Housing Dashboards · Milwaukee lens

Why the housing market isn’t a bubble

Sorry, first-time buyers.

All my first-time homebuying friends are waiting for the bubble to burst (no judgment). And since I spent 12 years in mortgage insurance, I’m often the bearer of bad news (for them): The data clearly shows this is not a repeat of the housing crash. 2008 was a credit bubble – easy loans pumped up demand on top of a building boom. This time, prices rest on a shortage of homes and on owners with cheap loans, plenty of equity and a habit of paying on time. Here’s the thing about a crash: it needs forced sellers, lots of people who have to sell right now at any price. In 2008 there were millions. Today there are almost none, and I’ll use the power of charts to explain why throughout the page.

The first time a technical term comes up, it has a dotted underline. Hover over it or tap it for a quick definition.

The puzzle

Yes, prices are higher than in 2006… even after inflation.

This is why so many would-be buyers think it’s a bubble, and I don’t blame them. After inflation, US home prices are — above their 2006 peak, and Milwaukee is — above its own. They’re way above the old trend line too. Not ideal! But high prices alone don’t make a bubble. A bubble has to be held up by something that can vanish, like easy credit. A shortage of homes can’t vanish overnight. Prices can still cool as they drift back toward what buyers can afford, but that looks very different from a crash.

Home prices after inflation

Index: 100 = prices in early 2000, so 160 means 60% higher after inflation. The FHFA all-transactions house price index divided by the consumer price index. Other measures put the US anywhere from — (Zillow) to — (FHFA’s purchase-only index) above its 2006 peak, but every one of them is above it.

    How far above trend? From 1980 to 2000, prices rose about — a year faster than inflation. Today they’re about — above that long-run trend. The crash only knocked prices back to the line, and more than half of today’s gap opened after 2019, with the shortage and record-low rates in 2020 and 2021.

    Another bubble test: buying vs. renting. The price-to-rent ratio is — above its 1983 to 1999 norm, close to the — of the 2006 peak (it briefly passed that in 2022). The difference is direction. In 2005 the gap was still growing fast. Since — the ratio has fallen — as rents kept rising while prices stalled. Milwaukee hasn’t turned yet.

    US prices against their long-run trend

    FHFA all-transactions index divided by the consumer price index, 100 = early 2000. The dashed line is the trend fitted to 1980 through 1999, extended to today.

      Why prices are high, 1 of 2

      2008 had too many houses. Now we don’t have enough.

      Same with great TV: in 2008, Mad Men, Breaking Bad, The Wire, 30 Rock and The Office were all on the air at once.

      Housing scarcity can be summed up in a nutshell: Before the bubble, the US approved about — homes a year; since 2008 it’s averaged — – roughly — of that pace – while the number of households has grown nearly 30% since 2001. That’s like hosting Thanksgiving for a bigger family every year and buying a smaller Tofurky (Vegan alert). Milwaukee never had much of a boom to come down from, and since 2008 it has built at about — of its early-2000s pace. Compared with that pre-bubble pace, that adds up to roughly — homes across the US that never got built, including about — in the Milwaukee metro. So, no glut: just — of owner homes sit empty and for sale, against — in early 2008.

      Homes permitted each year, compared with the pre-bubble pace

      Homes approved each year, compared with the average year from 2000 to 2002, before the bubble (100% = same pace). Milwaukee metro is Milwaukee, Waukesha, Ozaukee and Washington counties. Census Building Permits Survey.

        Why prices are high, 2 of 2

        The 3% mortgage is the new golden handcuffs

        I know because I’m shackled myself.

        If you locked in a 3% mortgage in 2021 like me, you already know this one. About half of US mortgages (—) carry a rate under 4%, and nearly one in five is under 3%. A new 30-year loan averaged — this summer, and it has climbed above 7% this fall. Moving would mean trading a great deal for a much worse one, so people stay put – and the homes they’d have sold never hit the market. Wisconsin owners are even more locked in: — are under 4%. For buyers, that means fewer homes to choose from and more competition for each one.

