Yesterday I showed a graph of two FRED datasets with exactly the same
name: "Monetary interest paid: Government: Federal". The names are the
same, but the numbers differ substantially. What to do, what to do?
The series with lower numbers starts in 1960, same as the "Monetary interest paid" (in total) series. When the components of interest paid are added up, it is probably this series (with lower numbers) that are used for interest paid by the federal government. However, as I pointed out yesterday, both datasets are current. It's not that one is the "old" data and the other is "new". Both sets of numbers are reported annually.
I prefer the series that starts before 1960. I always want all the data I can get.
Also (without knowing the facts) I prefer the series with the higher numbers because it is probably more honest.
Yesterday I was frustrated because
I couldn't find out why there were two different sets of numbers. Today
I'm using my judgement. If that's not good enough, then they should
provide better documentation. I looked. I didn't find any documentation at all. I didn't even find anybody wondering which series to use when.
Anyway,
here's what I did. I put FRED's interest paid (in total) on a graph, a
fat blue line. Then I added a thin red line showing federal government
interest paid (using the low-numbers data). Both these datasets start in
1960 -- that's how I know this must be the federal interest data that
they use when they figure the total. Then I added the interest paid by
state-and-local-governments, by households, by U.S. business, and by
"the rest of the world" to US creditors. Here is the result:
%20FRED%20Utn2.png) |
| Graph #1: The FRED series that starts in 1960 (blue); the Sum of Components (red) gives a good match |
I made the red line thin so I could see the blue line behind it. The red is centered on the blue from start to finish. So I'm satisfied that I have the right components, and all of them.
Next, I switched the federal government component from the one that starts in 1960 to the one that starts before 1960. Here's the result:
%20FRED%20Utm7.png) |
| Graph #2: The sum of components (red) now runs on the high side. |
The red line is no longer centered on the blue.
Obviously, FRED has all the data. They still track it all. I just don't know why they go with the lower number for interest paid by the federal government.
No matter. I can use the higher number and make my own
version of total Monetary interest paid. I can have two versions of it,
just like there are two versions of the "Government: Federal" data set
at FRED.
Anyway, the gap between red and blue is not so big on the graph.
4 comments:
I am going to make an educated guess and say the difference between the two lines are interest paid on intra-government debt in non-marketable securities owed to social security...One line is including it more than likely the one that goes back to the 30's and the newer on does not. I cannot remember when they started excluding SS debt out of the total debt owed number, but from the look of the series I would guess in the 60's.
Hey, O.T.
I thought about "gross" or total federal debt, versus the debt "held by the public". Didn't think of a good way to compare the numbers
The start-date relation you mention makes a lot of sense. For all my complaining about the 1960 start-date, it never occurred to me that the reason might be as you say. Good thinkin!
I'm looking at the ratios -- "held by the public" as a percent of the bigger number -- for the debt and (presumably) for the interest paid. The lines are by no means parallel, but they do cross more than once. This graph might support your educated guess. I can't say.
But there must be some explanation, and yours does make sense to me.
Hi Art
https://apps.bea.gov/iTable/iTable.cfm?reqid=19&step=3&isuri=1&nipa_table_list=87&categories=survey
The link above to Table 3.2. Federal Government Current Receipts and Expenditures
If you go the bottom of the page footnote 4 explains what changed in the accounting of interest paid in 1960. Apparently both accounting methods are graphed by FRED.
Just looking at the numbers that seems to account for only part of the difference. The rest may be interest collected from state and local governments that borrow through the Federal govt. Also interest that the Fed receives on its Treasury holdings is mostly given back to the Treasury so that may play a role also.
Thank you Jim.
Footnote 4 explains a lot: Prior to 1960, interest payments [are] shown net of interest receipts.
Okay, so my version of total Monetary Interest Paid, before 1960, is net of interest receipts. A little on the low side. Maybe all the way from start to finish, on the low side.
I'll have to look at Table 3.2 a few times to get familiar with it.
Post a Comment