Cash-Out Refinance on Investment Property: 99,035 Loans in 2025

What a cash-out refinance on an investment property looked like in 2025

The pages that rank for this search are lender blogs, forum threads and bank guides, and none of them counts the loans. The federal Home Mortgage Disclosure Act (HMDA) record has the count. Lenders above a size threshold report every loan and every application they decide, with its purpose, occupancy, note rate, combined loan-to-value ratio (CLTV), loan amount, lender and, for denials, the reasons, and the FFIEC publishes the file loan by loan. We downloaded the nationwide 2025 files on October 8, 2026: originated cash-out refinances (purpose code 32), cash-out applications that were denied or approved but not accepted, and, for comparison, originated investment-property purchase loans (code 1) and refinances without cash out (code 31). We kept occupancy code 3, investment property, and, for the homeowner comparison, code 1, principal residence, and applied the base filters of our investment property mortgage rates and BRRRR pages: conventional, first lien, one to four units, closed-end, not reverse. That is why the headline count, 99,035, matches theirs. The scripts and the loan-level extracts are in the data folder, and everything below is our arithmetic.

Investment-property cash-out refinances against three neighbors, originated in 2025

Three readings. First, a cash-out is a large slice of investor lending: 21.4% of the loans in the cash-out, purchase and no-cash-out groups, 99,035 against 290,263 purchases and 72,676 refinances without cash out. Second, the price is that of a purchase loan, not a homeowner's: the median rate was 7.25%, the same as on investment purchases, and 0.6 points above the 6.65% on principal-residence cash-outs. Third, the leverage is lower than on a purchase and higher than on a homeowner's cash-out: the median CLTV was 70% against 75% on investment purchases and 60% on principal-residence cash-outs, a gap of 10 points to the homeowner (our arithmetic).

HMDA reports a loan as a cash-out refinancing when the lender “considered it to be a cash-out refinancing in processing the application or setting the terms” (Regulation C, Supplement I, comment 4(a)(3)-2), so code 32 is the lender's own label. 13,606 of the 99,035 loans, 13.7%, were reported as sold to Fannie Mae or Freddie Mac in the same year; the rest were kept, sold to others or sold after December 31 (our arithmetic).

Is it more expensive than a cash-out on your own home?

The raw gap, 0.6 points, mixes products: most investor cash-outs are DSCR-type loans, which owner-occupants rarely use. Splitting by product and comparing the same lender with itself isolates the price of the property type.

Investment-property cash-outs by product, 2025, against the same product elsewhere

DSCR-type loans (business purpose, no debt-to-income ratio reported, 30 years or more: HMDA has no DSCR field, so this is our proxy, the same one as on our DSCR loan rates page) were 56.8% of investment cash-outs, 56,235 loans at a 7.375% median, and 68.2% of them came through brokers. The like-for-like comparison is the 30-year consumer-purpose loan: 7.125% on investment property against 6.875% on a principal residence, 0.25 points more. The short-term loans in the table are cash-outs on bridge-type loans of 36 months or less: 3,070 loans at 9.25%.

Same lender, two loan types

At the typical lender an investment cash-out cost 0.375 point more than that lender's cash-out on a principal residence, and 82.7% of the 150 lenders charged more. Against that lender's own investment purchase loans the cash-out was 0.04 point higher at the median, with 27.1% charging the same, and on DSCR-type loans the median gap was 0 points. Our DSCR refinance guide quotes a cash-out add-on of about 0.2 points from lender sources; the same-lender median gap on DSCR-type loans was 0, and a median of lender medians is not any one borrower's quote. The CLTV column is the sharper difference: at the same lender, the cash-out CLTV was 7.4 points below its purchase CLTV and 1.1 points below its no-cash-out refinance CLTV at the median (our arithmetic).

The limit: 75% on one unit, 70% on two to four, and what the loans did

Fannie Mae's Eligibility Matrix, “incorporated by reference into the Fannie Mae Selling Guide” and dated August 5, 2026, gives the maximum LTV, CLTV and HCLTV ratios for conventional first mortgages eligible for delivery to Fannie Mae. For loans run through Desktop Underwriter Version 12.1 the investment-property rows are below.

Fannie Mae maximum ratios, August 5, 2026

The cash-out rows are the ones that matter here, and they sit under the purchase rows: 85% and 75% to buy, 75% and 70% to take cash out. A homeowner can take cash out of a one-unit principal residence at 80%. For investment property, a rate-and-term (limited cash-out) refinance is capped at 75% too, so a conventional refinance of a rental needs about 25% equity even with no cash out (our arithmetic).

