Without the EAEU: Examining the Four-Country Report’s Figures and an Inclusive Development Hypothesis for Armenia

A report by Belarus, Kazakhstan, Kyrgyzstan and Russia projects a 15.6% fall in Armenia’s GDP if its membership of the Eurasian Economic Union (EAEU) is suspended. Primary data from Armstat, the Central Bank of Armenia, Armenia’s State Revenue Committee and international official sources paint a different picture: in the worst year of the transition, from −8.3% to −0.5% of GDP (−3.9% in the illustrative baseline scenario); after 10 years (the annual level relative to 2025), from −5.0% to +5.7%. Where exactly Armenia ends up within this range will be decided not by geography or by the goodwill of its neighbours, but by how inclusive its own decisions turn out to be: whether rural areas, the regions and wage earners get a place in the new growth model.

The article is in two parts. In the first, we examine the report and recalculate its figures. In the second, we develop our own hypothesis of three factors (leaving the EAEU, TRIPP and open borders, and rapprochement with the EU) and answer 11 questions about what is needed for these factors to work for the whole country, not just for Yerevan. The calculations were performed with the help of Claude (Anthropic); the formulas, assumptions and synthetic input-output table are published in the attached Excel file, and the source data are in the open tables of Armstat, the Central Bank and the State Revenue Committee.

Part 1. A response to the four-country report

What the four-country report claims

The report “On the Possible Consequences of Suspending the Treaty on the EAEU in Respect of the Republic of Armenia” was prepared following a joint statement by the four presidents on 29 May 2026 and will be presented to the Supreme Eurasian Economic Council in December. Its main estimates (sections 4.2–4.3):

Source: the four-country report, sections 4.2–4.3.

We were able to reproduce the report’s published totals to one decimal place. What raises questions is something else: the model from which the GDP and inflation figures were derived is not disclosed, and the initial assumptions are the worst possible. The Armenian working group (the Ministry of Economy, the Central Bank and the Ministry of Finance) has not yet published its calculations. There is only an oral summary. On 21 October 2026, Minister of Economy Gevorg Papoyan said: “The negative effect of leaving the EAEU is not very large; it is estimated at −0.65 percent”, that in the report “everything is overstated by about 50 percent”, and that the change in the gas price costs 5.7% in one scenario and more in another (Aravot). The minister did not specify what these percentages refer to (GDP, year, horizon), and the methodology has not been published. For comparison: in our calculation, the direct loss from the gas price in the baseline scenario is about 0.4% of GDP (excluding the second-round effect), and the worst year in the baseline scenario is −3.9%.

Where the report goes wrong: five checks against primary data

According to the report, the sum of the declines in goods exports to the EAEU ($2.09 billion), services exports ($0.54 billion) and remittances ($2.10 billion) comes to $4.73 billion, almost exactly the stated $4.6 billion GDP loss. This coincidence suggests that the report may have equated gross flows with GDP, but the published materials are insufficient to verify its methodology and rule out such a calculation. Yet GDP is value added, not turnover.

1. Real exports fell by almost a third, yet GDP grew. According to Armstat (2025, 2026, “Socio-Economic Situation of the Republic of Armenia”, table 1.1.2), in 2025 real exports of goods and services fell by 30.5%, while GDP grew by 7.2% and household consumption by 10.7%. Growth continued in the first half of 2026: +5.2% and +6.7% by quarter. This does not refute the report’s estimate for a hypothetical exit scenario: GDP grew primarily on the back of consumption. But it does show that a fall in gross exports does not translate into GDP one for one.

2. The report’s “re-export” group is Russian gold and electronics. According to State Revenue Committee data (customs), in 2025 gold worth $2.27 billion (EAEU Commodity Nomenclature of Foreign Economic Activity code 7108) was imported from Russia, while telephones worth $789 million (EAEU Commodity Nomenclature of Foreign Economic Activity code 8517) and televisions worth $142 million (EAEU Commodity Nomenclature of Foreign Economic Activity code 8528) were exported to Russia. In total, re-exports amount to about $3.0–4.1 billion out of $8.3 billion of goods exports. Armenia retains only the margin from them. And this flow is already winding down on its own: in the first half of 2026, imports of Russian gold fell by 79% and exports to the United Arab Emirates by 74%, while the economy grew.

3. The report’s “remittances” are 85% commercial payments. The report takes $3.95 billion of incoming transfers from the EAEU. But according to Central Bank of Armenia data, non-commercial transfers from the EAEU, which is what families live on, amounted to $598 million in 2025, of which $569 million came from Russia. Personal remittances to Armenia from all countries according to the balance of payments were $1.45 billion. The Central Bank classifies the remaining $3.36 billion as commercial transfers by individuals: in all likelihood, these are mainly payments for transactions rather than personal remittances, including re-export transactions. If the report counts them as household income, it risks counting them twice, both as trade and as remittances. The difference between the Central Bank’s two figures lies in coverage: $598 million is non-commercial transfers through banks from EAEU countries only, while $1.45 billion is all personal transfers and earnings from all countries according to the balance of payments.

4. A fall in imports is not a loss of GDP. When goods are bought from another supplier, what is lost is the price difference, not the entire amount. Switching is already under way: in the first half of 2026, imports from Russia fell by 27%, while imports from Turkey rose by 39%, from Germany by 44% and from the US by 53% (customs data). For the first time, imports from Azerbaijan appeared: $17.9 million, mainly petroleum products.

5. Gas: the price in the report is an upper bound. According to customs data, Russian gas cost $174 per 1,000 m³ in the first half of 2026 and $174.5 in 2025 (2.23 billion m³ for $389.0 million). Reinstating the 30% export duty gives about $227; before 2013, Armenia paid $270. The report assumes $300 and itself calls this price “indicative”.

Alternatives to Russian gas. We have added them to the model. Direct purchases of Iranian gas on top of the current “gas for electricity” swap (0.48 billion m³ in 2025 at a customs value of $169 per 1,000 m³, customs data) make sense only if they are cheaper than Russian gas; they are constrained by US sanctions against Iran. Azerbaijani gas is possible only after a peace treaty and the connection of networks, and Azerbaijan links the signing of the treaty to a change in Armenia’s Constitution (according to the Speaker of the National Assembly, Ruben Rubinyan, Armenpress, 27 October 2026). In January–July 2026, Azerbaijan sold Turkey 5.648 billion m³ of gas for $1.478 billion, at $262 per 1,000 m³ (State Statistical Committee of Azerbaijan, Trend). This is more expensive than both the current $174.5 and the baseline $227, so Azerbaijani gas will not become a cheap source; it can work as a price ceiling. An illustration, not an expected saving: if Russia raises the price to $300 and Azerbaijani gas at $262 becomes a real alternative, the price for the entire volume of Russian gas (2.23 billion m³ in 2025) will not be able to exceed $262; that is about $85 million a year, and together with abandoning the more expensive Iranian gas and the second-round effect, about 0.3% of GDP. What volumes of Azerbaijani gas are actually available to Armenia is not yet known. A gas pipeline to Nakhchivan along the TRIPP route, with transit paid for in gas, as promised by Prime Minister Nikol Pashinyan (RIA Novosti, 27 May 2026), would yield little: Nakhchivan is already supplied from Turkey via the Igdir–Nakhchivan gas pipeline with a capacity of up to 2 million m³ per day, about 0.73 billion m³ a year (Turkish Ministry of Energy). Even if this entire volume were routed through Armenia and it received 5% (the amount of gas Georgia receives for transit through the South Caucasus Pipeline; International Monetary Fund, Working Paper 04/209), that would be about 37 million m³, less than 2% of gas imports. Bottom line: the gas alternatives barely change the baseline estimates and shift only the optimistic bound, by 0.1–0.2 percentage points.

