Article Text

Download PDFPDF

What are the mental health consequences of austerity measures in public housing? A quasi-experimental study
  1. Chungah Kim1,
  2. Celine Teo2,3,
  3. Andrew Nielsen2,3,
  4. Antony Chum1,3,4
  1. 1 School of Kinesiology and Health Science, York University, Toronto, Ontario, Canada
  2. 2 Department of Applied Health Sciences, Brock University, St Catharines, Ontario, Canada
  3. 3 MAP Centre for Urban Health Solutions, Unity Health Toronto, Toronto, Ontario, Canada
  4. 4 Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada
  1. Correspondence to Dr Antony Chum, School of Kinesiology and Health Science, York University, Toronto, Ontario, Canada; antony.chum{at}


Background As governments around the world implement austerity measures to reduce national deficits, there is an urgent need to investigate potential health impacts of specific measures to avoid unintended consequences. In 2013, the UK government implemented the underoccupancy penalty (ie, the bedroom tax) to reduce the national housing benefits bill, by cutting social housing subsidies for households deemed to have excess rooms. We investigated the impact of the bedroom tax on self-reported psychological distress.

Methods Using data from the UK Household Longitudinal Study (2010–2014), the sample included those who received housing subsidies, aged 16–60, living in England. Control and treatment groupings were identified on their household composition and housing situation. We used matching methods to create an exchangeable set of observations. Difference-in-differences analysis was performed to examine changes across the prereform and postreform psychological distress of the treatment and control groups, using the 12-item General Health Questionnaire.

Results The implementation of the reform was associated with a moderate increase in psychological distress (0.88, 95% CI 0.06 to 1.71) among the treatment group, relative to the control group. However, the announcement was not associated with change in psychological distress (0.53, 95% CI 0.21 to 1.27).

Conclusion Our study provides evidence that the implementation of housing austerity measures can increase psychological distress among social housing tenants. As the use of austerity measures become more widespread, policy-makers should consider supplementary interventions to ameliorate potential negative health consequences.

  • policy
  • mental health
  • longitudinal studies
  • housing
  • public health

Data availability statement

Data are available in a public, open access repository. Data are available in a public, open access repository. This data can be accessed through the UK Data Service:!/abstract.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Reductions in government spending to address budget deficits, known as austerity measures, have been found to be associated with negative health outcomes in the general population. Austerity measures, such as the UK bedroom tax, may have a deleterious impact on the health of those with financial and health-related disadvantages. However, a few studies have addressed the effect of the bedroom tax on mental health.


  • Our study findings show that the bedroom tax was associated with an increase in psychological distress among social housing renters. As these measures tend to target already disadvantaged individuals, future studies should continue to investigate specific austerity measures to understand their impact on public health.


  • Evidence from our study suggests that governments should consider the psychological distress that may result from the implementation of austerity policies in public housing. Given that austerity policies continue to be applied to contract the welfare state, future research should investigate the health consequences of these measures.


The UK Welfare Reform Act 20121 introduced an austerity measure on social housing, officially known as the ‘underoccupancy penalty’ (ie, the bedroom tax),2 aimed at reducing the national deficit.3 The policy affected renters living in social housing whereby social benefits were reduced in the event that tenants were deemed to have too much living space on their rental property, relative to the number of occupants (ie, 14% reduction for 1 spare bedroom and 25% reduction for 2+ bedrooms).

The financial repercussions of the policy were detrimental for social housing renters as they are among the poorest in the UK, and are vulnerable to changes in welfare policies.4 5 In 2011–2012, the annual median income for social housing renters was estimated to be £8996.6 Approximately 660 000 social housing renters were affected by the policy, and had their weekly incomes reduced by £12–£22.7 In addition to the financial impact, the measure was criticised7 for human rights violations. For instance, under the policy, couples were expected to share a single bedroom; however, couples with a disability may not be able to share a room due to accommodations for hospital-style beds, or may need a spare room for medical supplies.8 In other cases, victims of assault or harassment, who had access to a panic room with reinforced windows and alarms linked to police dispatch, were also subjected to the underoccupancy penalty.9 A government-led impact assessment of the policy highlighted various negative consequences,10 including reports that (1) an estimated 35% of claimants would fall into arrears due to the bedroom tax and (2) claimants with disabilities would face difficulties where significant accommodations have been made.

