Private credit and direct lending to smaller services firms to 2028
What a $10 million-EBITDA company actually pays, and who will still be lending to it in 2028.
One listed lender's lower-middle-market portfolio averages $11.8 million of EBITDA, yields 12.6%, and comes with the lender owning an average 36% of the equity. Another's averages $247 million of EBITDA, yields 10.1%, and comes with 5.9 times net leverage and no equity at all. Both call what they do private credit. They are not the same business, and the difference decides what a smaller company can borrow and on what terms.
This is a forecast rather than a market summary: seven structural shifts with the evidence under each, the loop that drives them, five scenarios with odds at December 2027 and December 2028, what commoditises and what keeps its scarcity, and seventeen dated calls a named public source will settle. Every reading carries a source grade. Every judgment carries a confidence. The odds are judgments too, and the page says so.
Two markets wearing one name, 2.5 points and twenty-one times apart
Average EBITDA, one LMM portfolio
94 lower-middle-market companies at one listed lender, 30 Jun 2026, against $247M at an upper-market lender.
Weighted-average yield on that portfolio
Against 10.1% on accruing debt at an upper-middle-market lender, and 10.9% at a second LMM lender.
Average equity the lender also holds
Fully diluted, across all of that lender's lower-middle-market portfolio companies.
Median spread, 227 filed tranches
FD count over one LMM lender's filed schedule at 30 Sep 2026; median floor 2.00%, range 4.75% to 12.00%.
By the end of 2028 the capital has come down-market, and what a smaller company pays for it has stopped being a secret
The lower middle market's premium exists because nobody can see it. Every force now acting on this market makes it more visible.
We put the move-down-market shape at 34% by December 2028 against 26% a year earlier, and today's two-tier market down from 38% to 22%. The second panel is how a $5m to $25m EBITDA services firm's next deal actually gets funded. Each column sums to 100, and both are Frontier Desk judgments rather than model output.
The financing market
Which shape compute financing is in at each date. Frontier Desk judgments, not model output
| Scenario | Dec 2027 | Dec 2028 |
|---|---|---|
| Two-tier marketToday's shape: a visible upper market and an opaque lower one, with a 2.5-point yield gap between them | 38% | 22% |
| Move down-marketLarger funds compete for $10m-EBITDA borrowers and the premium compresses | 26% | 34% |
| Credit reckoningPIK, amendments and non-accruals rise enough to pull capital back out | 16% | 22% |
| Banks returnRates, rules or competition bring bank lenders back to small commercial borrowers | 12% | 10% |
| Equity-shaped creditThe lender-as-part-owner structure becomes the normal deal below $25m of EBITDA | 8% | 12% |
How a new large GPU purchase is funded
The dominant structure for financings signed in each year. Today the secured facility still leads
| Structure | Dec 2027 | Dec 2028 |
|---|---|---|
| Non-bank direct lenderA unitranche or first-lien loan from a fund, often with warrants or equity attached | 41% | 45% |
| Bank or government-guaranteed loanA commercial bank facility, in many cases with a federal guarantee behind part of it | 31% | 24% |
| Seller financing, earnout or rolloverThe seller funds part of the price and carries the risk of the first two years | 17% | 18% |
| Equity rather than a new debt layerA majority or control investor replaces the debt question with an ownership one | 11% | 13% |
Odds are Frontier Desk judgments at the two horizons, set against the triggers and early markers that decide each one in Scenarios, and the method is in Sources and method.
Bottom line
Eight forward-looking judgments, each with the evidence under it and the confidence we hold it at.
- 01Private credit is two markets wearing one name, and the gap between them is the whole opportunity.Main Street Capital's lower-middle-market portfolio at 30 June 2026 held 94 companies with average EBITDA of $11.8 million, a weighted-average annual effective yield of 12.6%, 99.4% of debt secured by a first priority lien, and the lender holding an average 36% fully diluted equity stake in every one of them. Blue Owl Capital Corporation's portfolio at 30 September 2026 held 233 companies whose weighted-average annual revenue was $1.07 billion and annual EBITDA $247 million, at a 10.1% yield on accruing debt, with interest coverage of 2.0 times and net leverage of 5.9 times. FD estimate: $247M ÷ $11.8M = 20.9 times the EBITDA, for 2.5 percentage points less yield. Anything written about private credit that does not say which of these two it means is not about your company.Confidence: high. Every figure is quoted from one of two registrants' own quarterly filings, and the ratio is shown. The two filings use different quarter ends, which we state rather than adjust.
- 02Banks are tightening on exactly this borrower, and told the Federal Reserve so.In the October 2026 Senior Loan Officer Opinion Survey, covering the third quarter, banks reported that standards for commercial and industrial loans "to large and middle-market firms were basically unchanged, while standards tightened for C&I loans to small firms", and specifically reported "tightening for collateralization requirements and use of interest rate floors for small firms". They cited "a less favorable or more uncertain economic outlook and a worsening of industry-specific problems". They also reported weaker demand across all sizes, attributing part of it to "customer borrowing shifting to more attractive bank or nonbank funding sources". A survey in which banks tighten on small firms and say their customers are going elsewhere is the clearest statement of this market's direction available from a public source.Confidence: high. One authoritative primary survey, quoted in its own words. What it does not give is the size of the shift, because SLOOS reports net percentages of respondents rather than volumes.
- 03The stress is visible at the bottom first, and it arrives as an amendment before it arrives as a default.Capital Southwest, a lender specialising in lower-middle-market companies, reported payment-in-kind interest income of $4,981 thousand for the quarter ended 30 September 2026 against $2,794 thousand a year earlier, and said the increase was "due to an increase in the number of amendments allowing portfolio companies to elect to pay interest in-kind". FD estimate: PIK was $4,981K ÷ $67,158K = 7.4% of total investment income, against $2,794K ÷ $56,945K = 4.9% a year earlier. Investments on non-accrual status were approximately 1.4% of the portfolio at fair value and 3.5% at cost, against 1.1% and 2.4% six months earlier. The Financial Stability Board, writing on 6 May 2026, identified reliance on payment-in-kind loans as something that "can also signal deteriorating credit conditions".Confidence: high. The PIK figures, the stated reason and the non-accrual percentages are all quoted from one registrant's own filing, and an international body independently names the indicator. One lender is not the market, which is why the signposts track it at three.
- 04A lower-middle-market loan is priced off a spread that is published, if you read the schedule.Frontier Desk counted every floating-rate tranche in Capital Southwest's consolidated schedule of investments as filed for 30 September 2026: 227 tranches referencing SOFR, with a median spread of 6.75%, a mean of 6.92%, a range of 4.75% to 12.00%, and a median interest rate floor of 2.00%. SOFR was 3.94% on 4 November 2026. FD estimate: 3.94% + 6.75% = 10.69% all-in at the median, against the 10.9% weighted-average yield the same filing reports. The comparison that matters: Blue Owl reports a weighted average spread on floating rate debt of 5.6% for upper-middle-market borrowers. FD estimate: 6.75% − 5.6% = 115 basis points, which is what being small costs, before the equity.Confidence: high on the count, which anyone can repeat from the same filing. Moderate on the comparison, because the two lenders define and weight their spreads differently and we are comparing a median of tranches with a weighted average of a portfolio.
- 05In the lower middle market the lender is also an owner, and that is the structure rather than an accident.Main Street describes taking "direct equity investments and/or receives nominally priced equity warrants in connection with a LMM portfolio company debt investment", and reports that at 30 June 2026 it "had equity ownership in all of its LMM portfolio companies, and the average fully diluted equity ownership in those portfolio companies was 36%". Its lower-middle-market portfolio is 71.1% debt and 28.9% equity at cost, and carries at $3,205.6 million of fair value against $2,547.7 million of cost — FD estimate: $3,205.6M ÷ $2,547.7M = 125.8% of cost. For an owner, this is the single most important fact on the page: the capital on offer below $25 million of EBITDA is not a loan with a rate. It is a part-sale with a coupon, and the coupon is the part people negotiate.Confidence: high. Every figure is quoted from one registrant's own filing, including the structural description in its own words. That this is the market norm rather than that lender's house style is an inference, supported by the Financial Stability Board's description of private equity and private credit interlinkages but not established by it.
- 06The capital above is coming down, and it arrives as smaller cheques before it arrives as lower prices.Ares Capital made 53 new investment commitments in the quarter ended 30 September 2026 against 80 a year earlier, at an average new commitment of $40 million against $49 million. Its new commitments funded at a weighted average yield of 10.4% against 10.0% a year earlier, while loans exited or repaid in the period carried 8.7% against 10.9%. Fewer, smaller cheques at a higher yield is what a large lender looks like while it is moving down-market and repricing as it goes. 63% (judgment) that a direct lender with over $10 billion of assets publicly states a lower-middle-market or small-company strategy by 31 December 2027.Confidence: moderate. The commitment counts, sizes and yields are quoted from one registrant's own filing. Reading a smaller average cheque as down-market movement rather than as a quieter quarter is an inference, and one quarter is a short series.
