Private debt in practice: direct lending, unitranche, mezzanine, loan structure and fund vehicles

Advanced12 min read · updated 24 September 2026 · 14 sources

Private debt means loans from funds and other non-banks to companies, projects and other funds that are not traded on an exchange. What private credit is in principle, why the market has grown and how it differs from bonds and leveraged loans is explained in the article Private credit vs. public credit. This article is about practice: which strategies exist, how a loan is structured, which fund vehicles hold it and what is left after defaults, leverage and fees.

Strategies: from senior loans to fund finance

Private debt is an umbrella term; the strategies differ mainly in seniority, collateral and the source of repayment.

Unitranche means that the company has a single loan with a blended interest rate instead of a senior and a subordinated loan (see direct lending). Among the lenders, it is often split by agreement (agreement among lenders) into a senior first-out and a junior last-out piece: first-out is paid first in a default and receives a lower margin, while last-out bears more risk for more interest. Mezzanine sits between senior loans and equity. Part of the interest is often not paid in cash but added to the loan (payment in kind, PIK), plus an equity kicker: warrants or a share of the sale proceeds.

In fund finance, the borrower is a fund. According to ILPA (2024), NAV financings are secured by the value of the fund's holdings, are usually used only once commitments have been largely called, and most often serve to fund early distributions; subscription lines, by contrast, rely on investors' outstanding commitments. According to FSB (2026), these credit lines are the most common way in which banks finance private credit funds.

How a direct loan is structured

• Floating rate: a reference rate (such as Euribor or SOFR) plus a margin, often with a minimum for the reference rate (floor).
• OID and upfront fee: the lender pays out, for example, only 98% of the face value (original issue discount) or receives an arrangement fee. Spread over the term, this raises the yield, all the more so with early repayment.
• Call protection: if the borrower repays early in the first years, a premium is due, for example 2% in the first year and 1% in the second (illustrative values). Or repayment is ruled out entirely at the start (non-call).
• Covenants: maintenance covenants are tested regularly, for example debt relative to EBITDA every quarter; a breach gives the lender negotiating power early on. Incurrence covenants only apply when the borrower wants to do something specific, such as take on new debt or pay out distributions. Loans without maintenance covenants are called covenant-lite.
• Sponsor-backed or non-sponsored: in sponsor-backed loans, the borrower is owned by a private equity fund that can inject additional equity in a crisis. According to Block et al. (2024), a sponsor helps funds lend more and design more effective covenants.

The managers surveyed by Block et al. (2024) (with private debt assets of more than US$390 billion) mainly make cash flow-based loans and, by their own assessment, finance companies and leverage levels that banks would not finance. According to FSB (2026), external sources put the leverage of typical private credit borrowers at 5 to 6 times EBITDA, compared with about 4 times for leveraged loans.

Fund vehicles and leverage

• Closed-end funds: according to Avalos et al. (2025), most private credit funds operate as closed-end funds that lock up capital over their life of typically five to eight years; Cai and Haque (2024) cite lock-up periods of up to ten years.
• Evergreen and semi-liquid funds: no fixed term, with regular but limited redemptions (see semi-liquid fund). According to FSB (2026), some funds received redemption requests above their limits, which in some cases are 5% of fund assets; in the euro area, around 20% of private credit funds are open-ended.
• BDCs: according to Cai and Haque (2024), business development companies are closed-end US investment companies created by a 1980 law. According to Avalos et al. (2025), they manage more than US$300 billion, around 20% of the US market; their debt-to-equity ratio has risen since 2011 from about 0.4 to more than 1 on average, and the statutory limit is 2 to 1.
• ELTIFs in the EU: European long-term investment funds may also originate loans and, since the reform (Regulation (EU) 2023/606 2023), can be marketed to private investors more easily.
• EU rules for loan-originating funds: the revised AIFM Directive (EU) 2024/927 contains specific requirements. According to the summary by the Malta Financial Services Authority MFSA (2024), it limits the leverage of open-ended funds to 175% and of closed-end funds to 300%, limits loans to financial undertakings, other AIFs or UCITS to 20% of the fund's capital per borrower, and requires the fund to retain 5% of each loan it sells on; member states had to transpose it by April 16, 2026.

Leverage at fund level magnifies gains and losses. According to Block et al. (2024), direct lending funds use leverage, but significantly less than banks and CLOs. On top of this comes the leverage of the borrowers themselves and, where applicable, NAV financings. According to IMF (2024), such stacked layers can amplify losses and are hard to monitor because of reporting gaps.

The return on a unitranche loan, worked through

The loss formula is simple: the expected loss per year is roughly the default rate times the loss rate. What this means for the net return is shown by an example with round, made-up numbers.

The example assumes evenly spread defaults, immediately received recoveries and a fixed reference rate; leverage only pays off as long as income after losses exceeds the borrowing costs. Whether 3% defaults a year and a 65% recovery rate are realistic depends on the cycle, seniority and manager; there are hardly any uniform public statistics. According to FSB (2026), actual loan defaults are around 1%, and around 5% when certain restructurings (“selective defaults”) are included.

Interest coverage, PIK and “amend and extend”

The floating rate protects the lender against price losses when rates rise, but shifts the burden to the borrower. Interest coverage (the interest coverage ratio, EBITDA divided by interest expense) shows the buffer; according to Cai and Haque (2024), a value below 1 means that the borrower cannot cover its current interest from cash flow.