        US mortgages by interest rate

        Share of all US mortgages people are still paying, by the rate on the loan. The dashed line marks the share under 4% today. A new 30-year loan averaged — in the third quarter of 2026 (Freddie Mac). FHFA National Mortgage Database, from 2013.

          Why it won’t pop, 1 of 3

          Today’s loans are a lot more stable

          That’s good for all of us. The Big Short 2 will have to wait.

          Remember “NINJA” loans: no income, no job, no assets? This is what their absence looks like. In 2006, — of new mortgage dollars went to borrowers with shaky credit (under 620, the usual subprime line). In 2025 it was —. The typical new borrower scored — out of 850 at the end of 2025, against — at the end of 2006. And only — of mortgages have a rate that can adjust, so almost nobody’s payment is about to reset higher – the thing that sank a lot of owners last time.

          New mortgage dollars by borrower credit score

          Share of each year’s mortgage originations, by Equifax Risk Score at origination. Labels show the under-620 share in 2006 and 2025. NY Fed Consumer Credit Panel / Equifax. 2026 is left out because the NY Fed switched to a different score that year.

            Good news first: most buyers don’t put 20% down. In 2025, — of US home-purchase loans came with less than 20% down, the same share as in 2006 (—), and in Wisconsin it was —.

            What changed is how those loans get made, and this one’s personal, since I spent 12 years in mortgage insurance. In 2006, a lot of buyers skipped mortgage insurance by taking out a second “piggyback” loan to cover the down payment, and — of regular, non-government loans had less than 5% down. In 2025 it was —. Today, low-down-payment buyers mostly use government-backed FHA or VA loans, or a regular loan with private mortgage insurance. That insurance covers the lender if you stop paying, so a second company checks that you can actually afford it.

            Home buyers putting down less than 20%

            Share of US home-purchase loans, by down payment. Second “piggyback” loans count against the down payment, so an 80/20 loan with nothing down counts as under 5%. “All loans” includes government-backed FHA, VA and USDA loans. FHFA National Mortgage Database.

              Why it won’t pop, 2 of 3

              Owners have a thick equity cushion

              So almost no one is underwater. In 2008, millions were.

              After what they owe, homeowners now own — of their homes’ value. In early 2012 it was —. Here’s why that matters: if you owe more than the house is worth and lose your job, you can’t sell your way out, and that’s how a layoff becomes a foreclosure. Today just — of US mortgages are underwater. In early 2013, the earliest this data goes, it was —. In Wisconsin it rounds to zero, which tracks for a state that never really did the boom in the first place.

              Owners’ equity as a share of home value

              US households. Federal Reserve Financial Accounts (Z.1).

              Mortgages that are underwater

              Share of mortgages where the loan is larger than the home’s estimated value. FHFA National Mortgage Database, from 2013.

                Why it won’t pop, 3 of 3

                People are paying their mortgages on time

                At the worst of the crisis, — of US mortgage debt was three or more months behind. Now it’s —. About — people got a new foreclosure in the second quarter of 2009; this spring it was —. In the Milwaukee metro, the share of mortgages seriously late or in foreclosure peaked at — at the end of 2010. Today it’s —.

                Mortgages seriously late or in foreclosure

                Share of mortgages 90+ days past due, in foreclosure or in bankruptcy. FHFA National Mortgage Database. The latest two quarters are preliminary.

                  People with a new foreclosure, per quarter

                  Thousands of US consumers with a new foreclosure on their credit report. NY Fed Consumer Credit Panel / Equifax.

                  Why it won’t pop, the recap

                  No forced sellers, no crash

                  Remember the forced sellers? This is where it usually clicks for my friends “waiting for the bubble to burst.” 2008 had a conveyor belt of them: prices dipped, owners went underwater, payments reset, banks foreclosed, and the foreclosures pushed prices down again. Today that conveyor belt is mostly empty. People with equity and a payment they can afford don’t have to sell, so a weak market just gets… quiet. Fewer sales, flat prices, no fire sale.

