Erdmann Housing Tracker · Housing & Cities
TIER 4 Mon, 25 Aug 2025 14:10:21 +0000
A recent paper quantified the effect of changing fertility on housing demand (via Arpit Gupta). ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ | | ---|---|--- | | | Forwarded this email? Subscribe here for more --- --- # Fertility and Home Prices | | Kevin Erdmann --- | Aug 25 --- | --- --- | | | --- | | --- | | --- | | --- | | READ IN APP --- A recent paper quantified the effect of changing fertility on housing demand (via Arpit Gupta). I thought it might be interesting to see how their findings might affect expectations in the American housing market. The abstract: > How do demographic changes affect asset prices? Based on centuries of data, this paper shows that demographic changes are a key and predictable determinant of house prices. A one percentage point increase in the current birth rate increases house prices about 25-30 years later by 4 to 5%. Using individual data on housing decisions, transactions and portfolios we argue this is driven by strongly age-dependent demand for home-ownership and the limited response of investors to such shocks. Correspondingly, we find large impacts of past birth rates on rent-price ratios, but negligible impacts on rent prices, bond yields, and dividend yields. Upgrade to paid I think maybe what it says about past prices is more interesting than what it says about future prices. I haven't found a non-paywalled link to the paper, so this is very much a "poor man's" analysis of the numbers. I've seen the abstract, some comments by the authors, and some charts. The main claim, which I will clumsily play around with below, is that a 1 percentage point rise in fertility (babies per capita x 1,000) is associated with a rise in home prices of about 5%, thirty years later, when those babies are forming their own families and buying homes. They find that fertility doesn't create an anomaly in housing demand 30 years later so much as an anomaly in first-time purchases. People tend to want to find a long-term home and settle down when they have children, so demand for purchasing homes rises anomalously, and price/rent ratios on single-family homes temporarily rise (if I understand the authors correctly). #### Past Prices and Fertility Of course, in the US, fertility has been and is generally declining. The main drop in fertility was from about 1960 to 1970, which means that American home prices should have been deeply depressed in the 1990s. And they were! | | ---|---|--- I think this is actually useful to think about. I have generally found the 1990s to be a bit of an anomaly. Prices _were_ too low, considering everything else. And, when I create benchmarks to analyze changing home prices over time, the 1990s tend to be below neutral. I feel like that's cheating a bit, when the conclusions I have reached about housing markets in the 2000s tend to push back against the "bubble" narratives. Pushing 90s prices below neutral also pushes the peak price levels of the "bubble" down so they aren't as high above neutral. In Figure 2, I have used their base claim to make a very broad estimate of the effect on relative home prices. Here, I compare the fertility rate from 30 years ago to the fertility rate from 40 years ago to estimate the relative shift in valuations that a market might be experiencing from unusual shifts in fertility. By this broad measure, home prices in the 1990s and early 2000s may have been about 30% lower than they otherwise would have been with stable fertility. | | ---|---|--- Figure 3 uses the broad estimate from Figure 2 to estimate the "fertility neutral" average price/rent ratio. Again, don't look at this to 3 decimal points. I have not tried to make this precise. But, surely, they are correct that in the 1990s, home prices were lower than they otherwise would have been because of the mid-20th century drop in fertility. The black line may be a more accurate estimate of the effect of worsening supply conditions in the 1990s and the credit boom of the 2000s than measured prices at the time were. | | ---|---|--- According to my broad adjustment, maybe the bubble in the 2000s was much _larger_ than we give it credit for. Maybe the mid-20th century drop in fertility was holding prices in the 2000s much lower than they otherwise would have been. Of course, price changes in the 2000s were highly regional, and were strongly associated with a lack of regional supply or a sudden spike in regional demand. If mid-20th century fertility hadn't dropped so sharply, would the migration out of the house-poor Closed Access cities have been even stronger? Would the inflows into the Contagion cities in Florida, Arizona, etc. have been even stronger? This also seems like it could be background color for some of the other findings in the Mercatus paper where I reviewed the factors associated with changing prices in the 2000s. I found that there were places where my methods attributed as much as 15% to 20% price appreciation in some ZIP codes to new credit access. But where that was the case, prices would generally have been soft without it. The credit effect was keeping prices stable, not driving them higher. Sensitivity to credit access was associated with price appreciation, but the other factors were pushing prices in the other direction, especially in the ZIP codes where prices were sensitive to credit access. So, ironically, ZIP codes with more credit-fueled price appreciation, on average, had less price appreciation than other ZIP codes that weren't sensitive to credit access. I think that is an interesting and useful finding. But, it does leave a nagging caveat. Were there really a lot of ZIP codes around the country that might have experienced relative price deflation at the time of 10% or more without the credit boom? This fertility data says, "Yes." So, maybe in the parts of the country that weren't dominated by the upheavals of Closed Access home prices and migration flows, the credit boom was smoothing out home prices that would otherwise have been depressed by these fertility effects. Anyway, it seems like an interesting nuance to consider. And