Look: I am eager to learn stuff I don't know--which requires actively courting and posting smart disagreement.

But as you will understand, I don't like to post things that mischaracterize and are aimed to mislead.

-- Brad Delong

Copyright Notice

Everything that appears on this blog is the copyrighted property of somebody. Often, but not always, that somebody is me. For things that are not mine, I either have obtained permission, or claim fair use. Feel free to quote me, but attribute, please. My photos and poetry are dear to my heart, and may not be used without permission. Ditto, my other intellectual property, such as charts and graphs. I'm probably willing to share. Let's talk. Violators will be damned for all eternity to the circle of hell populated by Rosanne Barr, Mrs Miller [look her up], and trombonists who are unable play in tune. You cannot possibly imagine the agony. If you have a question, email me: jazzbumpa@gmail.com. I'll answer when I feel like it. Cheers!

Tuesday, March 11, 2014

Equity Extraction and Personal Consumption Expenditures

In comments to my previous post I made this statement of historical fact: "home price bubble equity was an ATM that grew consistently as a fraction of PCE after 1996, and accounted for over 2% in 2005 – and I got this from Greenspan and Kennedy, Table 2." In my estimation, this implies that the collapse of the housing bubble eliminated the possibility of equity extractions, and was therefore a major contributor to the decline in personal consumption expenditures [PCE] that led into the Great Recession.

Here is the explanation.

Bill McBride at Calculated Risk has been following the equity extraction data. Here is the March, 2014 update. Dr. James Kennedy, mentioned above, wrote that for technical reasons, the data set that he and Greenspan were using was no longer valid after 2008, and presented an alternate calculation method [link at the linked CR post.] McBride uses this alternate measure, calculated from the Fed's Flow of Funds data and the BEA supplement data on single family structure investment. Also linked at the CR article is a spread sheet with the two data sets. Graph 1, from CR, shows how the two data set compare.

Graph 1 - Equity Extractions - G-K Data vs CR Calculation.

I'm looking at the correlation between equity extractions and personal consumption expenditures during the housing bubble and collapse to support my claim. The method is to compare McBride's calculated data, which extends past the end of '08, and FRED Series PCE. Graph 2 shows Equity Extractions [blue, left scale] and the YoY dollar change in PCE [green, right scale] from 1991 through Q1, 2010, both in billions, quarterly data.

Graph 2 - Equity Extractions and PCE Change

The traces start rising together after the 1991 recession. There's a disconnect during the 2001 recession, when PCE takes a dive, but extractions continue to increase. From 2003 until the crash, they are close to being in lock-step; but from the peak, extractions fall farther and faster leading into the recession. One objection to my claim is that extractions decline a year earlier than PCE. We'll get to that.

Graph 3 shows a scatter plot of Year-over-year PCE dollar change vs equity extractions from Q1 2001 through Q3 2008

Graph 3 - PCE Dollar Change vs Equity Extractions - '01 to '08

The period from Q1, '01 through Q2, '03 is highlighted in red. During this time, PCE falls and levels off without a correlation to extractions. After mid '03, extractions and PCE rise together into the peak values highlighted in yellow, then fall together into the crash. R^2 for Q3 '03 through Q3 '08 is .75. This includes the blue and yellow points.

Graph 4 is a scatter plot of the Q1 '91 through Q1 '10 period. I've color coded data subsets representing different coherent realms.

Graph 4 - PCE Dollar Change vs Equity Extractions - '91 to '10

Starting from the 1991 recession in red, PCE increased into the cluster of purple dots representing Q1 '92 to Q1 '98. From Q2 '98 until Q1 '01 is another [less tight] cluster representing a greater change in PCE but only slightly higher equity extractions. This is the peak of the dot com bubble. Q2 '01 to Q2 '03 is again in red, the slide into and climb out of that recession. The blue dots, as in Graph 3, represent the period from Q3 '03 to Q3 '08 - the housing bubble peak and decline into the crash. The green dots are from Q3 '08 through Q1 '10, when everything collapsed and the correlation fell apart. The R^2 for all the data points except the green is .49. Eliminate the red dots as well, and it rises to .70. Take out the top 4 yellow dots, Q4 '99 to Q3 '00, and it rises further to .81.

This suggests that outside of recessions and the peak of the dot com bubble, which can be considered as distortions to a underlying trend, from 1992 on, changes in consumption expenditures were strongly correlated to mortgage equity extractions.

I used an ATM analogy for equity extraction, but there's a big difference. You can go to the ATM as often as you like, but extracting equity is an event that is unlikely to be repeated very often. So it's not out of the question to expect that the flow of equity dollars into consumption expenditures would be spread over several months - possibly a year or more.

I also looked at the correlation between extractions and the PCE dollar change 4 quarters later. This is shown in Graph 5.

