Agefi Luxembourg - mai 2026

Mai 2026 25 AGEFI Luxembourg Fonds &Marchés By Dr. Sebastiaan HOOGHIEMSTRA, senior associate in the investment management practice of Loyens & Loeff in Luxembourg S ince the adoption of the AIFMD, the distinction between openended and closedendedAIFs has been one of the foundational taxonomies of European fund regulation. Yet the definition of openended funds, as laid down inCommissionDelegated Regulation (EU) No 694/2014 of 17 December 2013 (“CDR 2014”), reflects regulatory assumptions shapedmore than a decade ago by traditional fundmodels. Designed against the backdrop of classic openended funds and fully lockedup closedended vehicles, that definition is increasingly ill­ suited to today’smarket reality. The rapid rise of evergreen structures, the formaliza­ tion of liquidity management tools (“LMTs”) under AIFMD2,andthegrowingpresenceinEuropeofUS­ inspired tender and interval fund models have col­ lectively exposed the limits of a binary classification built on a single, formal criterion: whether investors may, at their request, have units or shares redeemed orrepurchasedoutoftheassetsoftheAIFpriortothe commencement of its liquidation or winddown phase. In an environment characterized by capped, staged or discretionary liquidity, this criterion risks obscuringmore than it clarifies. This contribution examines how AIFMD 2 has sharpened the regulatory consequences of theopen­ ended/closedended divide and argues that redemption rights, rather than liquidity as such, must remain the decisive classification criterion in evergreen structures. TheAIFMD“Openend Fund” Definition: ABinaryTest under Strain Under the regulatory technical standards on the determination of the type ofAIF, anAIF is classified as openended where, at the request of an investor, its units or shares are redeemed or repurchased out of the assets of the fundprior to the commencement of its liquidation or winddown, in accordancewith the procedures and frequency set out in its rules or instruments of incorporation, prospectus or offering documents. Any fund that does not meet this crite­ rion is closedended. Two clarifications follow from the legislative design. First, distributions are irrele­ vant for classification purposes, as they are neither investorinitiatednor liquiditycreating. Second, sec­ ondarymarket tradability is immaterial, since trans­ fers between investors do not involve a repurchase by the fund itself. That abstraction was workable when the definition was conceived, but it nowproves fragile, particularly in the context of evergreen strategies, where limited redemption windows, notice periods or gatebased mechanisms coexist with indefinite fund duration and continuous capital raising. TheRegulatoryStakes underAIFMD2 Under the original AIFMD, the principal regulatory consequence of an openended classificationwas the requirement to operate a liquiditymanage­ ment framework.AIFMD2, which entered into force on 16 April 2026, has materially altered that assessment. First, openendedAIFsmust nowhardwire at least two predefined LMTs into their constitutional documents, including redemption gates, suspensions, notice periods, swing pricing, antidilution levies and redemptions in kind. The manager is under a positive obligation to activate them where liquidity conditions so require. Liquidity man­ agement is no longer con­ fined to operational decision­ making, but nowdirectly constrains fund structuring at constitutional level. Second,AIFMD2introducedleveragelimitscalibrat­ ed to the openended/closedended distinction. For loanoriginatingAIFs, leverage is capped at 175% of NAV for openended funds and 300% for closed­ ended funds. Thedistinctionthuscarriesdirect,quantifiableconse­ quences for liquidity architecture, leverage capacity and the economic viability of certain fund structures. Refining the BinaryTest: AThreeQuestion Framework forOpenEndedClassification Before turning to specific structural edge cases, it is useful to articulate the conceptual distinction that lie at the core of the openended/closedended classifi­ cation, and that recent practice has brought into sharper focus. The dividing line is not liquidity as such,butthelegalnatureoftheinvestor’sentitlement. TheCDR2014definition,whileformallyclear,masks a structural limitation: it looks exclusively to the for­ mal existence of investor redemption rights, without qualifying their legal nature, degreeof enforceability, or practical operation. Liquidity risk, economic sub­ stance and themechanics of redemption are deliber­ ately abstracted away. What is required is interpretative recalibration: one thatremainsanchoredintheCDR2014text,butreads its criteria dynamically, by reference to their legal enforceability,practicaloperationandrelevancetoliq­ uidity risk, in light of contemporary fundmechanics and regulatorypurpose. Three sequential questions give operational content to that recalibration. Do investors have a redemption right before liquidation? This gateway question asks whether the fund docu­ mentation contemplates any form of preliquidation redemption at all. If not, the fund is closedended by design. Where a preliquidation mechanism exists, the analysis turns to its legal nature and