Two timing rules in Selling Guide B2-1.3-03 (version dated December 10, 2025) apply to every conventional cash-out. The first is ownership: “At least one borrower must have been on title for at least for six months prior to the disbursement date of the new loan.” The guide waives it for an inheritance or an award in a divorce and under the delayed financing exception for a property bought with no mortgage financing. If an LLC owned the property, its time counts, but ownership must be transferred to the individual borrower to close. The second is the age of the loan being paid off: an existing first mortgage “must be at least 12 months old at the time of refinance, as measured by the note date of the existing loan to the note date of the new loan.” The rule is why a conventional cash-out cannot take out a bridge loan before it is a year old; our BRRRR page works through that case.

Did the 2025 loans respect it? A test on loans sold to Fannie Mae and Freddie Mac

On loans sold to Fannie Mae the matrix is visible in the data: 3 of 5,247 one-unit loans were above 75% and 9 of 817 two-to-four-unit loans above 70%, and the median CLTV was 60%. Loans sold to Freddie Mac, tested against the same numbers (we did not read Freddie Mac's own limits), look the same. The loans outside the agencies do not follow it. 20% of consumer-purpose two-to-four-unit loans not sold to either agency were above 70%, and 46.6% of DSCR-type two-to-four-unit loans (5,678 of 12,191); for DSCR-type one-unit loans the share above 75% was 3.2%. The file does not say why a given loan went past the limit; what it shows is that the 70% rule on two-to-four-unit properties is an agency rule, and that investors who need more on a small multifamily property were in a different market.

Share of investment-property cash-outs by CLTV, 2025

The loans bunch at the limits. 26% of the loans with a CLTV were at exactly 75% and 14.6% at exactly 70%; 31.7% fell in the band just over 70% to 75%, and only 4.1% went above 75%. The median rate climbs with leverage, from 7% at 60% or less to 7.499% between 70% and 75% (our arithmetic). The principal-residence column explains the CLTV gap: 53.7% of homeowner cash-outs were at 60% or less, against 31.6% of investor loans.

What the cap does to the cash, an illustration. A rental worth $400,000 with a $200,000 first mortgage has $200,000 of equity. At 75% the new loan can be $300,000 and the cash before costs is $100,000; at 70% it is $280,000 and $80,000; at 80% it is $320,000 and $120,000 (our arithmetic). The 5 points between the one-unit and two-to-four-unit limits are worth $20,000 on this property before closing costs (our arithmetic).

Who makes these loans: non-banks, mostly through brokers

Investment-property cash-outs by lender type, 2025

Lender type follows the method of our hard money loan rates page: the federal agency code in each lender's FFIEC institution record, with agency 9 split by whether the name contains “bank”, and lenders with fewer than 10 loans of the group not classified. Non-banks made 86,501 of the 99,035 loans, 87.3%; banks and credit unions made 8,952 at a lower median rate, 6.99% against 7.375%, and a lower median CLTV, 66.7% against 70%. Only 7.4% of the bank loans were DSCR-type, against 64.1% at non-banks. If you hold a rental and your bank will not lend on it, the data says the market is non-bank and broker-led: 53.8% of the loans came through a broker or correspondent.

The 10 largest investment-property cash-out lenders in 2025 (names as filed)

1,620 lenders reported at least one such loan, 115 reported 100 or more and 24 reported 1,000 or more. The 5 largest made 22.8% of the loans, the 10 largest 36.4% and the 15 largest 45.1%. All of United Wholesale Mortgage's cash-outs in the file came through brokers. Velocity Commercial Capital's 9.865% median was far above the rest of the top ten.

The three lenders in the box below, and one more

How to read the lenders' rows. Kiavi's 2025 HMDA record is under Kiavi Funding, Inc.: no loan coded as cash-out, 6,679 refinances coded without cash out (against 18,839 purchases) and 5,209 of them DSCR-type. Regulation C lets a lender that “does not distinguish between a cash-out refinancing and a refinancing under its own guidelines” report all of its refinancings as non-cash-out (comment 4(a)(3)-2.iii), so Kiavi's refinances should be read as including cash-outs; the denial column says none for that reason, not because nobody was denied. Kiavi's website says that DSCR and rental loans are “originated and funded by Figure Lending LLC dba Figure effective 9/1/2026”, so 2026 data will sit under a different filer. Lima One and Investor Mortgage Finance LLC, the entity named in Visio's website footer, did report cash-outs. Velocity is shown because it is the other large name in investor cash-outs; all of its loans in the file came through brokers.

