Private credit’s test of patience

A lender’s difficult decisions can begin well before a borrower misses a payment. Investors commit more money to its funds, yet the companies it would most like to finance can borrow elsewhere on better terms. The manager can wait, defend its price, find different borrowers or make a concession. Fundraising figures reveal little about which choice eventually makes its way into the loan agreement.

On 15 October, Reuters reported US$129.15 billion of direct-lending fundraising in the first nine months of 2026, up 35% from a year earlier. That is the highest nine-month total in Preqin’s series, which starts in 2016. Estimated third-quarter lending, meanwhile, fell 28% year on year to US$49.77 billion, according to PitchBook/LCD. ^[S01]

The divergence makes competition among lenders worth investigating. It does not tell us how much cash is available today or how well the new loans have been underwritten. Those questions require following the money to the borrower and examining what the lender receives in return: its margin, the financial adjustments it accepts and the borrower’s capacity to repay without relying on another loan.

Method and scope

January–September 2026 fundraising and July–September estimated lending cover different periods and commercial universes, attributed by Reuters to Preqin and PitchBook/LCD. Proprietary extracts were unavailable. Subtraction cannot establish a cash balance. The refinancing panel covers the same third quarter: US$16.39 billion from private to syndicated lending and US$2.60 billion in the reverse direction. Those flows do not measure losses.

From investor commitments to usable cash

Private credit encompasses debt negotiated outside publicly traded markets and provided by nonbank lenders. In direct lending, a fund finances a company on its own or alongside a small group of partners. It can offer execution certainty, a structure tailored to an acquisition and a closer relationship with management. Loans are commonly held until repayment or refinancing, with limited secondary liquidity. These features help explain the premium that private borrowers have historically paid. ^[S03]

The fund’s own capital has a separate timetable. Institutional investors in a closed-end vehicle may commit money before the manager calls it. A promise to contribute and a transfer to the fund’s bank account are different events. Preqin’s methodological guide distinguishes commitments, capital calls, distributions and the residual value of investments. ^[S02]

Consider a deliberately simple example. Investors commit 100. The fund calls 40, lends 30 and retains 10 in cash. The other 60 remains with the investors. Total commitments, invested capital and available cash describe different things, even when they appear alongside one another in a fundraising presentation. Fees, fund borrowing, loan repayments and recycling are excluded from this illustration; in practice, each can complicate the reconciliation.

Method and scope

Hypothetical example excluding fees, fund debt, repayments and recycling: commitments 100, calls 40, loans 30, cash 10, uncalled commitments 60. Identities: 100 = 40 + 60 and 40 = 30 + 10. Uncalled capital remains with investors. Fees and recycling change the reconciliation in a real fund.

The October comparison has another limitation: fundraising covers nine months and lending covers three. The Reuters report does not supply all the geographical and measurement detail needed to reconcile the two datasets. We did not obtain the underlying Preqin or PitchBook/LCD extracts. Subtracting the figures would manufacture an excess-cash estimate that neither series measures. ^[S01]

There is still an economic tension. A manager has promised to invest within a strategy and must find suitable transactions before the fund’s constraints become binding. Waiting can protect prospective returns; a prolonged delay can also complicate deployment and the next fundraising round. The pressure depends on investment periods, fee arrangements, vehicle terms and investor patience. An industry total cannot establish how strongly any individual fund feels it.

Borrowers with another financing option

A more directly comparable pair of flows appears in the third-quarter data: US$16.39 billion of direct loans were refinanced in the broadly syndicated market, against US$2.60 billion moving in the opposite direction. Refinancing changes the financing channel. It does not, by itself, create a loss for the lender being repaid or reduce the economy’s total lending capacity by the same amount. ^[S01]

A syndicated loan is arranged by a bank and distributed substantially to institutional investors. Some of those loans are acquired by collateralized loan obligations, or CLOs, which pool loans and finance them through different layers of debt and equity. Calling this a return to banks can obscure where the exposure ends up. Banks organize the market without necessarily retaining the entire loan. ^[S04]