effects. Redemption right or redemption request: does investorinitiativecreatealegalobligationtoredeem? Thisisthecorequestion.Aredemptionrightisalegal­ ly enforceable entitlement exercisable by the investor at its own initiative, on a defined timetable and in accordance with procedures laid down in the fund’s offeringdocuments.Themanagerhasnogeneraldis­ cretiontorefuse;redemptionsmustbeprocessedsub­ ject only to specifically articulated limitations such as regulatoryconstraints,predefinedLMTs,orgatestrig­ gered by objective thresholds. Aredemption request is structurally different: a nonbinding expression of investor preference that the manager may, but is not obliged to, satisfy.Where themanager retainsuncon­ ditionaldiscretiontodeclineorpostponeredemptions indefinitely,withnofixedtimetableandnoobligation togenerateliquidity,noredemptionrightisconferred and the fund is closedended. Do LMTs in the fund documentation manage redemption pressure, or confirm its absence? The final practical check focuses onhowredemption mechanicsareoperationalizedthroughLMTs.Where LMTs are triggeredby investor behaviour andobjec­ tivelydefinedthresholds,theyevidencethatredemp­ tion rights can generate liquidity pressure that must be managed. Where LMTs remain entirely discre­ tionary and are not linked to investorinitiated exit, they confirm that redemption requests do not trans­ late into fundlevel obligations. Structural BorderlineCasesUnder theOpenEnded / ClosedEndedTest In practice, the adequacy of the AIFMD open­ ended/closedendeddefinitionistestednotinabstract, but against a limited number of recurring evergreen fund structures. Three archetypes dominate current market practice and frame the most commercially sensitive classificationquestions. RollingVintage Funds (ClosedEnded byDesign) Thefirstarchetypeistherollingvintagemodel.Under a rollingvintage model, investors subscribe to suc­ cessive vintages, each established for a fixed term.At expiry,investorsareautomaticallyrolledintothesuc­ cessor vintage unless they elect otherwise, in which case their capital is returned. During the life of a vin­ tage, liquidity is absent. Where vintages are structured as legally segregated compartments, as is efficiently achieved under the Luxembourg RAIF regime, each compartment con­ stitutesaseparateAIFwithnoredemptionrightsprior toitsownliquidation.Eachvintageisthereforeprop­ erly classified as closedended. The investor’s ability to decline participation in a successor vintage does not alter this conclusion, as capital is returned at the contractual end of the vintage term, not at the investor’s unilateral initiative during its life. Greatercomplexityariseswherevintagesaredefined contractuallywithin a single legalAIF through series of interests or share classes. In this configuration, a decisionnot to roll into the next vintagemay superfi­ cially resemble a redemption prior to liquidation, potentially suggesting an openended classification underAIFMD.Suchanoutcomewould,however,be misalignedwith the regulatorypurposeofAIFMD2. No intravintage withdrawals occur, no liquidity is requiredduringtheoperativelifeofavintage,andno liquidity stress can arise that LMTs are designed to address. The scheduled returnof capital at the endof afixedinvestmentperiodisnotthemischiefatwhich the LMT regime is aimed. RunOffFunds (ClosedEnded) The second archetype is the runoffmodel.Arunoff fund is anAIF that is not structured around periodic dealing or ongoing portfolio turnover, but rather around the progressive realization of a largely static poolofassetsovertime.Itsdefiningfeatureisthatliq­ uidity is generated exclusively through the orderly monetization of underlying investments in the ordi­ nary course of the strategy, rather than through investordriven redemption cycles. These funds permit investors to submit redemption requests. However, those requests do not give rise to any entitlement to liquidity at a predefined date or frequency. Redemptions are processed only as and when, and to the extent that, the relevant portion of the portfolio is realized in the normal course of asset management. Themanager is under no obligation to accelerate disposals, raise external finance, or other­ wise generate liquidity in response to investor requests. As a result, no liquidity mismatch arises. Investor requests do not dictate portfolio action. Rather, distributions follow asset realizations. This is therefore a redemption request mechanism, not a redemptionright,andthefundremainseconomically and legally closedended notwithstanding the possi­ bility for investors to signal exit intent. The structure of theAIFMD2 liquiditymanagement toolkit confirms this classification by negative impli­ cation.LMTs,suchasgates,noticeperiods,swingpric­ ingorredemptionsinkind,presupposetheexistence of a predefined dealing timetable against which liq­ uidity stress can arise and be managed. In a runoff model, there is no such timetable. Liquiditymaterial­ izesonlyuponassetmonetization,andtheimposition of LMTs would therefore serve no meaningful investorprotection function. SemiLiquid Evergreen Funds with Discretionary Redemptions Themost sensitive archetype is the semiliquid ever­ green fund operating on a NAV basis with discre­ tionary redemption mechanics, comparable to a US tender offer fund. These funds permit investors to submit redemption requests at periodic intervals, while reserving to theAIFMand/or GPabsolute dis­ cretionoverwhetherandwhenrequestsaresatisfied. The mere presence of such a mechanism does not render the fundopenended, provided thediscretion is genuine, substantive and temporally unlimited. Where governing documents make clear that the manager is under no obligation to generate redemp­ tion proceeds, no redemption right is conferred, and thefundisclosedended.Bycontrast,wherediscretion is confined to objectively defined constraints, regula­ tory limits,AML/KYCcompliance or predefined liq­ uidity thresholds, investors hold an enforceable enti­ tlement and the fund is openended. The decisive question is not whether liquidity is expected, butwhether investors can compel it. Outlook Classification as openended or closedended is no longer a technical formality. Under AIFMD 2, it directly affects leverage limits, liquidity structuring andriskmanagement obligations.AIFMsmust doc­ ument their analysis carefully, anchor it in economic substance, and be prepared to defend it to investors and competent authorities alike. The open­ ended/closedended distinction is a functional clas­ sificationdesignedtocalibrateregulatoryobligations to genuine investordriven liquidity risk. Until fur­ ther guidance emerges, disciplined interpretation remains essential. OpenEnded or ClosedEnded? WhyAIFMD 2 Makes Redemption Rights, not Liquidity, Decisive Par Nicolas ROTH, Head of Private Markets Advisory, UBP D epuis quelques tri­ mestres, la dette pri­ vée est devenue un sujet polarisé et polarisant. D’uncôté,lespluspessimistesévo­ quent des valorisations déconnec­ téesdelaréalité,uneexpositionau secteurdeslogicielssurlepointde vaciller et des files de demandes derachatquis’accumulentcomme autant de signaux d’une poten­ tielle crise systémique. Certains vont même jusqu’à pointer des analogies avec la crise de 2008. De l’autre, les gérants publient rapport sur rapport pour démon­ trer que les fondamentaux restent sains, que le levier demeure modeste et le risque systémique inexistant. Comme bien souvent, la réalité est nuancée, et les deux camps ont partiellement raison. Mais tous deux passent à côté de l’essentiel. La lecture de la situa­ tion par les pessimistes n’est pas totalement sans fondement. Dans le segment du «direct lending», l’allocationausecteurdeslogiciels représente une moyenne de plus oumoins 25%des BDC (Business Development Companies). Cette concentration est remise en ques­ tionàdeuxniveaux.Toutd’abord, le déploiement de l’intelligence artificielle met à mal un certain nombre de modèles d’affaires dans les logiciels. Le scénario catastrophiste d’une vaguededéfautsauseindecesec­ teuravecdestauxderécupération nuls reste une vue de l’esprit, pas unscénariocentral.Parailleurs,les demandes de rachat dans les structures de crédit privé ont récemment largement dépassé le plafond trimestriel de 5% sur les véhicules privés, alors que les véhicules listés s’échangent avec undiscountmoyend’environ20% par rapport à la NAV. Ces élé­ ments peuvent effectivement être constitutifs d’une potentielle cor­ rection,maisilssontnécessaireset non suffisants. La réponse apportée par les opti­ mistes est factuelle et se base sur des chiffres solides. Tout d’abord, le levier au niveau des fonds demeure inférieur à 1x pour les BDC, très loin des niveaux de 2008. Les taux de «nonaccrual» frôlent1%delavaleurdemarché, bien en dessous des moyennes historiques. Les pertes réalisées restent largement couvertes par des rendements avoisinant les 10%. Le problème de l’intercon­ nexion avec le système bancaire n’apparaît pas comme une com­ plication aujourd’hui. Quant à la comparaisonaveclacrisedessub­ primes, cet angle est hors sujet car la dette privée n’impacte pas la liquidité interbancaire. La ligne de faille actuelle se situe à un autre niveau, celui du véhi­ cule, entre la santé des actifs sousjacents et le stress des enve­ loppes qui les portent. Les inves­ tisseurs institutionnels disposant de capital de long terme ne sont pas concernés par les turbulences actuelles. En revanche, les véhicules mis en placepourlesinvestisseursprivés, qui sont conçus pour offrir une forme de liquidité partielle, subis­ sent une pression que leur struc­ ture n’a jamais été amenée à tester à grande échelle. La crise se joue donc au niveau du véhicule, ou plusprécisémentdanslaconfiance au véhicule, et ceci dans un contextedefindecycle,cequiam­ plifie chaque signal négatif. Pour un investisseur, le choixn’est pas entre panique et déni. La nuance utile consiste à dissocier l’analyseduvéhiculede cellede la stratégie. Un gérant de qualité dans unmauvais «wrapper» vaut moins qu’un gérant correct dans un véhicule bien structuré. Même si les gérants paient 5% comme convenudanslestermesdufonds, ilestindéniablequ’unstressdede­ mande de sorties aura un impact sur la capacité du fonds à croître. Sélectionnerdanslesegmentdela dette privée aujourd’hui, c’est d’abord comprendre comment le fonds se comportera quand la li­ quidité promise rencontrera l’illi­ quidité réelle. Dette privée : de la nuance dans un débat polarisé ©Magnific