Skip to main content

Does Volunteering Make Us Happier, or Are Happier People More Likely to Volunteer? Addressing the Problem of Reverse Causality When Estimating the Wellbeing Impacts of Volunteering


Evidence of the correlation between volunteering and wellbeing has been gradually accumulating, but to date this research has had limited success in accounting for the factors that are likely to drive self-selection into volunteering by ‘happier’ people. To better isolate the impact that volunteering has on people’s wellbeing, we explore nationally representative UK household datasets with an extensive longitudinal component, to run panel analysis which controls for the previous higher or lower levels of SWB that volunteers report. Using first-difference estimation within the British Household Panel Survey and Understanding Society longitudinal panel datasets (10 waves spanning about 20 years), we are able to control for higher prior levels of wellbeing of those who volunteer, and to produce the most robust quasi-causal estimates to date by ensuring that volunteering is associated not just with a higher wellbeing a priori, but with a positive change in wellbeing. Comparison of equivalent wellbeing values from previous studies shows that our analysis is the most realistic and conservative estimate to date of the association between volunteering and subjective wellbeing, and its equivalent wellbeing value of £911 per volunteer per year on average to compensate for the wellbeing increase associated with volunteering. It is our hope that these values can be incorporated into decision-making at the policy and practitioner level, to ensure that the societal benefits provided by volunteering are better understood and internalised into decisions.

This is a preview of subscription content, access via your institution.



  2. Recently, a number of studies have used more robust statistical strategies that provide a high degree of confidence in the findings. For example, Stutzer and Frey (2004) analysed trends in volunteering and SWB after controlling for other determinants of SWB and by assessing circumstances where volunteering status essentially became random due to policy changes (natural experiments). Binder and Freytag (2013) analysed six waves of British Household Panel Survey (BHPS) by using statistical matching to estimate the causal impact of volunteering on happiness. There is some evidence that in the context of charitable giving the pathway of happier people giving more dominates the pathway of giving making people happier; however, the effect does go both ways (Boenigk and Mayr 2015).

  3. An alternative suggestion is that it takes time to realize the benefits of volunteering.

  4. Panel data estimation techniques (FE and FD) control for time-invariant factors, including long-term health and psychological factors, as well as prior trends of pre-existing SWB.


  6. We also control for time-specific effects and seasonality by including dummy variables for the wave of the survey and the month of the interview and for local area characteristics via region dummies [unfortunately more detailed geographic information is not available in the BHPS data].

  7. The means over time of the respective variables for that particular individual are subtracted from each observation.

  8. Use of OLS and fixed effects regressions assumes that the SWB reporting scale (1–7) is cardinal. Research shows that the cardinal models (OLS regressions) and ordinal models (ordered latent response models, such as ordered logit/probit) give remarkably similar results, and hence for ease of interpretation we assume cardinality, as is standard in much of the literature (Ferrer-i-Carbonell and Frijters 2004).

  9. As a caveat, from a theoretical point of view, the coefficient \(\beta_{1}\) can be biased by the fact that those that are more likely to take up volunteering also naturally have an upward trend in wellbeing, although in practice that situation is somewhat harder to conceive.

  10. The drop in the magnitude of the coefficient for health when moving from OLS to FE/FD is likely to have occurred because selection effects have been partialled out in the panel data models. The OLS coefficient would have incorporated some selection of healthier people into volunteering.


Download references

Author information

Authors and Affiliations


Corresponding author

Correspondence to Ricky N. Lawton.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix: Model Equations Used for Estimation

Appendix: Model Equations Used for Estimation

Note. Xit includes:

  • The natural logarithm of equivalised household income (monthly)

  • Age in years

  • Age (in years) squared)

  • Gender (1 for being male, 0—female)

  • Marital status (dummies for married, divorced, widowed, separated, single [base group], cohabiting)

  • Education (dummies for no qualification [base level], A level or GCSE qualifications, other qualifications, degree or higher degree)

  • Employment (dummies for self-employed, in paid employment, unemployed [base level], retired, on maternity leave, looking after family or home, full-time student, long-term sick or disabled, on a government training scheme, unpaid worker in family business, doing something else)

  • Self-rated general health (dummies for 1—poor [base level], 2—fair, 3—good, 4—very good, 5—excellent)

  • Number of children (dummies for 0 [base level], 1, 2, 3, 4 or more)

  • Whether respondent is religious (dummies for No [base level], Yes, Unknown)

  • Whether respondent is a carer for someone in the household

  • Region of England/UK (dummy for each of the 9 regions + Wales, Scotland, NI)

  • House ownership (dummies for owned outright [base level], being bought on mortgage, shared ownership, rented, rent-free, other)

  • Whether the respondent would like to move house

  • Ethnicity broad categories (White [base], Mixed, Asian, Black, Other, Unknown)

  • Wave of survey (dummy for each included wave—6, 8, 10, …, 24)

  • Interview month (dummy for each month)

Xnohit includes the same as above but without self-rated health, as it is used as an outcome variable in those models.

Table 4.

Pooled OLS (column 2):

$$LifeSat_{it} = \alpha + \beta_{1} Vol12m_{it} + X_{it} \beta + \varepsilon_{it} \quad \left( {{\text{row }}2} \right)$$
$$SelfRatedHealth_{it} = \alpha + \beta_{1} Vol12m_{it} + Xnoh_{it} \beta + \varepsilon_{it} \quad \left( {{\text{row }}3} \right)$$
$$GHQ_{it} = \alpha + \beta_{1} Vol12m_{it} + Xnoh_{it} \beta + \varepsilon_{it} \quad \left( {{\text{row }}4} \right)$$

Fixed effects (column 3):

$$(LifeSat_{it} - \overline{{LifeSat_{i} }} ) = \alpha + \beta_{1} \left( {Vol12m_{it} - \overline{{Vol12m_{i} }} } \right) + \left( {X_{it} - \overline{X}_{i} } \right)\beta + \varepsilon_{it} \quad \left( {{\text{row }}2} \right)$$
$$\left( {SelfRatedHealth_{it} - \overline{{SelfRatedHealth_{i} }} } \right) = \alpha + \beta_{1} \left( {Vol12m_{it} - \overline{{Vol12m_{i} }} } \right) + \left( {Xnoh_{it} - \overline{{Xnoh_{i} }} } \right)\beta + \varepsilon_{it} \quad \left( {{\text{row }}3} \right)$$
$$\left( {GHQ_{it} - \overline{{GHQ_{i} }} } \right) = \alpha + \beta_{1} \left( {Vol12m_{it} - \overline{{Vol12m_{i} }} } \right) + \left( {Xnoh_{it} - \overline{{Xnoh_{i} }} } \right)\beta + \varepsilon_{it} \quad \left( {{\text{row }}4} \right)$$

First differences:

$$\left( {LifeSat_{it} - LifeSat_{i,t - 1} } \right) = \alpha + \beta_{1} \left( {Vol12m_{it} - Vol12m_{i,t - 1} } \right) + X_{it} \beta + \varepsilon_{it} \quad \left( {{\text{row }}2} \right)$$
$$\left( {SelfRatedHealth_{it} - SelfRatedHealth_{i,t - 1} } \right) = \alpha + \beta_{1} \left( {Vol12m_{it} - Vol12m_{i,t - 1} } \right) + Xnoh_{it} \beta + \varepsilon_{it} \quad \left( {{\text{row }}3} \right)$$
$$\left( {GHQ_{it} - GHQ_{i,t - 1} } \right) = \alpha + \beta_{1} \left( {Vol12m_{it} - Vol12m_{i,t - 1} } \right) + Xnoh_{it} \beta + \varepsilon_{it} \quad \left( {{\text{row }}4} \right)$$

Table 5.

Column 2:

$$LifeSat_{it} = \alpha + \beta_{1} Vol12m_{it} + X_{it} \beta + \varepsilon_{it} ;$$

Column 3:

$$LifeSat_{it} - \overline{{LifeSat_{i} }} ) = \alpha + \beta_{1} \left( {Vol12m_{it} - \overline{{Vol12m_{i} }} } \right) + \left( {X_{it} - \overline{{X_{i} }} } \right)\beta + \varepsilon_{it} ;$$

Column 4:

$$\left( {LifeSat_{it} - LifeSat_{i,t - 1} } \right) = \alpha + \beta_{1} \left( {Vol12m_{it} - Vol12m_{i,t - 1} } \right) + X_{it} \beta + \varepsilon_{it} ;$$

Table 6.

