Total US Debt to GDP Ratio – Deleveraging Analysis (2023)
A secular trend of borrowing
Many people may be aware of the ballooning US government debt, which is now approaching $20 trillion in 2017. What may not be obvious, however, is that since 2009 the total debt outstanding in the US (including consumer, business, and government debt) has actually dropped when compared to GDP. In fact, the ratio of total us debt to GDP peaked in 2009/Q1 around the 400% mark and has since steadily decreased.
I began researching this ratio after studying Ray Dalio’s theory on how the economy works. In his 30 minute video about “The Economic Machine,” he explains that the economy follows recurring cycles:
- a 5-8 year short term debt cycle which leads to economic expansions and recessions
- a 75-100 year long term debt cycle characterized by large swings in the ratio of total debt to GDP.
Completing the long-term debt cycle
Dalio believes that the long term cycle peaked in 2009. Debt burdens, he reasoned, had grown too high relative to income and were no longer sustainable. To resolve such a situation and purge unsustainable levels of debt, our economy had to (and still does) go through a period of deleveraging. This can happen in one of three ways:
- Deflationary deleveraging is characterized by tight credit and defaults. This is what happened during the Great Depression.
- Inflationary deleveraging occurs when prices rise rapidly while the country’s currency depreciates, eventually wiping away debts through hyperinflation. Germany experienced this in the 1920s after World War I.
- Beautiful deleveraging occurs when the government balances the deflationary tendencies of a deleveraging by borrowing and spending credit-based stimulus. GDP growth slows down as businesses and households consume less and use savings to pay down debt (thus reducing the debt to GDP ratio).
Tracking the ratio
In 2012, Dalio claimed that the US was executing a beautiful deleveraging and I wanted to confirm the validity of this statement. The only problem was that I couldn’t find any free, real-time analysis of such a the Total US Debt to GDP ratio on Google. Furthermore, I wanted to be able to analyze where the debt was concentrated. Obviously much of America’s debt is public government debt, but what about the proportion of mortgages, student loans, and corporate debt?
I took matters into my own hands by creating this analysis. It allows the user to easily track total debt in the US economy broken down by the main categories (Government, Household + Non-Profit, Business). Furthermore, it tracks the total US debt per dollar of GDP over time and shows how each sector contributed to the change in Debt:GDP.
Conclusions and Takeaways
As Ray Dalio describes in his video, the US has well-timed the increase in government debt to soften the blow of consumer/business deleveraging from 2009-2012.
Some of the big bubbles you hear about these days such as auto loans and student loans actually make up a very small fraction of total debt and likely won’t have a huge effect on the greater economy when those bubbles do pop.
Non-financial corporate debt has continued to grow. At ~$9 trillion, this is a similar scale as compared to the total value of mortgages outstanding when the housing bubble collapsed ($13.6t). It’s especially significant because corporations are issuing debt to buy back stocks that are overvalued and could be forced to deleverage their balance sheets quickly if future cash flows are lower than expected.
As time goes on, monitor the debt to GDP ratio to see whether is continues to drop steadily while the economy grows slowly. This would indicate a healthy deleveraging, regardless of the rhetoric coming from the media.
Technical Discussion
To create this dashboard I scoured the Federal Reserve Economic Data repository (FRED) to identify all the necessary debt-related economic data series. It’s relatively easy to calculate a bottom line number – just combine two series (ASTLL and ASTDSL) to get the sum of all debt – but it’s significantly more complicated to find the break down of the various child series for government debt, household debt, and business debt.
Once I identified the pieces of the puzzle, I used PowerQuery (the ETL module in PowerBI) to UNION the datasets for each series into a single fact table. I then added two dimension tables (dates and series) which I used for segmenting the data.
Because most of the data is available online, this report can be easily configured for auto refresh. New data is released by the government by the 3rd week of Mar, Jun, Sep, and Dec.
Nice Work. I really appreciate what you did here.
Came across your website after I started on the same road you did to verify Ray Dalio’s work. Great job on putting together an easy-to-digest set of graphics. Nicely done.
I am puzzled why an article about the debt/gdp ratio does not have a chart of said ratio so let’s take a look. A non current one since as you noted Google does not seem to supply a current one.
http://theeconomiccollapseblog.com/wp-content/uploads/2010/07/Total-US-Debt-As-A-Percentage-Of-GDP.jpg
(excuse the source)
So except for the Depression spike in the ratio caused by a shrinking GDP before the debt was eliminated, by default, the ratio averaged in the 150% area until 1980 when it began a doubling, and beyond, in 30 years. Now a decade later it has fallen by a few percent. This is deleveraging?
Now I will grant that the stupendous technological advances in information and communications technology has by itself made a higher ratio manageable and really possible. I won’t grant that these factors alone have allowed a 250% increase in 3 decades. Not when interest rates were in a secular downtrend during that exact same period.
Dalio’s sanguine analysis is based upon the unacknowledged assumption that interest rates will be tightly capped by central banks via massive monetization of long dated debt. Even that assumption is based upon another unacknowledged assumption, that the overall rapid growth of the amount of low quality debt doesn’t matter.