                  How a dip turned into a crash, then and now

                  1. Prices dip and owners go underwater

                    Mortgages owing more than the home is worth, US

                    Then
                    — 2013, earliest data
                    Now
                    —
                  2. Owners fall behind on payments

                    Mortgage debt three or more months behind, US

                    Then
                    — Q1 2010
                    Now
                    —
                  3. Banks force sales

                    People with a new foreclosure in one quarter, US

                    Then
                    — Q2 2009
                    Now
                    —
                  4. Empty homes flood the market and pull prices lower

                    Owner homes sitting empty and for sale, US

                    Then
                    — Q1 2008
                    Now
                    —

                  In 2008, step 4 fed back into step 1 and the loop kept running for years. Today the loop has little to run on: few owners are underwater, almost all have fixed payments, and there aren’t enough homes for sale to flood the market.

                  The honest part

                  OK, so what are the warning signs?

                  I’m not saying housing is risk-free – just that the risk moved, from shaky loans to stretched budgets. Builders have more unsold new homes than usual, a bigger share of credit card balances is going seriously late than before the pandemic, and buy now, pay later is growing in a corner credit reports barely see: — of adults used it last year, and — of them paid late at least once, up from — in 2021. Unemployment is higher than in 2023, though it has eased a little this year, and Milwaukee is still below the national rate.

                  If you’re buying right now, you’re feeling the worst of it. With 20% down, the payment on a typical $— US home takes — of average household income, about 1.5x what it was in 2019 (—), and not far from mid-2006’s —. With 5% down, like many first-time buyers, it’s —: about — a month in principal and interest plus — for mortgage insurance, before taxes and homeowners insurance. At October’s 7.3% rate it would be about 17%, nearly the 2006 level. That’s a slow squeeze, not a wave of foreclosures.

                  A new buyer’s mortgage payment, as a share of income

                  —

                  The monthly payment on a typical US home bought that quarter with 5% down, like many first-time buyers, including mortgage insurance, as a share of the average household’s income after taxes. It’s close to the — of mid-2006, at the peak of the bubble, and in late 2023 it passed it (—). With 20% down the picture is the same, just lower. First-time buyers usually earn less than average and have less saved, so for them it’s even tighter. Zillow, Freddie Mac, Federal Reserve, Census; insurance at an assumed 0.5% of the loan a year.

                    Unsold new homes

                    —

                    Months it would take to sell today’s new-home inventory. It was — at the end of 2008. Census.

                    Credit card balances going seriously late

                    —

                    Share of card balances moving into 90+ days late over the past year. That is above late 2019 (—) and late 2007 (—), but below the 2009 peak of —. NY Fed.

                    Unemployment rate

                    —

                    Milwaukee metro in Q2 2026, against — for the US. BLS; Milwaukee is not seasonally adjusted.

                      Buy now, pay later: more people using it, more of them paying late

                      —

                      Of US adults who used buy now, pay later in 2025, the share who were late on at least one payment. One in five users bought groceries or food delivery with it. The Fed’s yearly Survey of Household Economics and Decisionmaking; in 2023 it dropped brand names like Affirm and Klarna from the question and described the service as “pay in four,” so usage before 2023 isn’t a perfect match.

                        The worst case

                        If it goes wrong, expect a slow leak

                        Think week-old party balloon, not a pin.

                        Waiting for a crash is also a long bet. Even in 2008, with forced sellers everywhere, US prices took about — years to hit bottom, and they fell — in dollars. Anyone who waited for the bottom waited that long, paying rent the whole time, and prices were back above the old peak by —. Milwaukee fell — over about — years.

                        The leak may already have started. Since mid-2024, US prices are up — in dollars but only — after inflation. That is how a stretched market often cools: prices stop outrunning everything else and drift for years instead of dropping. Milwaukee, where building has lagged the most, is still climbing, up — after inflation over the same stretch.

                        Even the 2008 crash was slow

                        FHFA all-transactions house price index, not adjusted for inflation, so these are the prices buyers actually saw.

                          For first-time buyers

                          So what do I do?

                          Asking for all my friends.