- 07The sector concentration runs straight through services, and the Financial Stability Board says so.The Financial Stability Board's 6 May 2026 report defines private credit as "nonbank direct lending to medium-sized companies negotiated on a bilateral basis", sizes it at "between $1.5 trillion and $2 trillion" at the end of 2024 with the United States at around $1 trillion, and finds that "[t]he largest borrowing sectors in private credit are services, technology, and healthcare". It also reports that approximately 75% of borrowers have EBITDA below $100 million, and that borrowers are concentrated around the single-B rating bucket with leverage of 5 to 6 times debt to EBITDA against about 4 times in leveraged loans, with true leverage possibly "closer to 7x" once earnings adjustments are taken out. A services company looking for debt in 2027 is looking in the most crowded part of the most crowded market.Confidence: high on the attribution and the quotations, which are from an international body's own report. Moderate on the figures themselves, several of which that body attributes to commercial research it did not perform, and which it presents as estimates with ranges.
- 08The thing that will not commoditise is the lender's willingness to underwrite a business rather than a model.Below $25 million of EBITDA there is no public rating, no syndicated comparable, and in most cases no sponsor to stand behind the plan. The Financial Stability Board notes that borrowers predominantly in private credit "typically lack public ratings" and that this "tends to create transparency challenges". What that means on the ground is that the decision rests on judgment about a specific company — its customer concentration, its owner's role, its contracts, its records — and that is the one input that does not scale with fund size. Capital will keep arriving. Underwriting capacity at this size will not keep pace, and the gap is where the terms get decided.Confidence: moderate. The transparency finding is quoted from an authoritative report. The conclusion that underwriting capacity rather than capital is the binding constraint is Frontier Desk's judgment, and it is the assumption most likely to be wrong on this page.
The baseline
Where lending to smaller companies stands on 5 November 2026. This band is collection, not analysis: every row is a reading from a named public source or a count Frontier Desk performed over a filed schedule of investments, graded A, B or C, with the arithmetic shown and no inference attached.
The readings below are the position the rest of this report forecasts from: a market of $1.5 to $2 trillion whose largest borrowing sector is services, a 2.5-point yield gap between its two halves, banks tightening on the smaller half and saying so to their regulator, payment-in-kind income rising fastest at the bottom, and a lower-middle-market lender that takes 36% of the equity as a matter of course.
Average EBITDA, one LMM portfolio
94 lower-middle-market companies at one listed lender, 30 Jun 2026, against $247M at an upper-market lender.
Weighted-average yield on that portfolio
Against 10.1% on accruing debt at an upper-middle-market lender, and 10.9% at a second LMM lender.
Average equity the lender also holds
Fully diluted, across all of that lender's lower-middle-market portfolio companies.
Median spread, 227 filed tranches
FD count over one LMM lender's filed schedule at 30 Sep 2026; median floor 2.00%, range 4.75% to 12.00%.
Every reading, with its source
Publicly observableGrades describe the source, not the conclusion. A — record: a filing with a securities regulator, a central bank or international body's own published survey or report, or a published rate series. B — party: an organisation's own page or announcement about itself. C — crowd: aggregators and research summarised by a body that did not perform it, carried with its attribution.
Confidence describes Frontier Desk's inference from those readings: high where two or more independent sources agree and none contradicts, moderate where the direction is corroborated but the magnitude is not, low where a single source or an analogy carries it. The odds on this page are judgments of the same kind, not the output of a model, and they are stated as judgments wherever they appear.
| Reading | Source | Figure | Grade |
|---|---|---|---|
| Size of global private credit, defined as nonbank direct lending to medium-sized companies | Financial Stability Board, 6 May 2026 | Between $1.5 trillion and $2 trillion at end-2024; the US around $1 trillion | A · record |
| For comparison, the institutional leveraged loan and high yield markets | The same report | Approximately $1.5–$1.7 trillion and approximately $2 trillion | A · record |
| Largest borrowing sectors in private credit | The same report | Services, technology and healthcare | A · record |
| Share of private credit borrowers below $100m of EBITDA | The same report, attributing Fitch | Approximately 75% | C · crowd |
| Where private credit borrowers' ratings cluster | The same report | Around the single B bucket, against a broader range in leveraged loans and high yield | A · record |
| Private credit borrower leverage | The same report, attributing external sources | 5–6x debt to EBITDA, against about 4x in leveraged loans; possibly closer to 7x excluding adjustments | C · crowd |
| Middle-market CLO borrowers without cash flow to cover interest | The same report, citing a study | 10%, many rated B- with high debt to EBITDA | C · crowd |
| Direct bank lending to private credit funds | The same report | About $220 billion of drawn and undrawn credit lines in member data; commercial estimates more than twice as large | A · record |
| Bank lending standards for small firms | Federal Reserve, October 2026 Senior Loan Officer Opinion Survey | Tightened for small firms while unchanged for large and middle-market firms | A · record |
| What banks tightened specifically for small firms | The same survey | Collateralization requirements and use of interest rate floors | A · record |
| Why banks said they tightened | The same survey | A less favorable or more uncertain economic outlook, and a worsening of industry-specific problems | A · record |
| What banks said about demand | The same survey | Weaker across all sizes, partly from borrowing shifting to more attractive bank or nonbank funding sources | A · record |
| Lower-middle-market portfolio size and profile | Main Street Capital Form 10-Q, quarter ended 30 Jun 2026 | 94 companies, $3,205.6m fair value against $2,547.7m cost, 71.1% debt and 28.9% equity at cost | A · record |
| That portfolio's borrower size and yield | The same Form 10-Q | Average EBITDA $11.8m; weighted-average annual effective yield 12.6%, or 12.1% including non-accruals | A · record |
| Equity the same lender holds alongside the debt | The same Form 10-Q | Equity in every LMM portfolio company; average fully diluted ownership 36% | A · record |
| What that lender means by lower middle market | The same Form 10-Q | Annual revenues between $10m and $150m; investments of $5m to $125m; terms of five to seven years | A · record |
| The same lender's private loan portfolio, for contrast | The same Form 10-Q | 86 companies, revenues $25m to $500m, average EBITDA $39.3m, yield 10.2% | A · record |
| A second LMM lender's yield and portfolio growth | Capital Southwest Form 10-Q, quarter ended 30 Sep 2026 | Yield on debt investments 10.9%, from 11.5%; average debt cost basis up 25.4% to $2,152.8m | A · record |
| That lender's payment-in-kind interest income | The same Form 10-Q | $4,981k for the quarter against $2,794k a year earlier, from amendments allowing PIK election | A · record |
| PIK as a share of that lender's total investment income | FD estimate from the same Form 10-Q | 7.4% ($4,981k ÷ $67,158k), against 4.9% a year earlier | A · record |
| That lender's non-accruals | The same Form 10-Q | 1.4% of the portfolio at fair value and 3.5% at cost, from 1.1% and 2.4% at 31 Mar 2026 | A · record |
| Spreads actually charged to lower-middle-market borrowers | FD count over that lender's filed schedule at 30 Sep 2026 | 227 SOFR tranches: median spread 6.75%, mean 6.92%, range 4.75%–12.00%, median floor 2.00% | A · record |
| Upper-middle-market borrower profile | Blue Owl Capital Corporation Form 10-Q, quarter ended 30 Sep 2026 | Weighted average revenue $1.07bn, EBITDA $247m, interest coverage 2.0x, net leverage 5.9x, loan-to-value 47% | A · record |
| That lender's yield and spread | The same Form 10-Q | 10.1% on accruing debt at fair value; weighted average spread on floating rate debt 5.6% | A · record |
| That lender's scale and position size | The same Form 10-Q | 233 portfolio companies, $15.10bn fair value, average debt investment $59.3m, fund net leverage 1.13x | A · record |
| New commitment activity at the largest listed BDC | Ares Capital Form 10-Q, quarter ended 30 Sep 2026 | 53 new commitments against 80 a year earlier; average size $40m against $49m | A · record |
| Yields on new against repaid loans at that lender | The same Form 10-Q | Funded at 10.4% against 10.0% a year earlier; exited or repaid at 8.7% against 10.9% | A · record |
| The floating-rate base all of the above is priced off | Federal Reserve Bank of New York published reference rates | SOFR 3.94% on 4 Nov 2026 | A · record |
| The fixed-rate alternative, for comparison | US Treasury published daily yield curve | Five-year 4.69% on 4 Nov 2026, from 3.74% on 2 Jan 2026 | A · record |
One count on this page is ours rather than a quotation, and it is worth saying exactly what was done. A business development company files a consolidated schedule of investments with every quarterly report: one line per tranche, with the reference rate, the spread over it, any floor, the principal, the cost and the fair value. It is the most detailed public record of private credit pricing that exists, and almost nobody reads it as a dataset.
Frontier Desk extracted every floating-rate line in Capital Southwest's schedule as filed for 30 September 2026. There are 227 tranches referencing SOFR. The median spread is 6.75% and the mean is 6.92%. The lowest is 4.75% and the highest is 12.00%. The median interest rate floor is 2.00%. With SOFR at 3.94% on 4 November 2026, the median borrower in that book is paying FD estimate: 3.94% + 6.75% = 10.69%, which is close to the 10.9% weighted-average yield the same filing reports and confirms the count is reading the right column.