Customs duties: the EAEU customs-duty system already reduces Armenia’s budget revenue. In 2025, Armenia collected AMD 124 billion in import duties and, under EAEU rules, transferred AMD 122.5 billion to the other member states, of which AMD 104.3 billion went to Russia; under the distribution formula, it received back AMD 69.5 billion (of which AMD 1.5 billion was its own money). The Ministry of Finance itself states the result: the budget fell short by AMD 54.5 billion (about $140 million, 0.5% of GDP), and it attributes this to sanctions against Russia: the duty is paid in Armenia on goods that go on to Russia (2025 budget execution report, “Revenues” section and note on import duties, ARLIS). This exceeds annual budget spending on agriculture (AMD 43.1 billion). After leaving, Armenia will keep the duties on its own imports but will lose receipts from the distribution; the net result for the budget depends on how much re-export activity survives. In our model, this channel does not affect GDP: it is a redistribution between budgets.

The nuclear power plant, grain and debt: dependencies the model does not count. In 2025, the nuclear power plant generated 29.1% of Armenia’s electricity, and gas-fired thermal power plants 33.6% (Armstat, “Socio-Economic Situation of the Republic of Armenia”, January–December 2025, section 1.2.1); nuclear fuel worth $22.2 million was imported entirely from Russia (customs data). The budget is financing a further extension of the second unit’s operating life with a loan: AMD 20.1 billion over 10 years (Government Decision No. 995-Ա of 9 July 2026, ARLIS). In 2025, Armenia imported 333,000 tonnes of wheat worth $83 million, 99.4% of it from Russia, plus 50,000 tonnes of flour, 96% from Russia (customs data), whereas the entire domestic harvest of cereals and pulses was 253,000 tonnes (Armstat, section 1.2.2). Finally, debt: in 2025, Armenia paid Russia $45.8 million on government loans (of which $19.5 million was on the loan for the restoration of the nuclear power plant and the rest on loans for military products) and $33.3 million to the Eurasian Fund for Stabilisation and Development (2025 budget execution report, note on government debt, ARLIS). As of 30 September 2026, debt to the Russian government stood at about $79 million (1.8% of the government’s $4.3 billion in external loans), and to the Eurasian Fund for Stabilisation and Development at about $406 million (9.4%) (Ministry of Finance, monthly government debt bulletin, September 2026). The Fund was established by a separate agreement on 9 June 2009 (then as the EurAsEC Anti-Crisis Fund), and its members include Tajikistan, which is not part of the EAEU (the Fund’s founding documents). In the model, these dependencies do not affect GDP: EAEU membership does not determine them, but this is precisely where the risks that need to be insured against in advance are concentrated.

What the report gets right. There are real vulnerabilities: cognac (83% of exports go to Russia), fruit and vegetables (95%), tourism (40.7% of inbound tourists in 2025 came from Russia, Armstat), gas and Russian capital. The last point is important: in the first half of 2026, Russia accounted for AMD 53.5 billion of Armenia’s AMD 113.5 billion in net direct investment, almost half (Armstat, “Socio-Economic Situation of the Republic of Armenia, January–August 2026”, section 4.2).

Sources: Armstat, “Socio-Economic Situation of the Republic of Armenia” (January 2026 and January–July 2026), table 1.1.2; Central Bank of Armenia, individuals’ transfers by country and personal remittances according to the balance of payments; State Revenue Committee, foreign trade statistics.

Our recalculation: what leaving might actually cost

If the flows are converted into value added and the initial mitigating factors are taken into account, the estimated impact in the worst year of the illustrative baseline scenario is about $1.1 billion, or 3.9% of GDP.

BD calculation with the help of Claude; data from Armstat, the Central Bank of Armenia, the State Revenue Committee, the EAEU report; model in Excel

The method is simple and verifiable: for each channel, the flow is multiplied by its change and by the share of value added, and then the second-round effect is added. All 53 parameters, with sources, are set out in Excel (the model file is published together with the article), and any reader can substitute their own figures.

Source: BD calculation with the help of Claude; all parameters and references to official data are in the model file (Excel) attached to the article; the report’s figure is from the four-country report.

Even if we accept the report’s key percentage declines for exports to the EAEU, remittances and gas (the pessimistic scenario), our model estimates an impact almost half as large as the one claimed: −8.3% of GDP in the worst year versus −15.6%. This is the result of a different set of assumptions, not a proven size of the damage. In the comparable cases we found, the decline in a single year rarely exceeded 4–6% of GDP; the exceptions were war (Ukraine in 2014–2015) or the collapse of a system (the Baltic states in 1992–1993). Our calculation, like the report, starts from 2025 flows, meaning that it also includes Russian bans already in place in the effect of leaving. What the calculation does not capture: the dram exchange rate, inflation, the budget and the worst-case scenarios, such as a complete cut-off of gas or a blockade of Upper Lars.

Part 2. Our hypotheses: three factors and inclusive modernisation

Three factors and a fourth that decides the outcome

Our hypothesis: leaving the EAEU while building TRIPP, opening the borders with Turkey and Azerbaijan and moving closer to the EU gives Armenia a difficult transition and a cautiously positive result after 10 years, but only if investment really does arrive after peace and the switch to EU rules: without it, the baseline result after 10 years is about −1.0% of GDP. But on their own, the three factors deliver little.

Source: BD calculation with the help of Claude, baseline scenario; the breakdown by channel and the source of each parameter are in the model file (Excel) attached to the article.

Why so modest. Direct transit along the 43 km of TRIPP will bring $4.5–17 million a year (calculation by former State Revenue Committee chairman David Ananyan, Vectors). According to a World Bank estimate cited by the US International Development Finance Corporation (DFC), improved regional connectivity will give Armenia savings of up to $100 million a year and about +0.5–1% of GDP; in our assessment, this is achievable only if the route is connected to the domestic network (DFC). Exports to the EU grew by 80% in the first half of 2026, but according to customs data almost all of the increase consists of copper concentrate to Bulgaria (from $29 million to $177 million) and molybdenum to Belgium (from $39 million to $92 million). A reorientation of agricultural produce is not yet visible in the figures.