This competition is particularly relevant to companies large enough to use both channels. Federal Reserve research published in August classifies companies that borrowed between 2020 and year-end 2025 by their financing histories. Roughly 41% of larger companies had used both markets, against 25% of smaller companies. Conversely, around 57% of smaller borrowers had used private credit alone, compared with 16% of larger borrowers. The size threshold is US$250 million in revenue. ^{[S04] [S05]}

Method and scope

Financing histories of issuing companies borrowing from 2020 to year-end 2025, separated at US$250 million in revenue. Large: private only 15.9%, both markets 40.9%, syndicated only 43.2%. Small: 56.9%, 25.3%, 17.9%. Published small-company shares sum to 100.1% because of rounding; main labels use whole percentages. Shares count companies rather than loan dollars and do not guarantee future market access.

These are observed borrowing histories, not guaranteed access to an alternative lender at any future date. They nevertheless reveal an important asymmetry. A larger company may be able to solicit competing offers when conditions allow. A smaller company is more likely to depend on its negotiated relationship with a direct lender.

That creates a potential selection problem. If the easiest companies to refinance leave a portfolio, the loans remaining may require more attention or capital. The result is plausible rather than inevitable. Early repayment also returns liquidity and can create an opportunity for better reinvestment. Demonstrating deterioration would require comparing departing borrowers, remaining borrowers and new loans while controlling for company size, industry and financial strength.

Our analysis of borrowers able to leave examines reinvestment risk through a specific refinancing. This investigation extends the question to loan terms and lending vintages.

A lower spread can change risk in both directions

A floating-rate loan generally charges a market reference rate plus a credit margin, known as its spread. The reference rate transmits part of the monetary-policy cycle. The spread compensates the lender for factors including credit risk, illiquidity and the work involved in financing a company. A fall in the reference rate and a fall in the spread have different origins, although either can reduce the borrower’s interest bill. ^[S03]

The Federal Reserve Bank of Boston provides evidence beyond fundraising commentary. Its 5 August study of disclosures from 168 business development companies, or BDCs, documents increasing use of capitalized interest alongside spread compression. The researchers consider several explanations, including stronger competition and deliberate reductions in debt-servicing costs intended to limit defaults. ^[S15]

The second possibility matters. Accepting a lower return may help a viable business keep paying. Borrower risk can fall, at least in the near term, while the fund has less income available to absorb a future loss. The outcome depends on how much risk reduction the lender obtains for the revenue it gives up.

A lower margin can therefore be sensible on a safer loan and dangerous on one with weakening protections. Coupon comparisons alone cannot distinguish those cases. The borrower’s debt burden, cash generation, collateral and the lender’s intervention rights all matter.

The denominator is part of the negotiation

Loan documentation makes some of that trade-off observable. In the US transactions advised by Proskauer during 2025, average closing leverage reached 5.1 times EBITDA, against 4.8 times in 2024. 21% of deals were covenant lite; 91% of that group involved companies with EBITDA above US$50 million. These are observations from a law firm’s changing client sample. They describe 2025 rather than contracts being signed this autumn. ^[S06]

EBITDA measures earnings before interest, taxes, depreciation and amortization. It provides a reference for operating performance while excluding cash requirements that can be essential to running a business. Loan agreements may permit further adjustments, or add-backs, for exceptional expenses or expected savings. The evidence supporting those adjustments, their limits and the time allowed for savings to materialize become part of the credit decision. ^[S07]

Suppose a company has 120 million in gross debt and 20 million in annual EBITDA before additional adjustments. Its leverage is six times earnings. If the agreement allows it to add 5 million of expected savings, the denominator rises to 25 million and the ratio falls to 4.8 times. Under a hypothetical five-times limit, the same debt now passes the test. The 120 million still has to be repaid, and the savings still have to be delivered.