The use of this is not the median. It is the range. A spread of 4.75% and a spread of 12.00% were charged by the same lender in the same quarter to companies of broadly similar size. Almost all of that difference is about the specific business, and almost none of it is about the market. An owner who treats the first term sheet as the market price is misreading a 725 basis point distribution as a number.
Hold those readings together and the next two years are already implied. A market this size with this few borrowers at the bottom will send capital down the size curve. Banks that tighten on small firms hand those firms to somebody else. Payment-in-kind income rising from amendments is a credit cycle starting at the bottom, not a product innovation. And a 725 basis point spread range tells you the single biggest lever an owner has is not timing the market but being underwritable.
Structural shifts underway
Seven shifts are already running. Each separates what the sources say from what we infer from them: the evidence first with its grade, then the reading with its confidence, then how fast, then the specific thing that would reverse it. These are the inputs to the scenarios.
Capital moves down the size curve, chasing the spreadAccelerating
Speed: 12 to 36 months
What the sources say. One lender's lower-middle-market portfolio yields 12.6% on companies averaging $11.8 million of EBITDA; another's upper-middle-market portfolio yields 10.1% on companies averaging $247 million. The Financial Stability Board sizes the market at $1.5 to $2 trillion at end-2024 with the US around $1 trillion. The largest listed business development company made 53 new commitments in the quarter to 30 September 2026 against 80 a year earlier, at an average of $40 million against $49 million, funded at 10.4% against 10.0%.
Evidence grade: A on the portfolio figures, which come from three registrants' own filings; A on the market size, from an international body's own report.
What we read into it. A 2.5-point yield gap between two halves of the same asset class, with far more capital on the low-yielding side, is an arbitrage that capital will eventually close. It closes from the top down, because large funds can move small but small funds cannot move large. The early stage looks like smaller average cheques at larger lenders, which is visible now. The late stage looks like a $10 million-EBITDA company receiving a term sheet from a fund that would not have returned its call in 2024, and spread compression following about a year behind.
Confidence: moderate. The gap and the shrinking cheque size are both quoted. The causal link between them is our inference, and one quarter of commitment data is thin evidence for a trend.
What would reverse it. The yield gap widening rather than narrowing over two consecutive quarters, which would mean the risk premium on size is being repriced upward faster than capital can arrive.
Banks withdraw from small commercial borrowers, in their own wordsRunning
Speed: Already here
What the sources say. The Federal Reserve's October 2026 Senior Loan Officer Opinion Survey reports that standards for commercial and industrial loans "to large and middle-market firms were basically unchanged, while standards tightened for C&I loans to small firms", with tightening "for collateralization requirements and use of interest rate floors for small firms", citing "a less favorable or more uncertain economic outlook and a worsening of industry-specific problems". Banks also reported weaker demand, partly from "customer borrowing shifting to more attractive bank or nonbank funding sources".
Evidence grade: A · record — a central bank's own quarterly survey of its supervised institutions.
What we read into it. The survey is unusual in reporting divergence by borrower size in the same quarter: unchanged for the large, tighter for the small. Two of the specific tightenings — collateral and interest rate floors — are exactly the terms a smaller borrower has least room to argue about. The effect is not that credit disappears; it is that the marginal small borrower is handed to a non-bank at a price set by somebody with no deposit funding and a 12% return target.
Confidence: high on the finding, which is quoted. Moderate on the magnitude, because the survey reports net percentages of respondent banks rather than loan volumes, and does not say how many borrowers moved.
What would reverse it. Two consecutive surveys reporting easing standards for small firms, which would mean banks are competing for this borrower again and the non-bank premium has to come down to meet them.
Stress shows up as an amendment long before it shows up as a defaultRunning
Speed: Decisive through 2027
What the sources say. Capital Southwest reported payment-in-kind interest income of $4,981 thousand for the quarter ended 30 September 2026 against $2,794 thousand a year earlier, attributing it to "an increase in the number of amendments allowing portfolio companies to elect to pay interest in-kind", with non-accruals at 1.4% of the portfolio at fair value and 3.5% at cost against 1.1% and 2.4% six months earlier. The Financial Stability Board reports that default rates "are low but are showing an upward trend when using broader measures, such as selective defaults and distressed exchanges" and that reliance on payment-in-kind loans "can also signal deteriorating credit conditions".
Evidence grade: A on both — a registrant's own filing and an international body's own report.
What we read into it. The 2.1-point gap between non-accruals at cost and at fair value is the mark-down already taken on the loans that went wrong. The PIK increase is the one that has not. An amendment letting a borrower pay interest in kind converts a cash problem into a balance sheet problem and buys eighteen months, which is why it leads defaults rather than lagging them. Expect the gap between headline default rates and these broader measures to be the main argument in this market through 2027.
Confidence: high. One registrant states both the figure and the cause, and an international body independently names the indicator. One lender's book is not the market, which is why the signpost threshold is set at three lenders rather than one.
What would reverse it. PIK income falling as a share of investment income at two or more lower-middle-market lenders while non-accruals at cost also fall, which would mean the amendments were a 2026 rate artefact rather than a credit signal.
The lender becomes an owner, and the loan becomes a part-saleRunning
Speed: Already standard below $25m of EBITDA
What the sources say. Main Street "typically makes direct equity investments and/or receives nominally priced equity warrants in connection with a LMM portfolio company debt investment", had equity ownership in all of its lower-middle-market portfolio companies at 30 June 2026, and reports average fully diluted ownership of 36%. That portfolio is 71.1% debt and 28.9% equity at cost and carries at $3,205.6 million of fair value against $2,547.7 million of cost. The Financial Stability Board describes private equity firms acquiring stakes in or direct ownership of insurance companies that in turn engage in private credit lending.
Evidence grade: A on the lender's own disclosures; A on the interlinkage description, from an international body's own report.
What we read into it. A lender holding 36% of the equity is not pricing a loan; it is pricing a position. That changes everything an owner should negotiate. The coupon becomes the least important term, behind the equity percentage, the valuation it is struck at, the anti-dilution, the information and consent rights, and what happens on a sale. Owners negotiate the rate because the rate is the number they understand, and give away the part that will matter in five years.
Confidence: high on what that lender does, which it states. Moderate on it being the market norm, because we have one lender's explicit disclosure and the structure is not separately reported across the sector.
What would reverse it. A large lower-middle-market lender publicly moving to a cash-pay, no-equity structure and growing share with it, which would mean the equity is a return requirement rather than a structural feature of lending at this size.
Pricing becomes legible, because the schedules were always publicStarting
Speed: 2027 to 2028
What the sources say. Frontier Desk's count over one lender's filed schedule of investments at 30 September 2026 returns 227 SOFR-referenced tranches with a median spread of 6.75%, a mean of 6.92%, a range of 4.75% to 12.00%, and a median floor of 2.00%. The same filing reports a weighted-average yield on debt investments of 10.9%, down from 11.5% a year earlier. SOFR was 3.94% on 4 November 2026 per the New York Fed's published reference rates. The Financial Stability Board notes that borrowers predominantly in private credit "typically lack public ratings", creating "transparency challenges".
Evidence grade: A on the count, which is ours over a public filing; A on the rate series; A on the transparency finding.
What we read into it. The opacity of this market is a reading problem rather than a disclosure problem. Every listed business development company files the spread on every tranche it holds, four times a year. Nothing stops anyone building a price index for lower-middle-market credit out of those filings, and the first party to publish one will change how every term sheet at this size is received. The reason it does not exist yet is that the people who could build it are the people being paid for the opacity.
Confidence: moderate. The count is reproducible and the data is unambiguous. That somebody publishes an index from it inside the forecast window is a judgment, and the incentives argue both ways.
What would reverse it. Business development companies obtaining relief to report spreads in aggregate rather than per tranche, which would close the window this reading depends on.
Leverage layers up, and the regulators have started naming the layersRunning
Speed: 2027 for the first policy response
What the sources say. The Financial Stability Board reports that leverage in private credit "exists at multiple and varying levels, including within the portfolio companies, private credit funds, at the sponsor level, and investor financing", that "[t]his layering effect may amplify losses during market stress", that direct bank lending to private credit funds captured in member data is about $220 billion of drawn and undrawn credit lines with commercial estimates "more than twice as large", and that asset based lending to funds allows them "to take on more leverage than NAV facilities, with loan-to-value ratios up to 65%" — against net asset value facilities, whose loan-to-value ratios the same report says member data suggests are "generally limited to 30%". One upper-middle-market lender reports its own net leverage at 1.13 times debt to equity against a target range of 0.90 to 1.25 times.
Evidence grade: A on both — an international body's own report and a registrant's own filing.
What we read into it. A borrower at 5 to 6 times inside a fund at 1.1 times inside an investor using leverage of its own is carrying more total leverage than any single disclosure shows. That is a systemic observation rather than an owner's problem, but it becomes an owner's problem the moment the fund above them needs liquidity. Expect the first supervisory data collection aimed at the fund layer within two years, and expect the terms offered below to tighten while it is being designed.
Confidence: moderate. The layering and the figures are quoted from an authoritative report. That it produces a supervisory response inside the window is a judgment, and the report itself stops short of recommending one.
