No good answers

In On household debt at the FRED Blog, for household debt they use "consumer credit" plus "home mortgages" as a measure of household debt. I always use CMDEBT. What's the difference?

Graph #1: Blue is CMDEBT. Red is what the FRED Blog used.

Ooh, that's a fair amount. I was expecting to say "not much" but it is more than I thought. It varies, but my number is in the neighborhood of 9 or 10 percent higher than their number:

Graph #2 (This graph also shows more years.)

Maybe I should be using the FRED Blog number.

//

I wonder what the difference is. Just the nonprofit organizations? If it was some other household debt, the FRED Blog graph should have included it. So, maybe yeah. That's not a very good answer. But that's all I got.

//

Suppose we look at related things. Interest paid:

Graph #3

These are close. This, I've looked at before. Here's the ratio:

Graph #4

This one varies, too. But the interest paid by nonprofits is generally in the neighborhood of 2 or 3 percent of the amount of interest paid by households. Not 9 or 10 percent.

If our assumptions so far are good, then nonprofits borrow about 9 or 10% as much as households borrow, but pay only about 2 or 3% as much interest. That's odd. Do you suppose households on average pay interest rates that are four or five times as high as those paid by nonprofits? That doesn't seem right. Makes me think that the 9 or 10 percent difference we found is maybe only in part the debt of nonprofits, and in part also the debt of households. Not all of it nonprofit debt. But I still don't know.

The information should emerge from the numbers, but in this case it does not.

Suppose we look at effective interest rates: Monetary interest paid, relative to debt owed. Debt for households, using household debt as figured in the FRED Blog post. Debt for households plus nonprofits, using CMDEBT. And if we assume the 9 or 10 percent difference between CMDEBT and the other is the debt of nonprofits -- an assumption I do not trust -- then we can see the effective interest rate paid by nonprofits, too.

Blue for Households, red for Households and Nonprofits, and green for Nonprofits:

Graph #5

FRED has two different series for monetary interest paid: government: federal. There are also two basic versions of the Federal debt: the gross debt, and the one that excludes debt the government owes to itself. Using the lower measure of interest paid and the higher measure of Federal debt gives me the lowest measure I can get for the effective rate of interest paid on the Federal debt. That's the dotted line on the graph above.

Federal interest rates are low, compared to other rates. I've heard it many times. So it doesn't surprise me that the dotted line is lower than the red and blue.

It doesn't surprise me either, really, that the green line is lower than the dotted line. I take it to mean that the green line is wrong. The green line is too low. Either the interest paid by nonprofits is wrong, or the number I used for the debt of nonprofits is wrong.

The latter, obviously.

So here we are at the end of this episode, and all we know is that I still don't know about the debt of nonprofits.

And that means I still don't know about the debt of households.

Yield Curve, again

The other day I showed this graph:

Graph #1: Excel Polynomial Order 2 Trendline

What happens if we extend that trend line into the future?

Graph #2

The trend line bottoms out above 1% and starts to rise.

Debt and GDP on a Log Scale

Graph #1: TCMDO Debt (red) and GDP (blue) on a Log Scale

Accumulated Debt and Nominal GDP appear very much in sync from 1951 to 1981. Then GDP growth slows. Then debt growth slows, but not as much, and the two drift apart until the Great Recession.

Now they appear to be running parallel again. Like in the 1950s, except much farther apart. And at a much higher level.

And not so fast uphill.

Monetary interest paid: Domestic: Corporate business: Financial

A FRED search for monetary interest paid turns up 30 series. Look for "financial" in the results, and there are four hits. One of those is for nonfinancial corporate business; I skipped that one. Of the remaining series, one is for financial corporate business, and the other two are the components of that: interest paid "on deposits", and interest paid "on other liabilities".

I took a quick look: Add the interest on "deposits" and "other" together, and they do add up to the number for financial corporate business.

Here's a look at the two parts of financial corporate interest expense, each shown as a percent of the total:

Graph #1: Financial Corporate Business Interest Paid
on Deposits (red) and on Other Liabilities (green)

In the 1960s, 75 to 80 percent of that interest expense was on deposits. Now, 90 percent of it is not on deposits.

This looks like a change that started in the '60s, by the way.

Experimenting with ways to show the yield curve

An "area" graph:

Graph #1

It looks to me like the increases since mid-2016 are more parallel than not.

The skinny white vertical lines are not really lines. They are gaps in the daily data. I don't know why there are gaps; they're not "regular" enough to be mostly weekends. Anyway there are gaps in the data, and when FRED filled the area with color for the area graph, it kept the gaps as gaps.


This next graph uses the same daily data for the same time period, but shows it a different way.

Graph #2

The blue line here is the spread: the 10-year interest rate minus the Federal Funds interest rate. The red line is the Federal Finds rate, but with a minus sign to make it negative. So if you measure from the zero line up to the blue line, that's the size of the yield spread. And if you measure from the red line up to the blue line, that's the size of the 10-year interest rate. Plus, the red line lets you see when and how the Fed changed the Fed Funds rate. (But of course it has really been going up, not down.)

The changes in the red line are clearly visible in the blue. I had to hunt to find them, but they are there.

I like this way of showing the yield spread.


