Forecasting economic activity is generally a thankless task — unless you’re using demographics as a modeling foundation, which provides a surprisingly accurate means for looking ahead. That’s good news for analysts trying to develop robust estimates of GDP growth over a medium-to-long-term horizon. But it’s also bad news if you’re expecting economic activity to accelerate on a sustainable basis.

Many studies over the years have documented the relationship between demographics and economic activity. A 2016 Federal Reserve paper, for instance, connects the dots and finds that slower population growth and real GDP growth are correlated (“Understanding the New Normal: The Role of Demographics”). Some researchers also find evidence that demographic trends influence trends in financial markets (“Demographic Changes, Financial Markets, and the Economy”).

Ruchir Sharma, chief global strategist and head of emerging markets at Morgan Stanley Investment Management, summarizes the demographic-economic connection in “The Boom Was A Blip” in the May/June 2017 issue of Foreign Affairs. “The UN now predicts that worldwide, population growth rates will continue to decline through 2025 and beyond,” writes the author of The Rise and Fall of Nations: Forces of Change in the Post-Crisis World.

Such long-term forecasts, which are based on a relatively simple combination of birth and death rates, have an excellent track record. And the economic implications of that trend are clear: every percentage point decline in working-age population growth shaves an equally large chunk off the GDP growth rate.

Let’s kick the tires on this concept by comparing US GDP growth with the country’s working-age population. We’ll use data via the St. Louis Fed’s FRED database (tickers GDPC96 and LFWA64TTUSQ647N). The available demographic numbers only dates to 1960, which is too short to draw final conclusions. Nonetheless, it’s a reasonable way to begin, if only to get a sense of how real-world numbers stack up.

To filter out the short-term noise, 20-year annualized changes are used, which translates into a data set that runs from 1980 to 2017. Plugging the figures into a simple regression model produces the chart below.

It’s hardly a perfect fit – the adjusted R-squared is 0.36. But the analysis reveals a statistically robust relationship (P values, for instance, are close to zero, which suggests that the results aren’t random).

Given the data in this test, the model is still a blunt tool at best. Nonetheless, let’s ignore the caveats and use the regression results to predict GDP growth, based on plugging in the current 20-year annualized change in the working-age population (+0.9%). The resulting economic outlook: real annualized GDP growth of 2.7%, based on the point forecast.

That’s a slow pace compared with history. Sticking to the data in the model above, the high point for trailing 20-year GDP growth was in the early 1980s, when output was nearly 3.9% at the peak. There’s been a slow but steady decline in the 20-year trend ever since, with the latest 20-year GDP change slipping to 2.2% through this year’s first quarter. The latest working-age population change is a weak 0.9% increase, the lowest since 1980.

Also, our simple model, supported by a deep pool of economic research, suggests a low probability that growth will ramp up anytime soon. Exhibit A is the downshift in population growth. As Sharma advises, citing a longer data set, “In the United States, growth in the working-age population declined from 1.2% in the early years of this century to just 0.3 percent in 2016 – the lowest rate since the UN began recording this statistic in 1951.”

The future is still uncertain, of course. But to the extent that demography dictates economic growth, the case appears weak for predicting a significant acceleration in the macro trend. That doesn’t preclude an encouraging pop in growth for a quarter or two every so often. But assuming that output will rise on a sustainable basis for the longer run is probably assuming too much.