More recently, due to widespread unemployment and financial hardship caused by the COVID-19 pandemic, there is an influx of social housing renters who began claiming housing subsidies under the Universal Credit system.11 Given that approximately 50 000 additional households (between 2020 and 2021) are affected by the bedroom tax,11 our study is more important than ever to investigate its potential impact on psychological distress among those facing social benefit cuts.

Prior research has found that austerity policies, measured by reductions in aggregate government spending on public services, have an association with negative health outcomes.12–16 For instance, the implementation of stringent austerity in Greece, Spain and Portugal was associated with high epidemics of suicide and outbreaks of infectious diseases; while Iceland, which rejected austerity, had few discernible health effects.12 Likewise, a study found that increased provincial expenditures on health, social services, and education were associated with lower mortality rates across Canadian provinces.13 However, these prior studies only examined austerity as a reduction in aggregate government spending (rather than investigating the impact of a specific programme), which may not necessarily help to inform policy change. In other words, examining a specific programme using natural experiments may provide policy-makers with more concrete and practical information.

There have been studies of the Welfare Reform Act that used quasi-experimental methods to quantify the impact of the bedroom tax.2 7 17 A recent examination of labour market activity following the housing subsidy cuts found that the austerity measures were not able to increase employment among recipients of subsidies.17 Another study showed that although the bedroom tax incentivised downsizing, it penalised those who were unable to move and strained already financially unstable households.2 This study found that the implementation of the bedroom tax was followed by reductions in health satisfaction. However, the outcome measure has not been validated, and the clinical and policy relevance of health satisfaction may be easily challenged.

Although these studies found a fundamental shortcoming of the reform in addressing the rising social housing expenditures and undersupply of suitable housing, there is a lack of quantitative studies on the impact of the bedroom tax on psychological distress. In a qualitative study involving English social housing renters, participants reported extreme anxiety, stress and fear due to the unintended financial impacts of the bedroom tax.7 Due to the study design, these findings are not generalisable to the country level, and nationally representative evidence is needed to inform future welfare policy. This study will address the evidentiary gap in the implications of psychological distress, by using nationally representative longitudinal data and exploiting the exogenous variations to social housing subsidies through the policy shift. Our study will answer the following question: What is the impact of the bedroom tax on the psychological distress in social housing renters?


Study design

Our study used participants from the Understanding Society: Longitudinal Household Survey, a nationally representative panel survey that follows approximately 40 000 households from the UK.18 To examine the impact of the policy announcement and implementation in April 2012 and April 2013 respectively, we used data from 2010 to 2014 for the main analyses (and data up to 2017 used for sensitivity tests). The announcement and implementation effects of the policy are examined separately with different prepolicy and postpolicy periods, as their impact on psychological distress may have different mechanisms. For instance, the announcement of the policy may trigger stress associated with the anticipated loss of income and uncertainty about one’s future; whereas, its implementation would lead to reduced standards of living.

Study participants

We applied the following exclusion criteria: (1) those not living in social housing (not affected by policy), (2) those without valid responses (of the dependent variable) in the prepolicy and postpolicy period, (3) those under age 16 and retirees (based on state pension age) as the policy would only apply to working adults, (4) those without household information to construct eligibility criteria, (5) those lost to follow-up in the survey and (6) those living in Wales, Northern Ireland and Scotland. Only residents of England were included in the study to ensure that policy implementation and administration was consistent across participants. The devolved nature of housing policy across the UK means that approaches to addressing social housing (in addition to the bedroom tax) vary across countries.19 For example, Scotland introduced countermeasures to reduce the negative effects of the bedroom tax.20 In addition, housing availability and characteristics (eg, safety standards) significantly vary across nations.19 A total of 2078 individuals were included in our study, with 422 individuals who were living in underoccupied flats/eligible for bedroom tax (ie, treatment group) and 1926 individuals who were not affected by the bedroom tax (ie, control group). Our matching method identified 412 individuals in the treatment group and 412 in the matched control group. See online supplemental figure S1 for eligibility criteria for study cohort creation.

