Erdmann Housing Tracker · Housing & Cities
TIER 5 Tue, 21 Apr 2026 14:04:04 +0000
In yesterday's post, I discussed the various pathologies and blind spots that defined reactions to the 2008 financial crisis. ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ | | ---|---|--- | | | Forwarded this email? Subscribe here for more --- --- # Revisiting Aggregate Demand in 2008 | | Kevin Erdmann --- | Apr 21 --- | --- --- | | | --- | | --- | | --- | | --- | | READ IN APP --- In yesterday's post, I discussed the various pathologies and blind spots that defined reactions to the 2008 financial crisis. While poking around the data, I realized that there are some incremental connections in the story that I haven't previously explicitly laid out. Upgrade to paid Yesterday's post revolved around the debate between John Cochrane, who made a comment calling for recession in November 2008, months after calling for stimulus would have been tardy. And, while Brad Delong and Paul Krugman understood how crazy that was, their responses to it were neutered by the economics academy's complete blindness to the massive and regressive demand shock created by the unnecessary 2008 mortgage crackdown. #### Debt and Cash Increasingly, after the turn of the century, American household spending was being funded by debt. It wasn't inflationary because the Fed was being very tight with currency growth. So, total spending growth was normal, and nominal GDP growth was pretty normal - a bit above the 5% target I would prefer in a market monetarist approach, but about where recovery GDP growth generally had settled in the Great Moderation period. | | ---|---|--- Figure 1 One way to describe what happened in 2008 is that mortgage growth suddenly collapsed, and the Fed was unable or unwilling to replace that liquidity with enough cash to keep nominal spending stable. I think almost everyone would agree with that. And, the debate I mentioned in yesterday's post gets to the heart of the problem. In late 2008, the liquidationists were like Kylo Ren yelling "Moooaaaarr!" or rather "Leeeessssssss!" and the Keynesians saw the decline in mortgages outstanding as a correction from profligacy and recklessness. So, rather than stopping the purposeful downward overcorrection in mortgage access that was regressively destroying wealth, they wanted to replace that demand with fiscal stimulus. The scale of the mortgage crackdown was so large that the various forms of stimulus had to dwarf the scale of normal fiscal and monetary operations to even put a dent in it. #### Mortgages Outstanding Let's discuss mortgages for a minute. Figure 2 shows quarterly mortgages originated (red), net change in mortgages outstanding (black), and the difference between the two (orange). Mortgages are naturally repaid every quarter, so that mortgages outstanding naturally decline if new mortgages aren't originated. In fact, in any given quarter, most of the newly established mortgage debt replaces those repaid balances - either because a homeowner sold a home and paid the balance off or because the principal on every mortgage is slowly declining. Under normal conditions, new originations slightly outpace the declining principal. You can see with the orange line in Figure 2 that principal declines pretty regularly. When the orange line pops higher, that is an interest rate motivated refinancing boom. A bunch of borrowers pay off high interest mortgages and replace them with low interest mortgages. You can see that that is what was happening in 2003. It happened in 2020 and 2021, too, but in 2020 and 2021 there was also some cash-out refinancing. The new mortgages were a bit larger than the refinanced mortgages, so net mortgages (the black line) increased, too. And some of that rise might be from post-Covid payment deferrals. | | ---|---|--- Figure 2 Notice that the net change in mortgages remained pretty high from 2000 to 2007. Throughout that entire period, there was a lot of equity extraction. Some of the rising mortgages outstanding funded new homes, or the purchase of existing homes whose prices had risen. But, after subtracting out the construction of new homes, the net aggregate effect was that some home equity was being converted, in some way, to cash, and used for current expenditures. It wasn't all home equity loans or cash-out refinancing. Increasingly from 2000 to 2005, high-income borrowers were using Alt-A loans to purchase homes from Closed Access city homeowners (New York, LA, San Francisco, Boston) who then moved to cheaper markets. The net effect of that looked like equity extraction, but it was more just a rearranging of mortgages and equity among households. While the housing market was booming, aggregate leverage was flat. Home prices were growing fast enough that equity was growing as fast as mortgages outstanding, until 2006. Equity extraction slowed down after 2005. It just didn't slow down quickly enough to keep up with stagnating and then declining home