What would reverse it. The Financial Stability Board or a national supervisor concluding that current data suffices and closing the workstream, which would take the policy risk out of the forecast.
Services is the most crowded sector, and AI exposure is now part of the credit questionRunning
Speed: 2027 to 2028
What the sources say. The Financial Stability Board finds that "[t]he largest borrowing sectors in private credit are services, technology, and healthcare", and that "[o]ther commentators note concerns on the concentration, particularly in technology and its exposure to AI bust or AI disruption scenarios". One upper-middle-market lender's own portfolio description now includes "AI-related equipment such as graphic processing units". Approximately 75% of private credit borrowers have EBITDA below $100 million.
Evidence grade: A on the sector finding; A on the portfolio description; C · crowd on the EBITDA share, which that report attributes to a rating agency.
What we read into it. A services company borrowing in 2027 will be asked a question it was not asked in 2024: what happens to this business if the work it sells can be done by software. Lenders will not be able to answer it any better than owners can, which means the answer will be handled with structure rather than analysis — shorter terms, tighter covenants, more amortisation, more equity. Our companion forecasts on compute financing and on data cover the two sides of the shift that prompts the question.
Confidence: moderate. The concentration is quoted and the lender's own description is quoted. That it changes underwriting terms at the lower-middle-market size is Frontier Desk's judgment, and no public source yet documents it in term sheets.
What would reverse it. Terms at this size lengthening rather than shortening across two or more lenders, which would mean the disruption question is being priced as a growth story rather than a risk.
The causal model
The loop that produced everything above, in six steps, with the figure that shows each one is running. Three arms reinforce it and three brake it; the scenarios are what happens when one arm or one brake wins.
- 1
Allocate
Insurers and pensions buy the illiquidity premium
$1.5–2 trillion - 2
Raise
Managers raise more than the upper market can absorb
$15.1bn at one fund - 3
Withdraw
Banks tighten on small firms and hand them over
Tighter, Q3 2026 - 4
Lend
Non-banks price the gap and take equity with it
12.6% plus 36% - 5
Amend
Weak credits convert cash interest into PIK
$4,981k, from $2,794k - 6
Reprice
Losses or spreads re-rate the whole size band
Non-accruals 3.5% at cost
↻ Every dollar that cannot be placed in the upper market at 10.1% looks for the lower one at 12.6%, which compresses the 12.6% toward the 10.1% and sends the next dollar further down still. The loop runs until the credit at the bottom stops supporting the price.
The three arms, and the three brakes
Read the loop and the scenarios stop being moods. Move down-market is R1 and R2 running with B2 failing to hold. Credit reckoning is B1 arriving before the capital has finished arriving. Banks return is R2 reversing. Equity-shaped credit is R3 winning outright. The odds below are our judgment about which gets there first.
How it evolves, 2027 and 2028
Five phases to the end of 2028, each named for what changes in it, with the Frontier Desk judgments that attach to that window. Every figure here is one of the calls in band 13, carried unchanged.
The credit signal gets argued about in public
Fourth-quarter filings land in February and March, and with them the next readings on payment-in-kind income and non-accruals at every listed lender in this market. Expect the argument about whether rising PIK is a credit signal or a structuring choice to be had in earnings calls rather than in filings, and expect at least one lender to break out PIK from amendments separately from PIK written into the original terms. The distinction is the whole question.
PIK income rises again at two or more LMM lenders: 68% · a lender separately discloses amendment-driven PIK: 34%
The large funds announce the small strategy
The yield gap is too wide and too well documented to survive two more years unexamined. Expect at least one manager with more than $10 billion to announce a lower-middle-market or small-company vehicle, expect the pitch to be the 2.5-point premium, and expect the first fundraising materials to cite exactly the filings this report counts. For an owner the practical effect arrives as a second term sheet from a name you did not expect.
A lender above $10bn states a lower-middle-market strategy: 63% · the LMM-to-UMM yield gap narrows below 2.0 points: 41%
The premium starts to compress, or the credit turns
This is the window where the two halves of the forecast separate. On the move-down-market path the spread a $10 million-EBITDA company pays starts falling and the equity ask stays, because the equity is where the return goes when the coupon compresses. On the credit-reckoning path non-accruals at cost cross 5% at two or more lower-middle-market lenders, fundraising slows, and the premium widens instead. The same filings settle both.
Non-accruals at cost above 5% at two or more LMM lenders: 37% · median LMM spread below SOFR + 6.00%: 32%
Someone publishes the price
Every tranche in this market is filed with its spread four times a year, and the first party to assemble those filings into a published index for lower-middle-market credit changes the negotiation at every size below $25 million of EBITDA. Expect it to come from a data vendor or an academic rather than from a lender, and expect the first version to be badly constructed and still more useful than what owners have now.
A published index or dataset of LMM credit spreads exists: 29% · a supervisory data collection on private credit funds is proposed: 44%
Capital down-market, terms re-written
On the central path the capital has arrived, the spread premium has compressed by roughly a point, the equity component has not compressed at all, and the terms that matter to an owner at exit — consent rights, drag, valuation on the warrant — are where the whole negotiation has moved. The alternative is that the credit turned first, the capital left faster than it came, and the price of debt for a smaller services company in 2029 is higher than it is today.
Move down-market: 34% · credit reckoning: 22% · non-bank direct lender the dominant funding route: 45%
Scenarios
Five ways lending to smaller companies can go. Each carries Frontier Desk odds at 31 December 2027 and 31 December 2028, the trigger that puts you on it, the early markers that tell you first, and what it means on each side of a loan. Each column sums to 100, and every signpost in band 12 names the scenario it moves.
Two-tier market
Today's shape continues. A visible upper middle market at around 10% and an opaque lower one at around 12.6% with equity attached, separated by the fact that almost nobody compares them. The reason this falls from 38% to 22% is not that the segmentation is wrong — $11.8 million of EBITDA genuinely is a different credit from $247 million. It is that a documented 2.5-point gap with more capital on the low side has never stayed unexamined for long.
Trigger: None. This is the no-trigger path: large funds stay in their size band, the yield gap holds, and nobody publishes a comparison.
Early markers: The LMM-to-UMM yield gap holding near 2.5 points; large lenders' average commitment size stable; no new small-company vehicles announced.
For an owner: Your terms are set by who you can reach, not by the market. Running a real process across both kinds of lender is worth more than any negotiation on the rate.
For lenders: Lower-middle-market specialists keep an unusually durable premium. The risk is not competition; it is that the credit at the bottom is cyclical and the premium is not priced for it.
Move down-market
The modal destination by the end of 2028. Larger funds build or buy lower-middle-market capability, a $10 million-EBITDA company starts receiving term sheets from names that would not have returned its call, and the spread premium compresses by roughly a point. The equity component does not compress, because that is where the return goes once the coupon falls. The practical effect for an owner is more choice on price and the same pressure on ownership.
Trigger: A manager with more than $10 billion announcing a lower-middle-market vehicle, or acquiring a specialist lender outright.
Early markers: Average new commitment size falling at large lenders; new small-company vehicles announced; LMM yields falling toward 11%.
For an owner: The best window of the whole forecast, and a trap. Take the competition on price, and read the equity, consent and exit terms as carefully as if the rate were zero, because that is where the difference now lives.
For lenders: Specialists lose pricing power and keep their underwriting advantage. The ones that survive are the ones that can say no at the new price.
Credit reckoning
Payment-in-kind income, amendments and non-accruals keep rising until allocators stop. Non-accruals at cost cross 5% at two or more lower-middle-market lenders, fundraising slows, and the price of credit for a smaller company goes up rather than down. The Financial Stability Board already records default rates rising on broader measures, and one lender's non-accruals at cost have gone from 2.4% to 3.5% in six months. This is the second-likeliest 2028 outcome and the one least reflected in current borrowing plans.
Trigger: Non-accruals at cost crossing 5% at two or more lower-middle-market lenders, or a listed lender cutting its dividend on credit grounds.
Early markers: PIK rising as a share of investment income; non-accruals at cost rising faster than at fair value; amendment activity disclosed in earnings calls.
For an owner: Time is the risk, not price. Refinancing into this scenario is refinancing into a market with fewer lenders and worse terms, which argues for doing it before you need to rather than after.
For lenders: Whoever grew fastest in 2025 and 2026 discovers what they bought. Expect the gap between non-accruals at cost and at fair value to be the number everyone watches.
Banks return
Rates, capital rules or competitive pressure bring commercial banks back to small business lending, and the non-bank premium has to narrow to meet them. Note the shape of the odds: this flattens rather than compounding, because the structural reasons banks left — capital treatment, cost to serve, speed — are not things a cycle reverses. Banks returning would cap the price rather than reset the market.
Trigger: Two consecutive Senior Loan Officer Opinion Surveys reporting easing standards for small firms, or a material change to small business lending capital treatment.
Early markers: SLOOS reporting easing for small firms; bank C&I balances to small firms growing; government-guaranteed lending volumes rising.
For an owner: A bank facility is cheaper and slower and comes with covenants a fund would not ask for. If both are available, the right answer depends on how certain your next two years are, not on the rate.
For lenders: Non-banks lose the marginal borrower and keep the ones banks cannot underwrite, which are also the ones that default. Adverse selection is the whole risk here.