Supplemental material

Figure 1

Predicted GHQ score over the study period in matched and unmatched samples. Matched sample shows parallel trends in the prepolicy period (2009–2011). GHQ, General Health Questionnaire.

Participants were considered to be in the treatment group if they were eligible for the bedroom tax, otherwise they were in the control group. Eligibility was based on whether their number of bedrooms exceeded what is allowable beyond their household characteristics, where one room is allowed for: (1) every adult couple; (2) any other adults aged 16 or over; (3) two children under 10, regardless of their sex; (4) two children under 16 of the same sex; and (5) or any other child (ie, if there are three children under 10, two rooms can be taken). These criteria were applied to our study cohort using valid household responses preceding the policy in April 2013.1

Outcome variable

The psychological distress of study participants is measured using the 12-item General Health Questionnaire (GHQ). The GHQ is a validated21 survey tool to measure psychological distress in the past 2-week period. The survey questions rated various components (eg, ability to concentrate) of the participants’ mental health on a scale of 1 (not at all) to 4 (much more than usual). Based on the participant’s responses, the survey provided a derived score from 0 to 36, where a higher score indicated a greater degree of psychological distress.

Statistical analysis

We applied difference-in-differences (DID), a quasi-experimental technique widely used to evaluate public policy, based on the prepost policy effect differences between the treatment and control groups, adjusting for time-varying confounders.22–24 The DID uses a longitudinal individual-level fixed-effects regression model to estimate the effect of the bedroom tax on psychological distress. An interaction term is included for the cross-product of (1) the preintervention vs postintervention indicator and (2) the treatment assignment, to examine the potential differences in psychological distress across the treatment and control groups which can be attributed to the effect of the policy. To minimise endogeneity and calculate unbiased DID estimates, we need to ensure that the control and treatment groups are exchangeable, that is, to establish that the two groups are balanced with respect to covariates that are associated with the outcome of interest. Therefore, we used propensity score matching (PSM) to identify a subset of the control group that is as similar as possible to the treatment group to establish exchangeability.25 26 We used the one-to-one nearest neighbour matching method without replacement using the following characteristics: age, sex, marital status, education, ethnicity, employment status, presence of chronic illness and place of residence. We also applied a calliper (0.01) to improve the matching performance.

Another assumption in DID is that treatment and control groups have parallel trends in the outcome in the prepolicy period.27 We plotted the fitted prepolicy trends and conducted a test of the parallel trends assumption by performing a fixed-effect regression model that included the interactive term between year dummies and the treatment assignment status.

To further control for confounding variables at the modelling stage, our study controls for all observed and unobserved individual-level time-invariant characteristics such as immigration status and family history of illnesses by using within-person estimators (individual fixed-effects). A region-level fixed-effect was included to adjust for regional differences, and year dummies were added to adjust for trend effects. Since eligibility for the study excluded those with missing data on household information to determine bedroom tax status, there was negligible data missingness in the unmatched sample (ie, only one person had missing data for education and one person had missing data for marital status), and no missing data in the matched sample; therefore, complete-case analyses were conducted.

Sensitivity analysis

We conducted the following robustness checks. First, to ensure that specifying a calliper distance of 0.01 in the nearest neighbour matching was appropriate, we conducted matching using different calliper distances. Second, we used Mahalanobis matching to obtain the matched control group instead of one-to-one nearest neighbour matching, to ensure that the results were not dependent on a certain matching method. Third, we conducted stratified analyses by education and income groups, since the deleterious effects of the policy may be stronger among those with less resources to face the financial burden.28 29 Fourth, we extended the postintervention periods (up to 2017) to examine differences between the long-term and medium-term effects. Lastly, we ran the fixed effects logit model by using a clinical GHQ score cut-off of 6 and above, instead of using the continuous outcome, which has been used for the screening of common psychiatric morbidity including depression and anxiety.30


