Construction workers on the site in Rosslyn, Arlington, VA on Feb.22, 2012 . More than 216,000 people now live in Arlington, and the county expects another 36,000 to move to the inner suburb by 2040. (Jeffrey MacMillan)

All across Arlington County, cranes dot the skyline. Foundation pits pockmark the landscape, and construction trucks hold up traffic as they back into work sites along Wilson Boulevard and Columbia Pike.

County planners calculate that 1,380 new housing units were under construction in the third quarter of 2012. Five developers are awaiting county approval to build about 1,600 housing units, primarily rental apartments. More than 216,000 people live in Arlington, and the county expects an additional 36,000 to move to the inner suburb by 2040.

But a study by the University of Virginia’s Weldon Cooper Center for Public Service for the state employment commission says that’s not going to happen. Arlington, it said, will actually have almost 20,000 fewer residents a generation from now.

Arlington officials plan to meet with the university demographers to discuss why the county says the significantly lower population forecast is incorrect. A projection of fewer people could harm the county when the state apportions funds for everything from road projects to schools.

“Any forecast that says this county is going to lose population — we know that’s wrong,” said County Board member Chris Zimmerman (D) during a meeting with the local planning commission last week. “Everybody knows that’s wrong. . . . It’s off by orders of magnitude.”

(The Washington Post/Sources: Arlington County, University of Virginia)

“We feel Arlington is going to continue to grow because of the way we develop,” Elizabeth Rodgers, an Arlington County urban planner and demographer, said. “Our high-density areas are continuing to attract young residents, and we predict Arlington will grow about 22 percent by 2040.”

The dramatic difference in forecasts stems from how each organization calculates growth. The Weldon Cooper Center, which is in the first year of a contract with the Virginia Employment Commission, looked at the last 60 years of U.S. Census data and sought a single method to assess all 134 counties, towns and planning districts in the commonwealth. It sought the most reliable model statewide but acknowledged that forecasts 30 years away at the county level could err by more than 20 percent. “I do think we want to be clear that projections are in no way prophecies,” said Rebecca Tippett, the lead researcher who worked on the study. “Population is very dynamic and very responsive to intervention.”

What prompted the prediction that Arlington’s population would fall seems to be the loss of 20,000 residents in the 1970s, when Metrorail was under construction and urban areas around the country were losing residents to the far suburbs. That’s also the period when Arlington decided to focus high-density growth along major boulevards, such as the Rosslyn-Ballston corridor, the Jefferson Davis (Route 1) corridor and Columbia Pike. That focus preserved the single-family neighborhoods elsewhere and fueled the redevelopment of Clarendon, for example, where thousands of young, relatively wealthy residents walk to Metro, restaurants, shopping and nightclubs. Under the university’s statistical model, that 1970s population dip cost Arlington even though growth has been steady since the mid-1980s.

Arlington’s planners look at block-by-block surveys, examination of proposed development plans, existing vacancy rates and average household sizes to determine growth.

“We build from the base up,” Rodgers said. Developments do not always get built, and some complexes have been known to remain unfilled, she acknowledged, but “when we do these estimates, we err on the conservative side.”

It’s not just Arlington that has a problem with the Weldon Cooper Center’s forecast. Hampton Roads officials have also raised objections, Tippett said.

The forecast for Arlington has already been submitted to the employment commission, so it won’t be altered, but Tippett said the center will try to advise policy makers to take other factors into consideration when they look at its work.

“I wouldn’t feel comfortable saying that Arlington is going to lose population, given that its [previous] population decline is not recent,” Tippett said. “If we had based it in the past 30 years of data it might have been different.”

Scott Clement, a survey research analyst with Capital Insight, contributed to this report.