it is a plausible reason why prices, in general, in the 1990s were low, and why prices in the ZIP codes where credit access might have been inflationary in the 2000s remained stable even with that stimulus. #### Future Prices and Fertility As far as the effect of fertility on future American home prices, the changes since 1970 have been of a smaller scale. There was a bit of a rise and fall around 1990 that might have a small depressive effect on prices over the next few years. And, the decline in fertility after the Great Recession might, again, have a small depressive effect on prices in the 2040s. But, I doubt that will be particularly important. Total units required as we move through decades will be lower as a result of lower fertility. And, lower immigration might even push it further down. Spending on housing (when supply isn't unreasonably tight) tends to be highly sensitive to incomes. In other words, if there are two countries that each have total GDP of $10 trillion, but one of those countries has 100 million residents and the other has 200 million residents, they will both probably spend similar amounts on housing. Total spending on housing has more to do with income than population. Households make huge changes in the real characteristics of their homes to keep it that way. Lower population growth will lower GDP growth, all else equal. So, it all gets accounted for. But, I find that thinking of future residential investment that way - as mostly a product of GDP growth - simplifies the question for me. But, this paper does suggest that after accounting for those changes, there might be periods where home prices will be temporarily moderately depressed by dips in the rate of household formation. But most of the decline in fertility is well behind us, so the sorts of temporary fluctuations in valuations proposed by this paper should be relatively small in the future. Upgrade to paid #### More Homeownership and Fertility A couple more notes on the 1990s and 2000s: This paper seems to propose market frictions having to do with temporarily rising and falling homeownership as a substitute for renting, in which the investor market doesn't adjust immediately to changing yields from new homeowner demand, and so price/rent ratios fluctuate temporarily. First, another observation we might consider about the 1994-2004 period is that homeownership was rising sharply, even though the fertility effect was negative. As Figure 4 shows, 25-34 year-olds were definitely declining as a percentage of the population. | | ---|---|--- And, Figure 5 does show that within younger age groups, homeownership rates were rising from 1994 to 2004. | | ---|---|--- But, as Figure 4 also shows, by the 2000s, relative changes in homeowner demand were rising because of the compositional growth of the older age groups with higher homeownership rates. So, the decline in current homeownership among young adults was partially offset by the unusual new trends created by the aging of the baby boomers. You can see the effect of those compositional changes in Figure 5 by comparing the US aggregate rate of homeowners per capita to the 35-44 year-old age group. Homeownership in that group was lower than it had been in the early 1980s while the US total was higher. Over the long term, compositional changes in age demographics have been responsible for more than all of the increase in the homeownership rate since the early 1980s. Figure 6 shows the number of homeowners in each age-group over time. General population growth plus increased access to mortgages roughly countered the downward effect of the mid-20th century fertility drop among 25-34 year-olds. | | ---|---|--- I'm getting deja vous from the early days when I first started discovering data that turned into _Shut Out_ . There are all these empirical trends that everyone ignores. Just looking at age groups, the rise in homeownership from 1994 to 2004 was very much driven by older buyers who tend to be less credit sensitive. By 2005 - 3 years before the crash and before the rise of the famous CDO markets that are the focus of The Big Short and other popular literature - the rise in the number of homeowners was among 45+ year-olds (and almost all of that was among 55+ year-olds) who tend to buy homes with a very large down payment or no mortgage at all. And, just to drive the point home that nearly all the net additional owner-occupied homes purchased after 2004 were purchased by families that were not affected by credit markets, the number of homes owned by that age group just kept moving up in a straight line as if the 2008 mortgage crackdown never happened, while the absolute number of homeowners in every other age group declined. The bust was not a reversal of the boom, submission number 4,256. #### Homeownership and the Bust Finally, I wonder if the methods of this paper could be used to estimate the effect of the mortgage crackdown on home prices after 2008. My thesis is that the shock to the market for home purchases was of such a large scale that the traditional small scale investors simply couldn't fill the gap and so home prices fell into disequilibrium after 2008. The authors here are basically arguing the same thing - the investor market in housing has frictions. It doesn't immediately respond to temporary changes in rent yields when changes are large enough. In effect, their thesis works through changes in homeownership demand in the same way that my thesis about the mortgage crackdown does. I wonder if one could use their methods and the decline in homeownership from 2007 to 2016 to estimate the scale of the mortgage shock on home prices. I have my own estimates from my own methods, that tend to come out to about 20%-25% of a price drop from the credit shock. I wonder what estimate coefficients derived from their model would produce for that period. Upgrade to paid You're currently a free subscriber to Erdmann Housing Tracker. For the full experience, upgrade your subscription. Upgrade to paid --- | | | Like --- | | Comment --- | | Restack --- (C) 2025 Kevin Erdmann 548 Market Street PMB 72296, San Francisco, CA 94104 Unsubscribe