Denied: one application in four, and where the line falls

HMDA also records the applications that did not become loans. We counted each decision on a cash-out application: originated, approved but not accepted, or denied. The denial rate is denied divided by those three; withdrawn applications and files closed for incompleteness are not in the denominator (12 CFR 1003.4(a)(8)). A denial rate on public data is not a measure of how fair or careful a lender is: it depends on which applications a lender takes in the first place.

Cash-out applications on investment property, 2025: decisions and denial rate

35,476 of 143,127 decisions were denials, 24.8%, against 27% for principal-residence cash-outs. Business-purpose applications (24.5%) fared slightly better than consumer-purpose ones (26.1%), and applications on two-to-four-unit properties (21%) better than one-unit ones (25.7%). One lender distorts the headline: RCN Capital, LLC reported 1,685 denials out of 1,701 decisions and only 16 originated cash-outs, and every one of those denials carries the reason “Exempt” instead of a reason. Without it the rate is 23.9% (our arithmetic). We do not know why its applications are coded this way.

Denial rate by the CLTV on the application

This is the main finding on the denial side. On investment-property cash-outs the denial rate was 22.2%, 21.7% and 21.4% in the three bands up to 70% CLTV and 20.1% between 70% and 75%, then 37.8% between 75% and 80%, 66.7% between 80% and 85% and 71% above 85%. The jump sits where the agency limit and most DSCR-lender limits sit. Among DSCR-type applications the pattern is the same, 15.7% at 70% to 75% and 35.2% at 75% to 80%. Principal-residence cash-outs show it too, though later: 24.9% at 70% to 75% and 28.7% at 75% to 80% (the homeowner limit is 80%). Only 29,681 of the 35,476 denied investment applications reported a CLTV, so this table covers most but not all of them. The file shows the CLTV that was applied for, and not whether a smaller loan would have been approved.

Why they were denied

A lender may list up to four reasons; the average denied application lists 1.197 reasons. Collateral, the lender's view of the property or its value, came first on investment cash-outs, cited in 28.9% of denials (principal-residence cash-outs: 20.1%), and rose to 33.9% among business-purpose applications with no debt-to-income ratio reported. Debt-to-income ratio is the leading reason on consumer-purpose investment applications (37.8%) and on principal-residence cash-outs (32.6%), and only 7.7% of the DSCR-type denials. Credit history was cited in 19.3% and an incomplete application in 15.8%. The reasons are the lender's own coding, and the reason “Other” alone accounted for 15.7%.

Denial rate at the largest lenders by decisions (names as filed)

Among the 55 lenders with 500 or more cash-out decisions on investment property, the median denial rate was 24.2% and the middle half ran from 11.8% to 36.4%; individual lenders ran from 0% to 99.1%. Non-bank lenders denied 24.8% and banks and credit unions 25.7%, so the type of lender explains little; which products and applicants a lender takes in explains more. A high rate can mean a lender that quotes first and counts a declined quote as a denial, a lender that sends unqualified applicants elsewhere, or a lender that is simply stricter, and the file cannot tell them apart. Treat the table as a reason to ask a lender its approval rate on your kind of deal before you pay for an appraisal.

Agency or DSCR: the two doors, in the lenders' own words

A conventional (Fannie Mae or Freddie Mac) cash-out underwrites the borrower, with a debt-to-income ratio and, for DU loans, six months of reserves if the ratio exceeds 45%. A rental loan of the DSCR type is, in Kiavi's words, “qualified on the property's cash flow rather than your personal income”, and 82% of investment cash-outs were reported as business or commercial purpose. The three lenders in the box below publish the following on their own pages. They are the lenders' claims, saved as read on October 8, 2026, and they change.

What Kiavi, Lima One and Visio say on their websites

Kiavi's website says its cash-out is available after 90 days of ownership, or immediately on its bridge loans, and lists “No prepayment penalty after year 3”; the page does not give the schedule before then. Lima One's page lists a 75% limit on cash-out and 90 days to use the full market value. On the pages we saved, none of the three states an origination fee or points schedule, a recourse term other than Visio's personal guarantee, or a prepayment schedule: those come in the term sheet. If the loan you are paying off carries a prepayment penalty, that penalty is a cost of the refinance, which our DSCR refinance guide works through. Our reviews of Kiavi, Lima One and Visio have each company's record.

By state

The 10 largest states for investment-property cash-outs in 2025

The ten states are 61.3% of the loans. California is 13% with the highest median loan, $425,000, and a median CLTV of 60%, lower than the 70% of most states, and a denial rate of 21%. Ohio's median CLTV was 74% and its median loan $135,000. In every one of the ten states the median investment rate was above the principal-residence cash-out rate, by 0.375 to 0.875 points (our arithmetic).

What this data cannot see

Rules behind the file

• The file is last year's loans. HMDA 2025 covers loans made and applications decided in calendar year 2025 at a median rate of 7.25%; it carries no month, so it cannot say what a lender quotes today.
• Small lenders are missing. A lender that originated fewer than 25 closed-end mortgage loans in either of the two preceding years does not report (12 CFR 1003.3(c)(11)), so many local lenders and credit unions are not in the file.
• No fees, points or credit score on business-purpose loans. The public file shows the note rate and the CLTV, but not the points, lender credits or the borrower's score for these loans; the lender's term sheet has them.
• “Cash-out” is the lender's label. Kiavi coded no loans that way, so any count of cash-out refinances is a floor for what investors did.
• CLTV includes other liens. It is all debt secured by the property over the value the lender relied on, so a second mortgage counts, and the Fannie Mae matrix's LTV, CLTV and HCLTV can differ from it.
• Denials are applications, not people. One investor can file with several lenders, and withdrawn applications are not counted.

What you can do with this

1. Pick the door by the leverage you need, not by the rate. If your numbers work at 70% to 75% of value on one unit, the agency matrix and the DSCR lenders overlap. If you need 75% on a duplex, the agency limit is 70% and, in the file, DSCR-type lenders went past it on 46.6% of two-to-four-unit loans.
2. Do the cap arithmetic before you apply. Value times the limit, minus what you owe, minus closing costs, is the cash. At 75%, a $400,000 property with a $200,000 mortgage gives $100,000 before costs.
3. Keep the application at or under 75% if you can. Denial rates were between 20.1% and 22.2% in every band up to 75% and 37.8% from 75% to 80%. If an appraisal comes in low, a smaller loan may be the loan that gets approved.
4. Get the seasoning rule in writing. Fannie Mae: six months on title and a first mortgage 12 months old. Kiavi and Lima One publish 90 days. Ask which value the lender uses before then, because that sets the cash.
5. Ask for the approval rate on your kind of deal. Lender denial rates in this file ran from near zero to near 100%. Ask a broker which lenders have closed your property type and loan size.
6. Price the premium, then compare other routes. In 2025 the median was close to the investment purchase rate, 7.25%, and the other route to compare is the HELOC: a line of credit on a rental is a small market, as our HELOC on investment property page shows.

Kiavi pays us a referral fee when a loan closes through its button in the box below; Lima One and Visio pay us nothing. All three make closed-end investor loans, including cash-out refinances on rental property; their own pages, saved above, are the source for what they say about limits and seasoning, and the HMDA figures on this page are the same either way. This page does not tell you which loan to take.

FAQ

Loan-level data: FFIEC/CFPB HMDA Data Browser, nationwide 2025 files for loan purposes 32 (cash-out refinance, originated, denied and approved but not accepted), 31 (refinance without cash out, originated) and 1 (home purchase, originated), downloaded October 8, 2026 (request URLs in the sources folder; the API has no occupancy filter, so each nationwide file was streamed and filtered locally to occupancy codes 3 and 1, conventional, first-lien, one-to-four-unit, closed-end, non-reverse loans; the extracts, the scripts and their output are in the sources folder). Lender names and agency codes from the FFIEC public institutions API, read October 8, 2026; field definitions from the FFIEC public LAR data fields and the 2025 HMDA Filing Instructions Guide; Regulation C from eCFR (12 CFR 1003.3, 1003.4 and Supplement I), read October 5 and 6, 2026; the Fannie Mae Eligibility Matrix dated August 5, 2026 and Selling Guide B2-1.3-03 (December 10, 2025), read October 8, 2026; Kiavi, Lima One and Visio terms from their websites, read October 8, 2026, and they are the lenders' own claims. All counts, percentiles, medians, shares, gaps and the worked example are our arithmetic. This is analysis of public records, not investment, legal, lending or tax advice, and not a loan offer.