The Missing Horizon in Moldova’s 2027 Tax Reform

Why tax policy must be tested beyond the next budget year

Status of the draft: revised government proposal, publicly available as of 31 August 2026. Rates and implementation dates remain subject to the legislative process.

In June 2026, the Ministry of Finance published its first draft of tax and budget policy for 2027. Within weeks, central measures were being reconsidered under public and industry pressure. On 6 August, Prime Minister Vasile Tofan presented a substantially revised concept and returned it to public consultation.

The changes between the two versions were not cosmetic. The first draft did not simply adjust one rate. It proposed a broad redesign of labour taxation — a new income-tax scale of 7% and 15%, and a direct monthly payment of 500 MDL per employee and 200 MDL per child in place of personal allowances. These elements were withdrawn. The revised version returned to a simpler measure: raising the personal allowance from 29,700 to 40,000 MDL per year. In the VAT block, the initial move toward a uniform 20% rate across essentials was replaced by a differentiated approach — basic goods retained a reduced rate, HoReCa moved from the original 20% proposal to a proposed 12%, and energy taxation was redesigned around consumption thresholds.

Revising a draft is not, by itself, a problem. Governments should correct policy when evidence and feedback justify it. The issue is different: when measures with a direct effect on prices, household budgets and business costs require major correction within weeks of publication, and when whole architectural elements — an income-tax redesign, a replacement of the personal-allowance system — are proposed and withdrawn in succession, a basic question follows. Were the longer-term consequences of the first version fully calculated before it entered public debate?

The draft has also carried across a change of government. That makes the issue institutional rather than personal to any one minister. The question is not who proposed a particular rate. The question is whether Moldova is designing tax policy around the next budget year alone, or around the real economic effects that taxes create over several years.

What the revised draft proposes

The August 2026 package combines several important changes:

• VAT. The reduced 8% rate remains for basic goods including bread, selected dairy products, fresh fruit and vegetables, and medicines. A 12% rate is proposed for HoReCa (replacing the current 8%), accommodation, tourism, and agricultural products outside the protected list.
• Utilities. Natural gas VAT would remain at 8% for the first 150 m³ consumed monthly and rise to 20% above that. Electricity would remain at 0% for the first 100 kWh monthly and rise to 20% above that. Central heating would remain at 0%.
• Fuel excises. Diesel excise is scheduled to rise from 3,978.95 MDL/tonne in 2026 to 4,774.70 MDL in 2027, 5,729.70 MDL in 2028 and 6,875.60 MDL in 2029 — an increase of approximately 20% year on year.
• Personal income allowance would rise from 29,700 to 40,000 MDL per year. The proposed labour-tax redesign of the June draft (a 7%/15% two-bracket scale, and monthly per-employee and per-child payments) was withdrawn.
• Reinvested profit. The zero-rate facility for reinvested or undistributed profit would be extended through 2029. The threshold for access would rise from 100 million to 200 million MDL. Separately, the dividend-tax rate is proposed to increase from 6% to 8%.
• Real estate tax. Local authorities receive room to set rates within a proposed 0.05%–1% range from 2027.
• Tax-code reform. A broader rewrite based on the Estonian model is targeted for 2028.

Each measure has a defensible purpose. The problem is not that taxes should never change. The problem is that the package changes several major incentives at the same time, while its combined multi-year effect is difficult to see.

The regional contrast on fuel

Moldova, Romania and Ukraine responded to the 2026 fuel shock in different ways. The comparison does not assume identical conditions — Romania and Ukraine acted through temporary emergency measures during a market shock, while Moldova is legislating a multi-year excise path. Still, the direction of movement is instructive.

To translate the schedule into a per-litre figure, the National Energy Regulatory Agency (ANRE) applies a fixed conversion density of 0.845 kg per litre for diesel in its methodology for retail-price calculation (ANRE Board Decision No. 446/2021). At that density, one tonne corresponds to approximately 1,183 litres. On that basis, the excise component alone rises from roughly 3.36 MDL per litre in 2026 to about 5.81 MDL per litre in 2029 — a mechanical increase of approximately 2.45 MDL per litre in excise alone, before VAT and other price components (import or refinery cost, distributor margin, exchange-rate movements) are added on top. VAT is charged on a base that already includes excise, so the effect at the pump is proportionally larger than the excise change itself.

