Health Effects of Extending State Pension Age for UK Women

Ludovico Carrino, King''s College London
Karen Glaser, King''s College London
Mauricio Avendano, King''s College London

Prior studies have examined the impact of retirement on health, but there is no consensus as to whether longer working lives lead to better health. We investigate the impact of a recent pension reform which, since 2010, gradually extended the age of eligibility for the State Pension for women in the UK, from 60 to 66 years. Using a difference-in-differences approach on 8400 observations of women aged 60-64 participating in the Understanding Society study from 2009 to 2015, we find strong evidence that increasing the State Pension Age led to a statistically and clinically significant negative effect on mental and physical health. We also find evidence that women with a longer extension in the State Pension Age experienced worse health outcomes. We show that this effect is driven by women from routine-manual occupations and led to increasing health inequalities between occupational groups.

Over the last decades, manyGovernments have implemented policies to extend working lives of older workers.Driven by concerns about the sustainability of pension systems, several reformshave extended the State Pension Age (SPA), which requires older workers toremain employed until later ages before they can claim pension benefits.Delaying the SPA, however, might have important consequences for health. On theone hand, retirement may offerindividuals more flexibility in the allocation of time. After retirement, theopportunity cost of time declines, incentivizing individuals to invest moretime in their health, by exercising or cooking healthier foods. The “time-price”of medical care might also decline, incentivizing the use of preventive care. Thesechanges in time allocation flexibility and the price of time might mean thatextending SPA may not enable workers to reap these benefits of retirement,particularly for low-skilled workers, who are more likely to have lessflexibility and higher exposure to hazardous working conditions and job-relatedstress.

On the other hand,a vast literature suggests that social engagement and participation in olderage are associated with better health and cognition. Earlier retirement,therefore, may imply the loss of a daily time-structure, reduced social contacts,self-esteem and status, potentially leading to poorer health outcomes.Retirement may also generate changes in health-enhancing behavior, for example,by reducing physical activity and increasing consumption of smoking or alcohol.Under this scenario, extending the SPA may lead to positive physical and mentalhealth effects.

In this study, we investigate the health-consequences of a recent reformin the UK that gradually increased the women’s SPA from 60 to 66, for womenborn after March 1950. This represents one of the most significant increases instage pension age in high income countries. Because of the reform’s design,otherwise-similar cohorts were subject to very different pensionable ages. In apreliminary stage of our analysis we show that, as expected, extending the SPAsignificantly lowered the probability of reporting to be retired at the age of60, 61 or 62.

We use data from the UnderstandingSociety study (2009-2015), focusing on mental-health (the SF-12 MentalComponent Score -MCS- and the General Health Questionnaire -GHQ), physical-health(the SF-12 Physical Component Score -PCS). For GHQ, higher levels correspond toworse health, while the opposite occurs for MCS and PCS. We also explore biomarkersdata collected for a subsample of respondents between 2010 and 2012.

Our main sample consists of 8400observations of women aged 60-64 who have at some point engaged in paidemployment, interviewed between 2009 and 2015. We use a difference-in-differencesapproach that exploits the fact that women with similar characteristics wereexposed to a shorter or longer extension of their SPA based on their month andyear of birth.

Our baseline model specificationis defined as follows:

yict=α+β*belowSPAict+dct+X’ictφ+εict ,

whereby y refers to ahealth outcome at time t for individual i born in month-year cThe main variable of interest is anindicator function for being below SPA, i.e., an interaction between themonth-of-birth and the interview date. Due to the "gradual" nature ofthe SPA-shift, same-aged women may have different SPA-status, depending ontheir interview date. We account for the direct effect of age-in quarters(fixed-effects γi),calendar year-in-quarters (fixed-effects ηt) and cohort (a linear control for month-of-birth dc) (Cribb, Emmerson, & Tetlow, Labour Economics, 2016).Additional controls include living arrangements and marital status, number ofchildren, education, and occupation type (routine, intermediate or managerial).We then extend this model by interacting the “below SPA” variable withoccupation type, and by incorporating the number of months of SPA-postponement.

Our main results aresummarised in Table 1. Coefficients reported in Column 2 (referred to the“below SPA” variable) suggest that an increase in the SPA leads to a large andsignificant worsening of physical and mental health. In particular, the effectis found for mental health as measured with the MCS and the GHQ indices (firsttwo rows in the table). Results for the GHQ score are clinically relevant (evaluatedwith widely used thresholds in the literature). Detrimental physical and mentaleffects are also found for specific items of the SF-12 and GHQ scales (binaryvariables for the presence of bad-health).

Column 3 shows that routine-manualworkers are driving these results (the coefficient corresponds to theinteraction term), exhibiting large deterioration in mental and physical health.Column 4 reports the estimated change in health outcomes for an SPA increase of36 months or more, with respect to no increase: worse health outcomes areassociated with longer SPA-extensions.

Figure 1 shows the marginalchanges in the GHQ and MCS scores, for different extents of SPA postponement(with respect to no increase). Figure 2 shows how the marginal changes in theprobability of reporting selected bad-health outcomes are larger for routineworkers with longer SPA-extensions.


Presented in Session 1211: Policy Issues