Equity-shaped credit
The structure one lender already uses — first-lien debt at 12.6% plus an average 36% of the fully diluted equity — becomes the normal deal below $25 million of EBITDA rather than one house's model. Debt and equity stop being separate conversations at this size. For an owner this is the most consequential of the five, because it changes what you are selling rather than what you are paying.
Trigger: A second and third large lower-middle-market lender disclosing routine equity participation alongside debt.
Early markers: Equity as a share of LMM portfolios at cost rising; warrants disclosed on more tranches; lenders reporting equity ownership percentages.
For an owner: Negotiate the equity first and the coupon second. The valuation the warrant is struck at, the anti-dilution and the exit rights will matter more in five years than a hundred basis points ever will.
For lenders: Returns stop depending on the credit cycle and start depending on portfolio company performance, which is a better business and a different one.
Winners and losers
Who gains and who loses on the central path, and why. This band is analysis throughout: direction is stated against today, not against each other, and nothing here is a reading.
Owners and borrowers
| Segment | Direction | Why |
|---|---|---|
| Services firms with $5m to $25m of EBITDA, contracted revenue and clean records | Clear winner | This is the size band capital is arriving in, and the lenders arriving are large enough to compete on price. A company that can be underwritten quickly gets the benefit of the compression; one that cannot gets the old price in a better market. |
| Owners who treat the coupon as the main term | Clear loser | A lender holding an average 36% of the equity in every lower-middle-market borrower is not selling a rate. Negotiating the rate hard and the warrant lightly is the most expensive common mistake available at this size. |
| Firms with heavy customer concentration or an owner-dependent sales process | Loser | A 725 basis point spread range from one lender in one quarter is mostly about company-specific risk. Concentration and key-person dependence are the two factors that put a borrower at the wide end, and both are fixable before a process rather than during one. |
| Borrowers refinancing in 2027 and 2028 on loans written in 2021 and 2022 | Loser on the central path | They are refinancing at SOFR near 4% into a market where the median lower-middle-market spread is 6.75% with a 2.00% floor. The arithmetic is unforgiving and the date is already set. |
| Owners prepared to be bought rather than to borrow | Winner, selectively | In a market where the lender wants equity anyway, selling a stake deliberately is often better priced than giving one away as a warrant. The two routes should be run against each other rather than in sequence. |
| Firms whose work is plausibly automatable | Loser | Services is the largest borrowing sector in private credit and the question about AI disruption is now part of the file. Expect shorter terms and tighter amortisation rather than refusal, and expect the owner who can answer the question concretely to be priced differently. |
Lenders, funds and the layers above them
| Lender or layer | Direction | Why |
|---|---|---|
| Lower-middle-market specialists with their own origination | Winner, then squeezed | They own the premium until the capital arrives and the underwriting capability after it does. The ones that survive the compression are the ones whose sourcing does not depend on being the only call a borrower can make. |
| Large direct lenders moving down the size curve | Winner, with execution risk | The 2.5-point gap is real and the capital is there. What does not transfer is the ability to underwrite an $11.8 million-EBITDA company with no rating and no sponsor, and that is exactly what the lower middle market is. |
| Commercial banks in small business lending | Loser on the central path | They tightened on small firms while leaving large firms unchanged, and told their regulator their customers are going elsewhere. The business leaves faster than it returns. |
| Insurers and pensions allocating to the illiquidity premium | Winner until the credit turns | The premium is real and the maturities match the liabilities. What they cannot see is layered leverage across portfolio company, fund, sponsor and investor, which an international body has now named as the main systemic concern. |
| Advisers who run a real process for borrowers at this size | Clear winner | A 725 basis point spread range and a market where the lender also wants equity is a market where the outcome depends on who sees the company and in what order. That is a process problem, not a pricing one. |
Frontier Desk is the advisor to the company raising the money, never a lender and never a broker. The first conversation covers what a lender will read in your company, which kinds of capital fit it, and what to fix before anyone looks.
Commoditised or scarce
What becomes a commodity in this market and what keeps its scarcity value, with the date we think each crosses. This band is a forecast, and the band after it argues against itself.
Commoditising, and by when
| What | By when | Why |
|---|---|---|
| Senior secured debt to a sponsor-backed borrower above $50m of EBITDA | Already | One lender reports 233 portfolio companies at a 5.6% average spread and an average position of $59.3 million. That is a product, priced in a market, with many providers. |
| Knowledge of the market price at the lower-middle-market size | 2028 | Every spread in this market is filed quarterly, per tranche. The opacity is a reading problem, and reading problems do not survive long once somebody is paid to solve them. |
| The capital itself, at every size | 2027 | A market of $1.5 to $2 trillion looking for borrowers, with banks withdrawing from the small end, is not a market where capital is the scarce input. |
| Unitranche structuring | Already | A first-lien loan with one lender and one document is standard at every size now. Nobody will pay for the structure in 2028; they will pay for the decision. |
| Headline covenant packages | 2028 | Terms converge quickly once several lenders compete for the same borrower. The non-standard terms — equity, consent, exit — are where the differences will concentrate instead. |
Keeping its scarcity value, and why
| What | How long | Why |
|---|---|---|
| Underwriting judgment on a company with no rating and no sponsor | Through 2028 and beyond | An $11.8 million-EBITDA company has to be understood rather than modelled. That capability does not multiply with fund size, and it is the reason the premium survives the capital. |
| A borrower that can actually be underwritten quickly | Through 2028 | A 725 basis point spread range at one lender in one quarter is mostly company-specific. Clean records, contracted revenue and a business that runs without the owner are worth more than any negotiation. |
| Lender relationships that produce a second term sheet | Through 2028 | The price at this size is set by how many lenders see the company, not by a published rate. Competition is the only mechanism, and reaching it is a process rather than a phone call. |
| Willingness to lend without taking equity | Through 2028 | One lender holds equity in every lower-middle-market borrower it has. A lender that will take a coupon and nothing else is scarce, and worth paying more for if the exit matters. |
| Capacity to refinance a 2021 vintage loan on its original terms | Through 2027 | SOFR near 4% with a median spread of 6.75% and a 2.00% floor is a different world from the one those loans were written in. Nobody can give you back the old terms. |
The case against this table.
The strongest argument against this table is that we have built a market view out of four lenders. Main Street, Capital Southwest, Blue Owl and Ares are listed, regulated and therefore legible, and that is exactly what makes them unrepresentative: the private funds doing most of the lending in this market file nothing. If listed business development companies are the conservative tail of private credit — which is plausible, since their leverage is capped by statute and their holdings are marked quarterly in public — then every credit reading on this page is better than the market's, the stress is worse than we show, and the credit-reckoning scenario deserves more than 22%.
The second argument is that the yield gap we treat as an arbitrage may be a correctly priced risk premium that nobody should close. A $11.8 million-EBITDA company with no rating, no sponsor and one owner genuinely is riskier than a $247 million one with a private equity firm behind it, and 2.5 points may be too little rather than too much. If that is right, capital moving down-market is not arbitrage but mispricing, the move-down-market scenario is a prelude to the credit-reckoning one rather than an alternative to it, and an owner borrowing cheaply in 2027 is borrowing from someone who will not be there in 2030.
What to do about it
What the forecast implies for the two readers this report is written for: the owner of a company that needs capital, and the adviser or investor reading the paper. Frontier Desk is an advisor, never a lender and never a broker, and nothing here is a recommendation about any particular financing.
If you own the company that needs the money
| Decision | What the forecast says |
|---|---|
| What will you actually be quoted? | At $5m to $25m of EBITDA, the published evidence says a floating rate over SOFR with a median spread of 6.75% and a 2.00% floor, which was about 10.7% all-in in early November 2026. The range around that median at one lender in one quarter was 4.75% to 12.00%, and almost all of the difference is about your company. |
| Is it a loan or a part-sale? | One listed lower-middle-market lender holds equity in every borrower it has, averaging 36% fully diluted. Ask on the first call whether equity or warrants are part of the structure, and if they are, negotiate the valuation and the exit rights before you negotiate a single basis point. |
| Bank or fund? | Banks tightened on small firms in the third quarter of 2026 while leaving larger firms unchanged, specifically on collateral and interest rate floors. A bank is cheaper and slower and will want more security; a fund is faster and more expensive and may want ownership. Run both, and decide on certainty rather than rate. |
| What moves you from the wide end of the range to the narrow one? | The things a lender can check in a week: contracted or recurring revenue, customer concentration, a business that runs without you, clean monthly accounts and a clear owner's plan. These are worth more basis points than any negotiation, and they take months rather than days to establish. |
| When should you go? | Before you need to. The credit-reckoning scenario runs at 22% by the end of 2028, and in it the number of lenders willing to look at a company your size falls before the price rises. Refinancing early costs a fee; refinancing late costs the terms. |
| Should you borrow at all? | In a market where the lender wants equity anyway, selling a stake deliberately is sometimes better priced than giving one away. Run the loan and the sale against each other rather than in sequence, because they are now competing for the same thing. |
If you invest in or underwrite this paper
Read the schedules, not the summaries. Every listed business development company files the spread, floor and reference rate for every tranche it holds, four times a year. A median and a distribution for lower-middle-market credit can be computed from them in an afternoon, and almost nobody does it.
