March 2014:
 |
| Graph #1: The 'oido' graph as a scatter plot |
I was trying to update this graph from 2012:
 |
| Graph #2: http://research.stlouisfed.org/fred2/graph/?g=bLT from 2012 |
But the lines that connect the dots in the new graph are mostly vertical. In the old graph, they are not.
If I had put a straight-line trend on either graph, it would go from high on the left to low on the right.
To my way of thinking, if the dot-to-dot lines generally run in the same direction as the overall trend line, I'm happy with what I'm looking at. But if the dot-to-dots generally run in a different direction than the overall, then I have to think there's something fishy about the overall (as with the Phillips curve). So when I saw the discrepancy in dot-to-dot direction on the two graphs, it didn't sit right.
That was, what, two weeks ago. Well, this morning I woke up around three, as usual, had my coffee, and suddenly had an urge to see the 'oido' graph in Excel. Here:
 |
| Graph #3 |
As you can see on #3, the general direction of the lines is from high on the left to low on the right. Same as Graph #2, the old FRED graph. And definitely different from #1, the new FRED graph.
Graph #3 is drawn from the same data as Graph #1. Excel disagrees with FRED about what the dot-to-dot lines should look like.
Also, old FRED agrees with Excel, and disagrees with new FRED.
I am forced to conclude that the new FRED graph draws the dot-to-dot lines wrong.
1 comment:
Note that at the FRED Blog, in the post titled The Phillips curve after the Great Recession, the scatter plot is described this way:
"The data run from June 2009 to August 2015, and the line connects the points in the scatter plot in temporal sequence..."
The line that connects the dots connects them "in temporal sequence": in order by date. It matters.
As it happens, the FRED Blog post is a few years old now and lacks both the connect-the-dots line and the "customize" link that most FRED Blog graphs have. I recreated the graph from the instructions in the FRED post.
Unfortunately, when you get the "page short URL" for the scatter plot with the dot-to-dot lines visible, the URL returns a version of the graph that lacks those lines. This link shows a screen capture of FRED's oFtk graph with the dot-to-dot lines visible.
This line shows the data from FRED's oFtk graph, plotted in Excel for comparison.
The two graphs are similar because they show only one recovery. But the graphs are far from identical.
Post a Comment