The fourth factor is investment following peace and closer ties with the EU. In the model, the largest item after 10 years is not transit or preferences, but investment following peace and the switch to EU rules (+$0.44 billion in the baseline, up to +$1 billion in the optimistic scenario). It is this assumption that determines the sign of the result: without it, the baseline result after 10 years is −1.0% of GDP. It rests on an estimate by the Geneva Centre for Security Policy (GCSP): an increase in direct investment of at least 20% after peace yields +5–6% of potential GDP (GCSP); we take a third of the lower bound so as not to double-count the gains from the borders and the trade regime with the EU. At the same time, the peace treaty has not yet been signed. What determines whether investment arrives and who gets the growth is the subject of the following sections.

How the value-added shares and the multiplier were calculated. Armstat has published an input-output table only for 2006 (Input-Output table, 2006), so we constructed a synthetic table for 2025: the links between industries are taken from the 2006 table, while each industry’s shares of intermediate consumption and value added are taken from the 2025 national accounts (Armstat: output, intermediate consumption, value added); the import share of each product is based on customs statistics and the 2025 balance of payments, excluding re-exports. The result: of every dollar of traditional exports to the EAEU, 0.72 dollars of value added remains in Armenia; of every dollar of services exports, also 0.72; but of every dram of household consumption, only 0.48 dram: the rest goes on imports, directly or through the supplier chain. The second-round multiplier is 1.36: household consumption is 0.73 of the country’s disposable income, and the budget takes 25% of GDP (2025 budget execution report). This is an upper estimate: it does not take into account that the Central Bank and the dram exchange rate partly dampen the second-round effect. A robustness check (without updating the structure to 2025, and with imports 20% higher or lower) gives 0.68–0.76 for exports, 0.42–0.54 for consumption and 1.30–1.42 for the multiplier. The calculation and all source data are in the model file (the “Input-Output” sheet).

Caveats to the model. This is not a forecast but an illustrative calculation. The value-added shares, the domestic share of consumption and the multiplier are calculated from a synthetic input-output table, but the links between industries in it are those of 2006. The declines in exports, remittances and services and the investment effect are scenario assumptions, not data-based estimates; the baseline scenario is a set of average assumptions, not the most likely outcome. “After 10 years” means the annual level of the effect relative to 2025, not the sum of losses or gains over ten years. Re-exports are identified by product group (machinery and equipment, precious metals, jewellery), not by confirmed import-and-resale transactions, and their 3–8% margin is an assumption. Gas and the rising cost of imports are taken into account directly, without the response of demand, subsidies or pass-through to prices. Sensitivity check (the “Sensitivity” sheet in the model file; the values were calculated by a separate script, sens.py, whose text is on the “Scripts” sheet, and they are not recalculated when the parameters are changed in Excel): if one parameter is changed at a time (this check does not show a simultaneous deterioration of several conditions; that is described by the pessimistic scenario: −8.3% in the worst year and −5.0% after 10 years), the worst year stays within the range of −3.0% to −4.7% of GDP; a multiplier of 1.0 gives −3.0%, and 1.53 gives −4.4%. After 10 years, the result is fragile: with an investment effect of 0.5% of GDP instead of 1.5%, the baseline scenario turns negative (−0.5%), and with 3.5% it reaches +2.5%. Information technology is not modelled separately. Exports of telecommunications, computer and information services in 2025 were $1.775 billion, about 6% of GDP (Central Bank balance of payments), but neither the Central Bank nor Armstat publishes a breakdown by country, so it is unknown what share depends on the EAEU market. In the calculation, these services are included in total services exports to the EAEU; the four-country report estimates the decline in exports of information and communication technology services to the EAEU at 32.9% but does not disclose the base. We do not take into account the decline in services exports to third countries that the report assumes (−12.8%): the report does not explain the mechanism of such a decline. For a strong conclusion, the scenario coefficients need to be estimated from data, and Armstat needs to publish an up-to-date input-output table.

The inclusive modernisation hypothesis

Investment and new markets will become growth for the whole country only if those who are currently left out of modernisation are included in it: rural areas, the regions, small producers and wage earners. Otherwise, the gains will be concentrated in Yerevan, and the losses in the agricultural sector and in migrant families.

The starting point is two Armenias:

• Yerevan accounts for 65.8% of GDP (2023; 57.4% in 2021, meaning that concentration is increasing; Armstat, “National Accounts of Armenia, 2025”, table 6.2), while agriculture accounts for 7.9% of GDP (2025) and 17.6% of employment (2024; Armstat, “Statistical Yearbook 2025”, table 3.16; the FAO figure of 28% is outdated (FAO).
• Poverty stands at 21.7% (2024): 26.1% in rural areas and 16.9% in Yerevan (Armstat, “Poverty and Social Panorama of Armenia, 2025”).
• The wage gap between Yerevan and the poorest marz (province) is 2.3 times: AMD 328,911 versus AMD 144,908 in Gegharkunik (2024; Armstat, “Marzes of the Republic of Armenia and Yerevan City in Figures, 2025”).
• Agriculture has shrunk by 10% over 8 years (Armstat, “National Accounts of Armenia, 2025”), and the cattle herd fell from 559,500 to 468,000 head over 2022–2026 (−16%, Armstat, results of the livestock count as of 1 January 2026), while over the same period construction has been growing by 20% a year (Armstat).

The six areas of the hypothesis are examined below:

1. Who takes part in decision-making: elites, external players, trade unions.
2. Money for development: debt, targeted bonds, infrastructure instead of social programmes.
3. The Yerevan imbalance: construction, quality of life, lending to households.
4. The regions and labour: less complex manufacturing and rural areas.
5. EU standards as a tool: exports, import substitution, market protection.
6. Corridors and Central Asia: Armenia as a regional hub.

1. Who takes part in decision-making

Neither Armenian elites nor external players have yet made inclusiveness a goal: decisions are built around geopolitics, infrastructure and macroeconomic stability, while rural areas, the regions and wage earners have little involvement in them.