prices, and that is when average leverage started to increase. And if that had been the end of it, and agency mortgage access had remained stable in 2008 and the Fed had stimulated enough to cause a recovery in housing starts in 2007 and a return to normal population growth rates in Nevada, Arizona, and Florida, then all the seemingly naive propositions that get scoffed at would have been true. There would have been a correction from the subprime lending boom. There would have been some banking dislocations, some bankruptcies, and blown up CDOs. But, Ben Bernanke's claim that it would have been contained would have been true. There would have been no recession or a shallow recession. #### Mortgages and Consumption This might be something I have underemphasized in my previous writing. In general, I have documented how the lending boom had significant effects on the housing market before 2008. It probably juiced home prices temporarily by something a bit less than 10%. But, it was dwarfed by the effect of undersupply on homes in the Closed Access cities and the secondary effect on demand in the Contagion cities when displaced Closed Access refugees descended by the hundreds of thousands into the cities where they migrated to. And, I find that in much of the country, the stimulus from the lending boom was moderately stimulating home construction in slow-growing markets and moderating rents in a positive way that was, mostly, just raising real local incomes by keeping rents low. The lending boom did raise prices, on average, but it was negatively correlated with all the other factors that were raising prices. So, the lending boom had the strongest effect on prices in places where price trends had been the most moderate. tl;dr: I find an important role for lending in housing markets before 2008, but it is much less important than other factors stemming from the shortage. But, that lending _did_ have an important role in aggregate spending. I would _agree_ with the scale that the conventional story attributes to mortgage borrowing as a factor keeping the growth of personal expenditures going. Figure 3 compares net mortgage growth, new home sales, and the change in personal consumption expenditures. The first of these to break was residential investment. Mortgage growth broke next. It went negative by the 2nd quarter of 2008. And, _because of that_ , aggregate personal expenditures dropped in the 3rd quarter and went negative in the 4th quarter. | | ---|---|--- Figure 3 By the 3rd quarter of 2009, consumption had recovered, and the recession was officially over. It took a massive amount of fiscal and monetary stimulus _because_ the mortgage headwinds were sucking tens of billions of liquid dollars out of the economy each quarter and sticking them in illiquid home equity while the housing bust took that equity away. The downshift in mortgage growth was a really important reason why spending and GDP collapsed. Everyone agrees with that. I agree with that. But, where I would add an important asterisk is that the decline in mortgage access after 2007 was more important than the rise before 2007, and it was associated with a completely different change among a completely different group of mortgages. #### The Mortgage Crackdown And, here's where I have previously failed to connect the dots explicitly enough. Figure 4 compares net mortgage growth to gross mortgage originations by credit score. The drop in _originations_ to sub-760 credit scores after 2007 can account for the entire drop in net mortgages outstanding. | | ---|---|--- Figure 4 And, remember, reporting by Fannie Mae suggests that the composition of their new borrowers in 2007 was not significantly different than it had been a decade earlier. In the national data I am referencing here, shown in Figure 5, the percentage of mortgages going to sub-760 scores was as high in the late 1990s as it was in the 2000s. The dip and rise around 2002 is similar to the dip and rise around 2020 because high score borrowers tend to be over-represented during interest-rate driven refinancing booms. | | ---|---|--- Figure 5 I can reasonably estimate the rate at which those mortgages were being repaid, to get an estimate of net mortgage growth by credit score. When the subprime boom had ended in 2007, mortgage growth had slowed, but it had not gone negative. That happened because the housing boom in the Contagion cities and cash-out refinancing in the subprime markets had both ended. The decline below zero after 2007 was the result of the permanent retraction of purchase loan access through the federal agencies to 1/3 of the market. The subprime market before 2007 and federal agency lending after 2007 are 2 different markets that declined for different reasons. The first was a correction from an unsustainable boom, and _it had been an important factor keeping spending growth high even though the Fed had slowed currency growth_. The second was a mistaken and unnecessary over-reaction that