                          I’m not your loan officer, but here’s what the data says if you’re trying to buy:

                          • Don’t plan around a crash. Even 2008, the worst housing crash in modern memory, took about — years to bring prices down —.
                          • Watch the payment, not just the price. On a typical home with 5% down, a 1-point drop in mortgage rates cuts principal and interest by about — a month: — at this spring’s — rate, — at —.
                          • You probably don’t need 20% down. Most buyers don’t put that much down. Mortgage insurance or an FHA loan adds to the monthly cost, but it can get you in the door sooner.
                          • Look where the supply is. Builders are sitting on more unsold new homes than usual, so new construction may be where there’s room to negotiate.
                          • In Milwaukee, don’t expect prices to fall first. Building here has lagged the most, and prices are still climbing after inflation, up — since mid-2024.

                          The devil’s advocate

                          How this could still turn into a crash

                          Arguing with myself so you don’t have to.

                          A crash needs forced sellers, so the real question is what could create millions of them again. It would take a shock big enough to break several of the safeguards above at the same time. Here’s how that could happen.

                          • A deep recession with lots of layoffs. Equity and fixed rates protect owners who keep their jobs. They help a lot less if millions lose their jobs at once, burn through their savings and stop paying. In late 2009, US unemployment hit 10%. This summer it averaged —. A jump that big would test everything on this page.
                          • An AI bust could be the trigger. Counting Amazon and Tesla, tech companies are now — of the value of all US stocks, against — at the dot-com peak in early 2000. Stocks are nearly as expensive compared with earnings as in 1999, and spending on computers, software and data centers is a bigger slice of the economy than in 2000. If that spending stalls, the damage would come through layoffs in tech and construction, and through stock losses that make people spend less. History is mildly reassuring: from 2000 to 2003 the S&P 500 fell —, but mortgage rates fell too and US home prices rose —. The risk is a bust big enough to spread into layoffs everywhere.
                          • The newest buyers have thin cushions. The average owner has plenty of equity, but recent buyers who put 3% to 5% down don’t. About — of US mortgages owe more than 90% of the home’s value, so a 10% price drop would push those owners underwater. That’s a small share, but these are mostly recent buyers, and many of them are the first-time buyers this page is about.
                          • Insurance and property taxes. A fixed-rate mortgage doesn’t fix the rest of the housing bill. Homeowners insurance rose 8.7% faster than inflation from 2018 to 2022, and in the fifth of ZIP codes most exposed to climate damage, people pay 82% more than in the safest fifth and get dropped by their insurer about 80% more often (US Treasury). Mortgage escrow payments, which cover taxes and insurance, are up about 45% in five years and 70% in Florida since 2019, and in the states where they’re rising fastest, more borrowers are falling behind (Cotality). That’s one way owners with cheap, fixed loans could still become forced sellers.
                          • Places that built a lot. The extra unsold new homes in the warning signs above aren’t spread evenly. Prices are most likely to fall first in the fast-building areas, such as parts of Florida and Texas, and when builders cut prices to clear inventory, nearby owners feel it too.

                          Tech’s share of the US stock market

                          Market value of tech companies (computers, chips, software, internet and telecom equipment) plus Amazon and Tesla, as a share of all US stocks, at each month-end. The gray line leaves Amazon and Tesla out, since industry data files them under retail and autos. Ken French Data Library (12 industries, “BusEq”), from CRSP; Amazon and Tesla are share price × shares outstanding from their SEC filings.

                            2000 to 2003: stocks crashed, home prices kept rising

                            S&P 500 (quarterly average of monthly prices) and the FHFA all-transactions house price index, both before inflation and set so Q1 2000 = 100. Robert Shiller, FHFA.

                              What would change my mind

                              If these three move together, the forced sellers are coming back, and it’s time to take crash talk seriously. Each bar shows the latest reading against my line, with the worst of the 2008 crisis for scale.

                              US unemployment rate, quarterly average (BLS). Mortgage balances 90+ days late (NY Fed Household Debt and Credit Report). Months to sell all new homes for sale (Census).