Watch non-accruals at cost rather than at fair value. One lender's non-accruals were 1.4% at fair value and 3.5% at cost at 30 September 2026, against 1.1% and 2.4% six months earlier. The fair value number tells you what has been marked; the cost number tells you what went wrong.
Treat amendment-driven payment-in-kind as the leading indicator it is. An international body has named it as a signal of deteriorating credit conditions, and one lender has attributed a 78% rise in PIK income to amendments in plain language in its own filing.
Frontier Desk is the advisor to the company raising the money, never a lender and never a broker. The first conversation covers what a lender will read in your company, which kinds of capital fit it, and what to fix before anyone looks: Get your company ready.
See who lends to companies like yours, by name.
The spread range at one lender in one quarter ran from 4.75% to 12.00%. Almost all of that is about your company and who sees it. A short conversation covers where you would sit and who should be looking.
The evidence base
The present-state record the forecast rests on, as a reference band. Everything here is a reading from a named public source or a count over a filed schedule of investments, with its date and its arithmetic; the inferences are in the bands above.
The two markets, side by side
Publicly observable| Measure | Lower middle market | Upper middle market | What the difference means |
|---|---|---|---|
| Average borrower EBITDA | $11.8m (Main Street LMM, 30 Jun 2026) | $247m (Blue Owl, 30 Sep 2026) | FD estimate: $247m ÷ $11.8m = 20.9 times the EBITDA — not comparable companies |
| Average borrower revenue | $10m–$150m band (Main Street LMM definition) | $1.07bn weighted average (Blue Owl) | The band and the average are different measures; both lenders' own words |
| Weighted-average yield | 12.6% (Main Street LMM); 10.9% (Capital Southwest) | 10.1% on accruing debt (Blue Owl) | FD estimate: 12.6% − 10.1% = a 2.5-point gap, the widest comparison on the page |
| Spread over the floating base | Median 6.75% over 227 filed tranches (FD count) | 5.6% weighted average (Blue Owl) | FD estimate: 6.75% − 5.6% = 115 basis points, which is what being small costs before the equity |
| Borrower leverage | Not separately disclosed at this size | 5.9x net leverage, 2.0x interest coverage (Blue Owl) | The absence is the finding: nobody publishes leverage at the bottom |
| Equity held by the lender | 36% average fully diluted (Main Street LMM) | 4.7% of portfolio in common equity (Blue Owl) | The structural difference: at the bottom the lender is a part-owner |
| Share of portfolio in debt at cost | 71.1% (Main Street LMM) | 73.1% first lien at fair value (Blue Owl) | Similar headline, very different composition underneath |
| Average position size | $5m–$125m investment range (Main Street LMM) | $59.3m average debt investment (Blue Owl) | One position at the top is a whole relationship at the bottom |
| Non-accruals | 1.4% at fair value, 3.5% at cost (Capital Southwest) | Disclosed per investment rather than as a portfolio rate | The cost figure is the one that moves first |
| Payment-in-kind income | FD estimate: $4,981k ÷ $67,158k = 7.4% of investment income, from 4.9% | Reported within interest and PIK income by affiliate | Comparable disclosure does not exist; this is the gap in the evidence |
Every figure is quoted from the named registrant's own Form 10-Q as filed, read in full on 5 November 2026, except the two marked as Frontier Desk counts or estimates, whose arithmetic is shown. The quarter ends differ — 30 June 2026 for Main Street, 30 September 2026 for Capital Southwest, Blue Owl and Ares — and we have not adjusted for that. Frontier Desk holds no position in any security named here, acts for none of these companies, and names them because their filings are the only public record this market has.
Average borrower EBITDA, as each lender reports it
Publicly observableMillions of dollars, log scale. The top bar is roughly twenty-one times the bottom one, and both are called private credit
Source: Each lender's own Form 10-Q for its quarter ended 30 Jun or 30 Sep 2026, read 5 Nov 2026
Yields, as each lender reports them
Publicly observableWeighted-average yields, as each lender reports them
Publicly observablePer cent, from each registrant's own quarterly filing. The quarter ends differ and are shown in the evidence table above
Source: Each lender's own Form 10-Q and its own non-GAAP reconciliation, read 5 Nov 2026
What a lower-middle-market borrower pays, counted from the filing
Publicly observableWhat one lower-middle-market book actually charges
Publicly observable227 floating-rate tranches in one filed schedule, counted by Frontier Desk. Each bar starts at SOFR, so the bar's length is the spread
Source: Frontier Desk's count of 227 SOFR tranches in one filed schedule of investments, 30 Sep 2026
Frontier Desk extracted every floating-rate tranche in Capital Southwest's consolidated schedule of investments as filed for 30 September 2026: 227 tranches referencing SOFR, median spread 6.75%, mean 6.92%, range 4.75% to 12.00%, median floor 2.00%. SOFR was 3.94% on 4 November 2026. The filing the count was taken from is public, and the count can be repeated from it.
What the supervisors have published about the market as a whole
Publicly observable| Reading | Source | Figure | What it tells the forecast |
|---|---|---|---|
| Definition used | Financial Stability Board, 6 May 2026 | Nonbank direct lending to medium-sized companies negotiated on a bilateral basis | The definition excludes most of what gets called private credit in marketing |
| Size | The same report | $1.5–$2 trillion globally at end-2024; the US around $1 trillion | Comparable to the institutional leveraged loan market, and larger than some estimates of high yield |
| Largest borrowing sectors | The same report | Services, technology and healthcare | The sector this report is about is the most crowded one |
| Borrower size | The same report, attributing Fitch | Approximately 75% of borrowers have EBITDA below $100 million | Most of this market is smaller than the listed lenders' own averages suggest |
| Ratings | The same report | Concentrated around the single B bucket; borrowers predominantly in private credit typically lack public ratings | There is no rating to read at the size this report covers |
| Leverage | The same report, attributing external sources | 5–6x debt to EBITDA against about 4x in leveraged loans; possibly closer to 7x excluding adjustments | Headline leverage understates true leverage, by the report's own account |
| Default trend | The same report | Low but rising on broader measures such as selective defaults and distressed exchanges | The headline default rate is not the measure to watch |
| Payment-in-kind | The same report | Some borrowers relying more on PIK loans, which can signal deteriorating credit conditions | An international body naming the indicator this report tracks |
| Bank exposure to the funds | The same report | About $220bn of drawn and undrawn credit lines in member data; commercial estimates more than twice as large | Nobody knows the number within a factor of two |
| Layered leverage | The same report | Leverage at portfolio company, fund, sponsor and investor levels; asset based fund lending LTVs up to 65%, against NAV facility LTVs generally limited to 30% | The systemic concern, and the reason a supervisory response is likely |
| Bank standards for small firms | Federal Reserve, October 2026 SLOOS | Tightened for small firms; unchanged for large and middle-market firms | The divergence by borrower size, in one quarter, from the regulator's own survey |
| What banks said about where borrowers went | The same survey | Customer borrowing shifting to more attractive bank or nonbank funding sources | The substitution, stated by the banks losing the business |
Three gaps in this evidence base are worth stating plainly, because the forecast above is weaker where they are.
First, every portfolio figure comes from a listed business development company, and listed vehicles are the most conservative, most regulated and most transparent part of this market. The private funds doing most of the lending file nothing, so our readings are a lower bound on stress and an upper bound on discipline. Second, no public source reports borrower leverage or interest coverage at the lower-middle-market size, which is why the comparison table has a blank in it rather than an estimate. Third, the market-level figures on leverage, borrower size and default trends are ones the Financial Stability Board attributes to commercial research it did not perform; we carry them with that attribution and grade them accordingly.
Key assumptions
The assumptions the odds depend on, and the evidence that would break each one. Each is an analytical commitment, not a reading.
| Assumption | Breaks if |
|---|---|
| The yield gap between the lower and upper middle markets is wider than the risk difference justifies. | Lower-middle-market losses running materially above upper-middle-market losses over a full cycle, which would make the gap a correct price rather than an arbitrage. |
| Underwriting capacity at the lower-middle-market size, not capital, is the binding constraint. | A large lender scaling lower-middle-market origination quickly without a loss rate worse than the specialists', which would mean the capability transfers after all. |
| Listed business development companies are representative enough of this market to read as a proxy. | Private fund disclosures, a supervisory data collection, or a rating agency study showing materially different pricing or credit performance in unlisted vehicles. |
| Amendment-driven payment-in-kind leads defaults rather than substituting for them. | PIK rising while non-accruals at cost fall across several lenders for a year, which would mean amendments are resolving credits rather than deferring them. |
| Banks do not return to small commercial lending inside the window. | Two consecutive Senior Loan Officer Opinion Surveys reporting easing standards for small firms, or a change to the capital treatment of small business loans. |
| Equity participation stays a structural feature of lending below $25 million of EBITDA. | A large lower-middle-market lender growing share with a cash-pay, no-equity structure. |
| Spread data stays available per tranche in public filings. | A rule change or exemptive relief allowing aggregate rather than per-investment reporting, which would close the only window into pricing at this size. |
| Services sector concentration does not resolve into a correlated loss event inside the window. | A cluster of defaults in one services vertical across several lenders' portfolios in the same two quarters. |
| The supervisory response stays at data collection rather than reaching terms. | Any binding rule on private credit fund leverage, valuation frequency or disclosure, which would tighten the terms offered below while it is being written. |
| SOFR stays high enough that floors do not bind for most borrowers. | SOFR falling toward the 2.00% median floor, which would stop rate cuts reaching borrowers and change the whole refinancing calculus. |
Signposts
A signpost is a reading that moves the odds if it crosses the threshold shown. Each one names the scenario or scenarios it moves, so no reading on this page floats free of the forecast it is supposed to update.
