Was Chicago still building “too much” in 2018?

I’ve put up three previous posts¹ in which I pointed out that, given Chicago’s continued population losses, there was an enormous amount of residential building in the Chicago urban area, or at least an enormous amount of residential-building permit-filing.

This post is intended to be a sequel.

Data for 2018 (the latest available) suggest that there have been two major changes from the immediately preceding years.

[1] Building-permit filing in the Chicago Metropolitan Statistical Area (MSA) declined substantially. 17681 permits—valued at $3,516,676,000—were filed in 2018. The comparable figures for 2017 were 22132 and $4,186,156,000. In other words, the number of permits filed dropped 25%, and their value dropped 19%. There may have been a further drop in 2019, but data will not be released for some time. There’s still plenty of building going on, and there are plans to build still more, but there has definitely been a decline in the amount of planned building activity.²

[2] Chicago was not the only major metropolitan statistical area (MSA) to lose population between 2017 and 2018. It was joined by the only two MSAs that are larger, New York and Los Angeles. While individual cities in the United States have often lost population in the decades since World War II, MSAs have only rarely done so, and those that have (Pittsburgh, for example) have mostly been in regions of the country that were suffering major economic distress. For the three largest metropolitan areas in the country to lose population is quite unprecedented. It needs to be added that the losses in all three cases were small, and, of course, the Census Bureau estimates could be wrong. There is also the issue that all three MSAs are somewhat underbounded, and that the larger combined statistical area (CSA) of New York lost a smaller percentage of its population than its MSA, while the Los Angeles CSA gained population. (The Chicago CSA and MSA lost nearly the same percentage, which rounds to .2 percent.)

Here are two new graphs that show the same data for 2018 that I analyzed in earlier posts.³ As was the case with the earlier data, the figures are for new, privately-owned housing units only.

The first graph shows the relationship between residential building permits issued in 2018 and estimated change in population from 2017 to 2018 for American metropolitan statistical areas.

The second graph shows the relationship between the valuation (in thousands of dollars) of these 2018 residential building permits and (as in the earlier graph) estimated change in population from 2017 to 2018 for American metropolitan statistical areas.

At first glance, the results seem comparable to those for the preceding three years. There is, in general, a very high, and extremely significant (p = .0000) correlation between the number of building permits issued and the change in population (.816, r-squared = .666) as well as between the value of these building permits and the change in population (.833, r-squared = .693) (both correlations, however, are lower than in the preceding year, when they were .906 and .909 respectively). As in past years, a few urban areas are outliers, and Chicago once again seems to be building a great deal more than its population loss suggests it should be.

Note, however, that Chicago is joined by New York and Los Angeles in apparently building more than one might have expected. There’s a simple explanation for this. Large cities, whether their populations are growing or shrinking, are likely to build more than smaller ones, if only because old buildings sometimes get replaced even in cities losing population. And, in cities with a large class of prosperous people wishing to live in relatively new central-city housing, which is the case in all three of these urban areas, there can be a huge amount of new construction.

Thus, just as last year, I set up regressions in which housing permits and housing-permit valuations were dependent variables, and population change (2017-2018 this year) and population (2018) were independent variables. Both independent variables turned out to be highly significant, with population change more significant than population. Still, adding the latter to the equation increased the correlation considerably. For the two independent variables and permits, correlation = .965 and r-squared = .931. For the two independent variables and permit valuations, correlation = .971 and r-squared = .943.

Here are two graphs that show these relationships.

The first graph shows the relationship between the actual and predicted number of 2018 permit filings on the basis of population change from 2017 to 2018 and population in 2018.

The second graph shows the relationship between the actual and predicted value of 2018 permit filings on the basis of population change from 2017 to 2018 and population in 2018.

Note how close the points are to the regression line. This is of course just what would except when the independent variables come close to predicting the dependent variable.

There are still some slightly anomalous cases.

The five MSAs where the number of permits filed was more than expected by the largest amount were Riverside-San Bernardino-Ontario, Miami-Fort Lauderdale-West Palm Beach, Portland-Vancouver-Hillsboro, Phoenix-Mesa-Scottsdale, and Las Vegas-Henderson-Paradise. The largest positive residuals for valuations occurred in San Antonio-New Braunfels, Philadelphia-Camden-Wilmington, Riverside-San Bernardino-Ontario, Las Vegas-Henderson-Paradise, and Miami-Fort Lauderdale-West Palm Beach.

The MSAs where the number of permits filed was less than expected by the largest amount were Houston-The Woodlands-Sugar Land, Austin-Round Rock, Charlotte-Concord-Gastonia, Dallas-Fort Worth-Arlington, and New York-Newark-Jersey City. The largest negative residuals for valuations occurred in Dallas-Fort Worth-Arlington, Los Angeles-Long Beach-Anaheim, Charlotte-Concord-Gastonia, Nashville-Davidson-Murfreesboro-Franklin, and Santa Rosa (California).

Note that most of these MSAs are in the Sunbelt, where growth in the medium term has generally been substantial, but where growth has often been subject to booms and busts and where (in some cases) it was interrupted quite severely by the Great Depression. It’s also the case that the presence of these MSAs on these lists is in part a function of the fact that so much building has been taking place in them and that modest changes in percentage therefore end up being substantial in absolute terms. That could be one of the explanations for the appearance of Philadelphia and New York.

In Chicago in 2018, the number of permits filed and their value were, unlike in any of the previous years for which I have data, pretty close to the predicted value. For the moment Chicago could not be said to be building “too much.”