The promise of ACTRA: Preventing crime with cognitive behavioural therapy
Updates
Important Update
We mention the charity NEPI in this report. Since September 11th 2025, the Happier Lives Institute is no longer recommending NEPI, the NY non-profit organisation that works in Liberia, as the organisation has discontinued operations and is in the process of dissolving, due to the discovery of financial irregularities. This does NOT change our prediction about ACTRA.
Summary
ACTRA is a charity delivering CBT to prevent crime in Latin America (starting in Colombia). It uses a model similar to NEPI, a charity that combines cash transfers and CBT to prevent crime in Liberia. We evaluated and recommended NEPI as a ‘promising charity’ in 2024.
The key difference between NEPI and ACTRA is that ACTRA does not use cash transfers, instead only using the CBT component.
ACTRA is too new to have sufficient data for an evaluation. Thereby, we do a prediction of ACTRA’s potential cost-effectiveness – once it reaches the same scale as NEPI, which could be in a few years – by adjusting our model for NEPI. Only using the CBT component reduces the effectiveness of the intervention, but reduces the costs much more, leading to a higher cost-effectiveness.
We predict that ACTRA could have a cost-effectiveness of 37 WELLBYs created per $1,000 donated (or a cost of $27 to produce a WELLBY), which would be more cost-effective than our current cost-effectiveness estimate for NEPI of 22 WELLBYs created per $1,000 donated (McGuire et al., 2024f).
This higher cost-effectiveness is driven by two factors related to dropping the cash component:
- Decreases the effectiveness by much less 13.6 → 12.1 WELLBYS (11% reduction)
- But decreases the cost much more: $630 → $330 per person (48% reduction).
We also think that the ACTRA theory of change, characteristics of the ACTRA team, and their eye towards cost-effectiveness are reassuring. Critically, they have a credible plan for generating high quality evidence of how well their programme works in practice.
This reasoning provides the basis of giving ACTRA an ‘honorable mention’, a new evaluation category below our existing two categories of ‘top charity’ and ‘promising charity’. Honourable mentions are special cases that we think are likely cost-effective, but we don't have sufficient material to make a recommendation like our promising or top recommendation.
We think that ACTRA is potentially worth consideration by donors interested in more “hits based” approaches (i.e., less risk averse). At time of writing, they have a funding gap for 2025-2026 of around $500,000.
Our recommendations change over time as we produce new research, stay updated by consulting our website.
Notes and acknowledgements
Author note: Joel McGuire contributed to the conceptualisation, investigation, analysis, data curation, and writing of the project. Samuel Dupret contributed to the investigation, analysis, and writing of the project. Michael Plant contributed to the writing of the project.
The views expressed in this document do not necessarily reflect the perspectives of reviewers or employees of the evaluated charities.
Charity information note: We thank Laura Castro (co-founder) for providing us with information about ACTRA.
1. ACTRA and its context
ACTRA (Acción Transformadora) is a new organisation scaling the use of CBT to prevent criminal behaviour in Latin America. They were incubated by AIM in 2024. ACTRA’s plan is to adapt and evaluate NEPI’s Sustainable Transformation of Youth in Liberia (STYL) intervention in Latin America (instead of Liberia). They are starting in Colombia. If they are able to deliver a version of their intervention cost-effectively, they hope to scale its use across Latin America.
NEPI’s intervention targets young male offenders to provide psychotherapy that will help them with decision making, controlling emotions, and reducing criminal behaviour. They use both a cash transfer and CBT. The CBT is intensive with 3 sessions of 4 hours for 8 weeks (i.e., 24 sessions) for groups of 20 men. The therapy is facilitated by individuals from similar backgrounds. See page 9 of our report for more of a description of NEPI’s programme.
The core difference between ACTRA and NEPI is that NEPI uses both a cash transfer and CBT, while ACTRA only uses the CBT component. ACTRA is too new to have sufficient data for an evaluation. Thereby, we do a prediction of its potential cost-effectiveness by adjusting our model of NEPI (see Section 2).
Currently, ACTRA are running pilots with partners to test and refine their adaptation of the programme. They ran one pilot with the government of Cali in Colombia. They have more pilots in the pipeline. Currently they are planning to have 24 sessions of 1.5 hours, 2 to 3 times a week. The exact number of sessions and their duration will depend on the partners and target populations. The details of ACTRA’s programme may evolve over time and piloting.