Institutions for participation exist but work only formally:

• According to the 2025 assessment by the V-Dem Institute (Varieties of Democracy, University of Gothenburg), Armenia is the only electoral democracy in the region, while Georgia and Azerbaijan are electoral autocracies. Armenia’s Liberal Democracy Index is 0.37 (86th out of 179), Georgia’s is 0.28 (101st) and Azerbaijan’s is 0.05 (164th). But since 2020, Armenia has been undergoing re-autocratisation: the index has fallen by 0.249, the sixth-deepest decline in the world among such episodes; the authors link it to a personalist style of governance and restrictions on freedom of expression and civic space (V-Dem Institute, Democracy Report 2026, table 5 and table A2).
• Decisions of the tripartite commission (government, employers, trade unions) are not binding (EaP CSF).
• The e-draft.am portal for public discussion of draft legislation has been operating since 2017 and is genuinely used: 10,085 drafts have been published on it, citizens and organisations have submitted 18,809 proposals, and the site receives 450,000–820,000 visits a month. The weak point is feedback: government bodies failed to respond on time to 4,057 proposals (22%) (e-draft.am portal statistics, data as of October 2026).
• The 18 sectoral trade unions that make up the Confederation of Trade Unions of Armenia had 188,280 members as of 1 January 2026 (Confederation of Trade Unions of Armenia), about 16% of the 1,177,000 people employed in 2025 (Armstat, labour force survey; BD calculation). Armenia scores 3 out of 5 on the International Trade Union Confederation’s Global Rights Index (ITUC Global Rights Index 2025) (as reported by HAMK).
• The minimum wage has been frozen at AMD 75,000 since 2023, although in 2021 the government promised AMD 85,000 by 2026 (Government programme for 2021–2026, ARLIS; Law on the Minimum Monthly Wage, ARLIS). With the average wage at AMD 303,140 (2025), this is less than 25%.
• Trust between people is low: only 23% believe that “most people can be trusted” (Caucasus Barometer 2024). And without trust, neither cooperatives, nor trade unions, nor consolidated export shipments can work.

External players build their agenda around routes and large projects. The US holds 74% of the TRIPP joint venture and Armenia 26% (US International Development Finance Corporation, public summary of the TRIPP project), plus a $201 million TRIPP+ fund for the countries of the Trans-Caspian route (see section 8). The TRIPP framework agreement (text on the website of the Armenian Ministry of Foreign Affairs) was signed on 1 and 4 June 2026; the government submitted it to parliament on 29 September, and on 28 October the National Assembly ratified it by 63 votes to 32 (Armenpress). The agreement has not yet entered into force: the parties still have to exchange notes (Art. 11), and, according to Foreign Minister Ararat Mirzoyan, the joint venture’s charter and shareholders’ agreement have yet to be negotiated, so the joint venture and the tax benefits are, for now, terms of the agreement rather than an operating project. Under the agreement, Armenia ensures at its own expense that the land along the route sections is state-owned and free of encumbrances, and then grants the joint venture exclusive rights for 49 years, with a possible extension of a further 50 years (in which case Armenia’s stake would rise to 49%); the tax benefits in Armenia cover dividends paid by the joint venture to the American shareholder (if it is ultimately owned by the US government), dividends of its project companies, capital gains on the sale of their shares, and the transfer of land rights to the joint venture (Arts. 3, 6 and 9). The Ministry of Finance warns that the agreement may increase budget expenditure and reduce budget revenue, but that their size cannot yet be estimated (National Assembly, draft law K-18-29.09.2026-ԱՀ-011/0 and ministries’ opinions). The Ministry of Justice is already preparing amendments to six laws for the agreement: companies established under a ratified international treaty with a state stake of 25% or more are to be exempted from the laws on procurement, on protection of competition and on public-private partnership (draft on e-draft.am, public consultation until 31 October 2026). The EU offers investment of up to €2.5 billion and the “Resilient Syunik” programme. The exception with an inclusive component is the European Neighbourhood Programme for Agriculture and Rural Development (ENPARD, €25 million for cooperatives), but it has ended. Russia exerts influence through the market, gas and ownership.

What the hypothesis proposes:

• make decisions of the tripartite commission mandatory for consideration and introduce a formula for indexing the minimum wage;
• include inclusiveness indicators (employment in the marzes, rural incomes) in the conditions of external financing, as was done in the EU cohesion funds for Poland and the Baltic states. There, 6% of the funds of each programme formed a “performance reserve” and were released only if the country achieved its stated targets, while the money was allocated among regions according to their GDP per capita (Regulation (EU) No 1303/2013, Arts. 20–22 and 90, Annex II; the current rules for 2021–2027 are Regulation (EU) 2021/1060);
• give rural and sectoral associations a place in the working groups on approximation with the EU: it is they who will have to meet the new standards.

Comparison: in Sweden, Denmark, Norway and Finland 51–66% of employees are trade union members (2024), while in Estonia, Lithuania and Latvia the figure is 6–10% (2021–2023) (OECD, trade union database). The Baltic states grew rapidly after joining the EU but went through mass emigration; the link between this and weak social dialogue is our hypothesis. For Armenia, with its diaspora, this is a direct warning.

2. Money for development: debt, targeted bonds, infrastructure

Armenia has no national infrastructure bonds for three reasons: debt is at the threshold of the fiscal rule, borrowing is expensive, and the state cannot manage to spend even the capital expenditure already allocated.

Infrastructure instead of social programmes? This is a false choice. Capital expenditure in January–August 2026 accounted for only 12.6% of total budget expenditure and fell by a quarter, while social benefits and pensions rose by 27% (Armstat, table above). The bottleneck is the capacity to prepare and manage projects, not social programme money. Our hypothesis: infrastructure that changes the quality of life in the marzes (water, roads to markets, cold storage, kindergartens) is in itself the best social programme. But it must be financed in a way that does not cross the 50% threshold.

What can realistically be done:

1. Targeted bonds for residents of Armenia for specific projects with public reporting. There is money inside the country: household deposits in banks amount to AMD 3.9 trillion (about $10 billion), of which term deposits are AMD 2.6 trillion, as of September 2026 (Central Bank of Armenia, deposits by sector). Another source is money that goes into gambling: in 2024 bets in casinos and online totalled AMD 7.24 trillion and winnings AMD 7.09 trillion, meaning players lost about AMD 142 billion, a fifth of annual budget capital expenditure (State Revenue Committee data provided at the request of Factor.am: casinos, AMD 106.4 billion in bets and AMD 97.5 billion in winnings; online games, AMD 7.13 trillion and AMD 7.0 trillion). A bond for a clear-cut project in one’s own marz, with a yield above that of a deposit and below the 7% on Eurobonds, benefits both the depositor and the budget, while the debt remains in drams and inside the country.

Savers have little choice: as of 31 December 2024 Armenia had 725 registered open joint-stock companies, whose shares any investor can buy, and 3,156 closed ones, compared with 100,066 limited liability companies (State Register Agency of Legal Entities, 2024 report). In November 2025, 12 new closed joint-stock companies were registered, no open ones, and 789 limited liability companies (monthly report for November 2025). That is why a targeted bond is one of the few savings instruments other than a bank deposit.

2. An exemption for self-financing projects in the fiscal rule (with the approval of the International Monetary Fund), if a project generates a revenue stream: water charges, transit, leasing of logistics centres.
3. A project factory at the Ministry of Finance: a package of ready feasibility studies for the European Bank for Reconstruction and Development, the European Investment Bank and the EU’s Global Gateway programme. Money goes to those who have projects ready.

Comparisons. Israel Bonds have raised more than $46 billion over 70 years; India raised $4.2 billion and $5.5 billion in 1998 and 2000 (MPI). Failures are instructive too: in 2011 Greece failed to raise the planned $3 billion, and Ethiopian bonds for the GERD dam yielded only about 1 billion birr from the diaspora over 14 years. Conclusion: the diaspora buys trust and transparency, not patriotism. In 2024 Kenya placed tax-free infrastructure bonds that were 4.1 times oversubscribed (Central Bank of Kenya, results of auction IFB1/2024/8.5).

3. The Yerevan skew: construction, quality of life, household credit

Growth in recent years has been concentrated in Yerevan real estate and consumer credit. This adds to GDP now but does not create an export base that could help the country survive leaving the EAEU.

Construction. The volume of construction rose by 20.2% in 2025 (AMD 810.7 billion), by 22.9% in January–April and by 24.7% in January–August 2026 (Armstat, “Socio-Economic Situation of the Republic of Armenia”; see also Asian Development Bank). The apartment price index in Yerevan has almost doubled since 2018 (Central Bank of Armenia). The driver is the income tax refund on mortgages. In Yerevan it was abolished from 2025, in the neighbouring marzes it will disappear from 2027 and in the rest from 2029 (Grant Thornton). The Governor of the Central Bank of Armenia has acknowledged signs of overheating, and banks have built up provisions of about $500 million against real estate risks.

How long this can last. Our forecast: 2–3 years, no more. The pace is set by the tax benefit (its abolition shifts demand to the marzes until 2027 and 2029), inflows of money from Russia, and credit. All three factors will weaken if Armenia leaves the EAEU, and construction will be one of the first sectors to slump.

Quality of life and investors. The Prime Minister ordered the creation of a crisis group on traffic congestion in Yerevan (government session of 16 October 2025). Water losses in the Veolia Jur system were 65.5% in 2025, compared with 67.3% in 2024 (Public Services Regulatory Commission, 2025 report, p. 54). We found no investor surveys on the impact of quality of life.

Credit to households and businesses. In drams, commercial banks have lent 2.8 times more to households than to private non-financial corporations, that is, to businesses: AMD 4.01 trillion versus AMD 1.41 trillion at the end of August 2026. Including foreign-currency loans, the gap narrows to 1.1 times (AMD 4.23 trillion versus AMD 3.70 trillion; Armstat, “Socio-Economic Situation of the Republic of Armenia”, January–August 2026, section 3.1, commercial bank loans by sector). The structure is worrying:

• household debt is growing three times faster than earnings: loans to households grew by 19.6% in 2025 and by 21.9% in the year to September 2026 (Central Bank of Armenia, loans by sector), while the average monthly nominal wage in December 2025 and in August 2026 was only 6.5% higher than a year earlier (Armstat, “Socio-Economic Situation of the Republic of Armenia”, section 1.4.2: January–December 2025, January–August 2026; BD calculation);
• 62% of business loans are in foreign currency: AMD 2.37 trillion out of AMD 3.82 trillion as of September 2026 (Central Bank of Armenia, loans by sector).

What this threatens if Armenia leaves the EAEU. If remittances and capital from Russia decline, household debt burdens will become a channel that amplifies the shock. In the model, this risk sits in the multiplier. Comparisons: since 2019 Georgia has capped monthly loan payments by individuals at 20–60% of their income (National Bank of Georgia); in Hungary in 2009 foreign-currency loans to households and companies amounted to 47% of GDP, of which 30.6% of GDP was in Swiss francs, and borrowers were paying for them for a long time (Bruegel). In Armenia consumer loans are in drams, so the Hungarian scenario is not a threat, but the Georgian limits are worth considering.

4. Regions and labour: low-complexity manufacturing and trade unions

Cheap labour is available in the marzes, but it does not become an advantage without three things: training, logistics and an anchor customer. That is precisely why textiles are still concentrated in and around Yerevan.

What exists now:

• light industry (textiles, clothing, leather) employed 11,981 people in 2025, compared with 12,861 in 2024, with output of about AMD 54.8 billion (Armstat, “Main Indicators of Industrial Organisations…, 2025”); more than 90% of those employed are women, and vocational training is lacking, according to industry representatives (JAMnews);
• clothing exports were $280 million in 2025, compared with $328 million in 2024 (State Revenue Committee, commodity groups 61–62);
• Italian outsourcing: clothing exports to Italy were $61.5 million in 2025, compared with $50.0 million in 2024 (State Revenue Committee, groups 61–62).

Why companies do not move to the marzes. The wage gap with Yerevan is up to 2.3 times. The industry itself names shortages of staff and vocational training as the main obstacles. At the same time, construction and agriculture are increasingly recruiting workers from India (in 2025, 2,962 Indian citizens received temporary residence permits, of whom 1,959 for work; there is no breakdown by sector, Migration and Citizenship Service). In other words, the problem is not the number of people but the fact that factory work at the minimum wage of AMD 75,000 (about $195) cannot compete either with remittances or with seasonal work in Russia.

Hypothesis. Leaving the EAEU and the rising cost of working in Russia (a work patent in Moscow in 2026 costs 10,000 roubles a month: the Moscow coefficient of 2.9323 was set by Moscow City Law No. 41 of 1 November 2025) will bring part of the workforce home. This is a window for regional factories, if the country prepares for it: vocational training centres run with businesses, production premises with ready infrastructure in Gyumri, Vanadzor and Syunik, and a subsidy tied to employment in the marz. Gyumri, moreover, will be on the restored Kars–Gyumri line.

Trade unions are not an obstacle but a precondition. Under the EU directive on corporate sustainability due diligence, European brands are obliged to check working conditions at their suppliers. The 2022 strike at Sartex over unpaid overtime (AGH) is an example of a risk that will become a commercial one. A functioning trade union makes an Armenian factory more reliable for a European customer.

Comparisons:

• North Macedonia. The technological industrial development zones (TIDZ) employ more than 18,000 people with an average wage of €1,100, 10% above the national average (FEZ); textiles employ about 30,000 people.
• Moldova. Dräxlmaier operates two plants in Bălți employing more than 4,000 people and a third in Cahul employing more than 1,000 (DRÄXLMAIER, data as of March 2023). Electrical equipment has become Moldova’s largest export item to the EU: $497 million.
• Georgia. Turkey’s Kartli Textile in Rustavi provides 4,000 jobs, 90% of them for women (Enterprise Georgia, a government agency). With an open border, Turkish textile manufacturers are natural investors for Shirak.
• Serbia is a cautionary tale. Subsidies of up to €10,000 per job did not create additional employment beyond the direct jobs and partly displaced local employment (University of Greenwich study). They were abolished in 2018. The lesson: pay for training and infrastructure, not for jobs.

5. The countryside: who will persuade farmers to work together

What will persuade them is not campaigning but money for a ready shipment: a buyer who pays on time and pays more for volume and standards. Wherever cooperation has worked, it was launched by a sales market, not by a law.

Diagnosis:

• The cattle herd shrank from 559,500 head (1 January 2022) to 468,000 (1 January 2026), milk production from 590,900 tonnes (2023) to 565,500 tonnes (2025), and the sown area from 219,900 hectares (2023) to 203,600 hectares (2025) (Armstat: livestock census results, “Socio-Economic Situation of the Republic of Armenia”, “Statistical Yearbook 2025”, Table 13.5); in January–June 2026 agricultural output fell by 11.9% and crop production by 23.4% (Armstat, “Socio-Economic Situation of the Republic of Armenia”);
• less than half of arable land is sown: 208,000 hectares out of 441,600 hectares in 2024 (47%), compared with 337,500 out of 446,700 (76%) in 2015 (Armstat, Sustainable Development Goal indicator 2.4.1); according to the results of the 2026 sowing campaign, the sown area rose to 216,800 hectares from 196,400 hectares a year earlier (Armstat, sowing results; BD summary “Armenia in Figures”);
• the 2024 agricultural census counted 387,257 household farms, of which 334,741 had agricultural land; they hold 512,700 hectares, that is, about 1.5 hectares per farm on average, while of the 480 farms that are legal entities, 214 have land, holding 19,600 hectares, about 92 hectares on average, or 60 times more (Armstat, preliminary results of the 2024 agricultural census as of 10 October 2024; average size is a BD calculation);
• as of 31 December 2024 the state register listed only 176 agricultural cooperatives (22 new ones registered in 2024) and 141 collective peasant farms, against 335,000 farms with land (State Register Agency of Legal Entities, 2024 report);
• according to the report on the execution of the 2025 state budget, which the National Assembly approved on 17 June 2026, AMD 43.1 billion was spent on agriculture out of AMD 44.9 billion under the revised plan (96%). This is 1.3% of total budget expenditure (AMD 3.305 trillion). Including irrigation, forestry and fisheries, spending amounted to AMD 63.2 billion (90% of plan); irrigation was funded at only 79%: AMD 18.6 billion out of AMD 23.5 billion (resolution of the National Assembly and the 2025 budget execution report, ARLIS, note on expenditure by functional classification). The Ministry of Economy, which runs the support programmes, reported executing AMD 42.8 billion out of AMD 43 billion (National Assembly of Armenia, 12 June 2026). Of the AMD 39 billion of the ministry’s four agricultural programmes, almost half, AMD 18.4 billion, went on subsidising intensive orchards. On pilot agricultural insurance, AMD 1.3 million was spent instead of the planned AMD 1.5 billion, and on investment projects in livestock farming nothing was spent instead of the planned AMD 2.4 billion (ibid., annex “Expenditure by programmes and measures”). The ministry currently runs 24 programmes: interest subsidies on loans, leasing of machinery and equipment, “smart” cowsheds, greenhouses, subsidies for milk, fertilisers and exports, development of cattle breeding (2025–2029), sheep breeding (2024–2028), crop production (2026–2027), and agricultural insurance (Ministry of Economy, state agricultural support programmes). For small farms there is a pilot programme for orchards and irrigation on household plots: AMD 38 million was spent on it in 2025. There is no separate programme for cooperatives on the list.
• two measures shut out small farms with the same threshold of 3 hectares. The new crop production support programme for 2026–2030 (a draft government decree, public consultation until 3 November 2026) reimburses half the cost of sowing grain and perennial grasses (for winter wheat, AMD 80,000 per hectare), but only to farms cultivating at least 3 hectares, and only with certified seed (draft and justification on e-draft.am; BD analysis). Previous rounds involved about 5,000 farms with an average plot of 3.9–4.6 hectares, out of 335,000 farms with land. The programme to support the consolidation of agricultural land (2023–2026) sounds like a path to pooling, but is designed the other way round: the state reimburses 50% of the cadastral value of the land to whoever acquires at least five plots and assembles a holding of 3 to 200 hectares (sale and purchase contracts are attached to the application). Here the small owner is a seller, not a participant: the programme does not provide for owners themselves pooling land without selling it, whether in a cooperative or through joint cultivation. The entire programme covers roughly 400 hectares, the land of about 270 average farms out of 335,000, and even this money is not being spent: in 2025 nothing of the planned AMD 200 million was spent (Government Decree No. 1220-L of 4 August 2022, ARLIS; 2025 budget execution report, ARLIS). With the average farm at 1.5 hectares, both measures cut off most small farmers. This is a direct example of non-inclusive policy: the money goes to large players, and there is no incentive to pool;

Why leaving the EAEU hits the countryside in particular. The Russian market accepted non-standardised produce from small farms: 95% of vegetable and fruit exports went there, $190 million out of $200 million in 2025, while only $1.3 million went to the EU (State Revenue Committee, exports by country and product, commodity groups 07–08). The 2026 bans have already reduced apricot and plum exports to Russia to zero (from $23 million to zero in the first half of the year, customs data). A European buyer demands volume, traceability and GlobalG.A.P., and a single farmer with 1.5 hectares cannot provide that.

Hypothesis: consolidation around the buyer. The first consolidators will be private aggregators, cognac and canning plants, and freight forwarders; cooperatives will join later. The state mainly needs to do three things:

1. shift subsidies from the kilogram to the consolidated shipment and a long-term contract with the farmer;
2. adopt a law on producer organisations and warehouse receipts so that a harvest can be pledged at a bank;
3. act as an intermediary between the farmer and the bank. A pilot agricultural insurance programme with the participation of the German development bank KfW exists, but in 2025 AMD 1.3 million was spent on it instead of the planned AMD 1.5 billion (2025 budget execution report, ARLIS); it needs to be relaunched.

Comparisons:

• Georgia, ENPARD. About 1,500 cooperatives; participants’ incomes rose by 27% (ISET). Hazelnuts bring in $120 million of exports per season, mainly to the EU.
• Poland. Agri-food exports grew from €5.2 billion (2004) to €53.8 billion (2024) (Poland Tastes Good); the number of dairy farms fell from 600,000 to 380,000 during the transition to EU standards. The lesson: standards drive consolidation, and some small farms exit, so they need an alternative (work in regional factories, see section 4).
• Denmark, Arla. A cooperative owned by 11,000 European dairy farmers; group revenue in 2025 was €15.1 billion and net profit €415 million (Arla’s announcement of its 2025 results). In June 2026 Arla merged with DMK, Germany’s largest dairy cooperative (DMK). This proves that farmers can own processing rather than merely sell raw materials to it. Armenia has few forms of joint ownership: as of 31 December 2024 there were 725 open and 3,156 closed joint-stock companies and 176 agricultural cooperatives registered, compared with 100,066 limited liability companies (State Register Agency of Legal Entities, 2024 report).

6. EU standards as a tool: exports, import substitution, cognac

EU preferences on their own offer little. A temporary two-year removal of ad valorem duties (on top of the GSP+ regime, plus eight agricultural products within quotas) has already been adopted: Regulation (EU) 2026/2269 of 7 October 2026 was published on 9 October and applies from 10 October 2026 to 11 October 2028. According to the European Commission’s estimate in the financial statement accompanying the proposal, the waived duties amount to about €3.3 million a year in gross terms, and the net loss of own resources to the EU budget is €2.475 million for 2027 and €2.97 million for 2028, about €5.44 million in total; how much of this Armenian exporters will save depends on who bears the duty in a particular transaction (European Commission proposal COM(2026) 348, Council of the EU document 11475/26 of 2 July 2026). This is about 0.01% of Armenia’s GDP. The real value of EU standards lies within the country: they can bring order to the domestic market and make room for local raw materials.

Will Armenia be able to make use of the preferences: by four functions:

Import substitution through standards. In 2025 Armenia imported 16,000 tonnes of milk powder worth $26.6 million (from Iran, Belarus and Russia) and 7,100 tonnes of butter worth $45 million; in the first half of 2026 milk powder imports rose by a further 34% by volume (customs data). Less and less domestic milk is being produced while imports of powder are growing: processors are substituting one for the other. Since 2022, products containing milk fat substitutes have had to be placed on a separate shelf in large shops, and since 27 April 2025 in all shops and with colour labelling (Law ՀՕ-63-Ն, ARLIS), but there is no requirement to label products as “made from milk powder”. EU rules (honest labelling of ingredients, the “cheese” standard applying only to products made from milk) would give farmers a price premium for natural milk. As for alcohol: Armenia imports practically no ethyl alcohol (2207) ($150,000 in 2025), but it imports spirits and distillates (2208) worth $68.9 million, of which $31.4 million from Georgia (customs data). Some of these distillates may go into blends, and this is where the standard is a direct protection for grape growers.

Cognac, wine and geographical indications. Under the Comprehensive and Enhanced Partnership Agreement between Armenia and the EU (CEPA), the name “cognac” will be banned on Armenia’s domestic market from 2032, and it will be dropped for exports from 2043: the 13- and 24-year periods are counted from one year after the start of the agreement’s provisional application on 1 June 2018 (text of the agreement on the website of the Armenian Ministry of Foreign Affairs, Articles 237 and 385, pp. 213–215 and 354 of the file). In return, the EU undertook, no later than eight years on, that is, by June 2026, to provide technical and financial assistance in creating and promoting a new name; if no assistance is provided, Armenia is entitled to suspend its obligations regarding the name (ibid., Article 237, paragraphs 4–7). The EU project “Technical Assistance for the Development of the Geographical Indication ‘Armenian Brandy'”, worth €2.845 million, ran from 20 June 2024 to 19 June 2026 (EU project page); the European Commission calls this “a difficult renunciation of the names Cognac and Champagne” (EU Directorate-General for Agriculture). The Ministry of Economy filed the application for the geographical indication “Armenian Brandy” on 13 March 2026: brandy made from wine distillate from Armenian grape varieties only (Intellectual Property Office bulletin No. 03/2 of 16 March 2026, pp. 471–472); the EU project has ended, but there is no registration decision yet. Products with a geographical indication in the EU cost on average 2.5 times more for spirits and 2.85 times more for wines (EU study). The risk: the application allows ageing “in contact with oak wood”, which, according to Tigran Jrbashyan, director of management consulting at Ameria, could devalue the geographical indication (ARKA, 17 April 2026). Our answer: the country will win, but not all producers will. The losers will be those whose business rests on cheap 3–5-year-old cognac for Russia; the winners will be grape growers and premium brands. The trend is already visible: exports of spirits (HS heading 2208) to Russia in 2025 held at the 2024 level of $284.6 million, while volume fell from 40.5 to 34.0 million litres of pure alcohol, meaning the price per litre rose from $7.0 to $8.4; according to preliminary data, in the first half of 2026 exports to Russia grew by 24% to $137.6 million, volume by 13%, and the price per litre continued to rise (customs data).

Comparable cases. After the 2006 embargo, Georgia lost roughly 80% of its wine exports (from 57 to 11 million bottles, JAMnews), and in 2025 it exported $266.5 million worth of wine (Geostat). But even there Russia came back: 66% of exports in 2024 (JAMnews). The lesson: diversification reduces risk, but replacing the largest market is hard. Moldova raised the EU’s share of its exports to 68% — but its main export item is not wine but electrical equipment (EU Neighbours East).

7. Protecting the domestic market under WTO rules

Armenia makes almost no use of permissible protection measures. While it is in the EAEU, tariffs and anti-dumping are decided in Moscow; leaving would return these instruments to Yerevan.

The main caveat. Protection works only when time-limited and conditional: a benefit in exchange for growth in productivity or exports. This is exactly how South Korea and Taiwan acted in the 1960s–1980s: support went to those who entered foreign markets. Open-ended protection without such conditions is a path to expensive goods and rent-seeking, which is especially sensitive in a country with a poverty rate of 21.7%.

8. Corridors and Central Asia: Armenia as a regional hub

The US is building a “Central Asia — Caspian — South Caucasus — Turkey” chain, and TRIPP is its Caucasus link. For Armenia this is a chance to become not a pipe but a regional hub: a place where cargo and raw materials not only pass through but are also processed, insured and serviced.

What is happening in Central Asia:

• The C5+1 summit in Washington (6 November 2025): $25 billion in deals in the Deal Zone; according to the US Department of Commerce, all agreements exceed $100 billion (trade.gov).
• On 19 October 2026 in Samarkand, during the visit of US Secretary of State Marco Rubio, a package of 54 agreements and initiatives worth about $12 billion was prepared (Administration of the President of Uzbekistan). A bill has been introduced in the US Congress that would allow the Jackson–Vanik amendment to be lifted for Uzbekistan (H.R. 2329); Uzbekistan expects to join the WTO in 2026 (WTO).
• Kazakhstan joined the Abraham Accords and agreed on a $1.1 billion tungsten project and $4.2 billion of Wabtec locomotives (US Department of State).
• The TRIPP+ fund (Trans-Caspian Enterprise Fund) — $201 million to develop the route linking Central Asia and the South Caucasus with US and European markets — is a US Department of State grant running until 2036 (USAspending, grant SEUACE26GR0002). The Department of State announced the official launch of the fund during UN General Assembly week, together with a commitment to allocate, jointly with Congress, almost $200 million for TRIPP infrastructure (US Department of State, 29 September 2026). Investment through the fund was discussed on 23 September 2026 in New York at the US government event “Prosperity through Peace: Investing in the Trans-Caspian Trade Route and TRIPP”, attended by the foreign ministers of Armenia and Azerbaijan, US Special Envoy Steve Witkoff, and representatives of Georgia and the Central Asian countries (Armenian Ministry of Foreign Affairs).

But flows are still small. The Trans-Caspian route (Middle Corridor) carried 4.1 million tonnes in 2025 (−8.9%) against a capacity of about 6 million tonnes (according to the Kazakh side, Astana Times). Armenia’s trade with Kazakhstan was $124 million in 2025 against $106 million in 2024 (customs data). The first Kazakh wheat arrived in Armenia via Azerbaijan in November 2025 (1,000 tonnes), and additional Kazakh exports to Armenia could amount to about $350 million (estimate by Kazakhstan’s Deputy Prime Minister Serik Zhumangarin at the intergovernmental commission on 19 September 2025, Government of Kazakhstan). But in the first half of 2026 only 26.5 tonnes of wheat came from Kazakhstan, against 168,400 tonnes from Russia, or 99.9% of imports (State Revenue Committee, imports by country and product). That is why a backup route from the east is a matter of food security.

What a regional hub could deliver (our hypothesis, in order of realism):

1. Processing of critical minerals. Armenia already mines copper and molybdenum (mining accounts for more than half of goods exports, trade.gov), but ships out concentrate. Given US interest in mineral supply chains and cheap TRIPP logistics, it makes sense to look for an investor in processing in Syunik.
2. Corridor services: warehouses, customs brokers, cargo insurance, rolling stock repair. It is these, rather than the transit fee itself, that yield the +0.5–1% of GDP estimated by the World Bank (according to the US International Development Finance Corporation).
3. Information technology and finance for Central Asia: Russian- and English-speaking teams and banks can serve a region that American investment is flowing into.

Risks and competitors. The double-track Kars–Dilucu line (224 km, €2.4 billion of external financing) will link Turkey with Nakhchivan (Turkish Ministry of Transport). The war around Iran since February 2026 has lengthened delivery times from China from 45–60 to 90 days, while about 25% of Armenia’s foreign trade cargo went through Iran (Jamestown). The peace treaty has been initialled but not signed: Baku demands a change to the constitution, and a new constitution in Armenia is adopted by referendum. In the elections of 7 June 2026, Civil Contract won 49.9% of the vote and 64 of 105 seats (Central Electoral Commission of Armenia). Prime Minister Nikol Pashinyan set the goal of resolving the signing and ratification of the peace treaty by 2031 (government meeting of 15 October 2026). This means the gains from open borders will come later than the losses from leaving the EAEU — the most important risk for the entire construct.

Three modernisation scenarios and what is easy to miss

The same three external factors can lead to different outcomes depending on domestic policy. This is a qualitative framework on top of our calculation, not a separate model.

Source: BD estimates; the figures in the “After 10 years” column are model scenarios from the scenario table above; matching the scenarios to the model variants is our judgement.

What is easy to miss. We checked channels that appear neither in the four-country report nor in the public debate:

• The “exit” is already under way. From 12 June 2026, Russia introduced temporary restrictions on imports of products subject to quarantine control from Armenia and on their transit to other EAEU countries (decision of Rosselkhoznadzor) and suspended imports of Jermuk mineral water (Rospotrebnadzor). Armenia is already bearing part of the losses that the report attributes to a future exit. Both the report and our calculation start from 2025 flows; from 2026 levels the remaining blow would be smaller. The Central Bank is already recording the effect: in the second quarter of 2026 agriculture contracted by 15.3% and subtracted about 1 percentage point from GDP growth, due to a late season and restrictions on exports to Russia. By its estimate, the goods actually hit by the restrictions accounted for exports of about 1.2% of GDP in 2025, all “quarantine” goods for 2.4%, and all exports to Russia for 6.1% of GDP. The total amount at risk is small, but flowers, fish and stone fruit went almost entirely to Russia, the domestic market will not absorb them, and agriculture’s debt burden has reached about 52% of its output (Central Bank of Armenia, Monetary Policy Report, 2026 Q3, pp. 5 and 36, box Case B, chart 3.F.4).
• The budget gains from exit. The net outflow of customs duties to the EAEU budget was AMD 54.5 billion (about $140 million) in 2025 (2025 budget execution report, ARLIS). With its own tariff at current rates, this money would stay in Armenia (see section 7).
• The nuclear plant and energy. Metsamor provides about 30% of electricity generation and runs on Russian fuel; a “123” agreement with the US was signed in February 2026, and four American small modular reactors have been selected (WNA). Until they arrive, there is a risk regarding fuel and servicing.
• Bread. In the first half of 2026, 99.9% of imported wheat ($44.2 million), 96% of flour and 99.6% of sunflower oil came from Russia; wheat self-sufficiency in 2025 was 23.2% (132,700 tonnes of domestic harvest against 332,300 tonnes of imports; Armstat, per BD’s summary; imports by country — State Revenue Committee, imports by country and product). This is not a blow to GDP but a risk for inflation and poor families, and it needs to be insured against in advance: stocks, contracts with Kazakhstan.
• The only road north. Upper Lars is the only land route to Russia, and the war around Iran has narrowed the southern one. Opening the borders with Turkey and Azerbaijan is first and foremost insurance against a blockade, and only then transit.
• Protection of property. The transfer of shares in Electric Networks of Armenia to the state by 1 October 2027 (government decision No. 1510-Ն, ARLIS) is a signal to all investors about the protection of property.
• The cost of rapprochement with the EU. Adopting European rules costs money: the environmental part alone cost new EU members roughly €1,000 per person (Government of Ukraine); for Armenia this is about $3 billion over many years (our estimate). This is investment, not loss, but it has to be financed.
• People are returning. 2025 was the first year since 2010 with positive net migration: +14,300 people, and the population grew to 3.10 million as of 1 January 2026 (Armstat, population as of 1 January 2026); according to the Ministry of Internal Affairs, the balance of entries and exits of Armenian citizens in 2025 turned positive for the first time: +8,660 (discussion of the ministry’s report with the Prime Minister, 26 January 2026). This is the resource that the inclusive scenario can retain and the enclave scenario can lose.

Comparable countries: what this means for Armenia

In the comparable cases found, a change of trade bloc without war or system collapse rarely produced a GDP decline of more than 4–6% in a year; the gains, however, always came more slowly and more narrowly than promised.

Primary data on which the calculation is based:

• Armstat, “Socio-Economic Situation of the Republic of Armenia”, table 1.1.2: GDP 2025 and first and second quarters of 2026
• Central Bank of Armenia: transfers by individuals by country, personal transfers in the balance of payments, current account, financial account, loans by sector, Monetary Policy Report, 2026 Q3.
• State Revenue Committee: foreign trade by country and product, 2024–2025 and the first half of 2026.

Limitations. The calculation does not model the dram exchange rate, inflation or the budget; value-added shares and the multiplier are calculated from a synthetic 2025 input-output table (see the section on the model). Armenian statistics are taken from official sources (Armstat, the Central Bank, the State Revenue Committee, the Ministry of Finance, the Public Services Regulatory Commission); the media are cited only where a figure is the statement of an official, a company or an expert and does not appear in an official publication. Estimates of border opening rely on studies from 2001–2023. The model, with formulas and the source of each parameter, is attached; links to all secondary sources are in the text of the sections.