Sunday, December 7, 2014


"Net interest and miscellaneous payments on assets" is a measure of all the interest that is in GDP. Okay. Suppose we take it out of GDP:

Graph #1: GDP (blue) and GDP with Net Interest taken out (red)
Not much difference. "Net interest" is a small number.

Of course it is. It's "net".

By the way, we saw two measures of "net interest" as a percent of GDP last Thursday:

Graph #2: "Net interest and miscellaneous payments on assets" as a Percent of GDP
Less than 9% of GDP. On average, maybe 4%.

On day 18 I looked at the effective mortgage interest rate. A year and more back I looked at the effective interest rate on Federal debt. The next graph shows those two rates together. They share a path:

Graph #3: Effective Interest Rates: Federal (red) and Mortgage (blue) rates

"Effective" rates are calculated by working backward from dollars of interest paid and dollars of debt owed.

My numbers are not on the high side. Reproduced here, Steve Waldman's effective Federal rate peaks at 7.5%; my number matches his. And Mason and Jayadev (2012) show household debt peaking above 12.5%; my mortgage rate peaks at less than 11%.

A creditworthy government typically pays a lower interest rate than businesses do. A homeowner typically pays more than a business. So I want to take the average of the Federal rate and the Mortgage rate, and use that number as a ballpark estimate for the effective rate of interest on business debt.

Graph #4: The Effective Nonfinancial Business Interest Rate (green), an Estimate

There ya go. The green one is my estimate of the Effective Business Interest Rate.

Now we've got an interest rate for business debt. And we can get numbers for the level of business debt from FRED. So I can calculate estimate the total amount of interest paid by businesses. The gross amount of interest, I should say. Gross interest, as opposed to net interest.

I will exclude financial debt. As a rule, I never exclude any piece of debt. But surely there must be some double-counting if we count both financial and non-financial debt. I'll exclude financial. And I'll exclude government debt and household debt. That means we're excluding the mortgage debt that seems to be a large part of the interest that is counted in GDP, as we discovered yesterday.

According to my old shopping list, that leaves us with the debt of
1. farm business,
2. nonfarm noncorporate business, and
3. nonfinancial corporate business.

At FRED I find
1. Farm Business; Credit Market Instruments; Liability (FBTCMDODNS)
2. Nonfinancial Noncorporate Business; Credit Market Instruments (NNBTCMDODNS)
3. Nonfinancial Corporate Business; Credit Market Instruments (NCBTCMDODNS)

Item 2 there, nonfinancial noncorporate business debt, NNBTCMDODNS, that used to have a different name. Yeah yeah, well, it used to be called "debt" and now it isn't; now it's called "Credit Market Instruments; Liability". Yeah. But something else. The "NN" in "NNBTCMDODNS" stands for nonfinancial noncorporate. Back when they called it debt rather than liability, the NN stood for nonfarm noncorporate. I think that's odd.

I had trouble finding the Farm Business series. But I was trying to document my claim there that NN used to be "nonfarm noncorporate"... couldn't find even a suggestion of it on the internet... but in my old FRED downloads I found an Excel file dated 2 July 2013, containing the original FRED download worksheet and this header information:

That got me the Farm Business series name FBTCMDODNS. It documents that the NCB-, the NNB-, and the FB-prefix TCMDODNS series are all "domestic nonfinancial" series (as if the DNS suffix wasn't evidence of that). And it shows that FRED once used the NNB prefix to mean "Nonfarm" Noncorporate Business. That was satisfying.

So we can look at the debt of corporate and noncorporate nonfinancial business and farm business, tally up the debt, figure the interest paid on that debt at the estimated effective rate, and come up with a number for gross interest paid by nonfinancial business. Now we're getting somewhere.

Number one, I want to compare that to the "net interest" number, to see if net interest is more than just a joke. Number two, I want to compare it to GDP, to see the true cost of embedded business interest and its effect on prices.

So here are the three portions of nonfinancial business debt:

Graph #5: The Debt of Nonfinancial U.S. Business: Corporate, Non-corporate, and Farm
Now here they are in total, in blue, along with the Gross Federal Debt (red) and Home Mortgage Debt (green). Funny how these three group together:

Graph #6: Mortgage Debt (green), Gross Federal Debt (red) and Nonfinancial US Business Debt (blue)

The red shows the same Federal debt used to figure the Effective Federal Interest Rate in Graph #3 and #4 above. The green shows the same Mortgage debt used to figure the Effective Mortgage Interest Rate, shown on #3 and #4 in blue.

Now if I take these debt series and multiply each one by the appropriate "effective interest rate" I'll see the interest paid on each measure of debt. The Federal interest will be the actual number, as I am just reversing a calculation done for Graph #3. Same with the Mortgage interest number. The business interest number is an estimate, based on the estimated effective interest rate for business debt figured in Graph #4.

Graph #7: Mortgage Interest Paid (green), Federal Interest Paid (red),
and Estimated Nonfinancial Business Interest Paid (blue)

Remember, the red line includes Federal interest but not state and local government interest. The green line includes mortgage interest but no interest on auto loans, student loans, and credit cards. And the blue line includes interest on nonfinancial business debt, but no interest on financial business debt.

If I take the number for domestic nonfinancial business debt (the blue line on Graph #6) and multiply by my estimated Effective Business Interest Rate (the green line on graph #4), I can get an estimate of gross business interest expense. Let's see how that looks as a percent of GDP:

Graph #8: Gross Nonfinancial Business Interest Expense (blue) and Net Interest(red) as % of GDP
Gross is Less than Net, because Net includes Home Mortgage Interest
Well that's nasty. Gross interest is less than Net interest. That's gotta be because the "net" number includes home mortgage interest from those "fictional businesses" that Clopper Almon was talking about.

Okay, I'm gonna tie this one off.

I learned two things from this exercise. I learned that mortgage interest is a big number. And I learned that FRED is not designed for calculations of this complexity. (Not that it is a complex calc.)

Look at all the text in the title! Look at all the text in the vertical axis label! These are just ridiculous.

It would be bad enough if this was done before March 2014 when FRED went haywire. Back then, they used the series ID code in the graph title, rather than the series ID name. So you would have "NNBTCMDODNS" in the title, rather than "Nonfinancial Noncorporate Business; Credit Market Instruments; Liability, Level". When you've got several series all in the same calculation, the title becomes incomprehensible using series ID names.

FRED is not designed for calculations of this complexity. I think Clopper Almon's G7 might be much more useful. I could add up the three components of domestic nonfinancial business debt, think of it as a variable, and give it a nice short name. And then I could just re-use the short name later in the calculation and in other calculations. I think that would work. I think that would be a real benefit.

A short incomprehensible name is much easier to comprehend than a long incomprehensible description.

No comments: