Happiness for the whole household: accounting for household spillovers when comparing the cost-effectiveness of psychotherapy to cash transfers

By JoelMcGuire, Samuel Dupret, MichaelPlant @ 2022-04-14T20:05 (+51)

This is a linkpost to https://www.happierlivesinstitute.org/report/happiness-for-the-whole-family/

This post summarises our updated cost-effectiveness comparison of psychotherapy and cash transfers. It now includes an estimate of household spillovers and concludes psychotherapy is 9 times more cost-effective than cash transfers. 

Our original analysis estimated that the average psychotherapy intervention in our dataset would be 12 times more cost-effective than the average monthly cash transfer. We used this wider evidence base to estimate the cost-effectiveness of two highly effective charities at implementing each type of intervention: Give Directly (which provides $1,000 lump-sum cash transfers) and StrongMinds (which provides group psychotherapy). We estimated that StrongMinds was 12 times more cost-effective than GiveDirectly. 

Our analysis was critiqued (on Twitter and the EA Forum) on the grounds that omitting household spillovers would underestimate the effects of cash transfers compared to psychotherapy. We, therefore, decided to extend our estimate to include household spillovers. We now find that psychotherapy and StrongMinds are 9 times as cost-effective, a small reduction from before. 

We believe that our analysis demonstrates that high-quality mental health interventions can be as good or better than highly-regarded and well-evidenced economic interventions.  This is a surprising and important result that should inform our global priorities. 

A further implication is that we’ve shown it is feasible to compare interventions in terms of their effects on subjective wellbeing and doing so can reveal new funding opportunities. In short, we think we can do good better by measuring good better. 

In the rest of this post, we review the pathways through which household spillovers occur and briefly discuss our methods and main findings.

Pathways for interventions to spillover onto the whole household 

Some readers may be sceptical about the size of the spillover effects for cash transfers and/or psychotherapy. We think household spillovers can occur in at least two ways: economic contributions and emotional contagion (e.g. improved interpersonal interactions). Our aim here is simply to illustrate the mechanisms by which psychotherapy and cash transfers might plausibly produce household spillovers rather than providing a neat account of the nature and relative strengths of these dynamics. 

The evidence for household spillovers through economic contribution is clear for cash transfers. GiveDirectly recipients report spending most of their cash transfers (68.4%) on common household goods (GiveWell, 2016). Household spillovers via economic contributions also seem plausible for psychotherapy as meta-analyses of randomised controlled trials find that psychotherapy improves economic outcomes (Lund et al., 2020). 

Emotional contagion is likely to play a role in the spillover effects of both interventions as people’s mood is affected by the mood of their close connections (Headey et al., 2014; Rosenquist et al., 2011). Meta-analyses have found that psychotherapy improves mother-child relationships (Cuijpers et al., 2015) and cash transfers reduce domestic violence (Baranov et al., 2021). 

Methods

The key to estimating the total household effect for each intervention is to estimate the household spillover ratio, the share of the recipient’s benefit that the average household member receives. We estimate this by taking the ratio of the average recipient effect and the household member effect for every study we found that includes both. 

Results

For cash transfers, we estimate that each household member experiences 86% (95% CI: 43%, 154%) of the benefits experienced by the recipient. For psychotherapy, we estimate the spillover ratio to be 53% (95% CI: 11%, 108%). 

Accounting for these different spillover ratios reduces the cost-effectiveness advantage of psychotherapy relative to cash transfers; it is no longer 12 times more cost-effective. Instead, we find that psychotherapy is 8.8 times (95% CI: 1.7, 99.8) more cost-effective than monthly cash transfers. When comparing the charities that deploy these interventions, we estimate that StrongMinds is 9.5 times (95% CI: 1.3, 89.9) more cost-effective than GiveDirectly.

We conducted eight robustness checks and found psychotherapy (and StrongMinds) to be at least five times more cost-effective than cash transfers (and GiveDirectly) in all cases. These checks included a Bayesian analysis to see how different priors would affect our spillover ratio estimates, checking the sensitivity of our results to unusual studies, and testing how our estimates would change using different methods.

A serious and neglected gap in the literature

Our household analysis is based on a small number of studies, eight for cash transfers and three for psychotherapy, far fewer than we have for the effects on the recipients. This small evidence base means there is considerable uncertainty in our estimates. In Section 3 we discuss how we quantified our uncertainty using Monte Carlo simulations and Section 6 highlights some of the limitations of this approach. 

The lack of data on household effects seems to be a serious and neglected gap in the literature. Although the household effects did not substantially change our results in this case, we find it concerning that there is dramatically less data on how interventions impact household members. In our estimates for both interventions, the benefit to the recipient was considerably smaller than the total benefit to all the other household members. This implies that we need to estimate the total household effects to confidently compare and prioritise across interventions, particularly ones that plausibly benefit whole households, such as improvements to air, water, and housing.

Further information
Please refer to our full report if you would like to explore our reasoning, methodology, results, and robustness checks in more detail.