Why it matters to know who an anti-poverty policy will reach before approving it
Milagros Paniagua
20 November 2026
Key ideas
→ The problem. Spain’s regulatory impact assessment report (Memoria del Análisis de Impacto Normativo, MAIN) makes it possible to anticipate the consequences of regulatory decisions, but its existence does not guarantee that evidence actually feeds into policy design. The challenge is to use it while decisions can still be changed.
→ A specific case. A microsimulation of turning the personal income tax (IRPF) allowance for dependent children into a refundable tax credit makes it possible to estimate not only its cost, but also who would benefit, how much child poverty could fall and how it would interact with other benefits.
→ The opportunity. Ex ante evaluation can transform the way public policies are made: comparing alternatives, anticipating their impacts, identifying uncertainties and setting out from the start the conditions for evaluating their results later on.
Evidence: before or after deciding?
The quality of a public policy is determined, to a large extent, before it is approved. Decisions on the target population, eligibility requirements, amounts or the interaction with other interventions shape its results long before these can be observed. Yet a significant share of evaluation efforts is devoted to analysing policies once they have been rolled out, when changing some of those elements can be considerably harder.
In Spain, the MAIN is an instrument for anticipating the consequences of regulatory decisions. Its rules require an analysis of, among other things, economic and budgetary impacts and the effects on children, adolescents and families. But having this instrument does not guarantee that its analyses have actually contributed to the design of the intervention. The quality of regulatory impact assessment is precisely one of the dimensions examined by the OECD in its monitoring of regulatory policy in Spain. The challenge is not only to anticipate the effects of a decision, but to use that information while it can still be changed.
This question is especially relevant in redistributive policies, where the design of the instruments can be as decisive as the resources devoted to funding them. Two measures with an equivalent budgetary cost can have very different effects on poverty and inequality, depending on who has access to them, how their benefits are distributed and how they interact with the rest of the tax and benefit system. Having a redistributive goal does not, in itself, guarantee that resources reach those who need them most.
A family support instrument that does not reach every family
Child poverty is one of the structural problems of the Spanish economy. According to the 2024 Living Conditions Survey, 34.1% of children were at risk of poverty or social exclusion, compared with 24.6% of adults. Its persistence is due to many factors, among them the way public instruments for supporting families are set up.
In Spain, cash benefits aimed at low-income households coexist with tax benefits whose take-up depends on ability to pay. This architecture can leave gaps in protection between instruments which, taken one by one, pursue legitimate goals.
The analysis by Paniagua and Frutos (2026) makes it possible to examine this question empirically. Using the 2022 IEF-INE-AEAT Household Panel, the study estimates that the five main instruments of cash support for families with children mobilised around 8,700 million euros a year. Tax instruments accounted for 81%, and the allowance for dependent children made up 63.7% of the total.
Its distribution, however, was far from even. While households in the tenth income decile received 737 million euros through the allowance for dependent children, those in the first received just 55 million. The explanation lies in its design: as it depends on ability to pay, households without enough tax liability cannot take full advantage of the benefit.
The study identified around 464,000 households with children, 10.9% of the total, that were left without effective coverage: their incomes were above the eligibility thresholds for the minimum living income (IMV), but too low to benefit from the tax allowance.
The problem was not only how much money was devoted to supporting families, but how the existing instruments fitted together and which households were left unprotected between them.
What a simulation can tell us before legislating
Building on that diagnosis, the study simulated the transformation of the allowance for dependent children into a refundable tax credit. The reform would keep the amounts of the allowance, but would allow households without enough tax liability to receive the difference as a cash transfer.
The proposal did not start from scratch. It drew on the knowledge built up in earlier work, including the White Paper on Tax Reform, drawn up under the Recovery, Transformation and Resilience Plan (PRTR). Recovering and using that evidence is an essential part of ex ante evaluation: public decisions should not start from scratch every time. On that basis, the microsimulation made it possible to quantify the expected distributional effects of the reform using administrative records.
The estimated additional cost was around 1,910 million euros a year. A budgetary impact report could use that figure. But the analysis made it possible to answer other questions that are just as relevant for public decision-making.
Who would receive the resources? 54% of the incremental spending would be concentrated in the two bottom deciles of equivalised disposable income, while less than 2% would reach the tenth.
How many households would it reach? The share of households with children receiving some kind of cash support would rise from 89% to 100%. Just over two million households would receive the new credit, with an average amount of 937 euros a year, rising to 1,394 euros in the first decile.
What would happen to child poverty? Under the model’s assumptions, the child poverty rate measured at 60% of median income would fall from 23.8% to 22.0%. Using the 40% threshold, the reduction would be from 11.2% to 9.4%, equivalent to a relative decrease of 15.7%.
How would it interact with existing benefits? The exercise showed that the credit would complement the IMV in lower-income households and would cover part of the gap between social assistance and traditional tax benefits.
This last result is particularly telling. A reform does not operate in isolation: its effects depend on how it fits into the tax and benefit system as a whole. Two interventions with a similar budgetary cost can have very different distributional consequences. Microsimulation makes that architecture visible and makes it possible to anticipate where the effects of a regulatory decision are concentrated.
Evaluating before deciding also means comparing alternatives
The exercise above illustrates the potential of ex ante evaluation, but it also raises a further question: what would have happened if this analysis had been carried out while a real regulatory reform was being prepared? Its results could have served not only to quantify the effects of a proposal, but also to reconsider some of its elements.
For example, should the current amounts of the allowance be kept, or changed to strengthen support for certain households? Is it better to pay the credit through the annual income tax return or to set up advance payments? What rules on compatibility with the IMV and regional minimum income schemes would prevent part of the benefit from being cancelled out? How can it be ensured that households that do not file a return can actually access the credit?
Each alternative involves different design decisions and different consequences for coverage, the distribution of spending, incentives and administrative costs. In this sense, ex ante evaluation should not be limited to estimating the results of a single option. Its greatest value lies in comparing alternatives while they can still be changed.
Naturally, a static microsimulation cannot answer every question. This exercise estimates the immediate distributional effects under certain assumptions, but it does not incorporate possible behavioural responses, such as changes in labour market participation, nor does it fully anticipate problems with applying for and actually accessing the benefit. Nor is it a substitute for a later causal evaluation. That is precisely why a good ex ante analysis must identify both the effects that can be estimated and the uncertainties that remain open. Making them explicit helps to improve the design and to decide what information will need to be collected during implementation.
Deciding better while there is still time
Spain has an especially valuable information infrastructure for moving in this direction. Tax and Social Security administrative records, together with the microsimulation tools built on them, make it possible to anticipate consequences that can hardly be deduced from an aggregate budget estimate.
But ex ante evaluation is only the beginning. Once the policy is in place, monitoring must make it possible to check how it is being implemented, who is actually accessing it and what deviations from the plan are appearing. Later, ex post evaluation must analyse its results and, where possible, identify the effects attributable to the intervention.
Evaluation is thus a process of generating and accumulating knowledge that begins before the decision, continues during implementation and allows policies to be reviewed in the light of the evidence obtained.
The decision on whether or not to reform the allowance for dependent children belongs to the realm of democratically established public priorities. Evaluation does not replace that decision, but it can provide essential information on its foreseeable consequences, the alternatives available and the uncertainties that remain open.
Ex ante evaluation does not eliminate uncertainty: it helps to reduce it while there is still time to change decisions. Its real value lies not in justifying the policies we have already decided on, but in helping us to decide better.
Director of the Spanish State Agency for the Evaluation of Public Policies
View profile