Hi Rapier,
Thanks for your comment!
Take a look at the dashboard and review the bottom-right chart. You’ll see that the line represents the debt-to-gdp ratio using the left axis. We are currently at around 366%.
I agree that we haven’t really fully deleveraged. However, the fact that the ratio has been dropping over time and that we are experiencing slow GDP growth indicates that we are seeing the “beautiful” deleveraging that Dalio describes. Perhaps this won’t last, however, as the ratio has started to tick back up in recent years. Also, it’ll be interesting to see how Trump’s massive tax cuts cause the ratio to rise even faster.
Have you seen Rays latest book on big debt crises? His recent Bloomberg interview has him thinking we are in the 7 th inning and the next crises is going to be bigger in scope and time but not a single deep stock market drop. He expects rolling crises from asset prices to pension guarantees. I think the big deleveraging will really happen after that
You provide great data charts. Thank you
Hi Andrew,
Yes I just downloaded his latest book but haven’t had the time to get into it yet. What you say about a rolling deleveraging makes good sense, but it could potentially be offset by productivity gains from the solar energy/battery/autonomous vehicle/AI boom. It’ll be interesting to see!
This is great. I couldn’t see the tab(s) for Europe, Japan, China and all the other EMs though 🙂
That would be a great addition but I’m not aware of free data sources from those governments that would give me the same level of detail as the St. Louis Fed data source (FRED). Something to look into!
Hello,
just as Andrew above I totally randomly stumbled upon your site and this article while researching ‘Total Debt to GDP’ graphs after having read the first few pages of Dalios new “book/PDF” about big debt crises.
Originally I learned about Ray Dalio through Tony Robbins book “Money, Master the Game” and the introduction of Ray Dalios simplified version of an “All Weather/Seasons” Portfolio. Without being specific and hard to recreate for amateurs like myself, the original Bridgewater All-Weather Strategy (including used tools like leveraging etc.) was introduced in this open research paper: http://sdcera.granicus.com/MetaViewer.php?view_id=4&clip_id=75&meta_id=9141
I read it only recently and most noteably and interestingly within this paper the “buy and hold all weather strategy” gets put into perspective because, as is explained in the paper, Bridgewater has built a “depression gauge”. A tool that scans the current economic environment and thereby tries to predict a coming depression.
Using this tool, they did so prior to the downturn in 2008.
Their strategy then is to shift the entire portfolio into what they call the “safe portfolio”, which consists of 10% Gold, T-Bills 30%, IL Bonds 40%, T-Bonds 20%.
This information, together with the release of his book, together with ‘cryptic’ tweets like:
https://twitter.com/RayDalio/status/1041766017305911297 and the mentioned article:
https://www.linkedin.com/pulse/applying-my-debt-template-now-try-see-whats-ahead-ray-dalio/
where he says:
“At the same time b) the long term debt cycles of most developed countries are in their very late stages and their abilities to manage the obligations will be difficult. I say that because the total debt and non-debt (e.g. pension and health care) obligations are large/growing and the ability of central banks to reverse a contraction is limited (because interest rates don’t have much room to be lowered and because the ability to squeeze more growth from “quantitative easing” is limited). For these reasons and because of the size of the wealth/opportunity gap that exists, I am more concerned about the outlook for a couple years out than for the near term. I hope that the template that I explained in the first 64 pages of my template will help you figure things out for yourself. Of course, one has to run the numbers to get as accurate as possible in applying that template to form a good outlook.”
…
+interviews like these two:
https://www.youtube.com/watch?v=Nm0m62reFuY
https://www.youtube.com/watch?v=hFcnu_54zl8
where he loseley mentions a ‘hint’ that he is worried about what might happen in roughly two years…
makes me think and wonder wether or not it could actually be a good idea to shift my portfolio into his suggested “safe portfolio” in maybe one to 1 1/2 years..
Do I, as an amateur, interpret the global economic numbers as of today, together with his writing and his remarks correctly as a informed warning for a coming depressionary environment!?
Excuse this long post and the abuse of your comment section but I would really appreciate a discussion regarding this.
And maybe otherwise other people can use this information for their good.
Kind regards
Great articles, thanks for sharing! I like Ray Dalio’s all-weather strategy.
I am currently trying to reconcile several concerns: 1) I believe stocks are insanely overvalued as of Sept, 2018; 2) most of my friends who invest are telling me the same thing; 3) most economic articles I read project a recession around 2020. Since the majority is almost never right, I’m worried that stocks will “melt-up” from here for several years, leaving everyone who is out of the market in the dust (including me).
Thanks for the reply!
Well, aren’t you then just caught up in the old problem of market timing?! And at what %age of correction/recession from current prices would you be inclined to invest again? Have you read “Money Master the Game”? Maybe the proposed all-seasons portfolio would also be the right thing for you. Even in harsh downturns it has performed well and often only lost a small % compared to the S&P500. The philosophy there is protection of principal, while not missing out on the probable ‘melt-up’ and still a big enough ‘safe’ money to reallocate into the stock market after the elusive “coming depression/recession/downturn”.