I want to use this "spread and negative fedfunds" comparison graph to look at the 1990s now. But, oh, apparently the daily data only goes back to the year 2000. So I switched to weekly data for this one:

Graph #3

The Fed Funds rate (red) is shown here a little below -7.5% in 1990; really it was around 8%. By 1993 it appears to be a little below -2.5%; really, around 3%. The FedFunds rate fell (moved toward the zero line) in those years. And the spread (blue) went up.

In 1994 and early 1995, on the graph FedFunds goes from below -2.5% to below -5% (really, it increased from around 3% to around 6%). As FedFunds rose, the spread fell; it reached zero in July 1995, five years before the start of the 1990-91 recession.

You can see the Fed reducing the FedFunds rate from 6% to 5% in late 1995 and early 1996; and you can see the spread rising up from zero in 1996.  Then in early 1997 you can see the FedFunds rate increase to about 5.5% and stay there for more than a year as the spread falls to zero again.

I won't bore you with every excruciating detail of this. But maybe you can tell why I like the graph: It really lets you see what happened with the policy rate and the yield spread.


And the relevant detail? The spread hit zero five years before a recession hit the economy.

The yield spread since the end of the Great Recession

Looking at the yield spread since June 2009, an optimist might see this:

Graph #1: Excel Polynomial Order 5 Trendline

A pessimist might see this:

Graph #2: Excel Polynomial Order 2 Trendline

A panicking pessimist:

Graph #3: Excel Polynomial Order 3 Trendline

And a realist:

Graph #4: Excel Polynomial Order 6 Trendline

What do you see?

The yield curve (follow-up)

One more graph on the yield spread (again, figured as the 10-year rate less the FedFunds rate, based on what John Cochrane showed). Looking at "change from year ago" values for the spread, in blue. And in red, the Fed's upper target for the Federal Funds rate. Since January 2014:

Graph #1

In January 2014 the spread was one percentage point larger than it was in January 2013. In January 2015 the spread was one percentage point smaller than in January 2014. During that time there was no change in the FedFunds target. The change in the spread was not due to the Fed changing its target.

Over the course of 2015 the blue line moved toward zero. By December it was almost zero, meaning there had been almost no change from a year earlier. At that moment the Fed raised the target a quarter point and held it there for a year.

For six months after the Fed raised the target, the spread narrowed. By July 2016 it was one percentage point smaller than a year earlier. Then (with no additional change by the Fed) the trend turned upward. By the end of 2016, again, there was no change from a year earlier.

In other words, the first quarter-point increase by the Fed, the increase of December 2015, affected the spread for about six months. After that, the prior trend (narrowing less and less) resumed. And in the first two months of 2017 the spread actually started to widen.

As happened a year earlier, the Fed saw it coming. They raised the target once again at the end of 2016 and twice more in the first half of 2017. Despite all these increases, the spread held steady till October 2017, showing little or no change from a year earlier. Then, perhaps anticipating another increase by the Fed, the spread fell in November and again in December when the anticipated increase came to pass.

Those several increases together did hasten the narrowing of the spread. But the low point of the blue line this time was not as low as the two prior lows you see on the graph. And the third low occurred in the same month as the Fed's December 2017 target increase. By January, one month after the target hike, the trend was upward again, in the direction of a bigger spread.

The most recent increase in the target, in March 2018, probably slowed the move toward widening. But the move toward widening continues. And that's where we are as of the April data.


In summary:

• ONE target increase (Dec 2015) caused six months of increased narrowing. Then,
• THREE increases (Dec 2016-Jun 2017) stopped the widening of the spread but caused little narrowing. ANTICIPATION of the Dec 2017 increase did increase the narrowing; but despite
• TWO increases (Dec 2017, Mar 2018) the narrowing grows less, and the widening appears more ready to resume.

If I'm reading the graph right, the economy appears to be growing stronger and better able to withstand the increases in the Fed's target. The only bad thing I see in all of this is that anticipation -- or jitters, as Cochrane says -- seems to have more of an effect that two or three actual increases in the Fed's target rate.


Does this analysis stand up? I think so. Here's a graph of the yield spread plain and simple. No "change from year ago" view to complicate things, this time. The blue line is the spread. The red bumps indicate the Fed bumping up the Federal Funds target.

Graph #2

The December 2015 bump led to a 6-month fall in the spread, followed by a 6-month increase. The increase was interrupted by a series of three bumps up which brought the spread down again. But as soon as the Fed stopped raising the target, the spread started rising again.

The analysis stands up.

The December 2015 target change was a 0.25% increase that led to a 1% decrease in the spread, a decrease four times as big as the increase.

The target change that began in December 2016 was a 0.75% increase that caused a 1% decrease in the spread. This time, the spread changed only a little more than the target.

And the December 2017 target change was a 0.25% increase, despite which the yield spread continued to rise. There was no decrease in the spread.

The longer the Fed has continued this policy of increase, the more the economy has adjusted to the idea that interest rates are finally getting back to normal.


This seems like an appropriate time to remember that we still have many policies which encourage credit use, and no policy that encourages repayment of debt. A tax rate designed to vary with the taxpayer's debt-to-income ratio could encourage repayment of debt without being punitive. Such a policy could reduce the growth (and eventually even the level) of accumulated debt, bringing relief to debtors while encouraging economic growth and vigor.

The best time to implement such a plan would be in the very early stages of what will eventually be a long, strong boom. Now, for example. Right now.