Social-Purpose Microfinance in Hungary

Hungary had 890,046 active micro-enterprises with up to nine employees in 2023, 95.7 per cent of its 929,593 active enterprises (KSH STADAT 9.1.1.8.). The figure matters because the number of registered enterprises is roughly twice as high (1,839,940), so every enterprise count must name its base [CS I.3, X.2]. Small and medium-sized enterprises provide 68–70 per cent of employment but only 50–54 per cent of value added; micro-enterprise productivity is about 38 per cent of the domestic large-firm level, against 67 per cent for small and 81 per cent for medium-sized firms; early-stage entrepreneurial activity is in the EU mid-range, and the weak point is surviving the "valley of death" of the first years [CS X.2, citing the MNB Productivity Report 2024, KSH and Eurostat].

The HSCF study attributes five public functions to this layer – employment stability, crisis tolerance, social mobility, rural viability (in many villages the micro-enterprise is the only employer) and innovation flexibility – and observes that development policy has rarely addressed it: "support landed above the size threshold" [CS I.3; EXEC 1]. The financing gap is described as a market failure in the classic sense. For a profit-oriented lender, serving the smallest firm without collateral or credit history does not pay; this is "not a fault and not ill will – business logic" [EXEC 1]. Where there is no collateral and no credit history – the young starter, the re-starter, the returnee – there is market failure, and additionality is definitional: the programme's target group consists of those the market does not serve, so crowding out is excluded by design [FICHE-EN pt 2; CS II.7].

The evidence base for the gap is a series of client surveys run by the network since 2015 (300–500 responses per wave through the CREDINFO lending system; detailed 2018 report with 347 respondents). In 2018, 28 per cent of clients had received no loan offer from a commercial bank at all; among the excluded, the reasons were lack of collateral (39.3 per cent) and start-up status (32.1 per cent); 93 per cent said the financed project would not have happened, or only later, smaller or with difficulty, without the microloan (83 per cent in 2019) [SURV §4–5]. In 2022, firms with up to four employees had used the subsidised state loan schemes at about half the sample average (8 per cent), named an affordable interest rate of about 3 per cent and a useful loan size of about HUF 5.8 million [SURV §4]. The surveys are self-reported, biased towards active clients and designed as management information, as their analyst states; they are the best demand-side evidence Hungary has.

Three kinds of institution lend, or have lent, to Hungarian micro-enterprises with public support.

The first is the network of county and Budapest enterprise development foundations created from 1991 with Phare support – twenty in total – organised since 2013 in the Hungarian Microfinance Network. They lend under an exemption from the Credit Institutions Act (Hpt., Act CCXXXVII of 2013, §2(1)(f)), are not licensed or prudentially supervised by the central bank, and fall under the prosecutor's legality supervision as civil organisations [CS II.4; BMDF]. Their coordinating body since 2024 is HSCF, a public-benefit foundation registered in 2023 and led by the people who ran RVA, the Székesfehérvár foundation whose 34-year lending record is the network's track record. The network runs on a common technology: CREDINFO, a credit-information, appraisal and portfolio system developed with RVA, which HSCF states is used by all county foundations and by at least twenty licensed financial enterprises [STAFF 3.5].

The second is the state's own channel. In 2000 the government and MVA centralised the foundations' funds into the MVA-managed National Microcredit Fund (OMA); the foundations became commission-paid intermediaries of the National Microcredit Programme (OMP), which required double real-estate collateral – a requirement the 2018 national survey of the ATM for SMEs Interreg project found "does not allow the programme's main mission to be achieved" [CS III.3, III.6]. The programme still exists; in May 2023 the responsible ministry stated that it held no data or results on it [CS IV.7].

The third is the state-subsidised small-business finance system distributed through banks and for-profit financial enterprises – the Széchenyi Card Programme – whose "Széchenyi Mikrohitel MAX+" product lends HUF 1–150 million at a fixed 3 per cent with at least one real collateral mandatory and no accompanying service. Its outstanding stock is about HUF 4,500 billion and its interest subsidy about HUF 150–290 billion a year [SZM 1–2]. HSCF's evaluation, prepared for the development bank in August 2026, holds that this product is not microcredit under any accepted definition and that the two instruments serve different segments – a point developed in chapter 6.

Between 2008 and 2016 a fourth arrangement operated: JEREMIE microcredit refinanced through the holding fund of the Hungarian Development Bank group, in which the foundations, financial enterprises, savings cooperatives and banks all took part. Its record is the subject of chapter 5.

The history divides into periods that the committee will recognise from other countries, with one Hungarian peculiarity: the legal framework was never built.

1992–2000: the decentralised model. Phare money (an ECU 1.15 million set-aside within a ECU 21 million SME mandate given to MVA, later co-financed by the Hungarian government) reached the foundations as conditional, non-refundable earmarked grants given into their ownership; the foundations lent, recovered and lent again; banks contracted for a fee without bearing risk; advice accompanied the loan and was financed separately. Disbursement reached 2.7 times the grants on average and more than four times at the earliest lenders [CS III.1–III.2; MS 1992–1998]. In 1997 the supervisory authority confirmed that this lending was not a financial service under the banking act; in 1998 a working committee formed at the EU Delegation's request recorded full consensus that the foundations had acquired ownership of the funds after lending them out and that MVA had acted lawfully under the rules accepted by the Delegation [CS II.4, IV.2].

2000–2007: centralisation. A Government–MVA agreement of 17 May 2000 created the National Microcredit Fund under MVA management; the framework agreements with the foundations ended on 1 July 2000; the foundations became intermediaries paid by commission; the interest matrix inverted and performance fell "to a fraction" [CS III.3, IV.3]. In 2002, 48 per cent of operating commissions were absorbed by two central bodies while the twenty implementing foundations bore half the costs; the foundations' consortium recorded that operation had become substantially more expensive, bureaucracy had multiplied and implementation had "moved away from the original objectives and from international practice" [CS III.3]. In 2004 a ministry working committee minuted that a state fund "can only be created by law" and that the situation must be settled; nothing followed [CS IV.4]. Locally, RVA launched Hungary's first municipal microcredit programme (2001) and, after the closure of the IBM plant in Székesfehérvár, a fresh-start loan funded by a HUF 300 million corporate donation that financed about HUF 676 million of loans to 178 redundant workers over its life [CS III.4; MS 2001–2002].

2007–2016: JEREMIE. Under the Economic Development Operational Programme the foundations disbursed more than HUF 44 billion of refinanced microcredit; the model proved that a refinanced, decentralised architecture works, and RVA's digital lending technology won European recognition in these years (Milan 2009, JASMINE audit 2010, London 2011, transfer to Norway 2012, DIFASS 2012–2014) [CS III.5; MS 2009–2014]. The same period saw a litigation decade over one foundation's contracts and the deletion of MVA's public-benefit status in 2014 [CS IV.5].

2016–2023: dormancy and documentation. The development bank's 2016 intermediary system was built on banks only; the non-profit network was excluded and no mentoring accompanied the products [CS III.6]. The network turned to documenting its case: the Rome Directives (2016), the ATM for SMEs Interreg project with a national survey (2018) and a final study (2021/2022), a ministerial letter of 2017 confirming that no legal act defines foundation microlending, and the ministry's 2023 statement that it holds no data on the National Microcredit Programme [CS III.6, IV.6–IV.7; MS 2016–2022].

2023–2026: preparation of a relaunch. HSCF was founded in 2023, took over RVA's portfolio, offices, network rights and a documented succession plan in 2024, obtained public-benefit status in October 2025, submitted its Code self-assessment, prepared the Comprehensive Study, the draft Act and an EIF guarantee application in August 2026, and entered the Code's on-site evaluation in September 2026 [BMDF; MS 2023–2026]. Annex B gives the dated chronology.

A change of government in spring 2026 opened a policy window. On 11 May 2026 the ministry responsible for SME development requested full data from MVA on the National Microcredit Fund, including the quarterly reports due under the founding agreement for 2024, 2025 and the first quarter of 2026 [CS IV.7]. Government Resolution 1239/2026 (VIII. 3.) of 3 August ordered a review of state aid and guarantees linked to private-market lending and lowered the counter-guarantee ceiling, which HSCF reads as opening space for instruments that remedy market failure rather than distort the market [CS II.8; SZM 2]. The post-election government programme contains stated commitments on socially oriented enterprise financing, and the draft 2026–2036 national competitiveness strategy identifies a stepped, social-purpose microcredit portfolio with EIF guarantee cover and mandatory business development services as a central instrument of its SME pillar [BP §1.9].

Against this, the sector's capacity is thin. HSCF has four to five staff, no employed loan officers and an external appraisal chain "standing ready"; its 2025 lending revenue of HUF 0.82 million covered 3 per cent of its HUF 27.1 million operating expenses, the rest being operating grants [SAT; BMDF]. Several county foundations cannot meet obligations arising from JEREMIE-era loss claims, and their immediate enforcement "could cause the insolvency of several county foundations" – and with it the collapse of the network on which any relaunch depends [CS V.4]. The relaunch is conditional on the EIF guarantee decision, on the settlement of legacy funds through the draft Act and on national budget lines for operations and business development services, none of which was committed at the time of writing [BMDF].

5.1 Three funding models in thirty-four years

Hungarian social microcredit has been funded in three ways, and the outcomes track the funding model more closely than any other variable. In the first model (1992–2000) the funder – Phare, then the co-financing Hungarian government – expected no financial return, the intermediary's operating costs were largely chargeable to the fund, risk stayed with the fund, and the funds revolved locally; this is the model that produced the 2.7-fold disbursement multiple and the 1998 "best in the region" verdict [CS III.2]. In the second model (2000–2007) the funds were centralised in a fund managed by a national foundation, the local lenders became commission-paid intermediaries, and both cost and outreach deteriorated [CS III.3]. In the third model (2008–2016) JEREMIE refinancing was channelled through a holding fund to a mixed set of intermediaries; the architecture worked operationally, but the incentive came from interest income, credit risk was left with the intermediaries and mentoring was unfunded [CS III.5].

