October 31, 2017
October 31, 2017
Employment Reports, Preliminary Data, and Forecaster Bias
by Josh Lehner
A few thoughts that help interpret the recent Oregon employment reports showing job losses. Keep in mind Oregon went through something similar in 2015 and, like then, our office is not worried today for one very important reason. (Hint: it’s wages)
First, Oregon’s economy has clearly slowed over the past year or two. No longer are we seeing full-throttle rates of growth. An economy digging out of a recession behaves differently than one approaching full employment. Oregon is transitioning down from peak growth rates to something more sustainable. At least until the next recession when we start the cycle anew.
Second, economic data gets revised. All the time. The very first release, when the data is preliminary, gets the most attention. However the conversation is just getting started even if no one sticks around to finish it.
Third, unfortunately, the economics profession has a clear culture of chasing the most recent data point. Economists know data is preliminary and will be revised. Economists know that one data point does not make a trend. And yet, if you track forecasters via the Wall Street Journal, or the Philly Fed, or the like, it shows that the tail wags the dog more often than we care to admit.
These three points are intersecting right now in Oregon. Last quarter, as our office prepared the forecast, we were seeing a string of 6,000 jobs per month gains from April through July. Those weren’t sustainable and marked a return to the peak rates. That didn’t make a whole lot of theoretical sense, nor did it match the withholdings coming out of Oregonian paychecks. But that’s what the employer survey data showed and we discussed this with our advisors.
As we wrote in the forecast, we basically treated the preliminary data as upside risk. We tried to frame the discussion around what conditions we’d need to see to support these strong gains, including faster population growth and higher labor force participation rates. We did not build the return to full-throttle growth into our forecast. At the end of the day our forecast is a budgetary planning tool. We are not going to let preliminary data wag the dog, particularly not when actual tax collections are not following suit.
It turns out this was the correct forecast. As seen in the first chart below, for whatever reason, the Current Employment Statistics (CES, aka the employer survey) appears to have gotten off track a bit and is now on a course correction. The Quarterly Census of Employment and Wage (QCEW) is a near universal record of jobs and wages, but lack timeliness. Just last week we got data for April, May, and June. Given the QCEW’s accuracy, every year the CES is benchmarked, or revised to more closely resemble its trends. I’m not here to fault BLS too much. Survey work is hard. But I am here to say we should always take preliminary data with a grain of salt.
While the employment data appears to be getting back on track, the primary reason our office is not concerned about the preliminary job losses is the fact that wages continue to grow in Oregon. Withholdings out of Oregonian paychecks are considerably noisy, but two clear trends stand out. First, growth has slowed in the past year or two. Second, growth remains positive. There has yet to be any real concerning numbers that indicate economic growth is teetering on the brink just yet.
Bottom Line: The economic outlook remains intact. Preliminary data gets revised. The two recent employment reports showing losses, to a large degree, are offsetting huge gains in the previous months. So to those looking to pin the losses on the wildfires or the minimum wage, well, these are not the data you are looking for. Our office is currently working on the next forecast, which will be released Wednesday, November 29th.
no comments yet