Cai and Haque (2024) report a median interest coverage for 2023 of around 2.0 for private credit borrowers and about 2.7 for leveraged loan borrowers. For the euro area, Cera et al. (2025) find that, at the median, firms are more highly indebted and cover their interest less well after private lenders come in.

• PIK portions: if interest is added to the loan instead of being paid in cash, the debt grows, and the reported interest income includes interest that does not arrive in cash. According to IMF (2024), the share of PIK interest in BDCs' interest income has doubled since 2019. According to FSB (2026), PIK is used in about 12% of loans, in about half of cases as an option for the borrower (PIK toggle); such options are associated with a 1 to 2 percentage point higher probability of payment delinquency in the following quarter.
• Amend and extend: instead of declaring a default, lenders extend the maturity, often in return for a higher margin, fees, fresh equity from the sponsor or more PIK. This can be sensible, but it pushes losses into the future and props up valuations; what happens after a real default is described in Distressed debt. According to IMF (2024), covenant breaches often lead to renegotiated terms rather than an actual payment default.
• Valuation: funds value their loans infrequently, often only quarterly, using models; according to FSB (2026), the usual income-based approach leaves managers considerable discretion. On smoothing, see Private credit and Valuing private markets.

What the research shows

The return studies are summarized in the article Private credit; particularly important for practice:

• The right benchmark: in Munday et al. (2018), a leveraged loan index is usually the best fit; direct lending funds had a low beta and a positive alpha. Erel et al. (2024), by contrast, show that funds also contain equity risk: with equity and bond factors, no significant excess return remains after fees, while before fees it is positive.
• Manager and timing: according to Böni and Manigart (2022), results vary widely across strategies and quartiles, a manager's past results predict later ones, and credit market conditions strongly shape returns. Managers can time them only partially.
• Drivers of growth: Avalos et al. (2025) find that private credit grows when policy rates are low and banking systems weak: a policy rate one standard deviation (4.14 percentage points) lower is associated with around 12% more private credit.
• Europe is smaller and more externally financed: according to Cera et al. (2025), private credit funds in the euro area managed around €106 billion in the second quarter of 2024, compared with about €1.2 trillion in North America; in about 65% of 1,240 private credit deals with euro area firms, the sponsor came from outside the euro area.

Supervision: what the IMF, ECB, BIS and FSB are watching

• FSB: the FSB (2026) estimates private credit at US$1.5 to 2 trillion at the end of 2024. Banks' reported direct exposures are below 0.5% of their total assets; the report sees risks in interconnections between non-banks, liquidity mismatches and data gaps. Private credit, it says, has not yet been tested in a long downturn.
• IMF: IMF (2024) cites valuation, hidden leverage, interconnections and liquidity as vulnerabilities (see Private credit).
• BIS: according to Avalos et al. (2025), the young asset class has not yet been through a full credit cycle; low sector diversification and more private investors could make funds more vulnerable.
• ECB: Cera et al. (2025) do not consider euro area exposures alarming overall, but see risks in concentrated exposures, opaque interconnections and bank lending to private market players. In addition, data gaps prevent a complete assessment.

Advanced: loss, seniority and valuation

• Spread and loss: in intensity models (reduced-form models) following Duffie and Singleton (1999), the credit spread is approximately the default intensity (hazard rate) times the loss rate.
• Recovery rate and cycle: Acharya et al. (2007) show that creditors recover less when the whole industry is in crisis. Recovery rates are therefore low precisely when many firms in an industry are in difficulty; calculations with fixed averages underestimate the risk in a downturn.
• Calculating first-out and last-out: if a unitranche of 100 is split into 40 first-out and 60 last-out and 50 is recovered in a default, first-out receives 40 (100%) and last-out 10 (16.67%). The recovery rate of the whole unitranche remains 50%; the split only shifts the risk between the lenders.
• Spread duration: if a loan is discounted in a model using the spread of comparable loans, it loses around 3% of its value when the spread rises by 1 percentage point with three years of spread duration, even without a default.

Limitations and common mistakes

• Confusing the coupon with the return: after defaults, fees and non-cash PIK interest, much less than the coupon is left.
• Overlooking leverage: the leverage of borrowers, of the fund and of NAV financings stacks up.
• Liquidity of semi-liquid funds: redemption limits kick in precisely when many investors want to exit (redemption gate).

How Portfolio Lab calculates

Portfolio Lab has no time series for private debt itself. Asset classes offers exchange-traded proxies: Listed private credit (BDCs) from 2013, Leveraged loans (senior loans) from 2000, CLO AAA from 2020, as well as high-yield bonds in US dollars and euros. They are derived from exchange and fund prices and are therefore unsmoothed. In Look-through, the taxonomy assigns funds with “Private Debt”, “Private Credit” or “Direct Lending” in their name to the category Private markets → Private debt → Direct lending, with the risk driver Illiquidity premium (private debt) and the high-yield credit spread as a second driver; BDCs end up under Credit → Private credit (listed). The rule does not recognize funds without these keywords in their name, such as pure mezzanine funds; in that case, a custom mapping that an admin can save helps. The app does not calculate loan returns with OID, defaults and fund leverage as in the example. More in the asset classes methodology.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.