Fuel excise is not only a fiscal instrument. It affects transport costs, food prices, agricultural inputs, logistics and household purchasing power. Inflation reached 6.76% in May 2026, and the National Bank expects price pressure to remain elevated. In this environment, the relevant question is not simply how much excise the budget collects next year. It is how much of that projected collection is later offset by weaker consumption, higher operating costs, delayed investment and slower formal economic activity.

The Ministry’s case, in its own words

The Ministry of Finance has explained the proposed increase in agricultural VAT from 8% to 12% as an attempt to correct a distortion inside the production chain. Agricultural raw materials are taxed at 8%, while many processed products are taxed at the standard 20% rate. Grain, for example, is taxed at 8%, while flour is taxed at 20%. According to Finance Minister Victoria Belous, this difference shifts liquidity pressure onto processors and weakens the incentive to process agricultural raw materials inside Moldova.

The Ministry gives a similarly direct explanation for the diesel-excise increase. Moldova needs to move faster toward European Union minimum excise levels and needs more revenue for infrastructure. As Belous put it: “Nobody in Moldova likes the condition of our roads, but everyone likes the condition of roads in Europe. And if we see the result, we should understand what finances it.”

Each objective is legitimate. Domestic processing matters. Better roads matter. European alignment matters. But the statement also shows what remains missing from the public discussion. It explains why the state wants more revenue. It does not show, in a transparent and testable form, what happens across the full economic chain once that revenue is collected.

A completed impact assessment should show not only how much additional excise and VAT the budget expects to receive. It should show what happens to the cost of bread, freight services, agricultural production, domestic processing, household spending, business investment and the competitiveness of Moldovan producers. The Ministry’s argument is not disproved by these questions. It remains untested by a publicly visible multi-year calculation.

Two cases in short-horizon budgeting

Tax policy is often designed to close the coming year’s budget gap. But a tax base is not fixed. People and businesses adjust their behaviour when costs, incentives and risks change — where they invest, how they organise activity, how much they consume, whether they remain fully formal. When those adjustments are not modelled, the actual outcome can invert the projection.

Maryland, 2008. The state introduced a 6.25% surcharge on incomes above $1 million, projecting an additional $106 million in state revenue. In the first year, the number of returns declaring more than $1 million fell by 30% (from 7,898 to 5,529), and revenue from that group fell by 22%. Instead of collecting the projected $106 million, the state collected $257 million less than the previous year — a gap of $363 million against the plan. Part of this reflected the 2008–2009 financial crisis, and the case cannot be explained by the tax alone. That is precisely what makes the example relevant. The original projection treated the tax base as sufficiently stable to produce an additional $106 million. It was not. Whether the missing income reflected relocation, lower capital gains, changed reporting behaviour or a combination of all three, the fiscal forecast failed to anticipate how vulnerable the tax base was to changing conditions. The tax was allowed to expire at the end of 2010.

France, 2012–2015. A 75% marginal tax on incomes above €1 million was struck down by the Constitutional Council and restructured as an employer levy. It expired in 2015 after raising approximately €260 million in 2013 and €160 million in 2014 — sums that were small relative to the political attention, administrative complexity and economic uncertainty the measure produced. Economist Éric Pichet has calculated that the older Solidarity Wealth Tax (ISF), abolished in 2017, cost France roughly twice what it collected over its lifetime through capital outflow and reduced business activity.

Neither case is about the specific behaviour of wealthy taxpayers. Both are evidence of the same principle: a highly visible tax rate is not the same as a reliable long-term revenue source. When a change in policy shifts the incentive structure, taxpayers adjust — geographically, structurally, or by re-entering the informal economy — and the size of the adjustment is often larger than the projected revenue gain.

The Moldovan case Moldova has already run

Moldova does not need to import this question. It has a documented case of the same mechanism, in reverse, in one of its most cash-intensive sectors.