never should have happened, and it was responsible for the deep scale of the collapse in 2008, the slow recovery, and the need for massive stimulus. | | ---|---|--- Figure 6 I have generally focused on the effect that the mortgage crackdown had on home prices and wealth - collapsing home prices the most in neighborhoods with lower incomes. But, I haven't emphasized enough that the collapse in personal expenditures in 2008 was likely also entirely driven by the targeted loss of liquidity. Families in the neighborhoods with collapsing home prices also, in the aggregate, had a targeted liquidity shock. #### Regional Differences As the housing boom turned to bust, the borrowing that had been funding homebuilding increasingly funded consumption, and eventually, it was funding consumption in regions that otherwise would have been in recessions already. The conventional bubble/bust story just compresses all of these transitions into a single story of profligacy and overproduction leading to collapse. By, 2006, mortgage growth was making up for _underproduction_. This played out regionally. And, by 2008, mortgage contraction across the country was _causing_ underproduction. The Contagion cities all saw deep collapses in employment growth after 2005, growth started turning negative by the end of 2007. They were already in a condition normally associated with the _end_ of a recession. | | ---|---|--- Figure 7 The perennial and deep undersupply of homes in the Closed Access cities creates a countercyclical population pattern in the closed access cities. When Americans are better off at a pace that is higher than the pitiful willingness of the Closed Access cities to permit new homes, the Closed Access cities have to depopulate. That makes migration into and out of the cities closed access refugees move to pro-cyclical, so they have housing booms during expansions, and then they are first into recessionary conditions. (Now that the housing shortage is national, I think the destination regions still are the recipients of pro-cyclical domestic migration, but since there is so much local pent up demand for new homes, they don't experience the decline in homebuilding and construction employment that would have traditionally accompanied slowing migration. So, now, we get the pro-cyclical migration flows, but maybe we won't get the recessions. In Figure 8, notice how both Trump administrations have been associated with recessionary collapses in migration to Florida, but neither has been associated with a significant decline in construction. Compare that to 2005-2008.) | | ---|---|--- Figure 8 I wonder if we would have largely avoided a recession, even in 2008, if the mortgage crackdown hadn't happened. The late borrowing in the Contagion cities would have been a part of that process. As construction collapsed in 2006 and 2007, Contagion families still had ample home equity to borrow from and they attempted to use that to smooth out spending as they were already into recessionary conditions. That borrowing was equated with profligacy and recklessness, even though, by then, it was quite the opposite - it was an inevitable attempt at consumption smoothing. Everyone watches "The Big Short" and "Margin Call" as if the collapse of that borrowing was the inevitable cause of a financial crisis. Ironically, it was the public's immovable insistence on treating that borrowing as something it was not that caused the crisis. I suspect that recovery in 2007 and 2008 would have naturally been associated with moderation of debt in those markets if construction had increased and nominal GDP growth would have remained stable. Figure 9 shows cumulative changes to home prices and debt per capita in Midwest states after 1999. The diagonal line reflects stable leverage - debt and home equity rising at a similar pace. I have marked the end of 2003 with large yellow circles and 2005 with large green circles. During the boom period, families in the Midwest, on average, were increasing the leverage in their homes and using it to increase spending. That largely ended after 2005. The black line is the US average. Leverage in the Midwest had generally outpaced the US average. Debt grew at about the same rate as elsewhere, but home price appreciation tended to be lower than the US average. In other words, cross-sectionally, when construction was still growing and prices were rising, debt growth wasn't particularly correlated with higher home prices. Mortgage growth affected spending more than home prices. Also, this corroborates the finding that I noted above from an earlier Mercatus paper. The credit boom was related to price inflation in credit sensitive ZIP codes in the least inflated metro areas. In other words, the average home price appreciation in these Midwest states might have trailed the national average a bit more without that credit - not nearly enough to justify the eventual price collapses shown in the chart, but a bit. | | ---|---|--- Figure 