As shown in table 1, PSM improved the exchangeability of the treatment and control groups. Modelling results to follow are based on the matched sample. Given that unbiased DID estimates require the assumption of parallel trends, figure 1 shows the common trends across the control and treatment groups in the fitted outcome (GHQ) over the study period. The figure shows that the control and treatment groups have parallel trends in the prepolicy period. To provide further evidence for parallelism, we found no evidence that the changes in GHQ across treatment and control groups in the pre-policy period were different. Based on the results of a fixed effect model (with interactive terms between year dummies and treatment status) presented in online supplemental table S1, in the prepolicy period 2010–2011, the control group had a change in GHQ of 0.14 (95% CI −0.58 to 0.86), while the treatment group had a change in GHQ of 0.22 (95% CI −0.47 to 0.91). The difference in the change between the two groups was 0.08, which was not significantly different from zero (95% CI −0.91 to 1.07) with a p value of 0.87.

Supplemental material

Table 1

Descriptive characteristics of unmatched and matched study samples before the implementation of the bedroom tax

Table 2 shows the DID estimates of the impact of the implementation and announcement of the bedroom tax on GHQ (higher GHQ is greater distress). The bedroom tax announcement did not result in different GHQ across the treatment and control groups, where the difference between groups was not statistically different from zero (0.53, 95% CI −0.21 to 1.27). However, the implementation of the bedroom tax led to the treatment group experiencing a relative increase in GHQ of 0.88 (95% CI 0.05 to 1.71) compared with the control group.

Table 2

Fixed effect regression predicting the bedroom tax policy effect on GHQ of the treatment versus control group over a 2-year postintervention period

Many sensitivity analyses were conducted, and the results can be found in the online appendix. First, results using differently sized callipers all provide evidence for a small but statistically significant implementation effect on the treatment group (online supplemental table S2). Second, when using Mahalanobis matching as an alternative to one-to-one nearest neighbour, DID models (online supplemental table S3) indicate similar findings to the main analyses: while the bedroom tax implementation was associated with an additional 0.888-point increase in psychological distress (95% CI 0.02 to 1.75, p=0.04) compared with the matched control group, the announcement was not associated with a significant increase. Third, the results of our stratified analyses (online supplemental tables S4 and S5) suggest that the deleterious effects of the bedroom tax implementation were greater among individuals of lower socioeconomic status. Fourth, we tested for longer postimplementation effects at 3 and 5 years, where we found a 0.82-point increase (95% CI 0.05 to 1.59) in psychological distress was followed in 3 years of implementation, and the policy effect was fully attenuated after 5 years (online supplemental table S6). Finally, we replace the continuous outcome with a clinical cut-off of 6+score on GHQ (see online supplemental table S7), and we found that the bedroom tax policy increased the relative risk of being screened for common mood disorders by 42% (rate ratio 1.42; 95% CI 1.01 to 1.93).


We found evidence that the implementation of the bedroom tax led to increases in psychological distress among social housing renters. These policy effects were most pronounced in the time immediately after the implementation of the austerity, and were still significant 3 years after policy implementation. Compared with the control group, social housing renters who were affected by the bedroom tax had an additional increase in GHQ of 0.88 (2 years after policy implementation), and these findings were robust to multiple sensitivity analyses. To put into context the magnitude of the policy effect, we calculated the standardised mean difference for the treatment effect to be 0.16 (95% CI 0.02 to 0.29), which is considered small. Using a clinical GHQ cut-off of 6 and above, we saw the bedroom tax implementation was associated with a 42% increased risk for common mood disorders such as anxiety and depression. With over 491 765 households affected by the bedroom tax as of November 2020,11 assuming that the prevalence of common mood disorders are similar between social housing residents and the general population (~20%),31 a 42% increased risk of common mood disorders would translate to individuals from 42 291 households needing to be screened for depression and anxiety as a result of the intervention. This shows that while the effect size of the bedroom tax for targeted individuals is small, the bedroom tax as a population-wide intervention may have considerable impacts.32 Our findings are supported by established theoretical literature on the relationship between the welfare state and health.33 The welfare state plays an essential role in affecting health outcomes and health inequalities by either directly influencing the process in which social stratification occurs (eg, as a determinant of social class) or modifying the pathway through which social stratification affects health.34 35 Based on the welfare state theory, the UK’s move towards less generous welfare states through decreasing the key public services, such as housing benefits for the disadvantaged, may lead to deleterious effects on mental health of the affected populations.

