Filtering is an important process in housing. Homes should filter down.Filtering means that the typical income of new tenants in a given home differs from the previous tenants. When homes are filtering down, a neighborhood might be built for doctors and lawyers. Fifty years later, those homes will be sold to accountants and managers. Another fifty years later, they might be sold to teachers and administrators. Another fifty years later, they might be sold to receptionists and clerks. Of course, at any given moment, change is moving in a hundred different directions across a hundred different neighborhoods. But, where new housing is available, on average, downward filtering applies in the long run in the average neighborhood.In the US, homes haven’t been filtering down. Since 2008, they mostly have filtered up. I’m not sure we have fully come to terms with what this means. Figure 7 from my new paper at Mercatus gets at the issue, I think. The x-axis is the average filtering rate in each metro area from 1973 to 2018, as estimated by Liyi Liu, Douglas A. McManus, and Elias Yannopoulos. The orange plots show the typical price/income ratio of homes in each metro area in 1999. In markets with negative filtering, price/income ratios tended to be between 2x and 3x. But, where filtering was positive, there was a positive correlation between filtering rates and home prices.Filtering isn’t just on a gradient. There is a tipping point here. What’s going on?In the 20th Century, filtering was generally downward. Home price/income ratios tended to be in that 3x range in every city. And, they tended to be near 3x in both the poorest and the richest neighborhoods in every city.Cities that were successful grew. Where the growth of cities led to density and moderately rising costs, families sorted across the housing within those cities such that costs matched incomes. Where necessary, families kept costs in line by living in smaller units. Costs might be higher, per square foot. And there might be more of a difference in the value of different locations within those cities. But, families tended to make decisions on location and size that kept total spending near that comfortable zone. In a few locations, like New York City, where locational amenities were so dense that “housing expenditures” also included better access to transportation, spending on housing was a bit higher, especially in neighborhoods with lower average incomes, but nowhere near as high as scarcity has pushed housing costs today.As I mentioned in Part 2 of this series, that is why the real Case-Shiller price index was so flat for so many decades. Families make all sorts of decisions about their housing consumption that tend to keep home values around 3x incomes.So, in the cities with downward filtering, the changes in housing all happened on the real side (size, amenities, and maintenance) so that regardless of how quickly they filtered down, they were worth about 3x their tenants’ incomes on average. Maybe in the typical neighborhood in one city a home was more likely to have had the kitchen renovated or a room added than in another city. But in either case, homes were constructed and maintained to remain generally 3-4x tenant incomes across space and time.Where housing decisions are dominated by the question “How much would we like our housing to improve?” the answer is always to improve it such that the costs of improving it equate to homes worth about 3x incomes (with some variation due to property tax rates, etc.).That changes when filtering turns positive. When filtering turns positive, the decisions families face are compromises, not aspirations. Aspirations are easy. Spend what we can afford. Compromises are hard.Upward filtering means that as you remain in an unchanging house in an unchanging neighborhood, slowly, the families that move in have higher incomes than the families that moved away. They aren’t generally paying more for better or larger homes. They are paying more for the land under the homes. In today’s context - and the only context that could push price/income ratios into the double digits, as they are in parts of some cities - they are “paying a bribe to the land for permission to buy the house”.So, as I mentioned in an earlier post, where families own their homes, neighbors note to each other that they wouldn’t be able to afford to buy the homes they all live in. Maybe you’ve had that conversation at a neighborhood Christmas party. Or, they complain that property taxes and homeowner’s insurance are too high.Where families are renters, they are faced with the rising cost of housing with each lease renewal. Each year, their neighbors have higher incomes and the neighborhood rent reflects that. Of course, commonly, they blame the newcomers.We don’t move at the drop of a hat, so the rent ratchets up year after year and eventually, the compromise needed to re-normalize spending starts to sting. Do you move your kids to a worse school? A neighborhood with more crime?These are not decisions made lightly and without stress. And, so, there is a tipping point. Where filtering is downward, decisions are aspirational and comfortable, and we increase spending on housing to the comfortable amount. Where filtering is upward, we