The U.S. Federal Reserve’s Open Market Committee faces lots of crosswinds as it ponders whether to start raising short-term rates off of zero percent at its mid-September meeting. Turbulence in global financial markets alongside currency devaluations and plummeting commodity prices could mean the Fed will stay its hand, as University of Oregon economics professor Tim Duy points out on his Fed Watch blog. But as the central bank also considers “measures of labor market conditions,” how are developments going on that front?

When it comes to the labor market, the Federal Reserve is looking to see how close unemployment is to what the Fed calls its long-run rate, or the long-run rate consistent with full employment. Remember part of the Fed’s mandate is to promote “maximum employment.” This means the central bank needs to consider how low the unemployment rate can fall before the economy starts to overheat. High and accelerating inflation would be a sign of such overheating, but there are few signs of that happening.

Another way to determine the Fed’s long-run unemployment rate is to look at wage growth. Accelerating wage growth has yet to show up. Therefore the long-run unemployment rate is lower than the current 5.3 percent. The Fed projects the long-run unemployment rate is somewhere between 5.2 and 5.0 percent, which means the labor market is quite close to that mark. Yet the central bank has been moving down its estimate of the long-run unemployment rate since 2012 because employment gains haven’t turned into wage gains. The Congressional Budget Office has been doing the same, moving down their estimate to 5.0 percent for the 2017-2018 period in a recent report.

How low could unemployment get? In a recent interview, Columbia University economist Joseph Stiglitz said he thinks it could get below 4.5 percent, maybe even down to 3.0 percent.

Translating a 3 percent unemployment rate into another measure of labor market health, the employment-to-population ratio, may be helpful to see how feasible a target that could be. To make such a calculation, we have to make an assumption about the future path of the labor force participation rate, which is key in calculating the unemployment rate. A large chunk of the decline in that participation rate is due to the aging of the U.S. population. The President’s Council of Economic Advisers pegs its effect at about half of the decline from 2007 to 2014.

To make our calculation, then, let’s assume that the participation rate moves up to its current “potential” rate of 63.4 percent, according to the Congressional Budget Office. This 0.8 percentage point increase would grow the labor force by about 1.3 million workers. In addition, if we reduce the unemployment rate to 3 percent, this would mean 5.5 million more workers would be employed. (This calculation assumes that the increase in employment doesn’t affect the potential labor force participation rate.)

The end result would be an employment-to-population ratio of 61.5 percent. Currently, the ratio is 59.3 percent. 2.2 percentage points might not seem like that much, but consider that it took from January 2010 to July 2015 for the ratio to increase 0.8 percentage points. The increase of 2.2 percentage points to 61.5 percent is also more than twice as large as the 0.9 percentage point increase that would result from assuming we only hit an unemployment rate of 5%, as the Fed and CBO suggest our target should be.

Of course, some of the recent decline in the employment-to-population ratio is due to the workforce aging as the Baby Boom generation—now 51 to 69 years of age—enters retirement. But a 3 percent unemployment rate produces an employment-to-population ratio well below levels seen before the two recent recessions and the aging of the labor force.

This isn’t to say the U.S. unemployment rate can go as low at 3 percent. Wage growth might kick up and the Fed may raise interest rates before the economy could get to that level. Still, this exercise shows how much room there could be for the U.S. labor market to continue its recovery.