Method and scope

Hypothetical contractual example in one currency unit: gross debt 120, annual EBITDA before adjustments 20, expected-savings add-back 5. Ratios: 120/20 = 6.0 and 120/(20+5) = 4.8. Assumed covenant ceiling: 5. Changing the denominator repays no debt and generates no cash. The example does not determine whether a real adjustment is justified.

The arithmetic demonstrates the power of a contractual definition without establishing that an adjustment is improper. Savings from a documented restructuring already under way deserve a different assessment from highly speculative synergies. A useful question is what happens when the expected improvement fails to arrive: when must it leave the calculation, and what rights does that give the lender?

Covenants are contractual undertakings made by a borrower. Some require compliance with a financial ratio at regular intervals. Others restrict actions such as additional borrowing, distributions or asset transfers. A covenant-lite structure may remove a recurring financial maintenance test without eliminating every protection. Information rights, security interests and other restrictions remain matters for the complete agreement. ^{[S03] [S07]}

The location of collateral also matters. Permissions to move assets into particular subsidiaries, incur additional debt or amend agreements through majority votes can change an individual creditor’s position. A senior secured label describes a legal ranking and a claim over assets. It does not determine how much value will ultimately be available for recovery. ^[S07]

Interest that becomes another claim

Payment-in-kind interest, or PIK, is another way to trade current cash requirements against a future obligation. Instead of receiving all its interest in cash, the lender adds a portion to the amount owed. PIK can be agreed from the outset to give a growing company room to invest. It can also be introduced later when the company can no longer meet the original payment schedule. The distinction is essential to assessing the loan. ^{[S15] [S16]}

Consider an illustrative three-year loan of 100 carrying a fixed 10% coupon: 6% is paid annually and 4% is added to principal, both calculated on the balance at the start of the year. Principal reaches 104 after one year, 108.16 after two and 112.49 at the end of year three. Cash interest received over the period totals 18.73, compared with 30 if the full 10% coupon had been paid on an unchanged principal of 100. The example assumes no fees, defaults or interim principal repayments.

Method and scope

Hypothetical opening loan 100 over three years, with 6% cash interest and 4% capitalized on each year’s opening principal. Closing balances: 104; 108.16; 112.4864. Annual cash: 6; 6.24; 6.4896, totaling 18.7296. Display rounded to two decimals without intermediate rounding. All-cash comparison: 10% on constant principal 100, totaling 30. No defaults, fees or interim repayments; no superior-return claim without discounting and risk assessment.

The lender ends up with a larger claim that may itself earn interest, but it has collected less cash along the way. The value of that trade depends on eventual repayment, timing and risk. Adding future principal to coupons already received and calling the larger total a superior return would ignore those differences.

The relevant investigation follows contractual interest through recognized income, cash collection and any subsequent impairment. A receivable can generate accounting income before it is collected; recoverability still requires assessment. A high reported yield does not establish how much of that income borrowers have actually paid. ^[S15]

Valuation adviser Lincoln International’s US data illustrate why a single stress measure is inadequate. In Q2, its size-weighted covenant-default rate fell from 3.1% to 2.7%, while loans with PIK introduced after closing rose from 5.9% to 6.2% of all loans. The measures can move in opposite directions because they describe different events with different denominators. Lincoln’s description of the latter as a shadow default measure is an analytical convention, not a legal determination that every affected loan has defaulted. ^[S16]

Renegotiation can avert a failure while also revealing a weaker borrower. Judging the outcome requires examining what the lender receives in exchange, any fresh equity supplied by shareholders, and subsequent payments and recoveries. An extension can preserve value or merely postpone its loss.

A portfolio can change before its ratios explain why

BDC accounts offer a view of these developments, subject to their own limits. At 30 June 2026, Ares Capital reported loans on non-accrual status, on which ongoing interest recognition is suspended, equal to 2.4% of total investments at amortized cost, against 1.8% at year-end 2025. On a fair-value basis the figures were 1.4% and 1.2%. The denominator is total investments, rather than loans alone. ^[S09]

Cost-based and fair-value percentages answer different questions. Once a troubled asset has been marked down, it has less weight in a fair-value calculation. The resulting lower percentage does not necessarily mean that fewer borrowers are in difficulty. A stock of non-accruing investments also differs from the rate of new defaults occurring during a quarter or a year.