To the best of our knowledge, this is the first study using a quasi-experimental approach to evaluate the effect of the bedroom tax on psychological distress. Our findings are consistent with a prior qualitative study where social housing renters and service providers both stated that the bedroom tax led to heightened stress, symptoms of anxiety and depression.7 This study contributes to the literature by providing evidence that the adverse mental health consequences of the bedroom tax are generalisable to the English population. Likewise, our findings corroborate evidence from a previous quantitative study that reported decreased health satisfaction after the implementation of the austerity.2 Our use of a validated mental health outcome (GHQ-12) alongside a combination of matching +DID enhanced the rigour of the causal inference.26

Several limitations need to be considered when interpreting our results. First, there may be other time-variant covariates not captured in the data confounding the treatment effects of the intervention on mental health (eg, death of a friend). Second, while GHQ-12 is a validated measure for detecting psychological distress suitable for population health surveys,21 it is limited compared with a formal psychiatric assessment from a trained clinician. Third, lost to follow-up is a common problem in panel data analysis, and a differential lost to follow-up between the treatment and control groups may bias estimates. However, we found similar lost to follow-up across the matched treatment and control groups (ie, 11% vs 18%), which may be evidence that this is not a problem. Fourth, our matching approach may reduce the representativeness of the samples. Fifth, since the data do not contain information on the treatment assignment for the bedroom tax, we assumed the treatment status based on household characteristics, which may be inaccurate due to participant misreporting. Lastly, since our definition of the treatment assignment is based on household composition and housing situation in the preintervention period, changes in household composition or housing situation may change the households’ eligibility for bedroom tax. If household composition changes (eg, a new baby is born), these changes may bias our results in an unknown direction. However, if there is a change in housing condition as a result of downsizing to avoid the bedroom tax, the observed effects would be attributable to a mover’s effect (or a change in housing conditions that would decrease their overall well-being). In either case, any changes in mental health whether directly (as a result of the tax) or indirectly (ie, having to move which lead to a consequent change in living conditions), may be attributable to the bedroom tax policy. A major strength of this study is using rigorous quasi-experimental design applying matching +DID, thereby ensuring exchangeability between the treatment and control groups and obtaining an appropriate counterfactual to estimate the effect of the bedroom tax.22 25 Given the cost of performing randomised controlled trials in the topic and the possibility of the limited external validity, our research design provides the best available evidence on the evaluation of the bedroom tax and psychological distress.

The development and implementation of austerity measures that target disadvantaged populations must take into consideration the effects of such policies beyond economic benefit and budget reduction. Furthermore, our study findings suggest that clear directives and support for the changes need to be provided to the affected individuals when such policies are implemented. For example, the psychological impact of the bedroom tax implementation may be lessened if the affected families were provided with assistance and incentives to find alternative housing with the least amount of disruptions to family life.

Data availability statement

Data are available in a public, open access repository. Data are available in a public, open access repository. This data can be accessed through the UK Data Service:!/abstract.

Ethics statements

Patient consent for publication

Ethics approval

The Understanding Society survey was approved by the Ethics Committee of the University of Essex. The ethical approval for the secondary data analyses was granted by the Brock University Research Ethics Board (file number 21-257-CHUM).


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Twitter @antonychum

  • Contributors CK contributed to conception and design of the study. CK performed the statistical analysis. CT, CK, AN and AC wrote the first draft of the manuscript. All authors contributed to manuscript revision, read and approved the submitted version. AC as the gurantor had full access to all of the data in the study and takes responsibility for the integrity of the data and the accracy of the data analysis.

  • Funding Funding for the project is provided through research start-up funds from Brock University, Faculty of Applied Health Science, to the project principal investigator, Antony Chum.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.