Ned Meet FRED
Just how easy is it to use FRED ? I will find out starting four minutes ago, I am trying to set up a FRED account and generate a graph of interest (CPI inflation and nominal wage inflation plotted on time).
I have an account.
19 minutes in. I have downloaded data. Not the data I should have downloaded but that’s not FRED’s fault.
Don’t blame FRED. I can’t handle Excel.
OK here we are 58 minutes in. I can handle FRED. Unfortunately while I am a US citizen and, in theory, a macroeconomist, I don’t know anything about US macro time series (I haven’t dealth with them in over 20 years). I don’t understand why the FRED employment cost index starts in 2001 or so. I barbarically constructed something which has something to do with wages with the snappy name WASCUR-PAYEMS which is equal to the percent change since a year ago of WASCUR (Compensation of Employees: Wages & Salary Accruals) minus the percent change since a year ago of PAYEMS (total non-farm payrolls, all employees).
I still don’t know how to embed the graph which is called Ned Meet Fred.
I admit that WASCUR – PAYEMS is no good. WASCUR doesn’t include fringe benefits. Also it corresponds to quarterly pay not hourly pay, so fluctuations in hours worked per worker show up as if they were wage inflation.
Nonetheless my graph Ned Meets Fred does not fit my prediction (my record remains perfect). WASCUR – PAYEMS does not look like the Atlanta Fed’s index of sticky prices. It was highly correlated with the CPI in the late 40s back when unions were really strong, not correlated in the 70s back when US nominal wage rigidity was justly famous, and is now correlate again.
This feeble effort to deal with US time series data offers something less than no support for Paul Krugman’s Un-COLA hypothesis.
I went to the site and hit “Employment”. I expected their employment data series. Guess what I got? Yup, their job openings. I’m definitely not an economist. That should have bene obvious.
I think your problem is not with Fred.
First, none of the sticky or flexible prices in the Clevelend Fed decomposition consist of wages.
Next, prices are not really sticky, they just don’t always change in the way that we expect to get market clearing. I.e. the theory could be completely wrong in assuming exogenous restrictions on price-setting.
The Cleveland Fed piece makes the point that what we now see as “sticky” prices basically weren’t sticky in the earlier time periods, but are sticky now.
I think your data is basically right, in that wages continue to be more stable than prices of goods (whether we classify the goods as sticky price goods or not).
Nevertheless, it still makes sense to believe that in the COLA environment, inflation will be more volatile, but still less volatile than wages. So COLA can still be responsible for the change.
Here is a graph that looks a bit like Krugman’s, but is actually measuring the difference between hourly wages and the number of hours worked. If you imagine a traditional supply-demand relationship, hourly wages should go up along with hours worked with some of that spilling over into inflation. But with COLA, hourly wages can go up even though hours worked don’t, and this would create inflationary pressures, causing both “sticky” and “non-sticky” prices to increase together. Not an original interpretation 🙂
My interest in trying to get someting like wages out of FRED was to see iif sticky prices are sticky because they are markups on wages which are sticky. So I wasn’t trying for the sticky and flex price series but just a price inflation and a wage inflation series.
I was surprised by how much trouble I had with wages. My variable isn’t very good at all.