This piece originally appeared in the Wall Street Journal’s Think Tank blog.

The case against the Federal Reserve raising short-term interest rates at the end of the Federal Open Market Committee meetings Thursday is so clearly strong that is should carry the day. The point of raising rates is to rein in an overheating economy that is threatening to push inflation outside the Fed’s comfort zone. But inflation has been running below the Fed’s target for years—and its recent moves have been down, not up.

This subdued price inflation is not a puzzle; it’s the outcome of a labor market that remains so slack that nominal wage growth is running about half as fast as a healthy recovery would be churning out. And this slack is pretty easy to see so long as one is willing to look past the (welcome) progress in reducing the headline unemployment rate. The employment-to-population ratio of prime-age adults (25 to 54 years old) has recovered less than half of its decline during the Great Recession. Worse, progress in boosting this measure has stalled for all of 2015.

Many have asked: Would a 0.25 percent increase really do all that much harm? This is the wrong question. The literal, narrow-minded answer is: No, it wouldn’t do much harm. But the data above show that the Fed should not be tightening at all. A 0.25 percent increase is a small move in the wrong direction—but it’s still the wrong way to go.

More important is the bigger picture behind the decision of whether to raise rates. A rate increase in September is a clear signal that the Fed believes the U.S. economy is getting close to its binding floor on unemployment. In this view, attempting to push down the unemployment rate further by boosting aggregate demand (the goal of low-interest-rate policy) will result only in workers demanding wage increases that the economy isn’t productive enough to deliver. This, in turn, will lead to an upward wage-price spiral. Foregoing a rate increase, on the other hand, signals that the Fed is willing to continue boosting demand with low interest rates and to probe how low unemployment can get before sparking actual (not just hypothetical) increases in inflation. This would essentially follow the model of the late 1990s, when the Fed allowed a strong recovery to run its course, refusing to slow growth just because unemployment fell so long as inflation remained tame.

These two paths could lead to very different economic outcomes in 12 to 18 months. If the Fed pulls back demand in the short run this could leave unemployment significantly higher 12 to 18 months from now than if it had continued to let unemployment fall until an acceleration of inflation actually appears in the data.

The path that involves higher unemployment 12 to 18 months from now is one where wage growth for the vast majority of American workers is highly unlikely to have bucked a three-decade trend of lagging far behind economy-wide productivity growth. That path that allows the lower limits of unemployment to be plumbed, on the other hand, could replicate another key development of the late 1990s: strong across-the-board wage growth for American workers.

All in all, this week’s Fed decision is not just a referendum on what a 0.25 percent increase in short-term rates would do to the economy in the next month or two. It’s a referendum on how important it is for the Fed to test the limits of low unemployment, and not pull back just because models or theory or some need to show vaguely defined “independence” suggests doing so. The stakes are high over the Fed’s willingness to test the limits of genuine full employment.