Repayments introduce another effect. In a simple example, five troubled loans in a portfolio of one hundred represent 5% of its positions. If twenty healthy loans are repaid and not replaced, the same five represent 6.25% of the eighty remaining. No additional borrower has deteriorated, but the portfolio is more concentrated. This illustrative count-based calculation is separate from financial statistics that are often weighted by dollar amounts.

The reverse is also possible: rapid lending can dilute older problem assets as a share of the total. Meaningful comparisons therefore require tracking lending vintages, exits and realized losses rather than a single aggregate ratio. Ares is an observable case, not a proxy for the whole industry. Its September-quarter results were scheduled for 27 October and are not anticipated here. ^[S14]

Banks fund the lender as well as the borrower

A direct lender may itself borrow. BDCs combine equity, bonds and bank facilities to finance their portfolios. The operating company’s credit risk then passes through another balance sheet with its own financing costs and payment calendar. The Financial Stability Board identifies these layers of leverage and financial interconnectedness among the vulnerabilities that require monitoring. ^{[S10] [S15]}

Federal Reserve research published on 7 August examines how concentrated those banking relationships can be. Across its supervisory sample from 2012Q3 to 2023Q4, the five largest providers accounted for 64.75% of utilized bank lending to BDCs. This is a historical, reporting-bank sample, not an October 2026 share of the global market. It nevertheless shows why diversification should be examined on the funding side as well as across the loans a fund owns. ^[S11]

Method and scope

The 64.75% concentration covers utilized BDC bank borrowing among Y-14 reporting banks, aggregated from 2012Q3 to 2023Q4. It is neither a current nor a global share. The separate balance sheet is hypothetical: assets 200, debt 100, equity 100; a 10% asset loss leaves assets 180 and equity 80, a 20% equity loss. Debt unchanged, no other flows. The bank remains covered in this example; no automatic bank loss is calculated.

Our investigation of private-credit data gaps explains why measures of bank exposure can cover different perimeters.

The leverage arithmetic is straightforward. An illustrative fund owns assets worth 200, financed by 100 of debt and 100 of equity. A 10% decline in asset value, with debt unchanged at 100, leaves equity of 80: a 20% equity loss. With no other flows, the fund’s investors absorb the initial loss while the bank’s claim remains fully covered by asset value.

Funding pressure can still emerge before insolvency. A portfolio-backed credit line depends on a borrowing base: the assets that qualify for financing, after applying the agreement’s rules and discounts. A decline in that base may reduce borrowing capacity, making a loan harder to finance just when the fund would prefer to retain it. Ares’s financial disclosures explicitly describe borrowing-base restrictions on its facilities. ^[S09]

The Reserve Bank of Australia’s October review likewise describes largely idiosyncratic stress and limited links with the core financial system. It nevertheless highlights redemption risk and the potential for PIK to delay recognition of borrower difficulties. ^[S18]

The scale of potential bank losses should not be exaggerated. In the Boston Fed’s examined perimeter, direct BDC exposures remain modest relative to bank capital and benefit from seniority protections. A transmission channel is therefore not equivalent to an automatic threat to solvency. A higher funding price, reduced credit availability and an eventual principal loss are distinct outcomes that can occur separately. ^[S15]

Evidence that challenges the starting hypothesis

The claim that the whole industry is becoming less demanding runs into contrary evidence. Lord Abbett’s midyear assessment describes new lending conditions that are more favorable to creditors, including stronger protections. It is an interested manager’s view, and we did not obtain a comprehensive contract series to validate it. It nevertheless cautions against mechanically extending a 2025 trend through autumn 2026. ^[S08]

Vehicles also differ in how much time their investors give them. A closed-end fund with stable commitments can wait for a suitable loan. A vehicle offering periodic repurchases must manage subscriptions, loan repayments and investor withdrawal requests within the limits of its governing documents. Liquidity pressure in some funds can coexist with strong institutional fundraising across the industry. The FSB and Federal Reserve distinguish these structures. ^{[S04] [S10]}

