Subsidies have long been one of the main pillars of Tunisia's social policy, with the state providing basic food products and energy at subsidized prices. However, the sustainability of this system has become an increasing concern for public finances. In 2025, subsidy expenditure accounts for approximately 19.4% of total government spending, amounting to TND 11.593 billion, while the state budget records a deficit estimated at TND 11.02 billion.
One of the main limitations of the current system is that subsidized prices are available to all segments of society, regardless of differences in income and living standards. This study examines the distribution of subsidy benefits and identifies the groups that benefit from the current system using data from the 2021 Household Budget, Consumption and Living Standards Survey. The analysis goes beyond examining the distribution of subsidized consumption across households with different living standards by incorporating official fiscal costs to determine how each dinar spent by the state is distributed between household and non-household consumption. The results show that households with higher living standards benefit disproportionately from several subsidized products, particularly energy products. They also indicate that a substantial share of the fiscal cost of certain products is associated with commercial and institutional consumption, although measurement issues and differences in timing between the data sources may also contribute to this gap.
Based on these findings, the study develops three subsidy reform scenarios that link eligibility for subsidized prices to household living standards, with different degrees of protection and targeting across the scenarios. The recommended scenario maintains the diesel subsidy because of its role in transport, freight, and fishing, and the potentially significant inflationary effects of its removal, while redirecting subsidies for other products toward households with lower living standards and addressing non-household use of subsidized products. The recommended scenario is estimated to generate annual fiscal savings of approximately TND 5.4 billion. These estimates, however, represent potential fiscal space rather than the full economic impact of reform, as they do not capture all potential effects on inflation, production costs, employment, and economic activity, which fall beyond the scope of the present analysis.
Range across three scenarios: 4.71bn to 6.49bn DT a year, of which roughly 3.72bn DT in every scenario comes from recovering 95% of the residual, the part of each subsidy's cost currently going to businesses and institutions rather than any household. Source: author's calculations, EBCNV 2021 and Budget Citoyen 2025.
Share of each product's national household spending across the ten deciles of the per capita consumption distribution, from the poorest tenth of the population (D1) to the richest (D10). The dotted line marks 10%, the share each decile would receive if consumption were split equally. Bar colour deepens from D1 to D10.
Share of national household spending on the product, by decile · hover or tap a bar
Source: author's calculations from EBCNV 2021, household quantities annualised from the 7 day expenditure diary and weighted to the national population.
Share of national household spending on the product, by decile · hover or tap a bar
Source: author's calculations from EBCNV 2021, household quantities annualised from the 7 day expenditure diary and weighted to the national population.
For every product with a published official subsidy rate, this chart splits the total fiscal cost three ways: the bottom 60% of households, the top 40%, and the residual, the part of the official cost that cannot be matched to a household purchase observed in the survey. That residual is consistent with commercial and institutional use (bakeries, restaurants, transport fleets paying the same subsidised price as a household), but it can also reflect survey underreporting or the gap between the 2021 survey and the 2025 fiscal accounts, so it should not be read as directly observed non-household consumption.
Percent of total fiscal cost · hover or tap a segment
Source: author's calculations combining EBCNV 2021 household quantities with official Ministry of Finance subsidy rates, Budget Citoyen 2025. Coffee and electricity and gas are excluded here: no official per unit subsidy rate is published for either.
Share of each product's national household spending captured specifically by poor households and by extreme poor households. The dashed lines mark the population shares of each group. A product whose bar clears the dashed line is pro poor; one that falls short is not.
Percent of the product's national household spending · hover or tap a bar
Source: author's calculations from EBCNV 2021, using the survey's own official poverty and extreme poverty flags.
A government cannot observe a household's position in the national consumption ranking at the moment it needs to decide who is eligible for protection. So instead of ranking households by consumption, this paper builds a simple living standard score from a second EBCNV 2021 survey module on housing quality and durable goods, matched to the expenditure sample at a 100% rate. Items owned by almost everyone (a fridge, piped water, a washing machine) say nothing about relative comfort and are dropped entirely; items owned by a minority to a majority (a television, air conditioning, a computer) count for one point; genuinely rare, high-end items (a dishwasher, a freezer, a piped gas connection) count for two. The chart below shows exactly how common each item actually is today — the evidence behind which items count and which don't.