The HSCF authors reduce this history to an "interest matrix" with two questions: does the funder expect a return, and must the intermediary cover its operating costs from client interest? Where the funder expects a return, the social goal recedes; where operations must be earned from clients, risk aversion crowds out the target group [CS II.6, III.2]. The Rome Directives of 2016 – professional recommendations drafted by European national networks, not an EU act – put the same conclusion into principles that the Hungarian network has made a condition of membership: social impact takes precedence over preserving the nominal value of the fund; lending risk is not shifted to intermediaries; losses must be write-off-able from the funds; central fund-management costs should not exceed 5 per cent of programme implementation [HMN-RULES Annex 1].

5.2 JEREMIE and the lessons of working with Managing Authorities

JEREMIE microcredit under the Economic Development Operational Programme (GOP 4.1 and KMOP 4.1) is the most important Hungarian data point for the roadmap's paper on cooperation with Managing Authorities. To 31 December 2015 the programme recorded 16,944 transactions across all intermediaries. Financial enterprises made 47.34 per cent of transactions by count, enterprise development foundations 41.81 per cent, savings cooperatives 6.01 per cent and banks 4.83 per cent; on the per-intermediary indicator the foundations reached 59.61 per cent against 5.51 per cent for banks and 4.03 per cent for savings cooperatives, and disbursed more than HUF 44 billion in total [CS III.5; MS 2008–2013]. Only the foundations' average loan (HUF 6.1 million) stayed within the EU's EUR 25,000 microcredit band; banks and financial enterprises lent multiples of that, partly to a different clientele [CS III.5]. The ATM national survey concluded that JEREMIE "outperforms the National Microcredit Programme many times over" [CS V.2].

Portfolio quality was high. RVA's own GOP 4.1 sub-portfolio – 546 loans, HUF 3.0 billion – shows an expected loss of 1.23 per cent of disbursement on the current rating (1.8 per cent at disbursement), 0.00 per cent of principal actually written off and a programme-efficiency indicator of 98.77 per cent, defined as one minus expected and written-off principal loss over total disbursement; start-ups were 26.6 per cent of the loans with a 1.38 per cent expected loss (CREDINFO trend analysis, March 2026) [CoGC-DATA]. The whole JEREMIE microcredit portfolio showed an expected loss of 1.42 per cent of disbursement in the 2021 national evaluation [CS X.7]. These portfolios were originated under mandatory real-estate collateral; the new products prohibit it, so the figures are a lower bound, not a forecast, and are never used as a benchmark for collateral-free lending [CoGC-DATA B14].

The design lessons are recorded on both sides of the ledger. Positively, the refinancing-based, decentralised implementation proved workable and is the direct precursor of the Hub & Spoke model. Negatively, no concrete social objectives were set; the intermediaries' incentive came only from interest income; credit risk stayed with non-profit intermediaries and produced risk-averse behaviour and under-service of the neediest; mentoring went unfunded, against Rome Directives 6–8 [CS III.5, III.7]. The business plan's reading is that the configuration "obliged the non-profit intermediaries to fund some of the losses out of their own capital – leading in the long term to under-capitalisation, the inability to maintain operations and the disabling of the institutional system" [BP §5.3].

Two further lessons concern the recovery phase and must be stated with care. The call text promised a revolving fund; available information indicates that returned resources were later withdrawn or reallocated, which, if confirmed, raises questions under Article 78(7) of Regulation 1083/2006. Contract amendments after the coordinator change were repeated, non-negotiated and disadvantageous to intermediaries ("sign or be left out"), with a professionally contested methodology for calculating the "bad portfolio". The HSCF study itself says that any qualification cannot go beyond the possibility of a suspicion before itemised document review, and that no authority or court finding is known [CS V.3]. The transferable institutional lesson is that a coordinating body's unilateral amendment practice undermines a revolving instrument, and that expectations of "relentless collection" against micro-entrepreneurs are alien to a system whose banking-law exemption rests on a higher risk tolerance and a different collection culture [CS V.3].

The consequence in 2026 is concrete. Several county foundations cannot meet obligations arising from unpaid JEREMIE loans; the network's position is defensive, and its ask towards the development bank is a moratorium or settlement framework for the duration of statutory settlement, with legacy resources re-channelled into the new controlled system – "a settled exit route from a documented conflict" [CS V.4–V.5]. An internal legal note of August 2026, marked for lawyers' verification, finds no statutory prohibition on settling such claims and observes that EU law "provides for the earmarking of the returned money – not for the method or extent of recovery, nor for who bears the loss" [MFB-NOTE 3].

5.3 The dormant years: what "no EU instrument" has meant in practice

The roadmap asks for a per-country picture of what happens without InvestEU-type instruments. Hungary can supply one from experience rather than modelling. After the JEREMIE microcredit resources were reallocated into capital funds, "the HSCF was unable to launch a new disbursement wave"; new lending is recorded as "paused – due to funding constraint" [BP §3.2]. The legacy portfolio, originated by RVA in 2019–2023 at a fixed 6.5 per cent, fell from HUF 77.1 million and 32 loans at the end of 2023 to HUF 31.1 million and 18 loans a year later, HUF 9.4 million and 8 loans at the end of 2025, and HUF 5.3 million and 6 loans at 30 June 2026; there were no disbursements in 2024, 2025 or 2026, no restructurings and no write-offs; PAR30 was 1.9 per cent on a single loan at the end of 2023 and zero thereafter [MicPro; SAT F29–F36]. HSCF's operational self-sufficiency ratio for 2025 was 3.0 per cent: HUF 823,645 of lending income against HUF 27.1 million of operating expenses, the organisation being financed by operating grants of HUF 38.8 million in 2024 and HUF 21.9 million in 2025, part of which is the non-cash release of deferred income on the assets received from RVA [SAT sheet 2; annual statements].

The design conclusion drawn from this is that the guarantee is indispensable but not sufficient. Without a portfolio guarantee, collateral-free lending – the mission-critical feature – is not feasible, since the double real-estate collateral regime is documented as mission-defeating; the guarantee is the only external risk-sharing in the programme, because it draws nothing from the shrinking national counter-guarantee framework; and without the Code and EIF track there is no external quality certification [CS IX.2, X.6; EXEC 3]. The fallback without a guarantee is collateralised lending from domestic funds, which by construction cannot reach the excluded segment [BP §12.4; PAL §4–5].

5.4 The planned funding architecture and the role of the EIF guarantee

The programme fiche defines four sources that do not replace one another [FICHE-EN pt 6; CS X.8]:

• ring-fenced purpose-bound assets of OMA origin – the 1990s public microcredit endowment – re-dedicated by statute with purpose lock-in, a ban on diversion, a mandatory inventory and a duty to return funds placed contrary to purpose; "a legislative step, not a budgetary expenditure"; the 2004 ministry figure of "some HUF 4–5 billion" is a 2004 order of magnitude, and today's value is for the statutory inventory and forensic audit to establish;
• JEREMIE legacy resources under the SME re-use obligation of Article 78(7), redirected to their original purpose with the county foundations' documented own contributions as the allocation key, usable only for the revolving loan fund and network operations under segregated accounting; the European Court of Auditors (Review 06/2021) puts such legacy resources at EUR 8.5 billion EU-wide, to be reused by the Member State without time limit [CS IX.9];
• a budget envelope: an initial envelope in the order of HUF 0.5 billion, followed by annually itemised public expenditure with measurable social returns (planning assumption);
• the EIF InvestEU guarantee – credit-risk sharing from Union resources "without drawing on the national counter-guarantee system".

The fiche's formulation is the one HSCF asks partners to keep: "the guarantee shares risk, it does not provide capital – lending capital is assembled from national sources, in loan funds segregated by funding provider" [FICHE-EN pt 6]. Whoever reads the guarantee agreement as fund-raising over-estimates the programme's funding side by roughly HUF 60 billion [CS X.8]. The Strategic and Business Plan describes the same architecture as four "legs" – lending capital (a HUF 40 billion domestic revolving facility under discussion, not committed, with a HUF 4 billion staged alternative), a HUF 5 billion business-development envelope under formalisation, an operating base from settled legacy resources, and a National Capital Takeover Programme payable only on audited performance from the eighth year – with an explicit rule: no cross-subsidy, no own-contribution assumption, and the facility stays nominally sustainable only if operating and service costs are financed from their separate legs [BP §1.9, §11.1; FP §3]. The two descriptions differ in detail and are reconciled in Annex D.

The guarantee application was submitted to the EIF Board in August 2026 after a consultation on 3 August; it covers the three flagship products, with the collateral-free base product the subject of a supplement [BP §1.5; PAL §1]. The published terms of the InvestEU microfinance capped portfolio guarantee – guarantee rate up to 80 per cent, cap rate up to 30 per cent, maximum EUR 50,000 per transaction, free of charge to the intermediary with full Transfer of Benefit – and the planning anchors of a EUR 10 million portfolio, a three-year inclusion period and twelve-year loan maturities are published product terms; the actual parameters are fixed in the individual guarantee agreement [CS IX.2; EIF-v3.1 §6]. At the planning rate, EUR 10 million is about HUF 4 billion of guaranteeable disbursement; the conservative path places EUR 9.5 million in the inclusion period (94.9 per cent utilisation) while the base and ambitious paths would exhaust the frame in the second year, so the programme "plans to the frame, not above it" and treats an enlargement request as the natural next step on demand [CS X.6]. In the authors' assessment the guarantee is "the enabling and credentialising instrument of the collateral-free pillar, not its full risk backstop": in the business plan's base case cumulative claims reach the modelled EUR 1 million cap in the seventh year and residual risk sits with the domestic facility [BP §11.4]. The requirement that an intermediary bear own risk on its own balance sheet is one of the reasons the design vests ownership, not mere management, of the endowment in the coordinator [CS VI.4].

Two dates frame the timetable: the current InvestEU call closes on 30 June 2027, and InvestEU is centrally managed and accessed directly through the EIF, independent of the national operational-programme cycle, so risk cover can be obtained in parallel with, rather than after, the domestic asset settlement [CS IX.5, IX.9]. This is the basis of Hungary's interest in a directly accessible guarantee window in the next Multiannual Financial Framework.

5.5 Pricing and the cost of funding in a non-profit model

The planned client terms are fixed rates of 3.0 per cent (Start30, after 24 months at 0 per cent with no principal repayment), 3.5 per cent (Start30+, after 12 months at 0 per cent) and 4.0 per cent (Hazatérő, after a six-month grace), all over 144 months with a single registration fee of HUF 10,000 plus VAT and no other charge; the collateral-free base product is priced on the cost of funds within a 0–5 per cent band with a target at or below 4 per cent, and the collateralised Green line indicatively at 3–5 per cent over 120 months [BP §6.2; PAL §2–5]. An illustrative Start30 loan of HUF 10 million carries an indicative annual percentage rate of about 2.1 per cent, below the nominal rate because of the interest-free grace [SAT 1.6–1.7]. The legacy portfolio carries 6.5 per cent fixed.

The Pricing Policy sets a cost-plus formula – client rate equals funding cost plus operating cost plus risk cost plus a sustainability margin, less any interest subsidy passed on – in which the margin "does not serve profit maximisation but the security of the Foundation's operation"; the subsidy content must be disclosed as what the client saves against the market rate; default interest may not be punitive; and an annual percentage rate is disclosed voluntarily because Hungarian law requires none for business loans [PRICE §3–5]. The product palette adds the arithmetic that constrains the whole design: with an operating margin of one to one and a half points, funds costing no more than about 2.5 per cent keep the client rate at or below 4 per cent; at about 5 per cent "we do not expect meaningful demand from the target group", which is why a 5.5 per cent collateral-free product was discontinued on 8 August 2026 [PAL §1, §4]. The demand-side surveys point the same way: an affordable rate of about 3 per cent in 2022, and in 2020 a preference for loans below HUF 15 million at no more than 2.5 per cent [SURV §4, §6].

The financial logic is stated without euphemism. "The programme is intentionally not financially self-sustaining. It is a social-purpose instrument operated as a public task: interest income does not cover operations, and it is not meant to. The yield metric is the public cost per unit of social outcome" [FP §1]. The capital take-over, the interest-free phases and the mandatory services are planned public costs, budgeted on two lines – capital replenishment of the revolving fund, and operations and services – and the headline accountability figures are the public cost per financed enterprise and per job [CS VII.5, X.3]. The Comprehensive Study's three paths (planning assumptions, August 2026, fifteen years, 400 HUF/EUR) finance 3,449, 6,872 or 13,532 enterprises for a total public funding need of HUF 32.6, 60.0 or 119.0 billion, of which HUF 8.2, 15.2 or 29.0 billion in the first five years, at HUF 9.45, 8.73 or 8.79 million per enterprise and HUF 14.31, 8.95 or 6.98 million per job, with annual default assumptions of 10, 7 and 6 per cent [CS X.4]. The 7 per cent of the base path is "a prudential planning assumption, not a data-derived estimate"; the business plan and financial plan use 3 per cent base and 7 per cent stress for the guaranteed pillar, and the model generations must not be mixed [CS X.7; BP §11.3; FP §2]. Business development services are costed at about HUF 1.45–1.5 million per guaranteed-pillar client, some 16 per cent of the average loan [FP §2; PAL §6]. For scale, the five-year public funding need of about HUF 8.2 billion on the guarantee-aligned path is "3–5 per cent of the annual interest subsidy" of the Széchenyi Card Programme [SZM 2.3].

5.6 Capital markets, bank cooperation and capacity building

Capital-market access – securitisation, bonds – does not appear in Hungarian practice or planning; the only reference is a top rung of "equity finance (fund managers, BÉT Xtend)" in a proposed funding escalator [SZM 5]. Bank relations are designed as graduation rather than referral-in: the stepped architecture ends with commercial banks receiving the graduating client, the graduation rate is a central success indicator of the "micro-entrepreneur nursery" concept, and a "credit passport" is proposed under which the repayment history built in the social tier is transferred in standard form from CREDINFO to development-bank and commercial lenders in place of part of their collateral requirements [BP §6.2, §9.4; SZM 5]. Inbound referral runs through a second tier of sales points – accountants, chambers, local organisations – paid a success fee only after disbursement [SOIMP M1.2]. Investor-type funds envisaged for the base product are donations, capital contributions, concessional subordinated loans and municipal or corporate interest subsidies for near-zero local pockets [PAL §4].

Technical assistance of the SIFTA type is not mentioned anywhere in the Hungarian documents. Capacity building is planned domestically: an accredited postgraduate microfinance course (under negotiation with Pannon University), a public register of accredited microfinance experts with at least one registered expert per accredited lender, and an operating leg of the budget of about HUF 620–860 million in the first year [FICHE-EN pt 5; CS IX.4; FP §2–3]. The documented capacity gap – no employed loan officers, five staff, coach capacity at the county foundations to be financed "at programme launch" – is itself evidence for the roadmap's capacity-building strand [SAT F45–F46; STAFF 3.6].

6.1 A lending activity without a positive legal framework

The Hungarian case is instructive because the regulatory gap is not an oversight of recent years but a structure that has persisted through three banking acts. A 1997 supervisory position (ÁPTF Üsz. 2520/1997) found that the foundations' microlending was not a financial service; Act CXXV of 1999 exempted MVA's lending from the National Microcredit Fund; Act XXXIX of 2003 exempted the county and Budapest foundations; the current Credit Institutions Act keeps the exemption in §2(1)(f) [BILL general memorandum 1; CS II.4]. The exemption protected the activity from prudential rules written for deposit-takers, but it "also created a legal gap": no definition of microcredit, no defined circle of providers, no operating rules and no supervisory authority [BILL general memorandum 1]. In practice, HSCF's client-protection regime, its complaints procedure and its quarterly loan classification are each "a voluntarily assumed obligation", and a client has no supervisory financial-consumer-protection forum to turn to, "since no such forum is available in respect of the Foundation's activity" [CPR 1.4; COMPL ch. 7].

The state's own record documents the problem. Minutes of a ministry working committee of 18 February 2004 record that a state fund "can only be done by law. So this must be settled." The minister heading the Prime Minister's Office wrote on 28 April 2017 that "no legal act defines the conditions of the enterprise development foundations' microcredit activity" and that creating the legal background was "absolutely necessary". The responsible ministry replied on 4 May 2023 that "at present we have no data or results relating to the Programme" [CS IV.4, IV.6, IV.7]. Depending on the base, the framework has been missing for 34 years (since 1992) or 22 years (since the 2004 finding); HSCF asks that the base always be named [CS 0; EXEC 3].

The HSCF authors describe the application of banking logic to social microlenders as a category error: banking law protects depositors and investors, non-profit microlenders take no deposits, and the rules guard against risks that do not exist while preventing the risk-taking that gives the activity its social meaning. The remedy proposed is not to question the exemption but to complement it with a dedicated positive framework law [CS II.4].

6.2 The draft Act

The draft Act (v7.17, August 2026) is a stand-alone statute of fifty sections in three chapters with one linked amendment to the Credit Institutions Act. Its design choices are the material Hungary offers to the roadmap's best-practice set.

• Definitions. "Positive social-purpose microcredit" is a loan that puts social impact before profit, to which basic mentoring and business development services may attach; both service types are defined separately [BILL §2].
• Providers. County and Budapest enterprise development foundations accredited by the coordinator, holding public-benefit status, taking no deposits and lending for no profit-oriented investor; the system is open to new entrants [BILL §§8, 21–22].
• Public task. Microlending under the Act is a public task and a public-benefit activity by force of law [BILL §12].
• Ceilings. Not fixed in statute: the coordinator sets and publishes the per-transaction ceiling "having regard to the microcredit thresholds used in EU policy documents"; simultaneous exposure per client may not exceed twice that amount [BILL §§6–7].
• Priority clients. Those whose reduced creditworthiness stems from small ticket size, insufficient collateral, lack of credit history or track record, or other disadvantage [BILL §5].
• Social performance. Every lender must document its social mission, an outreach plan with expected impact and a measurement methodology [BILL §11].
• EU compliance by funding source. Accredited lenders are bound by EU microfinance compliance requirements only when EU funds are passed to them; the coordinator alone bears the duty when it runs EU programmes; nationally funded programmes carry only accreditation requirements [BILL §20].
• The coordinator. Designated exclusively but conditionally: a public-benefit foundation free of state, municipal or party control, with at least ten years of documented microfinance activity, membership of a European microfinance umbrella organisation and a commitment to EU compliance requirements, a national network and loan-registry technology; reviewed every five years; on persistent breach the court transfers the assets, purpose-lock maintained, to another eligible body [BILL §§24, 31].
• Transition. Twelve months for the accreditation procedure and twenty-four for foundations to qualify, with a transitional operating right so that no "lending vacuum" arises [BILL §§27–28].
• Non-repayable element. A defined part of the loan may become non-repayable if the debtor meets the contractual social and economic goals, only under de minimis, the General Block Exemption Regulation or another Commission-approved title, with an automatic successor-act clause; separate grants for services and capacity building are allowed [BILL §19].
• Revolving funds. "Use" of a fund is a one-time placement; repayments become the foundation's own mission assets, and finally transferred funds cannot be reclaimed [BILL §§15, 17–18].
• Depoliticisation. Founder rights of the state, municipalities and parties in the foundations pass to the coordinator, and the foundations continue as independent private foundations – a structure that Hungarian law has in any case not allowed to be created since 2006 [BILL §§35–38].
• Supervision. The minister responsible for economic development, not the financial supervisor, because the activity is outside the credit-institution regime [BILL §§23, 47].

The explanatory memorandum explains why a statute rather than a grant decision, litigation or supervision: the longest Hungarian case ran "some seventeen years across three court levels" and settled "one sub-question of one foundation's one contractual relationship" – "civil litigation adjudicates a legal relationship; it does not build an institutional system" [BILL general memorandum 3]. It also explains why the civil coordinator must own, not merely manage, the revolving endowment: only the creditor can waive a claim, so conditional write-offs require full disposal, and "an organisation built on managed third-party assets would not meet the own-risk-bearing requirement placed on financial intermediaries of EU guarantee instruments" [BILL general memorandum 4].

Three caveats accompany the draft. The statute itself contains no prohibition of real-estate collateral (that is a programme rule) and fixes no forint or euro amount; its §19 places the non-repayable element "as part of" the loan subject to the funding source, whereas the programme keeps the capital take-over in a legally separate national programme so that no guarantee call can arise from it – the two are consistent only under the "funding source permits" proviso; and the draft does not decide the ownership of the historic assets: it recognises and protects the county foundations' contractually settled property (§30(1a)) and orders restitution of third-party assets "in recognition of original ownership" (§30(5)) [BILL §§6, 19, 30; PAL §3.2]. The constitutional and state-aid opinions are outstanding, and the adoption timetable depends on the parliamentary calendar [BP §1.9; CS XI.7].

6.3 What "microcredit" means in Hungary: definitions and mislabelled products

Hungarian practice offers the committee two differentiations for its education work, neither of which is the microcredit-versus-consumer-credit line as such. The first is social-purpose versus profit-oriented microcredit: "not two points on the same continuum but two different logical frames" – financial-market correction versus public-good allocation; profit versus social return; the bank client versus the excluded; the bank or financial enterprise versus the public-benefit non-profit; prudential regulation versus a dedicated professional framework [BILL general memorandum 2; CS II.3]. The paradigm sentence recurs across the documents: "social-purpose microcredit is not a financial-market product but a social-policy instrument" [CS 0].

The second is genuine microcredit versus a state-subsidised small-business loan that carries the name. The "Széchenyi Mikrohitel MAX+" lends HUF 1–150 million (about EUR 375,000 at the planning rate – fifteen times the classic EUR 25,000 definition of COM(2007) 708 and 7.5 times the EUR 50,000 InvestEU ceiling) at a fixed 3 per cent, with at least one real collateral mandatory, working capital limited to a fifth of the loan and no accompanying service, through more than thirty banks and for-profit financial enterprises; the 2021 ATM national study had already found that these constructions are not microcredit under any accepted definition "apart from the name" [SZM 1]. In 2023 the scheme carried a gross rate of about 22.5 per cent against 5 per cent paid by the client, the difference borne by the budget, for clients who by definition are within bank risk appetite – which in HSCF's assessment produces deadweight, crowding out of market-priced bank lending and interest arbitrage; the government's June 2026 decision to raise the rates on the liquidity products to market level is read as an acknowledgement of the latter [SZM 2.1–2.2]. The label matters for policy: it creates the impression among decision-makers that the microfinance function is fulfilled, confuses international partners and blurs bankable firms with the financially excluded in statistics. The business plan records that "in Hungarian usage the term 'microcredit' is sometimes applied to constructions of HUF 100–150 million" and that HSCF "actively defends against the dilution of the term" [BP §4.6].

On the consumer-credit boundary itself, three regulatory facts are usable. Hungarian law requires no annual percentage rate for business loans (Act CLXII of 2009), while the statutory THM regime (Government Decree 83/2010) applies to consumer credit; HSCF discloses an APR voluntarily and aligns its terminology so as not to create the appearance of a statutory figure [PRICE §5.1; CPR 4.3]. The consumer-suretyship rules of the Civil Code do not apply to a surety who is an executive officer or majority owner of the borrowing legal person, which is the position of the only guarantor in the new products [DMC 4.2.1]. And the draft Act limits purpose to enterprises and to natural persons seeking income from independent activity, tying the enterprise definition to the SME criteria of the General Block Exemption Regulation [BILL §§2, 4]. HSCF's terminology proposal is to reserve "microcredit" for loans up to EUR 25,000 and to call transactions between EUR 25,000 and 50,000 "microfinance transactions"; the HUF 19.9 million product ceiling is calibrated to the InvestEU ceiling, while the programme's self-definition rests on the classic one [CS IX.6].

6.4 Microfinance, poverty and social rights; the social economy

The draft Act's purposes include self-employment and "getting on at home", job retention and creation, helping recipients of social benefits become taxpayers and reducing benefit dependency "and the mental decline that accompanies it" [BILL §3]. The impact plan names the European Pillar of Social Rights Action Plan 2030 and the InvestEU Social Investment and Skills window as reference frameworks and claims rural population retention and reduced poverty risk as long-term impacts – as contribution, not causation [SOIMP 1.2; CS VIII.4, VIII.6]. The evidence is of the client-survey kind: improved living conditions reported by 75 per cent of clients in 2014 and 2015, 85.7 per cent in 2017 and 83.7 per cent in 2018; the financed business the sole household income for 71 per cent (2015) and 74 per cent (2018); 83 per cent of clients with at least one dependent family member and 2.9 persons affected per client [CS VIII.9; SURV §3, §5]. The at-risk-of-poverty threshold used in outreach reporting is HUF 2,085,245 a year for a one-person household (KSH, 2024) [SAT D18].

The social-economy link is thin in Hungarian practice. HSCF presented Hungarian microcredit at the fi-compass conference on ESF+ financial instruments for microcredit and the social economy in Rome on 4 July 2024; the draft Act admits organisations providing incubation or business development services and condominium-type business-premises communities as target groups; and the Green Community concept addresses residential communities affected by energy poverty, with its legal status still open [MS 2024; BILL §4; PAL §5.2]. No data on lending to social-economy actors exist; the inclusion indicator set that starts with the 2026 fourth-quarter baseline could tag them.

7.1 How Hungarian practice defines its target groups

The Hungarian definition is financial before it is demographic. The primary criterion is documentable exclusion: viable enterprises and starters that banks reject for lack of collateral, lack of credit history or small ticket size. Four priority circles follow – starters and micro-enterprises without credit history or collateral; women entrepreneurs, people under 30 and over 55; enterprises in disadvantaged (LHH) districts; and capital-poor losers of the green and digital transition [CS VIII.2; SOIMP 3.1]. The labels are non-evaluative: each rests on a documented financing barrier, they serve targeting, monitoring and reporting only, and they may never become an automated credit criterion; creditworthiness is decided by the affordability test alone [CS VIII.2; SOIMP 3.3].

At application each client receives an inclusion score from 0 to 10 across five components – bank rejection or absence of real-estate collateral; settlement classification; being a woman, under 30 or over 55; an enterprise younger than one year or with low turnover; and a voluntary housing-cost and social index – with "prefer not to answer" as a full-value answer that does not affect the loan decision [SOIMP 3.2, 5.3]. The 2026 pilot target is that at least 40 per cent of new disbursements go to the target groups, with a mandatory gender breakdown [SOIMP M1.5]. The client-protection rules state that programmes targeted at "start-up entrepreneurs, women, young people, people living in rural and disadvantaged regions, and returning Hungarians" are not discriminatory when tied to a documented social objective, and guarantee paper administration, personal assistance and oral explanation to clients with disabilities and to non-digital clients [CPR ch. 8]. The draft Act names would-be entrepreneurs, students, career-starters, skilled workers in shortage occupations, entrepreneurial communities unable to offer real-estate collateral, and incubation and service providers among the groups the government may prioritise [BILL §§4, 48].

The honest limit is that no current outreach shares by group exist. Own lending has been paused since 2023, the inclusion indicator set starts with the fourth-quarter 2026 baseline, and the client surveys of 2015–2022 did not report gender or age shares [SAT F19–F27; SURV]. What Hungary can offer for the thematic briefs is design, historic and legacy data, and case studies.

7.2 Women entrepreneurs

There is no dedicated women's product. Women are reached through the three flagship products and a "WOMEN" message package that couples capital with mentoring ("capital and community in one place"); whether a separate product is needed is to be decided on data after the baseline year [CS VIII.3; SOIMP M1.3]. Two data points exist. In the 2015 client survey the share of women clients rose about four points on 2014, as did the share of clients in settlements below 5,000 inhabitants [CS VIII.9]. In the legacy RVA portfolio taken over by HSCF, 15 of the 32 loans (HUF 100.4 million of HUF 207.9 million) went to women, and three of the six loans live at 30 June 2026 are women's [Trend 2026/6; SAT]. The March 2026 trend analysis reports "positive results on women entrepreneurs, regional cohesion and territorial outreach" [BP §5.5].

7.3 Young people

Start30 is built around the under-30 starter. Its entry tier of HUF 1–5 million scales conditionally to HUF 19.9 million – "safe failure": a small first exposure, a larger tier only on proven operation – with 24 months at 0 per cent and no principal repayment, then 3.0 per cent over 144 months, a foundation training before disbursement, 24 months of mentoring and a performance-linked capital take-over check at 60 per cent of the term [CS VII.4, VIII.7; BP §6.2]. Start30+ serves starters, re-starters and career changers over 30 with a documented "honest failure" module – "the second chance is a right, not a free restart" [BP §6.2; PAL §3.1].

The HSCF authors grade their own evidence for the youth design in three levels: the primary barrier for a young starter is the capital need of the initial loss-making phase, and lack of credit history alone is a bank rejection reason (documented); credit alone is insufficient, so an integrated package is needed (consensus – Rome Directive 8, the Code); and the assumption that services, mentoring and the 24-month grace measurably improve survival is a planning assumption to be tested through the survival, mentoring and literacy indicators with a mandatory "young" breakdown. They explicitly do not claim any generational attitude research or any generalisation about young people's risk appetite [CS VIII.7].

A Hungarian data point on tax design is worth the committee's attention. The legacy portfolio contains no loan to an 18–26-year-old and one to a 27–35-year-old among 32 loans; the March 2026 trend analysis reads the under-representation of the 18–35 cohort not as an acquisition shortfall but as "a structural consequence of the under-25 personal-income-tax exemption and of the 2022 KATA simplified-tax tightening" – salaried employment for the young is subsidised, self-employment was made more burdensome [Trend 2026/6; BP §3.2]. The programme fiche adds that "a significant share of young people opt for employment or emigration rather than entrepreneurship, and succession remains unresolved in thousands of operating businesses" [FICHE-EN pt 2].

7.4 Returnees, migrants, refugees and minorities

Hungary's distinctive product in this field addresses its own emigrants. Hazatérő (Returnee) targets Hungarians returning after at least three years abroad: foreign income and credit history are accepted, the decision is taken within fifteen working days, the loan of up to HUF 19.9 million carries a six-month grace and then 4.0 per cent, the accompanying service is a legal, tax and market-entry integration network, and the capital take-over is conditional on job retention and Hungarian tax residency [CS VII.4; FICHE-EN pt 4; PAL §3.1]. These are returning nationals; the knowledge base contains no migrant or refugee programme.

For minorities, a "Roma Integrated Microcredit" – targeted microcredit combined with services and mentoring for the Roma micro-entrepreneurial layer – is listed as a shovel-ready initiative whose activation depends on a Horizon Europe social-cohesion call [BP §15.6]. Ethnic origin is special-category data under the GDPR and is not collected, so no Roma outreach share can be reported [SAT F19–F27]. A small transferable example of de-risking policy language: a previous nationality-based exclusion of foreign sureties was flagged as "problematic from the point of view of EU law" and reworded to the actual risk content, enforceability in Hungary [DMC 4.2.1].

7.5 Rural populations

Enterprises in disadvantaged districts form a priority circle scored through a yearly-updated settlement classification; outreach in such regions runs through mayors' offices, local press and civil-society organisations [SOIMP 3.2, 5.2, M1.1]. RVA's rural financing practice was included among EU good practices in 2019 and its agricultural financing experience was presented at the Commission's agricultural finance conference in Brussels in 2022 [MS 2019, 2022]. The base product allows preferential rates in disadvantaged regions [EIF-v3.1 §4]. The legacy portfolio, however, is urban – 27 of 32 loans in county seats, one in a village and one in a small municipality; none of the six live loans is rural – which is a caution against presenting Hungary as a rural-microfinance case on current data [Trend 2026/6; SAT D16]. Context: the 2025 drought damaged several hundred thousand hectares and most of the country is drought-prone, which the strategy treats as a planning precondition rather than a cyclical event [CS I.2].

7.6 Vulnerable groups and the household

"The business is the household" is the recurring finding of the survey series. The enterprise was the sole household income for 71 per cent of clients in 2015 and 74 per cent in 2018; 83 per cent said at least one family member depended on them, 2.9 persons per client; average headcount fell from 4.01 in 2018 to 2.93 in 2022; in April 2020 family-run businesses were hit hardest by the pandemic, with an average operating reserve of 1.79 months, while 63 per cent of all respondents held reserves for at most two months; by 2022 the average reserve had risen to 6.15 months but 63 per cent still felt their business in danger [SURV §3, §6–7]. Over-55s are a named vulnerable group; forced entrepreneurs and benefit recipients are addressed through Start30+ and the statutory purposes; the household-stability proxy (missed regular household payments) and the housing-cost burden question are collected on consent [SOIMP 3.1, 5.3; BILL §3].

The historic case studies are the 1990s transition-era self-employment wave that the network was created to serve and the IBM fresh-start loan of 2002, in which a HUF 300 million corporate donation, an agreement between IBM, the Székesfehérvár municipality and RVA, financed about HUF 676 million of loans to 178 redundant workers over the programme's life, later taught at university level in Spain and recognised as EU good practice in 2019 [CS III.4; MS 2002].

8.1 The framework

HSCF's Social Outreach and Impact Management Plan (Hungarian abbreviation SEIMP, English SOIMP – one document, version 1.21, a draft for 2026–2030) applies a Theory of Change from inputs (capital, staff, local AI, the guarantee envelope, advisory capacity) through activities and outputs to outcomes (turnover, digital transition, survival) and impact (rural population retention, reduced poverty risk) [SOIMP §2]. Its stated approach answers the concern the committee itself raised at its kick-off: "we measure not the number of loans placed, but the change", with "a 'minimum viable' indicator set that does not overburden clients but satisfies the EIF's needs" [SOIMP §2, §4]. The year 2026 is the pilot and baseline year; the baseline is locked in the fourth quarter; target bands for 2027–2030 are set only afterwards on actual data [SOIMP §8]. The EU anchors cited are the European Pillar of Social Rights Action Plan (COM(2021) 102), the InvestEU Regulation (EU) 2021/523, the ESF+ Regulation (EU) 2021/1057, the EU AI Act and Part 4 of the Code [SOIMP 1.2]. The draft Act would make a documented social mission, an outreach plan with expected impact and a measurement methodology a statutory duty of every accredited lender – a national example of social performance management written into law [BILL §11].

8.2 The indicators

Sixteen indicators are defined verbatim. Outreach: active clients (monthly); target-group ratio – women, young, over-55, disadvantaged districts (quarterly); depth of outreach as average disbursed loan divided by GNI per capita (annual); average inclusion score (quarterly); bank-exclusion proxy – the share of entrants rejected by a commercial bank in the past 24 months (at application). Economic outcome: business survival at 12 and 24 months; banded turnover change; banded employment change (retained and created as a proxy); portfolio quality – PAR30 and an NPL proxy of the stock over 90 days past due, with the default definition to be adopted from the EIF guarantee agreement after signature; take-up of the three service layers. Social and green: financial-literacy change measured by entry and exit mini-tests (phase 2); a 0–3 digital-maturity self-assessment; the share of disbursements with a green or digital element; a Net Promoter Score; a household-stability proxy; and, from 2027, the difference in survival and PAR30 between client cohorts with and without advisory services [SOIMP §4]. Every relevant indicator is broken down by target group, programme, sector, region and partner. Data are collected at four points – application (T0), twelve months after disbursement (T1), twenty-four months (T2, possibly sampled) and closure (T3) – through five standardised social questions on bank rejection, housing-cost burden, missed household payments, digital maturity and the green or digital purpose of the loan [SOIMP §5].

8.3 Design features and limits

Two design features are transferable. The first is a firewall: the inclusion score and the social questions are not part of the borrower assessment, "a staff member involved in appraisal may see the SOIMP data in the system but may not use them in the credit decision – neither in the client's favour nor to their detriment", rejection texts contain no social reference, and an internal-audit control point enforces the separation [SOIMP 3.3; CPR §6]. The second is the data-protection design: household questions are consent-based and withdrawable, AI-assisted extraction from monitoring notes runs only on the on-premise route under pseudonymous identifiers with the key stored separately, every AI-drafted summary passes human approval, at least 10 per cent of records are validated quarterly against documents or on site, and a data-protection impact assessment is recommended before the baseline closes [SOIMP 5.4, §6–7; CS VIII.6].

The HSCF authors are candid about what the framework can and cannot show. Outreach depth and composition are administrative, near-complete and auditable – fit for EIF and Code reporting. Banded economic outcomes are weaker; self-reported household stability is biased upward by selective non-response; the Net Promoter Score measures service, not impact; macro impact is claimed as contribution, not causation; displacement effects are not measured [CS VIII.6]. Additionality is captured through coded counterfactual statements in monitoring interviews, which can rank cases but "not quantify net effect"; a control-group design – near-threshold rejected applicants, matched public company data or a staggered roll-out – must be chosen before the baseline year closes, and the English fiche commits to one [CS VIII.5, VIII.10; FICHE-EN pt 5]. The headline accountability metric is deliberately simple: two transparent budget lines and the public cost per financed enterprise and per job, comparable with other development-policy instruments [CS X.3]. "What cannot be measured cannot be defended" [EXEC 1].

8.4 The baselines Hungary already holds

Eight years of client surveys give consistent self-reported indicators: additionality of 93 per cent (2018) and 83 per cent (2019); an employment effect reported by 63.1 per cent of respondents in 2018 – 283 jobs in a sample of 347, computed as retention cases times an average of 2.6 jobs plus creation cases times 2.1 – and 60 per cent in 2019; improved living conditions for 85.7 per cent (2017) and 83.7 per cent (2018); satisfaction of 92 per cent and complaints in 1.94 per cent of cases (2018); ratings above commercial banks in all six service dimensions, by 1.85 points on perceived cost of borrowing [SURV §5]. "The consistency of the impact indicators across years is the strongest feature of the series", its analyst writes, while warning that the series is active-client-biased and self-reported [SURV §2, §5]. Earlier, RVA's 2014 survey of 407 clients found that loans had contributed to retaining 208 and creating 174 jobs [CS VIII.9]. At portfolio level the reference values are the RVA JEREMIE sub-portfolio (546 loans, 1.23 per cent expected loss, 0.00 per cent written off, 98.77 per cent programme efficiency, March 2026) and the legacy book (zero write-offs, PAR30 zero since 2024) – both collateralised [CoGC-DATA; MicPro].

9.1 A small provider under evaluation

HSCF is going through the Code evaluation in 2026 as a pre-launch "Small Provider" (fewer than 4,000 active clients and fewer than 35 employees – in fact four employees at the end of 2025 and five at mid-2026) [BMDF]. The initial self-assessment tool was submitted on 15 September 2025; MicroFinanza Rating's feedback in May 2026 asked for evidence on two priority clauses, the annual external audit (2.38) and the size-adjusted internal audit (3.21); the evaluation cut-off date is 30 June 2026; an English package of twenty-five documents was prepared in August 2026 – thirteen core documents of about 133 pages and twelve further policies approved on 24 August; the updated self-assessment of 27 August shows 87 per cent weighted compliance, 114 of 134 clauses and 30 of 33 priority clauses met, with area scores of 96.3 per cent for client protection and governance, 100 per cent for risk management and management information, and 50.8 per cent for reporting, "the gap concentrated on public online disclosure" [BP §1.5, §13; IDX; CoGC-DATA]. The three open priority clauses each carry a time-bound plan tied to the operational launch of guaranteed lending: a voluntary annual external audit from the 2027 financial year (the Foundation is below the statutory audit thresholds), annual public disclosure of the operational self-sufficiency ratio, and annual public disclosure of complaint numbers [CoGC-DATA]. The evaluator requested the business-model description form on 8 September 2026 for its Steering Group at the end of the month; accreditation is targeted for the first quarter of 2027 and re-assessment for 2030 [MFR-NOTE; BP §14.6].

Several clauses have visibly shaped HSCF's rules: clause 2.34 (a documented succession plan for the Director General and Managing Director, inherited from RVA and written into the operational rules); clause 3.21 (an internal audit function as a condition of accreditation and of the EIF relationship); clause 4.4.7 (annual publication of the self-sufficiency ratio – about 3 per cent in 2025 because of the run-off); Area 1 (client protection, price transparency, complaints and pricing rules mapped clause by clause); and the Code's prohibition of tying and hidden costs, which shaped the whole design of the mandatory services – free mandatory elements, a service default that is never a credit event, a one-page "full burden sheet" before contracting, free choice of provider, and the service quota as a right rather than an obligation [SZMSZ 17.2, 18; CPR Annex 3; BDS 12].

Two positions taken in the process matter for the Code's revision. HSCF finds the current Code "built primarily around a financial-undertaking organisational model" and works with a functional-equivalence table for foundation governance (Board of Trustees to Board of Directors, Executive Board to Senior Management) [BP §4.6]. And it applies the proportionality principle through a responsibility matrix in which every dormant function – internal auditor, risk officer, coach network – has a defined activation point at programme launch [STAFF 2, 5].

9.2 Hungary's part in the Code

RVA's fully digitised microcredit technology was audited under the EU JASMINE programme in 2010 ("first of its kind in Europe"); in 2012 Tibor Szekfü represented Hungary as an expert in the preparatory development of uniform European microcredit standards – the later Code – at the Commission's invitation, and is a participant in the current review; HSCF attended the Commission's expert meeting on the Code review in Brussels on 21 October 2025; a new framework is expected in 2027 [MS 2010, 2012, 2025; BP §4.6; CS IX.3]. The Budapest Directives (2013, under the EMN's IPFI initiative) and the Rome Directives (29 September 2016, Microfinet) were drafted with Hungarian authorship; several of their principles are stricter than the Code – social impact before nominal preservation of the fund, no risk-shifting to intermediaries, blameless clients not worse off, mandatory advice, sustainability "interpreted primarily on the basis of the humane approach of a sustainable society, and not on the basis of absolute financial sustainability" [CS II.6, IX.8; HMN-RULES Annex 1]. The HSCF study's own critical view is that the Code "has not lived up to the hopes attached to it" where voluntariness and formal-compliance logic sideline substance; the conclusion drawn is not to resist compliance but to treat preparation as substantive quality development, and to use the 2027 review as a window for Hungarian experience [CS IX.3]. "Code compliance is not the adoption of a foreign yardstick but the formalisation of our own, earlier-developed norms" [CS IX.8].

9.3 National recognition of the Code

The draft Act does not name the Code. It uses a generic hook – "EU microfinance compliance requirements" that an EU financier may prescribe – and makes membership of a European microfinance umbrella organisation and commitment to those requirements a continuous condition of the coordinator's designation, allocates the compliance duty by funding source and ties the rules on the non-repayable element to those requirements [BILL §§2, 19, 20, 24]. The approach lets national law anchor adherence to EU standards without freezing a particular Code version into statute. Under InvestEU the Code is contractual – banks endorse, non-banks must comply within 18 months (brownfield) or 36 months (greenfield) – and HSCF, as a non-bank, aims at substantive compliance before the guarantee agreement [CS IX.3]. Inside the network, acceptance of the Rome Directives is already a membership condition, and the coordinator may recognise qualifications already obtained when accrediting members – a hook for making Code certification a recognised qualification [HMN-RULES 2.2; NOP 2.1.2].

10.1 An affordability-first regime built outside banking law

Because the activity is exempt from the Credit Institutions Act, every element of client protection in Hungarian foundation microlending is voluntary, and HSCF has built it to the standard of Area 1 of the Code. The first principle of its Client Protection and Price Transparency Rules reads: "The Foundation grants only loans which the client is demonstrably able to repay. Lending may not lead to the over-indebtedness of the client." An application must be rejected when repayment would exceed capacity "even if the collateral offered would otherwise be sufficient"; refinancing is allowed only where it demonstrably improves the client's position; "lending whose sole purpose is to conceal a default or to improve portfolio quality is prohibited"; recommending a loan of an amount, term or structure the client does not need is prohibited; staff incentives may not reward volume at the expense of quality; and the credit decision is always taken by a human, the scoring system being decision support [CPR §2–5; LP responsible-lending principles].

The over-indebtedness indicator is specific to micro-enterprises. The "debt repayment period" divides total liabilities by net profit after tax plus depreciation and expresses the result in years; total liabilities deliberately include trade payables, tax and social-security arrears and wage arrears, because "the real debt burden of a micro-enterprise is not bank debt". Bands run from five years or less (best) to more than eleven years, which excludes the applicant under the lending policy; the note records benchmarks of about three years for well-managed Western European firms and eight to ten years for the Hungarian average; for start-ups the projected debt service may not exceed 70 per cent of conservatively adjusted free cash flow; both tests run automatically in CREDINFO as blocking control points before the Credit Committee [INDEBT §2–6]. The indicator was designed for the Phare Microcredit Programme in the late 1990s and has been used in network scoring since; the collateralised portfolios to which it was applied show the 1.23 per cent and 0.00 per cent figures already cited, which the note itself calls "a lower bound rather than a forecast for the unsecured products" [INDEBT §7].

A credit-information enquiry is mandatory and system-blocked before any assessment can close. Because CREDINFO is used across the county foundations and, per HSCF, by at least twenty licensed financial enterprises, it "creates debt visibility in the microfinance sector outside the scope of the Credit Institutions Act – an area in which the coverage of traditional credit information systems is incomplete" and is therefore "also an infrastructure of client protection" [CPR 3.2; STAFF 3.5]. Exposure is capped at HUF 20 million per transaction (about EUR 50,000) and twice that per client under the draft Act, applied as a voluntary prudential regime until the Act enters into force; the classic EUR 25,000 definition is reported separately [CPR 3.4; LP Annex 1].

10.2 Price transparency, complaints and collection

The pricing rules are simple by design: one registration fee of HUF 10,000 plus VAT and no other charge; hidden fees prohibited and fee clauses not in smaller print; written pre-contract information itemising the rate, the 360-day day-count convention, every fee, known third-party costs, the total amount payable and an annualised comparison indicator with an illustrative calculation; a reflection period of at least two working days during which the offer may not be withdrawn or worsened; written decisions within five working days with objective rejection grounds; and prohibited selling practices including tying to unnecessary ancillary services and exploiting inexperience or language difficulty [CPR ch. 4–5; PAL §2]. Complaints are acknowledged within three working days and answered with reasons within thirty calendar days, non-extendable; network partners must forward complaints within one working day; quarterly reports and an annual root-cause analysis are required; and "the absence of complaints may not in itself be regarded as evidence of proper operation" [COMPL §5, 4.4, 9.3]. The only historic statistic is the 2018 survey's 1.94 per cent complaint rate with three quarters resolved satisfactorily [SURV §5].

Collection follows Rome Directive 7 – clients in payment difficulty through no fault of their own may not be left worse off. The first reminder must inform the client of payment-relief options; requesting relief "may not in itself result in any detriment"; intimidation, disclosure to third parties, contact outside 8 a.m. to 8 p.m. on working days and untrue legal threats are prohibited; collateral enforcement is "a last resort" after a documented relief examination; and the Foundation is liable to the client for the conduct of outsourced collectors [CPR 7.2–7.5; DMC 1.2]. The draft Act adds statutory debtor protection in creditor succession – a double written notification, the validity of payments made to the old creditor before notice, and the rule that no repayment may be left "pending, unsettled" during a transfer – a template for portfolio transfers between institutions [BILL §§29(5)–(7), 33].

10.3 Financial education through the lending process

Financial education in Hungarian practice is embedded in the loan rather than delivered as a separate course. Start30 requires a foundation training before disbursement; every guaranteed product carries an intensive preparation module at the end of the grace period – a repayment stress test, a pricing review, reserve building and a trial repayment – because "the months before repayment starts are critical for portfolio quality"; a monthly eight-question mobile micro-report on turnover, liquidity, receivables, overdue debt and orders feeds a green–amber–red early-warning system; and a rescue intervention on red status is "not a sanction but an escalated service" [EIF-v3.1 §5; BDS 4.1, 4.3]. Financial-literacy change will be measured with entry and exit mini-tests from the second phase [SOIMP §4].

The network's track record in education is longer than its lending relaunch. In 2009 RVA made its internet-based lending system available free of charge as a teaching aid to any school or teacher providing financial education; the National University of Public Service has offered a "Microcredit from Public Funds" elective since 2019; the University of Cádiz runs a fully electronic international master's programme with a curriculum written by RVA professionals (launched 2018); a textbook, "Microcredit – From Theoretical Foundations to Practice", appeared in Hungarian, English and Spanish; and a postgraduate microfinance-expert course is under negotiation with Pannon University [MS 2009, 2018, 2019, 2021; CS IX.4].

10.4 Referral and access to funding

No bank-to-microfinance referral scheme of the Dutch type exists in Hungary. The flow is designed the other way: graduation of clients to commercial banks is an explicit goal and success indicator, and the "credit passport" proposal would transfer the repayment history built in the social tier from CREDINFO to development-bank and commercial lenders in standard form, replacing part of their collateral requirements, so that the non-profit tier becomes "the banking system's client-producing channel" – the opposite of crowding out [BP §6.2, §9.4; SZM 5]. Inbound referral is organised through accountants, chambers and local organisations paid a success fee after disbursement [SOIMP M1.2]. The regulatory gap identified at the kick-off by Baltic members – non-bank lenders unable to make tax-deductible provisions – has no counterpart in the Hungarian documents; the accompanying rule HSCF proposes is different, a client-side tax exemption of written-off microcredit principal [CS 0].

11.1 A digital lending platform with a twenty-year record

The Hungarian network's digital practice predates the current AI debate by more than a decade. RVA's internet-based appraisal and management system won a place among the five best European practices at the Microfinance – European Good Practices Award in Milan (2009), was audited under the EU JASMINE programme as a fully digitised microcredit technology, "first of its kind in Europe" (2010), won the innovative-technologies category of the Investment and Innovation in Microfinance Europe Awards in London (2011), was introduced in Norway (2012) and was ranked among the three best enterprise-financing practices of sixty examined in the DIFASS Interreg IVC project across sixteen countries (2012–2014) [MS 2009–2014; CS III.6]. Its successor, CREDINFO, is a cloud-based credit-information, appraisal-management and client-administration system developed with Credinfo Kft. (a disclosed related party): it covers the Code's management-information clauses, provides fully online application with eligibility pre-screening, electronic voting of the Credit Committee, automated delay counting and a coded reminder ladder, client-level de minimis records, quarterly EIF reporting and the four-point impact data model; qualified electronic signature is in technical and legal validation; the system ran uninterrupted online lending through the pandemic [BP §7.3, §8.1; FICHE-EN pt 5; DMC 2.1; SOIMP §9; STAFF 3.5]. The counter-example is the state channel, which the ATM survey found lending "on the basis of a manual written 26 years ago and 'modernised' 18 years ago", paper-based and without impact measurement [CS III.6].

On the client side, the share of surveyed clients planning digitalisation rose from 16 per cent in 2016 to 36 per cent in 2019, and "new ways of reaching customers" was a resilience key in 2022 [SURV §3, §7]. Digital maturity is recorded at application on a 0–3 scale and the green or digital purpose of each loan is tracked; Start30+ carries a digitalisation and AI mentoring module with a final exam; the service catalogue includes a digitalisation and AI package; and non-digital clients are guaranteed paper administration and personal assistance [SOIMP §4; PAL §3.1; BDS 5; CPR ch. 8].

11.2 The AI Use Policy

HSCF's Artificial Intelligence Use Policy (MFKA-SZAB-14, version 2.10, public) is, in effect, an existing set of ethical AI guidelines for a microfinance institution, and it binds the network partners that join as Microcredit Points or Sales Points [AI-POL 1, 7.1]. Its principles are an "AI first" stance for internal processes and the production of educational and service content; compliance with the EU AI Act through a risk-based approach; data sovereignty through a self-operated, network-isolated Local AI Workstation for any client, secrecy-protected or personal data; and Code conformity – "the use of algorithms and machine-learning models may not result in algorithmic bias or discriminatory credit assessment. AI use must support objective equal opportunities and the financial inclusion of vulnerable target groups" [AI-POL 2].

The risk classification has three tiers. Unacceptable and prohibited: subliminal manipulation, social scoring and the exploitation of clients' vulnerabilities. High-risk: any AI supporting credit assessment or debtor scoring and any system processing banking secrecy, business secrecy or sensitive personal data, which run exclusively on the on-premise workstation – "external cloud services (e.g. public ChatGPT) may not be used for processing these data". Minimal-risk: market research, translation, marketing copy and drafts of public educational material, for which cloud tools are allowed under an anonymisation procedure [AI-POL 4.1]. The controls are an approved-systems register kept by the Technology Lead; a mandatory workflow of data classification, tool selection, data-minimised prompting, human quality assurance "with particular attention to identifying algorithmic bias and factual errors", and logging in CREDINFO of "both the AI-recommended result and the human decision approving or overriding it"; process-owner approval before any AI-assisted client communication; an AI use-case register; an incident procedure that suspends the module until investigation concludes; a glossary entry defining "automation bias"; and mandatory training before any staff member, including network partners, gains access to high-risk processes [AI-POL 3, 4.2, 5, 6.4–6.5]. Accountability is unambiguous: "The AI system provides decision preparation only, and may not take any decision producing legal effects for the client without human oversight"; final responsibility for disbursements, pre-qualification and contracts stays with the Director General or Managing Director through the Credit Committee [AI-POL 6.1].

Two applications illustrate the policy at work. In impact measurement, AI extracts social indicators and counterfactual statements from monitoring notes under pseudonymous identifiers on the local route, every AI-drafted summary passes human approval, run logs and pipeline versions are recorded, and a data-protection impact assessment is recommended before the baseline closes [SOIMP §6–8; CS VIII.6]. In risk calibration, a March 2026 AI analysis of the historical JEREMIE portfolio found sector patterns – a modelled loss of about 15.9 per cent in catering against 0.04 per cent in trade – that feed scoring; it is "a decision-support instrument; it does not replace human-in-the-loop credit decision-making" [BP §5.5].

11.3 Digital capacity of small network members

The hub holds the technology; the roughly ten county foundations expected to act as Microcredit Points need to be equipped and trained before 2027. Network services include IT assistance, provision of IT systems and professional training on a fee basis; network staff obtain access to the Local AI Workstation only after training in ethical and secure AI use; the Network Operations Policy is scheduled for revision to codify the coaching and digital role of the Microcredit Points; and coach capacity is to be financed "at programme launch" [HMN-RULES 6.1; AI-POL 6.4; NOP status note; STAFF 3.6]. This documented gap between a technologically advanced hub and thinly staffed spokes is the Hungarian case for the roadmap's digital capacity-building strand.

Hungarian green microfinance is a mandate and a design, not yet a portfolio. HSCF's deed makes "a climate-neutral 'GREEN' Continent" and a resource-efficient economy primary objectives, foresees Green Service and Information Points, an international register and qualification system of green service providers, and a Green Point Network alongside the microfinance network [DEED II.1, II.3; NOP 2.1.1]. Capital-poor losers of the green and digital transition are a formal priority group; the green or digital element of every loan is recorded at application; ESG guardrails include an exclusion list (fossil extraction and the InvestEU sustainability exclusions, to be reconciled with the guarantee agreement), a restricted-activity list, ESG factors in scoring and anti-greenwashing documentation; paperless operation is an environmental objective in its own right [SOIMP 3.1, 5.3; LP Annex 3; BP §2.3, §10.3].

Two products are designed. Green (Zöld) finances energy investments of operating micro-enterprises able to provide collateral – energy efficiency, renewable energy, energy independence – at HUF 1–19.9 million over 120 months without grace, at a cost-of-funds-based rate of indicatively 3–5 per cent, secured by real estate and the majority owner's suretyship and outside the EIF guarantee ("the risk is borne by the collateral and the loan fund"); its readiness is "developed concept" and its performance incentive is under review [PAL §5.1, §7]. The Comprehensive Study describes the complementary line with an early check at 20 per cent of the term and a partial write-off element [CS VII.4]. Green Community (Zöld Közösség) addresses joint energy investments of residential communities affected by energy poverty; its target group is being narrowed, its collateral on community property and legal structure are being clarified, it is excluded from the main funding figures, and its authors note that a residential community is not a micro-enterprise under Recommendation 2003/361/EC, so office-building communities cannot be justified under the programme [PAL §5.1–5.2]. In the business plan the collateralised window – "e.g. the Green Microcredit line" – is modelled at 4 per cent over 120 months with a 5 per cent cumulative final net loss and would scale to about HUF 6 billion a year by 2030, contingent on the uncommitted HUF 40 billion domestic facility [BP §6.1, §11.2–11.3].

Demand signals exist. In 2022 surveyed firms were "preparing increasingly for energy costs" and their free-text suggestions were led by energy-efficiency support; the outreach plan's GREEN package ("cut your utility costs with your own energy development") targets enterprises with high utility costs and its 2026 calendar opens with "planning and energy" [SURV §7; SOIMP M1.3–M1.4]. The country context – about three quarters of primary energy imported, industrial electricity among the most expensive in the EU, a 2025 drought that damaged several hundred thousand hectares in a mostly drought-prone territory – is treated by the strategy as a planning precondition [CS I.2]. RVA's agricultural financing experience was presented at the Commission's agricultural finance conference in Brussels in 2022 [MS 2022]. What Hungary cannot yet offer is green portfolio data or climate-adaptation lending results.

Hungary has no housing or home-improvement microloan and no demand study; what it has is the household dimension of enterprise microcredit and one energy-poverty concept. The deed lists the prevention of homelessness among the public tasks to which the Foundation contributes and states that its microcredit "pays particular attention to preventing the homelessness of vulnerable groups by promoting income generation from individual activity and the transition to self-employment" [DEED II.2–II.3]. The inclusion score carries a voluntary housing and social index; the household questions record the burden of housing costs (rent or loan plus utilities) on a 0–3 scale and missed regular household payments, on consent; the household-stability indicator follows these over the loan life [SOIMP 3.2, 5.3, §4]. From the 2026 fourth-quarter baseline these give a household-level view of housing-cost stress among micro-entrepreneurs, usable for the roadmap's demand evaluation.

The survey evidence on the household–business link is the material Hungary can bring to the social-housing research: the enterprise was the sole household income for 71–74 per cent of clients, 83 per cent had dependent family members, 2.9 persons were affected per client, and family-run businesses held 1.79 months of reserves in April 2020 [SURV §3, §6]. Green Community, addressing residential communities affected by energy poverty, is the only home-improvement-adjacent construction, with its legal and consumer-protection status open [PAL §5.1–5.2]. In enforcement, Hungarian law protects a debtor's only residence: a valid auction bid may go down only to 70 per cent of the opening price, against 50 per cent otherwise [DMC 3.3]. Two adjacent items must not be mis-cited as housing microfinance: the draft Act's condominium clause concerns business-premises communities, and the Széchenyi programme's real-estate purposes are business investment [BILL §4(1)(b); SZM 1].

14.1 Services as the first line of risk management

In Hungarian practice business development services are not an add-on to the loan but the condition on which its social logic rests. Historically, the first model attached advice to the loan and financed it from a separate programme; JEREMIE left mentoring unfunded; the 2016 development-bank channel had no mentor programme – and where advice was cut, outreach worsened [CS III.2, III.5, III.6]. The current design states: "BDS is not an ancillary service but an integral part of the design: pre-disbursement preparation, mentoring, digitalisation and market-entry modules – completion is a condition of the takeover review, financed from the national component at no cost to the client" [FICHE-EN pt 4]. Completing the mandatory programme is a condition of the capital take-over option, not of the loan: "a BDS default is never a credit event", never a ground for termination and can in no way affect a guarantee call; "the incentive is the capital takeover, not punishment" [PAL §6; EIF-v3.1 §5; BDS 3].

14.2 The three-layer model

The BDS Programme Concept (version 1.0, August 2026, an internal design with open implementation questions) gives every client three layers. Layer A is the lender's own free accompaniment, delivered by the coach at the Microcredit Point: a monthly eight-question mobile micro-report, a quarterly consultation, the intensive preparation module at the end of the grace period, a three-level early-warning system and a quarterly coach evaluation report on five dimensions that becomes the basis for further support. Layer B1 is a basic advisory quota for every client – a diagnostic within thirty days of disbursement, an individual mini development plan and one or two targeted services – of HUF 300,000 plus 2 per cent of the loan, capped at HUF 700,000, 70 per cent directed to the diagnostic target areas. A rescue variant (B1-M) of HUF 300,000–500,000 with no own contribution is prescribed on red early-warning status or two consecutive amber signals; three suitable experts are invited by objective criteria and the client chooses within five working days. Layer B2 is enhanced advisory for market entry, business-model change, digitalisation and AI, product development or export readiness in the first 24 months – HUF 1.5–3.0 million, planning average HUF 2.5 million – with eligibility decided by a human on the coach's report under "governed discretion", an own contribution of 0 per cent if prescribed and 10 per cent (at most HUF 150,000) if freely chosen, a project price rather than an hourly rate, and a verifiable written deliverable with a performance certificate [BDS 4.1–4.4, 5–7].

Three principles run through the model. "Money never passes through the client: the programme pays the expert directly after certified performance." Free choice from an open expert register on competitive online offers – "conflict of interest would be created by exclusivity, not by participation" – with a concentration cap of 10 per cent of annual payments per provider and the exclusion of related providers from prescribed rescue work. And market price at market quality, because "a below-market band chooses between two bad outcomes: the quality expert declines, or accepts and cuts corners" [BDS 3, 7.2]. The public-procurement analysis relies on the open-register logic of the Court of Justice's Falk Pharma and Tirkkonen judgments and is flagged for lawyers' review [BDS 11].

14.3 Cost, funding and evidence

The five-year budget on the "generous" path is about HUF 2.83 billion net (about 3.2 billion with VAT for clients who cannot deduct it) for 1,900 transactions – roughly HUF 565 million a year, HUF 1.49 million per financed enterprise and 16 per cent intensity against a HUF 9 million average loan; the authors note that this is 0.38 per cent of the Széchenyi Card Programme's estimated annual interest subsidy, and also that it is non-refundable money that "increases the politically most sensitive part of the programme package; this must not be glossed over" [BDS 9]. Indicative aid content per client is about EUR 8,500 for services and about EUR 14,250 for a Start30 capital take-over, together some 7.6 per cent of the EUR 300,000 de minimis ceiling and falling in different three-year windows [BDS 10]. The funding source is the national component – the fiche and the EIF product description require national resources fully separated from Union programmes, and a ministry-level envelope of HUF 5 billion is "under formalisation" – while the financial plan names a running EU programme and the concept flags ESF+ as a channel to be checked against the ring-fencing decision and cleared with the EIF; the draft Act would let revolving repayments fund basic mentoring and services and allow separate non-repayable grants for them [FICHE-EN pt 6; BP §1.9; FP §3; BDS 10; BILL §§15, 19(4)]. Until public funding exists, market-priced services are delivered by innoGold Pro Kft., a company owned by the Director General and Managing Director, on a client-ordered basis through a signposting structure recorded in the conflict-of-interest register [BP §6.6].

The candour of the concept on evidence is unusual and useful. To the question "where is the evidence that BDS reduces default?" it answers that the effect is treated "honestly as a planning assumption – but this is the intensity at which the effect can be measured at all"; "either we fund it at effective intensity, or we do not call it advisory"; "with an under-sized programme we will never have evidence" [BDS 2, 12]. From 2027 an indicator compares survival and PAR30 between client cohorts with and without advisory services; a pilot of 50–100 clients with a six-month review is planned [SOIMP §4; BDS 14]. The proxies that exist – additionality of 83–93 per cent and employment effects of 60–63 per cent among mentored clients in the 2018–2019 surveys, and zero write-offs across 578 agreements and HUF 3.2 billion since 1992 – are collateralised-era results and not causal evidence for services [SURV §5; BMDF].

14.4 Partnerships

The delivery model is Hub & Spoke: the hub (HSCF) owns policy, funding, the credit decision, CREDINFO and reporting and lends as the single nationwide lender; the spokes – county enterprise development foundations as accredited Microcredit Points – acquire, prepare, monitor and coach, paid a success fee at origination and a monthly per-client fee only for default-free clients; a second tier of sales points (accountants, chambers of industry and agriculture, local organisations) receives a success fee after disbursement; the 2026 pilot targets fifteen partner organisations, 200 qualified leads, more than 20 per cent conversion and a client-acquisition cost below HUF 50,000 [BP §7.1; SOIMP M1.1–M1.5; CS VII.3]. Partnerships on record include the Erasmus for Young Entrepreneurs programme (2019–2021), the ATM for SMEs Interreg Europe project with the Prime Minister's Office as a partner (2016–2022), membership of EMN and partnership with MFC, university cooperation (Cádiz, the National University of Public Service, Pannon University), and a network platform open to service providers, sponsors, IT suppliers and business associations [BP §3.1; MS 2016–2022; HMN-RULES 3.6]. HSCF's evaluation of the state programme proposes a joint working group with the development bank on graduation gates and the credit passport, and the separation of chambers' advocacy role from state-aid intermediation [SZM 5–6].

The Hungarian record is long enough to separate the effects of design from those of circumstance, and the HSCF study draws five conclusions from it [CS III.7, citing Vonnák 2016].

First, the social purpose was served where five conditions held together: a non-profit organisation lent; the funds were decentralised and revolving; the resources came as earmarked grants without a return expectation; risk and operating costs could be charged to the fund; and mandatory advice accompanied the loan. Where these conditions were violated – in the centralised model after 2000 and in part under JEREMIE – costs rose, incentives were distorted and outreach worsened. "The failures came not from the model but from deviations from it – centralisation, incentive distortion, risk shifting."

Second, the proven model is not an experiment. Hungary had internationally recognised, working social microfinance; what brought it to its present state was an institutional reorganisation in 2000 that was never given a legal basis, not a professional failure.

Third, expertise and infrastructure exist but are being eroded. "Hungarian microfinance expertise exists today, but it is tied to a narrow circle – and with the generational change it may be lost if it is not institutionalised." This is why the reform package includes an accredited postgraduate course and a public register of microfinance experts [CS III.6, IX.4].

Fourth, JEREMIE proved that a refinanced, decentralised architecture works – and that it fails its social purpose when risk is left with non-profit intermediaries, when no social objectives are set and when advice is not funded.

Fifth, every settlement attempt – 1998, 2004, 2017 – stalled at the same point: the absence of a statutory-level settlement. The explanatory memorandum of the draft Act puts the general lesson in one sentence: "civil litigation adjudicates a legal relationship; it does not build an institutional system" [BILL general memorandum 3].

Three further observations concern institutions rather than models. The three-way recording contradiction around the historic public microcredit assets – published accounts showing them as the manager's equity, the manager's own pleading calling the state the owner, the ministry stating it has no data – blocked both the "grant it" and the "litigate it" routes and left legislation as the only systemic remedy; the lesson for any Member State is that public microfinance funds parked in private-law vehicles need a statutory basis before problems arise, not after [CS IV.8, VI.1]. The programme's insistence on ownership rather than management of the revolving endowment follows from the EIF's own eligibility logic – an intermediary must bear risk on its own balance sheet – and from the Civil Code: only a creditor can waive a claim [CS VI.4]. And the JEREMIE recovery phase shows how quickly a revolving instrument can be undermined when a coordinating body amends intermediary contracts unilaterally and expects a collection culture alien to the social purpose [CS V.3].

Finally, what remains open should be stated as plainly as what is known. Constitutional and state-aid opinions on the draft Act are outstanding; the qualification of the coordinator as a service of general economic interest and the need for notification are undecided; the current value of the legacy assets awaits the statutory inventory and forensic audit; the EIF decision and the final guarantee parameters are pending; the sources of the operating and service budgets are not committed; and the new programme has, as yet, no outcome data of its own. "A proposal that knows and states what it does not yet know is more defensible than one promising ready answers to everything" [CS XI.7].

Unless stated otherwise, sources are documents of the HSCF knowledge base as of 18 September 2026; English policy texts are working translations and the Hungarian originals are authoritative. Where documents differ, the draft Act governs legal questions, the Comprehensive Study governs professional and numerical questions, and the English programme fiche governs terminology.

External references reproduced with their original attribution: KSH STADAT 9.1.1.8. (2023) and 5.1.1.3 (2024); COM(2007) 708; Regulation (EC) No 1083/2006, Article 78(7); European Court of Auditors Review 06/2021; Regulation (EU) 2021/523 and the EIF InvestEU microfinance guarantee term sheet as checked on 31 July 2026; Regulation (EU) 2021/1057; Regulation (EU) 2023/2831; Commission Recommendation 2003/361/EC; European Code of Good Conduct for Microcredit Provision, Update 2022 (KE-07-22-571-EN-N); Government Resolution 1239/2026 (VIII. 3.); Act CCXXXVII of 2013 §2(1)(f); Act CLXII of 2009; Government Decree 83/2010 (III. 25.); ministry working-committee minutes of 15 January, 18 February and 24 February 2004; letter KABHÁT 6/2/2017 (28 April 2017); ministry reply FEF/48/3/2023 (4 May 2023); Kúria Pfv.V.20.341/2016/5 (29 September 2016); P. Vonnák, Microcredit in Hungary (2016); ATM for SMEs Interreg Europe national survey (2018) and final study (2021/2022).

Abbreviations: BDS – business development services; CoGC – European Code of Good Conduct for Microcredit Provision; EIF – European Investment Fund; EMN – European Microfinance Network; ESF+ – European Social Fund Plus; GBER – General Block Exemption Regulation; HMN – Hungarian Microfinance Network; Hpt. – Hungarian Credit Institutions Act; HSCF – Hungarian Sustainability Centre Foundation; JEREMIE – Joint European Resources for Micro to Medium Enterprises; KSH – Hungarian Central Statistical Office; LHH – disadvantaged (least developed) districts; MAC – Microfinance Advocacy Committee; MFC – Microfinance Centre; MFKA – Magyar Fenntarthatósági Központ Alapítvány; MVA – Hungarian Enterprise Development Foundation; NRPP – National and Regional Partnership Plan; OMA – National Microcredit Fund; OMP – National Microcredit Programme; PAR30 – portfolio at risk over 30 days; RVA – Székesfehérvár Regional Enterprise Development Foundation; SEIMP/SOIMP – Social Outreach and Impact Management Plan; SGEI – service of general economic interest.