In October 2018, HoReCa VAT was reduced from 20% to 10%. The regime changed several times in the years that followed. By the time the revised 2027 proposal was published, the standard HoReCa VAT rate stood at 8%. According to sector data presented by the National Association of Restaurants and Leisure Venues (MĂR), between 2018 and 2025 the sector’s revenues grew by more than 300% and profits by nearly 276%.

These figures do not attribute all of the growth to VAT. The period also included post-pandemic recovery, general inflation, changes in tourism, and shifts in consumer behaviour. But they still matter: the sector became larger, more visible and more formal during the period of reduced VAT.

The likely mechanism is straightforward. In a sector with many small transactions and high cash turnover, a lower VAT rate reduces the incentive to hide sales. Card payments, reported payroll, formal investment and declared turnover become more economically viable at 8% than they were at 20%. This is not an argument for permanent sectoral privileges regardless of circumstance. It is an argument that a rate increase should be tested against the behaviour it may change.

The revised proposal of 12% is materially better than the initial 20%. That correction is not trivial. But the point is not whether a 12% HoReCa VAT rate is automatically wrong. The point is that the country first considered 20%, then moved to 12%, while withdrawing other central parts of the package. That sequence makes one question unavoidable: what was calculated before the first version was published, and what was only calculated after public reaction began? Before raising the rate under which the sector expanded and formalised, the government should show a multi-year calculation of how much additional VAT it expects to collect, how much turnover it expects to retain, how many jobs it expects to preserve, and what level of formal activity it assumes will remain in the system.

The Estonian model, read as a system

The draft’s own reference point deserves attention. The Ministry of Finance has stated that the new Tax Code, targeted for 2028, will be built on the Estonian model. Elements of that model are already embedded in the 2027 package — specifically, the zero corporate tax on reinvested profit, now extended to companies with turnover up to 200 million MDL.

This is the right direction. The Estonian design taxes distributed profits only, and leaves retained earnings untaxed — a structure introduced in 2000 and copied in modified form by Georgia in 2017. Its purpose is to reward capital formation inside the business and defer tax to the moment when capital leaves it.

But the Estonian model is not defined by a single rate. It is a system: stable rules, simple administration, predictable enforcement, and a clear reward for keeping capital inside the business, working alongside a high standard VAT (24% in Estonia since July 2025) and reduced rates for specific sectors like accommodation. Georgia’s successful transplant included complementary changes; the corporate-tax design alone was not sufficient.

The question for Moldova is therefore not whether it has copied Estonia. It has adopted an important component, and that is a positive development. The question is whether the different parts of the reform work together in time. Changes in consumption taxation and fuel excises affect households and operating costs immediately, from the first month of implementation. The investment incentive works differently: it benefits companies that already have profits to reinvest, meet the eligibility conditions, and are confident enough in the system to plan several years ahead.

A reform becomes coherent when its higher taxes, lower taxes, investment incentives and compliance rules are designed as one sequenced system. It becomes fragile when the pain is delivered immediately and the incentives arrive later, conditionally, for a narrower subset of participants — because the immediate pressure suppresses the very activity the later incentives are meant to stimulate.

What a completed impact assessment would contain

In most OECD tax jurisdictions, and increasingly in Romania and Estonia, a reform package of this scope is accompanied by a published fiscal impact analysis. It typically contains: elasticity estimates for the taxed base, sensitivity ranges around the central revenue projection, scenarios for behavioural response (informal migration, structural relocation, changes in consumption), and a multi-year projection of net effect against a baseline.

The June draft published by the Ministry of Finance contained headline revenue projections but did not, in its public version, contain any of the above. The August revision has repriced several measures but has not published such an assessment either. This is not the same as saying the reform is wrong. It is saying that its consequences have not been publicly modelled at the horizon on which they will be felt.

The missing element is not purpose. It is a transparent multi-year calculation showing whether the reform’s objectives reinforce one another or create offsetting costs elsewhere in the economy.

Moldova has, in its own recent history, a case demonstrating both the mechanism and the magnitude of what an unmodelled reform can miss. Rebuilding that model — before, rather than after, the next revision — is the least the process should include.