| Indicator | Now | Threshold | Scenario it moves | What it would mean | Checked |
|---|---|---|---|---|---|
| The LMM-to-UMM weighted-average yield gap | About 2.5 points (12.6% against 10.1%) | Below 2.0 points, or above 3.0 | Move down-market · Two-tier market | Below 2.0 means the capital has arrived and the premium is compressing; above 3.0 means risk is being repriced upward | Quarterly, from 10-Qs |
| Median lower-middle-market spread over SOFR | 6.75% over 227 filed tranches — FD count | Below 6.00%, or above 7.50% | Move down-market · Credit reckoning | Below 6.00% is compression arriving; above 7.50% is the credit turning before the capital does | Quarterly, from filed schedules |
| Payment-in-kind as a share of investment income at LMM lenders | 7.4% at one lender, from 4.9% — FD estimate | Rising at three or more lenders | Credit reckoning | The amendment wave is market-wide rather than one book, and the credit brake is the binding one | Quarterly, from 10-Qs |
| Non-accruals at cost at LMM lenders | 3.5% at one lender, from 2.4% six months earlier | Above 5% at two or more | Credit reckoning | Allocators slow, fundraising slows, and the price of credit for a smaller company rises rather than falls | Quarterly, from 10-Qs |
| Average new commitment size at large direct lenders | $40m, from $49m at one lender | Below $35m, or above $50m | Move down-market · Two-tier market | Below $35m says large funds are writing lower-middle-market cheques; above $50m says they are staying home | Quarterly, from 10-Qs |
| Large managers announcing a small-company strategy | None above $10bn on the public record | Any one | Move down-market | The arbitrage has been named by the people who can close it, and compression follows within about a year | Company announcements and fund filings |
| Bank lending standards for small firms | Tightened in Q3 2026; unchanged for larger firms | Easing in two consecutive surveys | Banks return | The structural withdrawal was cyclical after all, and the non-bank premium has to narrow to meet it | Federal Reserve SLOOS, quarterly |
| Equity or warrants alongside LMM debt | 36% average fully diluted at one lender | Two or more lenders disclosing routine equity participation | Equity-shaped credit | The part-sale structure is the market rather than one house's model, and owners should negotiate ownership first | Quarterly, from 10-Qs |
| A published index or dataset of LMM credit spreads | None | Any one published | Move down-market · Two-tier market | The opacity that supports the premium is gone, and every term sheet at this size gets read against a number | Vendors, academic publications |
| Supervisory action on private credit funds | An international body's report, May 2026 | Any proposed data collection or rule | Credit reckoning · Two-tier market | Terms tighten while the rule is being written, regardless of what it finally says | FSB, Federal Reserve, SEC |
| SOFR against the median LMM floor | 3.94% against a 2.00% median floor | Below 2.50% | Banks return · Move down-market | Rate cuts stop reaching borrowers because the floor binds, and the refinancing calculus changes entirely | NY Fed published reference rates |
| Default measures beyond headline defaults | Low but rising on selective defaults and distressed exchanges | Any published broad measure above 5% | Credit reckoning | The gap between the headline rate and the real one has closed, in the direction nobody wants | FSB, rating agencies, supervisors |
| Services sector loss clustering | No clustering on the public record | Defaults in one services vertical across three or more lenders in two quarters | Credit reckoning | The sector concentration an international body named has become a correlated loss rather than a statistic | Quarterly, from 10-Qs |
Calls
Each call is dated, carries a Frontier Desk judgment as a percentage, and names the public source that settles it. No call on this page is settled by a social post, an unnamed source or a private record. We score each one when it resolves, with the Brier score: the squared gap between our odds and what happened, where 0 is perfect.
| Call | Resolves | Odds | Settled by |
|---|---|---|---|
| Payment-in-kind income rises again year on year at two or more lower-middle-market lenders | 31 Dec 2026 | 68% | SEC EDGAR, 10-Q and 10-K |
| A lender separately discloses payment-in-kind arising from amendments rather than original terms | 31 Dec 2026 | 34% | SEC EDGAR, 10-Q and 10-K |
| A direct lender with over $10 billion of assets publicly states a lower-middle-market or small-company strategy | 31 Dec 2027 | 63% | Company announcements and fund filings |
| The lower-to-upper-middle-market weighted-average yield gap narrows below 2.0 points | 31 Dec 2027 | 41% | SEC EDGAR, the lenders' own 10-Qs |
| Non-accruals at cost exceed 5% at two or more lower-middle-market lenders | 31 Dec 2027 | 37% | SEC EDGAR, 10-Q |
| The median lower-middle-market spread in a filed schedule falls below SOFR plus 6.00% | 31 Dec 2027 | 32% | SEC EDGAR, filed schedules of investments |
| Two consecutive Senior Loan Officer Opinion Surveys report easing standards for small firms | 31 Dec 2027 | 23% | Federal Reserve SLOOS |
| A second large lower-middle-market lender discloses routine equity participation alongside its debt | 31 Dec 2027 | 44% | SEC EDGAR, 10-Q and 10-K |
| A supervisory data collection or rule on private credit funds is formally proposed | 31 Dec 2028 | 44% | FSB, Federal Reserve and SEC publications |
| A published index or dataset of lower-middle-market credit spreads exists | 31 Dec 2028 | 29% | Any publisher's own published methodology and data |
| A listed lower-middle-market lender cuts its dividend and cites credit performance | 31 Dec 2028 | 31% | SEC EDGAR and company announcements |
| Average new commitment size at the largest listed BDC falls below $35 million for a full year | 31 Dec 2028 | 36% | SEC EDGAR, 10-Q |
| SOFR falls below 2.50%, bringing the median 2.00% floor into play | 31 Dec 2028 | 21% | Federal Reserve Bank of New York reference rates |
| Defaults cluster in one services vertical across three or more lenders inside two quarters | 31 Dec 2028 | 27% | SEC EDGAR, 10-Q |
| A broad default measure for private credit is published above 5% | 31 Dec 2028 | 33% | FSB, rating agencies and supervisors |
| Lending market: two-tier / move down-market / credit reckoning / banks return / equity-shaped credit | Jan 2028 / Jan 2029 | 38/26/16/12/8 · 22/34/22/10/12 | Triggers and markers in Scenarios |
| How a $5m–$25m EBITDA services firm funds its next deal: non-bank / bank or guaranteed / seller / equity | Jan 2028 / Jan 2029 | 41/31/17/11 · 45/24/18/13 | SEC EDGAR, Federal Reserve SLOOS and published deal records |
Questions to test this
Questions a sharp reader would use to break the argument.
- 01We read four listed lenders as a proxy for a $1.5 to $2 trillion market whose largest participants file nothing. If listed vehicles are the conservative tail, how much worse is the unlisted middle — and is there any public source that could ever tell us?
- 02A 2.5-point yield gap between $11.8 million and $247 million of EBITDA: arbitrage, or a correctly priced risk premium that capital will destroy itself trying to close?
- 03One lender holds 36% of the equity in every lower-middle-market borrower it has. Is that lending with an equity kicker, or private equity with a coupon — and does any owner negotiating a term sheet at this size know which one they are in?
- 04Every spread in this market is filed quarterly, per tranche, by every listed lender. Why does no index exist, and is the answer that the people who could build one are paid by the opacity?
- 05Payment-in-kind income rose 78% at one lender, from amendments. At what threshold does that stop being prudent portfolio management and start being extend-and-pretend, and who would be the first to say so?
- 06Banks tightened on small firms and left large firms unchanged in the same quarter. If that divergence is structural rather than cyclical, what is the terminal share of small business credit that sits outside the regulated banking system?
- 07Services is the largest borrowing sector in private credit, and the AI disruption question is now in the file. Which services verticals get repriced first, and will anyone be able to tell a repricing from a credit event when it happens?
- 08Our central assumption is that underwriting capacity, not capital, is the constraint at this size. It is also the assumption with the least direct evidence on this page. What reading would settle it before 2028?
Sources and method
Where the direction comes from
Direction comes from Frontier Desk's standing coverage of the private credit market and of the services companies that borrow in it — who lends, by type, on what terms, and in which sectors. The public surface of that work is at /markets and in the guides at /guides.
The standard every claim had to clear
A figure, a quotation or a market fact is carried only if an independent primary or reputable named source states it — a filing with a securities regulator, a central bank's own survey, an international body's own report, or a published rate series — and is dropped otherwise. Every source on this page was opened and read on 5 November 2026. No figure on this page comes from a private record, from a company Frontier Desk advises, from any piece of work it is doing, or from an unnamed source.