spend whatever we are forced to spend to avoid uncomfortable compromises. And, for hundreds of thousands of families each year, eventually the uncomfortable compromise is regional displacement - moving away from long-standing connections to friends, family, jobs, public services, and infrastructure.The more homes filter up in a given city, the more of those decisions are forced, the more domestic out-migration accumulates. The population that remains are the families for whom displacement would be most painful - the families willing to pay more to avoid it.Where filtering is downward, the correlation between filtering and home prices is flat. Where homes filter up, the correlation is positive. Families pay more to avoid difficult compromises. Under positive filtering, “We are not as wealthy as we thought we were” because all that extra cost is just the toll to the troll under the bridge. A naive measure of household wealth under these conditions is like reading a fairy tale about a village with a troll under the bridge as a story of success, because the bridge is so valuable to the troll.The story of American cities before the 21st century is a story about making locations valuable by building proverbial bridges. Families moved to those valuable places, and when they did, they spent 3x their incomes to do it. The story of American cities this century is that we block the building of proverbial bridges and the trolls collect tolls on the bridges that were already there. Now families are moving away from those places because they can’t afford the tolls.Economists frequently are making 2 systematic errors about this change. First, they look at how expensive it has become to live in the cities where the trolls are, and they say, “Wow. If people are willing to spend that much to live there, those places must be superstars.” But, they were superstars when they didn’t have trolls. Austin is a superstar. New York City has trolls.As a nation, we Americans are not wealthier because 60-year-old, 1,500-square-foot homes in suburban Boston sell for a million dollars. We are wealthier than our housing stock. Our collectively high net worth is a result of our housing poverty.Second, they say, “It is wrong to claim that Americans are worse off over time. Bridges have become very valuable. If you add up all the wealth of our cities (including the wealth of the trolls) we are richer than ever.”The families that live in those cities, who have lived the experience of slowly having the advantages of the location be sucked out of their hands into the hands of the trolls, may not understand exactly what has happened, but they know that that claim about wealth and well being doesn’t seem accurate.Renters feel this palpably. But I think even homeowners have a sense of stress about it. Homeowners are the trolls. They might pocket capital gains when they sell their overvalued homes, but in the meantime, they live with an impending sense that they don’t belong in the home they used to belong in.Looking back at Figure 7 from the paper, the blue plot shows price/income ratios in 2022. It looks like price/income ratios have increased, even where filtering was negative. But, in the chart, I am using stable filtering rates, estimated over a 45 year period.What is really happening is that filtering rates are becoming positive everywhere. Think of the typical price/income ratio in a given city as the real-time measure of the current rate of filtering. The trendlines in 1999 are the natural correlation between filtering and home prices. If price/income in a given city has gone from 2.5x to 4x, the city has likely moved from downward filtering in 1999 to annual positive filtering of 0.4% today. Dots are moving up because they are moving to the right, into positive filtering territory.Figure 8 from the paper compares the typical price/income ratio over time in cities with lower than average filtering versus cities with higher than average filtering. From 1999 to 2005, there was some cyclical elevation, but more importantly, the upward filtering in cities where millions of families are forced to make compromises each year and where regional displacement is the compromise that hundreds of thousands must make, pushed home prices much higher in those cities. From 1999 to 2005, a cyclical boom combined with more upward filtering was associated with rising prices in general, and especially rising prices in the cities that weren’t growing. From 2005 to 2012, a large cyclical collapse was engineered that lowered home prices but that didn’t fix the filtering problem. From 2012 to today, upward filtering continued to push price/income ratios higher, but now it was pushing price/income ratios higher in almost every city because almost every city now has upward filtering.Families are paying higher rents under duress in an attempt to simply stay put.So, the difference between the cities with the lowest filtering and the cities with the highest filtering increased from a difference in home prices of 1x local incomes to a difference of more than 2x local incomes, and then stayed there as homes became more expensive everywhere. Homes became more expensive everywhere because everywhere stopped growing as much as they previously had grown.