Household debt as an indicator of secular bull and bear markets

Submitted by Edward Harrison of Credit Writedowns

In my last post, I presented you with a bunch of data on debt levels broken down by sector of the economy (see “A brief look at the Asset-Based Economy at economic turns”).  I found it interesting that a secular pattern seemed to be at play when looking at the household debt charts.

Notice the three areas boxed in red on the chart to the right.


The chart measures the differential between the year-on-year change in household debt and nominal GDP.

The three areas show three distinct periods of household debt accumulation.

  1. 1951-1966. The first shows household debt changes generally outstripping nominal GDP by a wide but decreasing margin. This period coincided with a secular bull market in equities.
  2. 1966-1982. This second period is more volatile, but with the overall numbers lower.  In general, debt was accumulated less rapidly compared to the growth in nominal GDP. And when recession hit in 1970, 1974 and 1980, it induced a retrenchment (at least relative to nominal GDP growth). This period coincided with a secular bear market in shares.
  3. 1982-?. This last period shows an enormous increase in debt growth relative to GDP growth during the 1980s followed by minor retrenchment after the 1990-91 recession and strangely also in 1997 (could this be a butterfly effect to the Asian Crisis?). But after that it was off to the races right through the 2001 recession until mid-2007.  This period coincided with a secular bull market in equities.

The pattern seems to indicate that there is a relationship between debt build-up in the household sector and stock prices.  The build-up in debt relative to nominal GDP troughed in Q3 2008 at -0.4%. As of Q2 2009, the number was +1.2%.

I see this as evidence of the so-called Wealth Effect. The data suggest that the secular bear market may not have begun in 1998 or 2000 as I have generally believed. And they also suggest that, despite the recent rise in shares, a new secular bear market may have just started in 2007. I will be curious to see what the data look like for the second-half of 2009.


Z1 Data Series – Federal Reserve

Print Friendly, PDF & Email
This entry was posted in Guest Post, Investment outlook on by .

About Edward Harrison

I am a banking and finance specialist at the economic consultancy Global Macro Advisors. Previously, I worked at Deutsche Bank, Bain, the Corporate Executive Board and Yahoo. I have a BA in Economics from Dartmouth College and an MBA in Finance from Columbia University. As to ideology, I would call myself a libertarian realist - believer in the primacy of markets over a statist approach. However, I am no ideologue who believes that markets can solve all problems. Having lived in a lot of different places, I tend to take a global approach to economics and politics. I started my career as a diplomat in the foreign service and speak German, Dutch, Swedish, Spanish and French as well as English and can read a number of other European languages. I enjoy a good debate on these issues and I hope you enjoy my blogs. Please do sign up for the Email and RSS feeds on my blog pages. Cheers. Edward


  1. Bob_in_MA


    How is the differential calculated?

    Are you using nominal or real changes?

    The two periods of rising debt differ in that one began at a time of historically high savings rates and was a period of slowly rising inflation, and the second began at a time of a more normal savings rate and was a period of slowly falling inflation.

    1. Edward Harrison Post author

      I calculated the series by comparing the year-on-year change in nominal GDP as reported by the BEA to the year-on-year change in household debt as reported by the Fed. I then subtracted the debt percentage change from the GDP percentage change.

      What this does is compares the rise in nominal debt to the rise in nominal output with the view that nominal debt cannot rise faster than output indefinitely.

      I also suggest you see the linked post as well because I take this same methodology to the entire Flow of Funds series:

  2. Kievite

    You should be vary to trust an indicator based on two metrics that are definitely politically charged (especially GDP).

    If we assume that the error in measuring debt is 10% and the error in measuring GDP is 20%, you metric has max 37% error.

  3. Simon

    If it were possible you could look at household debt level differentials for the period before and after the great depression. You would probably find a similar pattern. Looking at Japan post 1990 and you would probably find a very similar pattern for household debt. Government debt of course expanded to try to offset private deleveraging just as now in the US. In my opinion as it is for many commentators the bull market in everything except the dollar is readily explained by the massive liquidity supplied to banks by central bankers. I almost wrote “central governments”. Having no where else to go it goes into those assets. As to what happens when the last newly minted dollar has been spent…The rush to the exits is likely to be at least as dramatic as last time…

  4. Nostradoofus

    This seems like correlation, not causation. The underlying force is interest rates.

    Debt levels and stock prices both likely reacted to the direction of interest rates, which generally fell in the first period, rose in the second, and fell again in the third.

    This is logical for debt levels, because debt affordability varies directly with the cost of capital. It is also logical for stock prices, because the yields of different asset classes compete with each other, causing earnings yields to rise (i.e., P/E to fall) when money yields rise.

    Hence, as interest rates now have nowhere to go but up, we can anticipate both a sluggish market and falling consumer debt.

    1. Edward Harrison Post author

      Right. One can’t say its causation at all. That is why I was careful to use the term ‘coincided.’

      The question is: where are we now both in terms of debt as a percent of debt and stock prices?

  5. rd

    This analysis makes some sense to me, especially given how important the consumer has been to the economy since WWII.

    On the question of dating the secular bear market, there is an argument to be made that the secular bear did just start recently. The S&P 500 has had negative returns for a decade but much of that is because of the tech bubble that blew up. The S&P 500 tech firms got hammered from 2000-2002 which is why the NASDAQ graph looks like the Nikkei graph from 1990 on.

    Non-tech S&P firms really only got hit hard in about 2002 when it seemed everything went down 20% but they didn’t suffer much in 2000 and 2001 when the tech stocks imploded. Many stocks didn’t go down anything close to 50% during that period, so large segments of the market just had a normal bear 2002 and more than recovered over the next few years.

    However, everything but Treasuries got whalloped last year with most stock sectors in the 50% or more club. So that could have been the start of the massive secular crash where virtually everything lost value, similar to the 1929-32 time frame.

  6. Fed Up

    I think you need to look at positive/negative real earnings growth for the lower and middle class, positive/negative real earnings growth for most businesses, price inflation, and whether there was a trade surplus or trade deficit.

    I would be especially interested in the period since 1982. Was there negative real earnings growth for the lower and middle class, positive real earnings growth for most businesses, lower rates of price inflation, and an increasing trade deficit?

    Has the fed “tricked” the lower and middle class into going further and further into debt because they have actually been fighting price deflation since about 1982? Has the fed been so successful in fighting price deflation because of their claim to be fighting price inflation due to the experience of the 1970’s and poor assumptions by the lower and middle class about their lifetime wage income?

  7. Rudi

    Ed — Interesting post. It appears that these three periods roughly coincide with secular changes in the distribution of income. The early post-war period was characterized by a high (but falling) share of profits in GDP. The second period featured a high wage share. Since Reagan, the profit share has again made advances. The data is pretty rough, so it’s hard to really infer anything from such coarse observations, but these periods do roughly coincide.

Comments are closed.