How the spread count was done
The spread figures are Frontier Desk's own count, not quotations. Every listed business development company files a consolidated schedule of investments giving the reference rate, the spread and any floor for each tranche it holds. We extracted every funded tranche in Capital Southwest's schedule as filed for 30 September 2026 — every line carrying a current interest rate, which is 227 of the 301 lines that reference SOFR with a stated spread, the other 74 being undrawn revolver and delayed-draw commitments — and took the median, mean, minimum, maximum and median floor across them. Including the unfunded commitments would price tranches nobody is paying on, and would break the cross-check against the filing's own 10.9% weighted-average yield. The count is unweighted by principal, which means it describes the typical tranche rather than the typical dollar, and the report says so wherever the figure is used.
How a reading was graded
Processing graded every reading A for a record, B for a party's own published statement about itself, and C for a crowd source. Several market-level figures in the Financial Stability Board's report — borrower leverage, the share of borrowers below $100 million of EBITDA, and the middle-market CLO cash flow study — are research that body attributes to others rather than performed itself, so they are graded C and attributed in the sentence that carries them.
How the readings became a forecast
Analysis turned the readings into structural shifts, a causal model, phases, scenarios and odds, and every judgment on this page carries its own confidence and the reason for it, kept visibly separate from the grade on the evidence under it. Where a figure is derived rather than quoted it is marked FD estimate and its arithmetic is shown beside it.
What the odds are, and what they are not
The odds on this page are Frontier Desk judgments at two fixed horizons, not the output of a model, and the page says so wherever they appear. They are internally disciplined rather than simulated: each column sums to 100, a call that is a scenario arriving is held at or below that scenario's odds, and a call that is a necessary marker of a scenario is held at or above it. They are intended to be scored and to be wrong in public.
What comes back, and when
Dissemination is this page. Feedback is the signposts, each of which names the scenario it moves, and the Brier-scored calls. The spread count is re-run from the next quarter's filed schedule before every update. The next update is 5 December 2026.
What we dropped, and why
Several claims that would have strengthened this report did not clear the standard above, and are not on the page in any form.
Market-wide pricing from deal databases
Commercial databases publish average spreads and leverage multiples for middle-market loans. Their samples and methods are not public and we could not check them, so none of them is on this page. The pricing here is either a lender's own disclosed weighted average or our own count over a filed schedule.
Default and recovery rates for private credit as a whole
Widely quoted default rates come from rating agency samples of rated private credit, which is a small and unrepresentative slice. We carry the Financial Stability Board's qualitative finding that broader measures are rising, with its attribution, and no number of our own.
Fundraising and dry powder figures
Totals for private credit fundraising and undeployed capital come from data vendors whose coverage we cannot verify. The capital-abundance argument on this page rests instead on the Financial Stability Board's market size and on one lender's own disclosed portfolio growth.
Anything said by an unnamed lender or borrower
Practitioner commentary about where terms really are is routine in this market and unverifiable by a reader. Every price on this page is quoted from a filing or counted from one.
What we still cannot verify
No public source reports borrower leverage or interest coverage at the lower-middle-market size, which is why that row of the comparison table says so rather than estimating. No public source reports payment-in-kind on a comparable basis across lenders. And nothing in the public record describes what the unlisted private funds doing most of this lending actually charge.
- 01The supervisory view of the market as a whole, 6 May 2026: private credit defined as “nonbank direct lending to medium-sized companies negotiated on a bilateral basis”, sized “between $1.5 trillion and $2 trillion (as at the end of 2024)” with the US “around $1 trillion”; “[t]he largest borrowing sectors in private credit are services, technology, and healthcare”; borrowers “concentrated around the single B credit rating bucket” with “5-6x debt-to-EBITDA” leverage and “‘true' leverage … closer to 7x”; about $220 billion of bank credit lines to private credit funds in member data; NAV facility loan-to-value ratios “generally limited to 30%” against asset based lending, which lets funds “take on more leverage than NAV facilities, with loan-to-value ratios up to 65%”; and the finding that borrowers “appear to be relying more on payment-in-kind loans, which can also signal deteriorating credit conditions”; read 5 Nov 2026: Financial Stability Board, Report on Vulnerabilities in Private Credit
- 02What banks told their regulator about small firms, covering the third quarter of 2026: standards for C&I loans “to large and middle-market firms were basically unchanged, while standards tightened for C&I loans to small firms”, with tightening “for collateralization requirements and use of interest rate floors for small firms”, reasons given as “a less favorable or more uncertain economic outlook and a worsening of industry-specific problems”, and weaker demand attributed partly to “customer borrowing shifting to more attractive bank or nonbank funding sources”; read 5 Nov 2026: Federal Reserve, October 2026 Senior Loan Officer Opinion Survey
- 03The lower-middle-market portfolio, in the lender's own words: “annual revenues between $10 million and $150 million”, investments “from $5 million to $125 million”, terms “between five and seven years”, 94 portfolio companies at $3,205.6m fair value against $2,547.7m cost, 71.1% debt and 28.9% equity at cost, 99.4% first lien, weighted-average annual effective yield 12.6% (12.1% including non-accruals), average EBITDA $11.8m, and “equity ownership in all of its LMM portfolio companies, and the average fully diluted equity ownership in those portfolio companies was 36%”; quarter ended 30 Jun 2026, read 5 Nov 2026: Main Street Capital Form 10-Q
- 04A second lower-middle-market lender's pricing and credit: weighted average yield on debt investments 10.9% from 11.5%, average monthly cost basis of debt investments up 25.4% to $2,152.8m, PIK interest income of $4,981k against $2,794k “due to an increase in the number of amendments allowing portfolio companies to elect to pay interest in-kind”, total investment income $67,158k, and non-accruals “approximately 1.4 % of our total investment portfolio at fair value and approximately 3.5 % at cost” against 1.1% and 2.4% at 31 Mar 2026 — and the consolidated schedule of investments from which Frontier Desk counted 227 SOFR tranches at a 6.75% median spread; quarter ended 30 Sep 2026, read and counted 5 Nov 2026: Capital Southwest Form 10-Q
- 05The upper-middle-market comparison: 233 portfolio companies at $15.10bn fair value, “weighted average yield of accruing debt and income producing securities at fair value … 10.1%”, “weighted average spread of floating rate debt investments was 5.6%”, average debt investment “approximately $59.3 million”, net leverage 1.13x, and portfolio companies at 93.3% of the debt portfolio with “weighted average annual revenue of $1.07 billion, annual EBITDA of $247 million, interest coverage of 2.0x, net leverage of 5.9x, and net loan-to value of 47%”; quarter ended 30 Sep 2026, read 5 Nov 2026: Blue Owl Capital Corporation Form 10-Q
- 06Activity at the largest listed business development company: 53 new investment commitments against 80 a year earlier, average new commitment $40m against $49m, 85% at floating rates, funded during the period at a weighted average yield of 10.4% against 10.0%, and exited or repaid at 8.7% against 10.9%; quarter ended 30 Sep 2026, read 5 Nov 2026: Ares Capital Form 10-Q
- 07The floating-rate base every spread on this page sits on top of: SOFR at 3.94% on 4 Nov 2026, with a 25th percentile of 3.93% and a 75th of 4.00%; read 5 Nov 2026: Federal Reserve Bank of New York reference rates
- 08The fixed-rate alternative, for comparison: five-year US Treasury at 4.69% on 4 Nov 2026 against 3.74% on 2 Jan 2026, across 212 published business days; read 5 Nov 2026: US Treasury daily yield curve
- 09Our own public work on what lenders read in a smaller company, which this forecast is the two-year view behind; each page checked 5 Nov 2026: The markets we map; Loan or recap for a 25 to 75 person company; How AI changes what a services firm is worth; See who lends to companies like yours
- 10Our companion forecasts on the two markets this one borders, both published 5 Nov 2026 and read the same day: GPU and neocloud compute financing to 2028; The company-data and RL-environment landscape to 2028
Find out what a lender will actually quote you, before you ask one.
Frontier Desk is the advisor to the company raising the money. We read your company the way a lender will — the contracted revenue, the customer concentration, the owner's role, the records — and tell you which kinds of capital fit, what each will ask for, and what to fix first. Answer what you have and the first conversation covers the rest.
If a lender wants equity as well as a coupon, the loan and a sale are competing for the same thing. The same review prices both routes against each other rather than one after the other.
The rate is the number you will negotiate and the equity is the number that will matter. Read them in that order and you will lose.
Frontier Desk outlook, 5 November 2026. Frontier Desk acts as the advisor to the company raising capital, through its relationships with lenders and investors, and a cybernetic system of its own — which is why the forecast stays current. Frontier Desk is not a broker-dealer, lender or investment adviser. Every figure is quoted from a named public source read on the date stated, or marked as a Frontier Desk estimate or count with its arithmetic shown. The odds are Frontier Desk judgments, not model output, and may be wrong. Nothing here is an offer, a recommendation to buy or sell any security, or advice on any particular financing. Lenders are named because their public filings are the only record this market has, not because they are connected to Frontier Desk in any way.