ACTRA are also considering leveraging partner organisations to run the ACTRA programme (on top of, or instead of, less well-evidenced programmes these partner organisations might be running). This would save them from having to hire facilitators directly and train them from the ground up. This could lead to cost savings and a greater scale. However, for our modelling (see Section 2), we assume that ACTRA will hire the facilitators and bear those costs.
1.1 ACTRA’s strengths
We have had a few conversations with them, and reviewed some of their public and private plans.
We are impressed with:
- Their theory of change is simple, coherent, and persuasive.
- The ambition and carefulness of their plan to generate evidence.
- Their eye towards cost-effectiveness.
- Their team: Laura Castro (co-founder of ACTRA) has had experience working at IPA, thereby, we think they are well placed to generate useful evidence.
- Their pitch deck: They have one of the better pitch decks we have come across. It is rich in detail, easy to follow, and efficiently conveys the necessary information.
- Their advisory team: Which includes support from the original NEPI team and related academics like Dr Blattman (who ran the main RCT about using CBT to reduce crime in Liberia).
1.2 Funding gap
ACTRA has told us (early 2025) they have a funding gap for 2025-2026 of around $500,000. With this money they plan to:
- Run multiple pilots of their adapted programme.
- Improve their monitoring and evaluation to collect more and better data.
- Fund an evaluation of their programme.
- Continue operations and implementation into 2026.
2. Potential cost-effectiveness of ACTRA
The basic premise of ACTRA is to adapt the CBT component alone of the intervention delivered by NEPI to the Latin American context. We can make slight tweaks to our model for NEPI to estimate how this broadly changes the cost-effectiveness.
We evaluated NEPI in 2024 (McGuire et al., 2024f). We assume the reader is familiar with our evaluation of NEPI. If they are not, we suggest they at least skim that report. Briefly, our model for NEPI worked this way: We combine (a) benefit on the recipients (criminal offenders) and spillovers on their households with (b) benefits on the wider society due to the reduction in crimes.
Our modelling is largely based on an RCT of the programme NEPI is based on (Blattman et al., 2023; and follow-up 10 years later in Blattman et al., 2023). There were four arms with participants nearly equally divided into cash only, CBT only, cash + CBT, and the pure control condition. We assume that ACTRA works identically to the CBT only arm in the RCT (instead of the CBT + cash estimates we used for our NEPI evaluation).
Here is how the figures change when we change the estimates we use:
- The recipient and household effects decrease from 3 → 1.04 WELLBYs because:
- The recipient effects in the RCT for the only CBT condition are smaller (0.08 SDs versus 0.19 SDs for the CBT and cash option).
- The average household size in Colombia is smaller. We reduced the non-recipient household size from 2.74 → 1.66The average non-recipient (i.e., -1 to not count the recipient) household sizes, as predicted linearly for 2024 using UNPD data, is 3.29 for Liberia and 1.99 for Colombia. However, for the NEPI evaluation, we used the non-recipient household size of 2.74 as reported in the Blattman et al. studies. This is 2.74/3.29 = 83% of the size calculated in the UNPD data, suggesting that the target population for NEPI has a smaller household, potentially related to the individuals' criminal activity. Therefore, we also adjust the Colombian non-recipient household size by 83% to account for this pattern in our analysis..
- The benefit to victims increased from 10.7 → 11.1 WELLBYs because the effect on the number of thefts is slightly higher for therapy alone (350 over 10 years instead of 338).
- The total benefit thus decreases from 13.6 → 12.1 WELLBYS (89% the size)
- However, the biggest driver is the cost reduction. The cash transfer is $300 of the $630 we estimate for the cost of NEPI, which means the naive cost of therapy is $330.
- Notably, other charities have demonstrated that therapy can be delivered for less than $50 (e.g., StrongMinds and Friendship Bench; albeit in AfricaAlthough these are costs for 6 sessions instead of 3*8 = 24 sessions. Do note that the costs of delivering psychotherapy is not only driven by the cost of the number of sessions, and the marginal cost of adding more sessions might be lower. Even taken in a naive linear calculation, at $50 for 6 sessions we would predict at most $200 for 24 sessions.), suggesting that $330 is very achievable and likely reducible.