Population-weighted ownership rate · items above 75% are dropped from the score entirely
Source: author's calculations, EBCNV 2021 living-conditions module. Air conditioning (39.6%) and connected sanitation (60.7%) fall well short of the near-universal threshold, so both count as ordinary modern comfort, not luxury.
Despite using no consumption data whatsoever, the resulting score, from 0 to 17, separates households by true economic welfare about as cleanly as a far more complex model would: households it excludes show official poverty rates as low as 1.2%, against 19.8% among the households it protects, and consume roughly twice as much per person. Three cutoffs are defined directly on the score, each independently defensible, not picked to hit a target: a cautious Scenario A (protects 82% of the population), a recommended Scenario B (protects 69%), and an assertive Scenario C (protects 48%).
Population-weighted distribution of the living standard score, with the three scenario cutoffs
Source: author's calculations, EBCNV 2021. Scenario A (score ≥ 8), Scenario B (score ≥ 6, recommended), Scenario C (score ≥ 4).
Every scenario removes the subsidy from the households it excludes, and, separately, recovers 95% of every product's residual, the part of its official fiscal cost that never reaches any household purchase at all — treating only 5% as unrecoverable survey measurement error rather than genuine institutional use. Diesel is the one exception throughout: its own subsidy, including its residual, is kept in place for everyone in the recommended variant, since removing it risks second round price increases through transport, freight, and fishing that could hurt the same households a reform is meant to protect. The residual recovery, not the subsidy removed from excluded households, is the larger part of the total in every scenario below.
Million Dinars · recommended (diesel kept) variant · split into subsidy removed from excluded households and residual recovered from businesses
Scenario A: 4,711 Million Dinars (46.4% of the projected 2025 deficit). Scenario B, recommended: 5,400 Million Dinars (53.2% of the deficit). Scenario C: 6,488 Million Dinars (63.9% of the deficit). A mechanical variant that also recovers diesel's own residual would add 1.2 to 1.3 billion Dinars more — but is not what this paper recommends. Source: author's calculations, EBCNV 2021 and Budget Citoyen 2025.
For the recommended scenario, the chart below splits every product's official fiscal cost three ways: the households it protects, the households it excludes (one source of savings), and the residual recovered from businesses and institutions (the larger source). Diesel's residual dwarfs everything else, exactly why it stays out of the reform. Gasoline shows the largest excluded-household share of any product, confirming that comfortable households really do capture a disproportionate share of gasoline spending specifically.
Share of each product's official fiscal cost, by who receives it
Source: author's calculations, EBCNV 2021 and Budget Citoyen 2025. Households classified by the recommended living-standard scenario (B).
This paper's numbers cannot settle everything on their own, and we want to be direct about that rather than let it pass in silence. Recovering 95% of the residual means a pastry shop pays more for flour, a restaurant pays more for bread, and a shared-taxi driver pays more for bottled gas — and some of that cost will reach households through higher prices elsewhere. This is not a hypothetical extension of our recommendation; it is the larger part of the savings figure itself. We use the word agony deliberately: this is a genuinely hard, contested, and costly process, not a clean technical fix, and it will have real effects on inflation and employment we do not attempt to force into a number this household survey cannot support on its own. Taken abruptly and in isolation, a reform of this kind is genuinely risky. Taken gradually, and alongside a higher minimum wage, competitiveness reforms, and a serious effort against rent-seeking, it has, in other contexts, produced successful and durable outcomes — and we believe starting with public finance reform of the kind proposed here is a defensible first step on that path.
We also want to name the concern a reader may reasonably have: this can sound like an anti-social policy. We believe it is the opposite. A universal subsidy that a comfortable household, or a business, captures on the same terms as a poor one is not an equal way to spend public money; it only looks that way because it is visible and universal. Freeing this fiscal space for infrastructure, public health, and public education reaches the same underserved households through channels that are harder to see, but, in our view, considerably more consequential for their long-run welfare than the price of sugar or coffee. A more equal subsidy system and stronger public investment in health and education are, we believe, the same project.
