On Friday, when we get the latest jobs numbers from the Bureau of Labor Statistics, I’ll have my eye on a few measures of slack in the labor market. As the economy continues to add more jobs than are needed to simply absorb working-age population growth, I expect to see increases in both the overall and the prime-age labor force participation rates. This in turn should lead to a continuation of the slow but steady rise in the employment-to-population ratio, a metric that has been rising but which remains far below full employment levels.

Besides these key quantity measures of labor market slack, I’m also monitoring key price measures of slack—particularly nominal wage growth—which the Federal Reserve will also be looking at when it meets later this month. The year-over-year measures of nominal wage growth we get with each month’s employment report are the most timely measures of labor market slack, and while there has been some slight uptick recently, wage growth is still far below what we should be targeting for a truly healthy economy.

Year-over-year private-sector nominal wage growth was 2.5 percent in April 2016. After maintaining relatively slow growth, in the 2.0 to 2.2 percent range for several years, an increase of 2.5 percent is a welcome improvement. However, given the Fed’s target for price inflation of 2.0 percent, long-term trend productivity growth, the length of slow growth, and the failure of labor’s share of corporate income to rebound, the labor market should produce wage growth in excess of 3.5 percent for a consistent period before concerns over wage led inflationary pressures lead the Fed to increase rates. Tightening before a period of strong wage growth will lead to price and wage targets hardening into ceilings, and that will cause grave damage to the economy. Such premature tightening will be avoided if the Fed’s decisions truly are data driven.

Throughout the recovery, any time some particular measure of wage growth seemed to be indicating strong growth, it would attract attention—particularly from those arguing that rate hikes should come sooner rather than later. The latest such measure is an alternative measure of wage growth produced by the Federal Reserve Board of Atlanta. In recent months, their wage tracker has edged up towards 3.5 percent, hitting 3.4 percent in April 2016.

However, I can’t stress enough that 3.5 percent is not the appropriate wage target for their data series. There is nothing incorrect about the Atlanta Fed’s individual wage tracker; it’s just that the relationship between their series and the Fed’s 2.0 percent price inflation target is different. To put it simply, the Atlanta measure should show growth appreciably above 3.5 percent to be compatible with the Fed’s price inflation target.

First, let me explain the difference between the two wage series. The Atlanta Fed’s individual wage tracker does just that—it tracks individuals’ wages from one year to the next. To be tracked in both periods, an individual must be in the survey with valid wages reported in both periods. An individual has to be at the same address in both years to be in the survey both years. There’s clear evidence that lower income people move more than higher income people, which has the potential to create upward bias in the wage series. To have valid wages in both periods, an individual needs to be employed in both periods. Unemployment is higher among those with fewer credentials (highly correlated with wages) thereby, again, creating a sample that is biased upwards.

Proponents of this series point out that restricting the sample to only those individuals who are employed in both years removes any compositional shifts—excluding those who may enter or leave employment. But, there’s no evidence that composition matters much.1 And the price of holding composition constant is that the Atlanta Fed’s measure only looks at workers with increasing amount of experience.

By definition, the wages in the Atlanta Fed’s series are buoyed the fact that they’re reflective of an extra year of experience (or even tenure in the same position). Additional experience (and even more so tenure) in and of itself is likely to cause substantial wage growth from one year to the next. Using State of Working America data for 2007, one more year of experience can mean an additional increase in measured hourly wage growth that ranges from 0.6 to 2.4 percent, depending on gender and age. In an admittedly rough calculation, taking the mid-point of 1.5 can more than fully explain the difference between the latest wage differential in the individual versus private sector year-over-year wage changes (3.4 versus 2.5 percent). That’s a huge correction factor on the growth rate of the private-sector wage series. As you can see in the figure below, the individual series generally runs about a half a percentage point faster than the overall series. The private-sector series only began in 2006, so I track back using the production/nonsupervisory wage series (which has followed the same trend as overall wages over the entire common time-period under discussion).

When the Fed is looking for signs of wage led inflation, they should be interested in economy-wide wage growth, not the wage growth of any particular subset of workers. Unlike the Atlanta Fed’s individual wage growth measure, year-over-year changes in the private-sector wage series looks across the entire economy. And, when using that appropriate measure, the evidence is clear. The Fed should continue to hold off on raising rates until the wages are consistently above 3.5 percent, a strong indication that the economy has finally reached full employment.

1. Composition adjusted measures such as the Employment Cost Index are also showing relatively slow growth over the year.