This blog post is based on: Christl, M. and Navarro Berdeal, S. (2026), “Tax Expenditures in Portugal: Fiscal Costs, Redistributive Effects and Cost-Efficiency,” JRC Working Paper JRC146412, Publications Office of the European Union.
Tax expenditures, such as deductions, exemptions, credits, and deferrals, are a key component of personal income tax systems. Typically introduced with social or economic objectives in mind, they carry a real price tag: forgone revenue that governments could otherwise spend on public services or transfers. Yet how well do they achieve their stated goals? And who actually benefits? In a new paper, we take a close look at Portugal, a country that stands out as one of the few EU Member States where personal income tax expenditures are associated with an increase in income inequality (Turrini et al. 2024). Our findings reveal a system that is fiscally costly and largely regressive, with one important exception that points the way toward reform.
A Hidden Budget of Nearly 40%
Using EUROMOD, the EU’s tax-benefit microsimulation model, and 2022 EU Statistics on Income and Living Conditions (EU-SILC) data, we assess the first-round fiscal and distributional effects of personal income tax expenditures in Portugal in 2023. Our analysis covers seven categories of provisions, those focused on family, work, housing, education, health, pensions, and the Net Minimum Income Guarantee (NMIG – Mínimo de Existência).
The results reveal a striking fiscal cost. Removing all modelled tax expenditures simultaneously would increase personal income tax revenues by nearly 40%, equivalent to EUR 717 million. Work-related provisions are the most expensive single category (EUR 138 million), followed by pension-related deductions (EUR 95 million), the NMIG (EUR 81 million), and family-related provisions (EUR 73 million). Housing, health, and education-related instruments are comparatively minor in fiscal terms.
This highlights the scale of public resources channelled through the tax code rather than the expenditure side of the budget, even though such measures largely escape the scrutiny that direct spending programmes face. As the IMF and others have argued, tax expenditures deserve to be evaluated alongside direct expenditures for a complete picture of public finances.
The Middle Gets the Most
The fiscal cost, however, is only part of the story. Equally important is understanding how these benefits are distributed across the income scale. If all tax expenditures are removed simultaneously, the proportional income loss is smallest for the poorest households (6.3% for the bottom decile), rises to a peak around the third and fourth deciles (roughly 10–11%), and then declines again towards the top (5.1% for the top decile). In other words, Portugal’s tax expenditures are neither strongly progressive nor strongly regressive in aggregate — they are predominantly a middle-income phenomenon and follow an inverted U-shape.
At first glance, this result may seem surprising, but it largely reflects how personal income tax systems operate. Deductions and non-refundable credits can only benefit households that pay income tax in the first place. Many low-income households fall below the tax-free threshold and are therefore largely excluded. At the upper end, while high-income households receive larger absolute amounts, these represent a smaller share of their overall income. As a result, the system provides the largest proportional support to those in the middle – not necessarily those most in need.
Looking at specific population groups, the self-employed and retirees are among the largest beneficiaries, reflecting the structure of work-related allowances and the pension deduction. Persons with disabilities stand out as particularly dependent on the system, drawing on both health-related and pension-related provisions, with a simulated total income loss exceeding 11% when all tax expenditures are removed.
Most Tax Expenditures Increase Inequality
Our instrument-level distributional analysis is where the results become most striking. The Portuguese personal income tax schedule is formally progressive, but the tax expenditures layered on top of it largely work against that progressivity.
Removing work-related tax expenditures would reduce the Gini coefficient by 0.33 percentage points, and removing pension-related deductions would reduce it by a further 0.30 percentage points. This means that these two categories — the most expensive in the system — are regressive: their presence in the tax code increases income inequality relative to what the system would produce without them. This is consistent with earlier cross-country evidence by Barrios et al. (2020), who found that Portugal’s pension-related tax expenditures are an EU outlier: despite their large fiscal cost, they generate negligible redistributive returns. Education-related provisions are also mildly regressive, while family, housing, and health-related instruments are broadly neutral.