                              How this was built

                              • Sources. Every housing number comes from the agencies’ own files rather than a data portal, mostly multi-tab workbooks, zipped XML and fixed-width text: FHFA’s House Price Index, National Mortgage Database and Monthly Interest Rate Survey; the Federal Reserve’s Financial Accounts (Z.1); the New York Fed’s Household Debt and Credit Report; Census new construction, new home sales, building permits, construction spending and housing vacancies; BLS consumer prices and national and local unemployment; Freddie Mac’s mortgage rate survey; and Zillow. The devil’s advocate adds the Ken French Data Library, Robert Shiller’s stock data, BEA, SEC EDGAR filings and Yahoo Finance prices. Buy now, pay later comes from the Fed’s household survey (SHED) and a CFPB report.
                              • Timing. Quarterly values average the months in each quarter; balances are end of quarter. “Then” values use the year before the crash (2005 to 2008) or the crisis peak, as the text says. In the forced-seller chain, each “then” is that measure’s worst quarter on record, so the four come from different dates between 2008 and 2013. Q4 2025 BLS and Census survey values cover November and December only, because October 2025 data wasn’t collected during the federal funding lapse.
                              • Gaps. FHFA doesn’t publish loan rates, loan-to-value or credit scores for metro areas, so Wisconsin stands in for Milwaukee. Underwater and rate data start in 2013, after the worst had passed, so the underwater “then” understates 2008–2011. Credit scores at origination are Equifax Risk Score through 2025; the NY Fed moved to VantageScore 4.0 in 2026, so 2026 isn’t compared.
                              • Prices. “After inflation” is FHFA’s all-transactions index divided by CPI-U. The long-run trend is a log-linear fit to 1980–1999 (about 0.4% a year), extended to today; a 1% a year trend would still leave prices about 23% above it. The 2008 decline uses the unadjusted index: the highest quarter before 2010, the lowest after it, and the first quarter back at the peak. Price-to-rent divides by BLS rent of primary residence (seasonally adjusted from 1981), with the 1983–1999 average = 100; owners’ equivalent rent gives a similar picture. Milwaukee uses Midwest rent, and its own CPI rent, which ended in 2017, agrees. CPI rent tracks what existing tenants pay, so it lags new-lease rents.
                              • Payments. The new-buyer payment is principal and interest on Zillow’s typical US home value (seasonally adjusted, so a little below the spring figures in Zillow’s monthly reports) at Freddie Mac’s 30-year rate, divided by average disposable income per household (Fed personal income over Census households). Average income runs above the median, so compare years rather than read the level. The 5%-down case adds mortgage insurance at an assumed 0.5% of the loan a year, a round figure for a strong-credit borrower; real premiums vary with credit score and coverage. The “watch the payment” tip compares the Q2 2026 rate with one point lower, before mortgage insurance, taxes and homeowners insurance, rounded to the nearest $10. The October figure uses Freddie Mac’s October 22 rate (7.32%) with the Q2 2026 home value and income.
                              • Down payments. The National Mortgage Database covers all purchase loans from 1998 and counts piggyback second loans. The older Monthly Interest Rate Survey covers conventional first mortgages from 1973 to 2018. The two aren’t spliced.
                              • Devil’s advocate. Tech is Ken French’s 12-industry “BusEq” group (computers, chips, software, internet services and electronic equipment): number of firms × average firm size from CRSP, shifted to month-end, through July 2026. Amazon and Tesla add month-end split-adjusted price × split-adjusted shares outstanding from each 10-K and 10-Q cover page. Tesla’s share count rose about 22% from mid-2025 to mid-2026, mostly restricted shares granted to its CEO, which lifts its recent value. Tech spending is BEA private investment in information processing equipment and software, plus Census private data center construction (annual rate, from 2014), as a share of GDP. The S&P 500 and CAPE are Shiller’s monthly series through September 2026. Tech share is not an “AI share”: no public series splits stock value by AI exposure. Insurance figures are from the Treasury’s Federal Insurance Office (January 2025, 2018–2022 data on about 50 million policies a year); escrow growth is from Cotality, as reported in March 2026.
                              • Buy now, pay later. Fed SHED, data years 2021–2025: use in the past year among all adults, and being late on a payment among users. The late-payment question has the same wording every year; the usage question’s description changed in 2023. The CFPB’s data from six lenders counted 53.6 million borrowers in 2023, which likely overcounts people, since some borrow from more than one.