- However, it is very plausible that operating costs in Latin America will be higher due to general higher incomes and salaries than Sub-Saharan Africa. Although, we also think that this cost, based on the RCT, will be lower in practice“Now that we have an estimate of the effect, let’s return to the program we’re discussing. The CBT and cash transfer for high-risk men in Liberia. The Blattman et al. studies claim a cost of $530 per person treated. NEPI doesn’t report a cost, nor have they appeared to scale and thus exploit the associated cost reductions. But in their emails, they mention increasing cash transfers from $200 to $300 to account for inflation. When I asked them if this implied the current costs were around $630 per person treated, they confirmed that. However, I’m sceptical. Rarely do the costs in practice reflect the RCT costs. I suspect that often costs in practice will be lower, especially at scale. At the time of writing this they are currently collecting more precise cost data (which I expect to represent lower cost per person figures), but they were unable to share anything further.” (McGuire et al., 2024f, p. 24)..
- Putting all of these into our existing model nearly doubles the cost-effectiveness from 22 → 37 WELLBYs created per $1,000 donated.
Naturally, there are large open questions about how these results may generalise to Colombia, which is a very different context from Liberia. For example, one of the reasons the results in Liberia were surprising is because it seemed that the group nature of the psychotherapy encouraged accountability. Prisons are well known to be criminogenic in part because they afford networking with more experienced criminals (Cullen et al., 2011). The fact groups worked in Liberia may not generalise to Colombia where different cultural factors could tip group dynamics back to an unfavorable equilibrium. Although, do note that departing from group settings would presumably increase the cost of therapy.
However, one reason why we are excited about ACTRA is that we think that they intend – and are capable of – providing the data necessary to answer these questions.
A key question underlying this modelling tweak is whether psychotherapy is driving the effects in Blattman et al.? Or is the cash transfer also a necessary component? We do believe psychotherapy is driving the effects.
NEPI chose the cash + CBT results because the CBT + cash results had more outcomes with statistically significant results. It is the safer, more robust choice. Here is a key passage from Blattman et al. (2023, p. 19).
“After 10 years, Therapy Only reduces the index of all 7 antisocial behaviors by 0.20 standard deviations (p = 0.055), and Therapy+Cash reduces it by 0.25 standard deviations (p= 0.016). Although the Therapy+Cash estimate is larger and more robust, we cannot reject that it is equivalent to the effects of Therapy Only (p = 0.61).”
Nevertheless, Blattman et al. (2023)’s explanation of why the combination of CBT + cash transfer seemed to be more effective than either alone does not suggest that cash transfers are necessary for the intervention to work. They explain on page 18 that:
“Receiving cash was akin to an extension of therapy, in that it provided more time for the men to practice independently and to reinforce their changed skills, identity, and behaviors. The therapy helped participants change their intentions, identity and behavior, and provided almost daily commitment and reinforcement. After eight weeks of therapy the grant provided some men with the cash they needed to maintain their new identity—to avoid homelessness, to feed themselves, and to continue to dress decently. Thus they had no immediate financial need to return to crime. The men could also do something consistent with their new identity and skills: execute plans for a business. This was a source of practice and reinforcement of their new skills and identity.”
Hence, the authors seem to think the effect was driven by psychotherapy, but the cash bought more time to practice, let the lessons sink in, and removed financial pressures to return to crime in the very short term. This suggests that ACTRA’s model with only CBT is plausible. Furthermore, if the cash transfer’s role is to buy time for practice, then it seems plausible that the structure of the psychotherapy-only intervention could be modified to directly include more practice. Giving cash may be a blunt way to provide more practice and accountability, but other sharper tools may be thought of for this role.
Conclusion
ACTRA seeks to implement the same design as NEPI, except without the cash transfer and in Latin America. We tweak our original NEPI model to predict the cost-effectiveness of ACTRA and find that it might be even more cost-effective, at 37 WELLBYs created per $1,000 donated. ACTRA has a good team and theory of change. For these reasons, we put forward ACTRA as an ‘honourable mention’, which might interest donors who are happy to fund very certain opportunities. At time of writing, ACTRA has a funding gap for 2025-2026 of around $500,000. Our recommendations change over time as we produce new research, so we advise interested readers and donors to stay updated by consulting our website and signing up the mailing list.