Private credit has genuine tools for adaptation: relatively few lenders, direct engagement with shareholders and the ability to amend financing without assembling a dispersed bondholder base. A well-managed restructuring can preserve a business and improve recovery. More negotiations do not necessarily translate into greater ultimate losses. ^[S03]

The mechanisms also matter in Europe, but US percentages cannot be transferred without checking their relevance. European supervisors highlighted interconnectedness, liquidity and transparency risks in September. Their warning supports examination of exposures; it does not provide an equivalent European frequency of PIK usage or contractual concessions. ^[S13]

This lending vintage will supply the answer

The evidence supports three different levels of conclusion. Divergent fundraising and lending trends are documented, within the limits of the commercial datasets cited. The mechanisms through which lenders can surrender income or protection are identifiable. The proposition that record 2026 fundraising is already causing a general deterioration in new-loan terms remains unproven.

Testing it requires following newly originated loans: spreads above reference rates, leverage before and after adjustments, maintenance tests, permitted asset transfers and PIK introduced after closing. Those terms need to be assessed alongside repayments and the quality of companies moving between markets. Deterioration that persists after allowing for company size, industry and loan vintage would strengthen the excessive-competition hypothesis. Preserved protections and capital deliberately left available would point the other way.

The Federal Reserve Banks of Dallas and New York are preparing a US direct-lending pilot survey. The Dallas FAQ, updated on 8 October, schedules fieldwork for 9–18 November 2026, with initial aggregate findings expected in Q1 2027. Participation is voluntary and the survey is separate from individual supervisory activity. That series had not yet produced findings when this investigation was compiled. Existing filings, research and advisers’ samples already provide useful evidence, but they do not all cover the same borrowers or periods. ^{[S12] [S17]}

Fundraising measures a manager’s commercial success. Maintaining terms that properly reflect risk will determine the quality of the loans that success leaves behind. The gap between those achievements will become visible in contracts signed now and cash eventually recovered.

Sources and method

Documentary analysis as of 16 October 2026. Observation dates appear in the text and figures; no future release is anticipated. Numerical examples are teaching calculations, not calibrated fund models or forecasts. Assumptions, formulas and limitations appear in the native notes below the six figures. Full commercial datasets remain unavailable and their aggregates are attributed to their producers. Our guide to reading private-credit risk explains the documents and risks to examine.

1. Reuters / Preqin / PitchBook-LCD · Private credit faces deployment squeeze as fundraising hits record · 2026-10-15

   Reuters reports commercial-provider aggregates for January–September and Q3 2026. The underlying database extracts and full perimeter definitions were not obtained. Nine-month fundraising cannot be netted against quarterly lending.

2. Preqin · Private Capital Performance Data Guide · 2021-03-03 (guide page; current PDF undated)

   The guide page is dated 3 March 2021; the current official PDF is undated. Commitments, paid-in capital including fees, distributions and residual value are distinct. Recycling can make cumulative calls exceed commitments. This is methodology, not a 2026 fundraising extract.

3. Federal Reserve Board · Private Credit: Characteristics and Risks · 2024-02-23 ; source correction dated 2024-02-26

   Historical research used for market structure, floating-rate lending and negotiated workouts. Its historical market-size and return estimates are not treated as current.

4. Federal Reserve Board · Private Credit and Leveraged Loan Markets: Similarities, Differences, and Substitution · 2026-08-11 ; figure notes revised on 2026-08-13

   Borrower histories establish different use of financing markets by company size. The threshold is US$250 million in revenue, not EBITDA. Past use does not guarantee current access; the unit is borrowers, not lending volume.

5. Federal Reserve Board · Private Credit and Leveraged Loan Markets: Accessible Data · 2026-08-11 ; associated note revised on 2026-08-13

   Accessible figure descriptions give the exact published one-decimal percentages. The small-company values total 100.1% because of rounding. Whole-number chart labels are rounded independently and sum to 100%.