The standout exception is the Net Minimum Income Guarantee. Removing the NMIG would increase the Gini coefficient by 0.38 percentage points, making it the only personal income tax expenditure in Portugal with a clearly progressive profile. Its removal would also raise the at-risk-of-poverty rate by 0.77 percentage points, confirming its role as the primary tax-based safety net for low-income households. The NMIG was substantially reformed in 2023, restructured from a floor on net income into a direct deduction from taxable income linked to the Social Support Index. It benefits taxpayers with gross incomes below a defined threshold and is designed explicitly to prevent low-income individuals from being pushed below a subsistence level by their tax obligations.
A complementary assessment using Portuguese tax administrative data for over 9 million taxpayers corroborates that the NMIG redesign was progressive, improving the Gini coefficient (Cost, 2023). It also corrected two flaws in the old rule. First, income just above the threshold had effectively been taxed at a 100% marginal rate, since the old rule capped disposable income at a fixed point regardless of how much a household earned beyond it; the new rule instead lets the benefit shrink gradually as income rises, so additional earnings are no longer entirely offset. Second, eligibility was previously assessed only on aggregated income, letting some high-income households qualify by leaving capital or rental income unaggregated; the reform closed this by basing eligibility on total declared income instead.
Notably, the NMIG has been excluded from harmonised cross-country assessments of EU tax expenditures on the grounds that it affects too large a share of the tax base to be considered a deviation from the benchmark tax system — a classification choice that, as our results show, substantially alters the picture for Portugal.
Cost-Efficiency: Getting the Most from Forgone Revenue
Beyond the direction of redistribution, we also assess how efficiently each instrument reduces inequality and poverty per euro of forgone revenue, defined as the change in the Gini coefficient or at-risk-of-poverty rate per EUR million of fiscal cost. Importantly, the scores are relative: they rank instruments against each other within this sample rather than against any absolute benchmark of adequacy. The NMIG ranks first on both dimensions: it achieves its progressive and poverty-reducing effects at a relatively modest fiscal cost of EUR 81 million. By contrast, pension-related tax expenditures rank last. A note of caution is warranted here: the Pensioner Tax Allowance was not designed primarily as a redistributive instrument but rather to provide income security in retirement — so measuring it against an inequality yardstick is somewhat unfair. That said, the instrument’s flat-rate structure means it disproportionately benefits those with higher pension income, and EUR 95 million in foregone revenue generates very little distributional return by any measure. Work-related and education-related provisions score similarly low. Family-related provisions occupy an intermediate position: their near-neutral effect on the Gini combined with a modest fiscal cost produces a relatively favourable inequality cost-efficiency score, though their impact on poverty remains limited.
The Case for Reform
Taken together, our findings reveal a fundamental tension at the heart of Portugal’s personal income tax expenditure system: the government devotes the largest fiscal resources to the most regressive TEs, while the most progressive and most cost-efficient instrument accounts for a comparatively small share of foregone revenue.
This points clearly towards a reform agenda, one that, notably, the IMF’s 2026 Article IV mission for Portugal has also called for. The most immediate implication in our analysis is to reallocate resources within the tax expenditure system itself – phasing out the costliest, most regressive in favour of targeted instruments like the NMIG. But we would go further. We believe that the deeper lesson is that redistribution is often better achieved through the expenditure side of the budget altogether, rather than by reshuffling resources within the tax expenditure system. The reason is structural. Deductions and non-refundable credits can only benefit households that owe tax in the first place – a limitation that holds even for the NMIG, the best-performing instrument in our analysis – which is precisely why the poorest decile gains so little from the existing system.
Direct transfers do not share this constraint: they reach households regardless of tax liability, are easier to means-test, more transparent and more straightforward to evaluate and monitor. Phasing out regressive provisions, particularly those related to work and pensions, and redirecting the freed fiscal space towards well-targeted direct transfers – such as the Social Integration Income (RSI – Rendimento Social de Inserção), Solidarity Supplement for the Elderly (CSI – Complemento Solidário para Idosos) for low-income pensioners, or an expansion of housing allowances for low-income renters – would likely deliver greater redistributive returns than any rearrangement of the tax code. Portugal already reports annually on the revenue cost of its tax expenditures; the next step is to ask whether that money would do more good spent directly.
Portugal is not alone in this challenge. Across the EU, tax expenditures often serve as an opaque substitute for more targeted social spending, with distributional consequences that are neither well understood nor regularly debated. Making them visible, and accountable, is a first step towards making them work better.