Graph 5 - PCE Dollar Change vs Equity Extractions - '91 to '10

Color coordination, based on the PCE values, is the same as in Graph 4. The blue dots now make a more spiky array, but there is almost no loss in the coherence of the data sub sets. Only the green dots, now extended though Q1 '11, look substantially different.  R^2 for the blue dots slips from .75 to .56, but for the entire data set [except the green dots] it increases slightly from .49 to .55.

Extraction data is seasonal, with local peaks in Q's 2 or 3, and valleys in Q's 4 or 1.  This accounts for some of the data scatter.  PCE dollar change data is smoother, with no consistent seasonal pattern.  Graph 6 shows a scatter of 4Q averages of each variable, same color coding as graphs 4 and 5.  Not including the recession-related red and green dots, R^2 is .75.  Include the red dots and R^2 drops to .588.

Graph 6 - PCE Dollar Change vs Equity Extractions - '91 to '10 - 4 Q Avgs

It's now very easy to see the two variables rise and fall together from the time after the 2001 recession into the Great Recession.

I know correlation is not causation, but I have a coherent narrative that is completely consistent with the data. The behavior of the [blue dot] data from 2003 until deep into the collapse is striking, either with or without a 4 Q lag. Non-conforming data [other color dots] are explainable variances. I think the counter assertion that equity extractions had not a darned thing to do with the collapse into the great recession is not supported by real world data.


Cirze said...

You should publish this, sweetie.

Fantastic writing.

Great reporting!

Jazzbumpa said...


You got here a little early. I've been editing for the last hour.

Should be just a little bit better now.

I'll put it up at Angry Bear in the morning.


The Arthurian said...

I like it. Sounds right to me: As long as consumption spending was getting a boost from equity extraction, the economy was getting a boost from consumption spending. And then the collapse of the housing bubble toppled the dominoes.

Jazz, I think you've answered a question Steve Roth asked at Asymptosis a couple years back: Why did it take almost a decade, this time, for the big drop in Federal debt to create a depression?

Nice work.

Jazzbumpa said...

Thanks, Art -

Steve's post is very thoughtful.

I'm not going to claim I've discovered the ultimate answer here, but I do think I've identified a significant contributor.

P.S. When do you sleep?


State of Thought said...

While you've got a rather good demonstration of correlation there and it's a fairly plausible contributor, I'd be leery of presuming that it is a primary trigger for the crisis. Consider the leveling and then decline in employment that went along with the decline in availability of home equity. Reduction in home sales sacked a lot of jobs, not just in home construction but also in related industries (e.g. furnishings). It seems highly probable that the job losses from reduced home sales contributed as much as or more than reductions in potential for equity extraction.

jim said...

I don't really see the point in focusing on the relationship of equity extraction and personal spending. I'm not really disagreeing. It just seems that second liens have about the same effect as first liens.

All loans contribute to increased spending. The reason you usually get a loan is to spend the money. If the loan is equity extraction you might spend it on big screen TVs or on a new car or you might pay someone to remodel your house and then the contractor does the spending. And when the borrowing stops the spending will drop also.

That said, It is important to understand the huge extent to which home equity extraction was part of the sub-prime borrowing of the 2001-2007 era. 70% of sub-prime loans were second liens. And home equity extraction could be set up to work like an ATM as can be seen from this 2007 ad.

The popular myth is the housing bubble was about expanding home-ownership, but it was really the opposite. The home-ownership rate now is several percentage points below what it was in 2001 and a lot of the losses were due to people who got in trouble by taking out sub-prime home equity loans.

Jazzbumpa said...

jim -

Th point is that 2nd liens go to consumption spending and first liens go to mortgage payments.

I think this ties in well with the rest of your comment.



jim said...

I understand your point.

My point was that I suspect if you did the same analysis for the same period of total household debt increase vs PCE increase you would see a similar correlation.

Troy said...


blue is YOY job gains
red is YOY consumer credit growth (mostly mortgage)

home borrowing did not just fund HELOCs, the cash-out refis paid off credit cards and funded all kinds of business investment, car purchases, vacations, etc etc.

Plus all the transaction volume was making the FIRE sector immensely happy, which also trickled out into the wider economy.

The credit influx peaked at $100 BILLION PER MONTH for a couple of years.

It was really great while it lasted -- it was a money injection DIRECTLY into the middle quintiles.

Nobody gets this.

The Arthurian said...

I spent some time with your graph and data here, here, and here.

Troy said...


the extension back to the 1960s was appreciated. I did leave in the 1990s in my original to compare 1990s vs 2000s.

We can save a lot of analysis and just take TCMDODNS over GDP, which is our non-financial debt leverage.


when I first made this graph my jaw dropped. SO much had become perfectly clear.

Anonymous said...

Using ingenious technology and innovation the Romans made an Empire that
withstood the test of your time. An attractive attractive website is essential inside the technologically advanced and highly competitive market of current age,
for the achievements business. it may also be useful for headhunting and employment.

my page http://nasebook.com