/pdf_versions/index.html.
The association between social capital and mental health and behavioural Tessler, R., Mechanic, D., 1978. Psychological distress and perceived health status.
problems in children and adolescents: an integrative systematic review. BMC J. Health Soc. Behav. 19, 254e262.
Psychol. 2, 7e7. Tsuboya, T., Tsutsumi, A., Kawachi, I., 2015. Change in psychological distress
Muntaner, C., 2004. Commentary: social capital, social class, and the slow progress following change in workplace social capital: results from the panel surveys of
of psychosocial epidemiology. Int. J. Epidemiol. 33, 674e680. the J-HOPE study. Occup. Environ. Med. 72, 188e194.
Murayama, H., Fujiwara, Y., Kawachi, I., 2012. Social capital and health: a review of Uslaner, E., 2002. The Moral Foundations of Trust. Cambridge University Press, New
prospective multilevel studies. J. Epidemiol. 22, 179e187. York, New York.
Murayama, H., Nishi, M., Matsuo, E., Nofuji, Y., Shimizu, Y., Taniguchi, Y., et al., 2013. van Ingen, E., Bekkers, R., 2015. Generalized trust through civic engagement? Evi-
Do bonding and bridging social capital affect self-rated health, depressive mood dence from five national panel studies. Polit. Psychol. 36, 277e294.
and cognitive decline in older Japanese? A prospective cohort study. Soc. Sci. & Verduin, F., Smid, G.E., Wind, T.R., Scholte, W.F., 2014. In search of links between
Med. 98, 247e252. social capital, mental health and sociotherapy: a longitudinal study in Rwanda.
Pearce, N., Davey Smith, G., 2003. Is social capital the key to inequalities in health? Soc. Sci. & Med. 121, 1e9.
Am. J. Public Health 93, 122e130. Watson, S., Mackin, P., 2006. HPA axis function in mood disorders. Psychiatry 5,
Pildes, R.H., 1996. The destruction of social capital through law. Univ. Pa. Law Rev. 166e170.
144, 2055e2077. Waverijn, G., Wolfe, M.K., Mohnen, S., Rijken, M., Spreeuwenberg, P.,
Poortinga, W., 2006. Social capital: an individual or collective resource for health? Groenewegen, P., 2014. A prospective analysis of the effect of neighbourhood
Soc. Sci. & Med. 62, 292e302. and individual social capital on changes in self-rated health of people with
Portes, A., 1998. Social capital: its origins and applications in modern sociology. chronic illness. BMC Public Health 14.
Annu. Rev. Sociol. 24, 1e24. Whitley, R., McKenzie, K., 2005. Social capital and psychiatry: review of the liter-
Putnam, R.D., 2000. Bowling Alone: The Collapse and Revival of American Com- ature. Harv. Rev. Psychiatry 13, 71e84.
munity. Simon & Schuster, New York. World Values Survey, 2005. Last accessed 1st Feb 2016 at. http://www.
Putnam, R.D., 2001. Social capital: measurement and consequences. Isuma Can. J. worldvaluessurvey.org/WVSOnline.jsp.