6. Proskauer · Proskauer Releases 15th Annual Private Credit Insights Report · 2026-02-09

   The public summary supports 5.1x versus 4.8x US closing leverage and 21% covenant-lite deals, 91% involving EBITDA above US$50 million. These are the firm’s 2025 client transactions, not a random or October 2026 sample. The full report downloaded successfully but its encrypted contents could not be read.

7. Proskauer · Private Credit Prepares for the Next Cycle · 2019-09-18

   Used only for the mechanics of EBITDA adjustments, asset-transfer permissions and creditor voting rights. None of its 2019 percentages is presented as current evidence.

8. Lord Abbett · 2026 Midyear Investment Outlook: Private Credit’s Lender-Friendly Reset · Mid-2026; publication day not displayed

   A manager’s midyear assessment describes tighter new-loan terms. This interested view is treated as counterevidence to test, not an independently established market-wide trend.

9. Ares Capital Corporation · June 30, 2026 Financial Results · 2026-07-29

   Official company disclosure, not a sector estimate. Non-accrual shares use total investments as the denominator, once at amortized cost and once at fair value. They are stocks, not annual default rates.

10. Financial Stability Board · FSB warns on private credit vulnerabilities · 2026-05-06

    The official release and full report’s executive summary were consulted. Layered leverage, interconnectedness and redemption structures are documented vulnerabilities, not proof of bank insolvency or industry-wide losses. Historical market-size estimates are not used.

11. Federal Reserve Board · The Price of Bank Funding Behind Private Credit: Evidence from Business Development Companies · 2026-08-07

    The top-five share is 64.75% of utilized BDC bank lending aggregated over the 2012Q3–2023Q4 supervisory sample. It is neither a current market share nor a measure of expected bank losses.

12. Federal Reserve Banks of Dallas and New York · New York Fed and Dallas Fed Release Frequently Asked Questions on the Private Credit Survey · 2026-09-02

    The 2 September official update confirms voluntary participation and initial aggregate findings expected in Q1 2027. It is a future pilot, not a published data series or an individual supervisory assessment. The Dallas FAQ supplies the subsequent fieldwork dates.

13. EBA / EIOPA / ESMA · ESAs call for vigilance over external dependencies, cyber threats and private credit risks · 2026-09-23

    EU supervisors identify vulnerabilities while assessing broader resilience. US borrower and fund statistics are not transplanted to the European market.

14. Ares Capital Corporation · Schedules Earnings Release for the Third Quarter Ended September 30, 2026 · 2026-10-02

    Third-quarter earnings were scheduled for 27 October 2026. The article’s company balances therefore remain explicitly dated 30 June.

15. Federal Reserve Bank of Boston · Early Warning Signals in Private Credit? What BDC Portfolios Reveal about Emerging Risks · 2026-08-05

    Research on 168 BDCs documents active PIK usage and spread compression, with alternative interpretations and important perimeter limits. Approximate endpoints are not used to invent a continuous time series.

16. Lincoln International · The Lincoln Private Market Index: Earnings Growth Drove a Q2 Rebound, While Private Markets Became More Selective · 2026-08-13

    A valuation adviser’s sample distinguishes PIK introduced after closing from original terms. Covenant-default rates and shares of loans with added PIK measure different things and must not be added together.

17. Federal Reserve Bank of Dallas · Private Credit Survey FAQs · 2026-10-08 (updated)

    Updated on 8 October: the US direct-lending pilot is scheduled for 9–18 November 2026. Participation is voluntary and publication aggregated. These are announced fieldwork dates, not results or an individual supervisory investigation.

18. Reserve Bank of Australia · Financial Stability Review, October 2026, section 1.3 · 2026-10

    October assessment: stress has largely been idiosyncratic and links with the core financial system remain limited. Redemption pressure and deferred recognition of borrower stress through PIK warrant attention. No additional numeric estimate or fundraising-to-deterioration causal claim is inferred.