9 In the Contagion states, in Figure 10, during the boom period, homes weren't becoming more leveraged than average. Home prices were rising at an above average pace, and new debt was rising at a pace pretty similar to the US and the Midwest. The unusual increase in debt in the Contagion states happened during the slowdown. After 2005, debt per capita in the US and the Midwest states generally increased by another 20% or so. In the Contagion states, it increased 30%-50% more. | | ---|---|--- Figure 10 Why? Because they had been flung into recessionary conditions and they still had elevated home equity they could utilize to try to make it through it. Then, after 2007, in almost every state, the lines moved sharply to the left. Mortgages outstanding started to decline while home prices collapsed sharply (and they collapsed in a regressive pattern in neighborhoods with lower incomes and lower credit scores). That happened because what had changed in the mortgage market was that millions of households with 700-760 credit scores suddenly couldn't get prime or near prime mortgages to buy homes they had been buying for decades. So, the effect was sharpest on home prices. But, also, it sapped trillions of dollars of liquid assets out of the economy. There was the direct loss of liquidity from the collapse in mortgage funded cash payments to home sellers. But look at Figures 9 & 10\. The _collapse_ in mortgage originations in 2008 was, by a long shot, the most significant cause of rising leverage. Growth of mortgages outstanding started to slow in 2006 and 2007, with some lag, after home prices peaked. It had completely disappeared by 2008. And, the reason mortgages outstanding continued to fall in the aggregate, at a rate that was crisis-inducing, was because prime borrowers with scores under 760 were paying off mortgages on schedule, and families without mortgages were prevented from buying homes with new ones. The mortgage crackdown caused the Great Recession and the 2008 financial crisis. It was the worst thing to do well after doing anything better than the worst thing had been well overdue. Upgrade to paid The Lucas Critique (if I understand it correctly) says that if, for example, the government has a policy targeting a harvest of 2 billion bushels of wheat each year, then research on what causes wheat harvests to rise and fall will give specious results. Where we might normally study a variable and "hold all else equal", the policy itself is holding the policy target equal, and all those other variables are now changing in service of that. I think the Financial Crisis Inquiry Report might be the most interesting example of that problem. We can't determine what caused the 2008 financial crisis because, explicitly or implicitly, the crisis was policy. It just so happens that we hit our policy target by suddenly and permanently removing $1 trillion in annual mortgages from families who had had access to it for decades before. So that the scale of any anti-recessionary fiscal or monetary policy that could have maintained economic stability made stability controversial. The crisis is treated as if it was inevitable. It was inevitable once the mortgage policy that would create it was in place. The debates about federal debt, bailouts, Federal Reserve balance sheets, etc. are seen as the aftermath of the inevitable crisis, but really they were a necessary acting out of the cause of it. That was just the inevitable drama playing out - the acting out of reasonable frictions against the scale of response necessary to reverse the liquidity event we had stumbled upon and implicitly chosen to communally ignore and maintain so that it would create the crisis that would make us all so satisfied later as we watched The Big Short and Margin Call. As noted in the previous post, high profile economists like John Cochrane were worried that there were too many homes in Nevada in November 2008 and looked forward to how a recession might fix that problem. Other high profile economists disagreed about the need for a recession but agreed that there were too many homes in Nevada. The economist who had developed the very useful model that said trends in home construction were a leading recessionary indicator, and that recessions could be avoided by stimulating declining home construction before the recession arrived, was still saying there were too many construction workers in 2015! Maybe I am failing the Lucas Critique when I say that the mortgage crackdown caused the financial crisis and the Great Recession. If it hadn't caused it, we would have found something else that would have. And not a single word in the Financial Crisis Inquiry Report would be different if we had. You're currently a free subscriber to Erdmann Housing Tracker. For the full experience, upgrade your subscription. Upgrade to paid --- | | | Like --- | | Comment --- | | Restack --- (C) 2026 Kevin Erdmann 548 Market Street PMB 72296, San Francisco, CA 94104 Unsubscribe