A recent report from Fannie Mae finds that baby boomers are not leaving their comfortable suburban homes for lively inner-city communities with walkable streets. As a news article about the report observes, this challenges the “conventional wisdom that ‘empty nester’ baby boomers would eventually downsize from the homes where they raised families, flocking instead to apartments or condos.”

Rather than conventional wisdom, it would be more accurate to say that this notion was wishful thinking among urban planners who believe more Americans should be packed into high-density “compact cities” where they will get around by foot, bicycle, or transit rather than by automobile. In contrast, demographers have known that populations of virtually all age groups, whether millennials or empty nesters, are growing faster in the suburbs and exurbs than in the cities. After all, the baby boomers’ parents overwhelmingly preferred to “age in place” rather than move when their children left home; why should baby boomers be any different?

Despite this, regional planning agencies all over the country are writing plans that presume America will need no more single-family homes, especially on large lots, and instead will need lots of apartments, condos, or townhouses. Many of these plans effectively zone away the possibility of new single-family homes on large lots while they subsidize construction of high-density housing. For example, the San Francisco Metropolitan Transportation Commission’s Plan Bay Area mandates that 80 percent of all new housing be in high-density urban centers.

To justify these plans, the planning agencies often hire Arthur C. Nelson, the University of Utah urban planning professor who in 2006 predicted that the U.S. will soon have 22 million surplus single-family homes on large lots. Nelson wrote a 2011 report predicting that the Bay Area, which has one of the most acute housing shortages in America today, would have a surplus of nearly 572,000 single-family homes by 2040; Plan Bay Area relied heavily on this report to justify its strict land-use policies.

More recently, the Twin Cities Metropolitan Council hired Nelson to do a similar analysis for its Thrive 2040 plan. “Demand for attached and multifamily housing in the Twin Cities will continue to grow,” trumpets the council’s press release about Nelson’s report on Twin Cities housing. That, of course, is what the Met Council wanted Nelson to “prove,” which is why they hired him. However, his report can’t really justify the Met Council’s plans.

Nelson’s report predicts “a shifting mix of housing products demand for the next 30 years” such that the share of single-family homes on medium and large lots will decline from 37 percent to 26 percent; while homes on small lots will increase from 25 to 33 percent; and townhouses and multifamily will increase from 38 to 41 percent. Based on this, he says, “to meet housing demand by type in 2040 all new residential units will need to be attached options (apartment, townhouse, condominium) or small-lot detached homes.”

That sounds dramatic at first glance. But even if you believe his numbers, most of the change is from medium and large lot to small lot, not from single-family to multifamily, as the Met Council’s press release implies. Multifamily growing from 38 percent to 41 percent is not that big of an increase, and one that could easily be attributed to measurement error.

There are a lot of potential sources of measurement error in Nelson’s report. Most of the report is based on his interpretation of realtor surveys of people’s housing preferences in which lots of people said they wanted to live in “walkable neighborhoods.” Of course, if you ask people, “Would you like to live in a neighborhood where you can walk to shops?” a lot of people will say yes. But if you ask, “Would you prefer spending $400,000 on a 1,000-square-foot condo in a congested, noisy neighborhood or $200,000 on a 2,000-square-foot home on a large lot in a quiet suburb?” few people would pick the condo.

Nelson thinks housing demand is changing because of “sweeping demographic changes” including an aging population, increasing numbers of ethnic minorities, and declining numbers of households with children. Yet, as shown by other studies, his assumption that these groups will necessarily prefer apartments, townhouses, or houses on small lots is not well grounded.

The report’s biggest problem is that Nelson repeatedly uses the word “demand” but apparently does not know what this word means. Demand is not a point, like the 619,000 Twin Cities households that he predicts will want to live in attached homes in 2040. Demand is a relationship between price and quantity, and prices never enter into Nelson’s analysis.

One reason why Nelson may not understand demand is that he seems to be arithmetically challenged in the first place. Page 26 of the report admits that “in the near term, 2020, demand for more homes on larger lots may still seem robust. The overall demand for such lots will increase by about 25,000 between 2010 and 2030—nearly 1,000 units annually.” Whatever you make of this “demand analysis,” 25,000 divided by 20 years is 1,250, not “nearly 1,000.”

A second problem is that the big change that Nelson predicts–a decline in the share of homes on medium and large lots from 37 percent to 26 percent–isn’t carefully measured by most of the surveys Nelson cites. Page 28 of the report notes that, in the surveys that do distinguish between lot size, a “small lot” is a quarter acre or less. Yet when many urban planners talk about small lots in walkable neighborhoods, they typically mean 25’x50’ lots, nearly nine of which would fit on a quarter acre. Do the people who answer vague questions about their housing preferences really understand this difference?

A third problem is that a lot of the data cited in the report have one source: Arthur C. Nelson. The report includes ten figures and thirteen tables, five of each of which say, “Source: Arthur C. Nelson.” The citations don’t even say, “Arthur C. Nelson, [year],” which would allow readers to pick out which of the ten papers in the reference section by Arthur C. Nelson the tables or figures are from. This makes the report even less persuasive than it already is.

Perhaps the most important self-citation in Nelson’s reference section is a 2006 article in the Journal of the American Planning Association where Nelson first predicted the future surplus of single-family homes. The article issued a clarion call to planners to lead the way to prevent this by forcing builders to focus on apartments instead. The Journal was honest enough to attach a critique by University of North Carolina planning professor Emil Malizia that pointed out that Nelson’s predictions were based on unreliable surveys whose results could have been “heavily influenced by the data collection method.”

In sum, Nelson predict fairly small changes in housing preferences, especially between multifamily and single-family, and those predicted changes are based on specious data. Yet based on those predictions, the Metropolitan Council wants to make large changes in the Twin Cities’ housing mix, mainly a large increase in multifamily along its various rail lines. (Did I mention that the Met Council also wants to increase taxes so it can build more rail transit–which in 2012 carried all of 0.3 percent of Twin Cities commuters to work–thus providing rail stations near which it can locate high-density housing?)

The real question is: Just why should regional planning agencies such as the Bay Area Metropoltan Transportation Commission or Twin Cities Metropolitan Council, have anything to do with determining future housing supplies anyway? These agencies were created to hand out federal transportation and low-income housing grants to cities in each region, not to dictate housing choices to a region’s middle-class residents. As Malizia pointed out in his critique of Nelson’s 2006 paper, if people’s preferences change, and housing is left to the market, the market will respond to those changes.

Not satisfied with that, regional planners want to dictate future housing choices by restricting low-density housing and subsidizing high-density housing near rail stations. These policies will make housing less affordable which (perhaps deliberately) will make Nelson’s prediction of an increasing desire for multifamily housing a self-fulfilling prophecy.