Column 2:

$$(LifeSat_{it} - \overline{{LifeSat_{i} }} ) = \alpha + \beta_{1} \left( {Volfreq1_{it} - \overline{{Volfreq1_{i} }} } \right) + \beta_{2} \left( {Volfreq2_{it} - \overline{{Volfreq2_{i} }} } \right) + \beta_{3} \left( {Volfreq3_{it} - \overline{{Volfreq3_{i} }} } \right) + \beta_{4} \left( {Volfreq4_{it} - \overline{{Volfreq4_{i} }} } \right) + \left( {X_{it} - \overline{X}_{i} } \right)\beta + \varepsilon_{it} ;$$

Column 3:

$$\left( {SelfRatedHealth_{it} - \overline{{SelfRatedHealth_{i} }} } \right) = \alpha + \beta_{1} \left( {Volfreq1_{it} - \overline{{Volfreq1_{i} }} } \right) + \beta_{2} \left( {Volfreq2_{it} - \overline{{Volfreq2_{i} }} } \right) + \beta_{3} \left( {Volfreq3_{it} - \overline{{Volfreq3_{i} }} } \right) + \beta_{4} \left( {Volfreq4_{it} - \overline{{Volfreq4_{i} }} } \right) + \left( {X_{it} - \overline{X}_{i} } \right)\beta + \varepsilon_{it} ;$$

Column 4:

$$\left( {GHQ_{it} - \overline{{GHQ_{i} }} } \right) = \alpha + \beta_{1} \left( {Volfreq1_{it} - \overline{{Volfreq1_{i} }} } \right) + \beta_{2} \left( {Volfreq2_{it} - \overline{{Volfreq2_{i} }} } \right) + \beta_{3} \left( {Volfreq3_{it} - \overline{{Volfreq3_{i} }} } \right) + \beta_{4} \left( {Volfreq4_{it} - \overline{{Volfreq4_{i} }} } \right) + \left( {X_{it} - \overline{X}_{i} } \right)\beta + \varepsilon_{it} ;$$

where Volfreq is:

  • 1—volunteering at least once a week

  • 2—volunteering at least once a month

  • 3—volunteering several times a year

  • 4—volunteering once a year or less

  • 5—volunteering never or almost never [base group]

Table 7.

Pooled OLS (column 2):

$$LifeSat_{it} = \alpha + \beta_{1} Vol12m_{it} + X_{it} \beta + \varepsilon_{it} ; \, \quad \left( {{\text{row }}2} \right)$$
$$LifeSat_{it} = \alpha + \beta_{1} Vol12m_{it} *Male_{it} + \beta_{2} Vol12m_{it} *Female_{it} + X_{it} \beta + \varepsilon_{it} ;\quad \, \left( {{\text{rows}}\;3{-}4} \right)$$
$$LifeSat_{it} = \alpha + \beta_{1} Vol12m_{it} *Age16\_34_{it} + \beta_{2} Vol12m_{it} *Age35\_54_{it} + \beta_{3} Vol12m_{it} *Age55\_74_{it} + \beta_{4} Vol12m_{it} *Age75plus_{it} + X_{it} \beta + \varepsilon_{it} ;\quad \, \left( {{\text{rows }}5{-}8} \right)$$
$$LifeSat_{it} = \alpha + \beta_{1} Vol12m_{it} *HINC1_{it} + \beta_{2} Vol12m_{it} *HINC2_{it} + \beta_{3} HINC3_{it} *Age55\_74_{it} + \beta_{4} HINC4_{it} *Age75plus_{it} + X_{it} \beta + \varepsilon_{it} ;\quad \left( {{\text{rows}}\;9{-}11} \right)$$


  • HINC1—household income below £1500/month

  • HINC2—household income between £1500/month and £3000/month

  • HINC3—household income between £3000/month and £5000/month

  • HINC4—household income above £5000/month

Fixed effects (column 3): same as above but all variables demeaned.

First differences (column 4): same as above but all dependent variables and the volunteering variable taken as first differences.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Lawton, R.N., Gramatki, I., Watt, W. et al. Does Volunteering Make Us Happier, or Are Happier People More Likely to Volunteer? Addressing the Problem of Reverse Causality When Estimating the Wellbeing Impacts of Volunteering. J Happiness Stud 22, 599–624 (2021).

Download citation

  • Published:

  • Issue Date:

  • DOI:


  • Volunteering
  • Altruism
  • Subjective wellbeing
  • Wellbeing valuation
  • Compensating surplus
  • First difference