…at least somewhere along those lines is my “plan” for the coming years, which of course, could be naive.
—
A question as an amateur:
Let’s say we could confidently say that the onset of a downturn starts in exactly two years, but we can’t place bets on that or leverage any investments. (hypothetically) And right now we have a mixed portfolio of Stocks, Bonds and Gold. Would it be better to pull everything out of the investments, lets pretend: 1 day before the downturn starts, or should we be only pulling the money out of stacks – and stay in… Gold?! (I suppose?!) and..?? something else?
In short: is there an asset calls that you would want to be in in a downturn that you had 100% clairvoyance of, but without the option to short anything!?
I ask this because once Bridgewaters “depression gauge” is going off they transfer the funds into their “safe portfolio” instead of having it all in cash.
Appreciate any input!
Good question! When you find the answer to that, let me know!
A recent thought I’ve had is that the best potential place to park your cash during a turbulent recession would be your own house by paying down your mortgage. I personally like the idea that no matter what happens in the financial world, my shelter is completely paid off. If I lose my job, I still have a place to live.
This of course is predicated on several assumptions: 1) your net worth is not too large compared to the value of your house (if it is, paying off your mortgage might not account for much of your wealth); 2) you’re not in an overpriced market; and 3) it relies on the basic assumption that government will never confiscate your property. I’d also probably make sure to have enough cash left over for an emergency fund of cash/gold.
I’m sure people will disagree with me on the basis that you can theoretically earn more in other investments since paying off your mortgage only entitles you to save the 4-5% interest you pay for the loan. However, 4-5% risk free, post tax isn’t terrible compared to the 3% (pre-tax) you earn on a 10-year treasury. What do you think?
Since your prior comment the market has gone down a lot – do you think it is still insanely overvalued? What are the key metrics you consider?
I really enjoy reading John Hussman’s take on the market. He has a proprietary valuation method that indicates that the stock market is priced to return just above 0% over the next 12 years. At the market peak in Sept 2018, it was priced to return -2.5% over 12 years.
Have a look at his analysis and decide for yourself if you’re interested in following his research: https://www.hussmanfunds.com/comment/mc181101/
Hi, I was wondering if you could shed some color on both the “total federal gov debt (non-security)” and “other (misc)” items? Would love to take a look at the underlying for those data as well, thanks!
Great piece, definitely a good read.
If you add up the two main FRED debt series, ASTDSL and ASTLL, you get a total of ~$70T in debt as of Sept, 2018. That’s the easy part — the harder part is to find each component series that adds up to that total of 70T.
Let’s start with your question about “Total Federal Gov Debt (Non-Security).” If you look at US government debt, you’ll notice two things: 1) the total reported national debt (GFDEBTN) is currently around $21T. However, if you simply use that series and add it up with the other component series, you end up with total debt of $74T (as seen in my live report). So where is the discrepancy coming from? After much research, I realized that ASTDSL and ASTLL only count loans and debt securities and the only way I was able to get the components to add up to $70T was by only including the FGDSLAQ027S series (Federal government; debt securities; liability, Level). However, that is only $17.1T in debt securities. Where is the other 4T? That is the question I was unable to answer, and I was unable to find a series that matched it. Hence, I made an assumption that the remaining government debt is non-security debt, and calculated it as the difference between GFDEBTN and FGDSLAQ027S. As a result, I’m able to reconcile both the component series adding up to the total debt as well as showing a total of 21T in federal debt.
Similarly, the second series “Other (Misc)” is also calculated to account for a minor difference between household debt components and the total debt levels reported in ASTDSL and ASTLL. I was unable to figure out what caused the discrepancy and I figured that it was such a small amount that it wouldn’t detract from the overall analysis.
Since both those series don’t exist, you won’t be able to see the underlying data anywhere. They are calculated for the purposes of reconciling the totals with the components.
I hope this helps, and thanks for your question!
Somewhat flawed analysis. The level of debt today is far higher and more dangerous than it was before the financial crisis.
Both debt series you mention (ASTLL and ASTDSL) have increased since 2009. A 11% increase in ASTLL and a 38% increase in the ASTDSL.
Thus, the only way you can make your argument that debt has been reduced relative to GDP is to use the low point in GDP (2009) and compare it to today’s high point.
Plus, if you add the $4 trillion or so of debt, the Federal Reserve owns which in many cases needs to be replaced with new public and private debt for the Fed to continue reducing their balance sheet. You can see why debt is more dangerous today.
Fair analysis. The $4T in debt owned the Fed will certainly have to be absorbed by the market, which should make things interesting if the appetite for those bonds disappears.
Regarding GDP, the data shows that it has grown faster than debt since 2009, and that’s why the ratio of debt:gdp is dropping. I agree, if we have a deep recession and GDP drops faster than debt levels, we would easily see the ratio rise again, but defaults and other restructuring could cause debt to drop too.
Narayana the critical ratio is the total debt to GDP before the financial crisis started. That ratio is what broke the economy not the rate at the bottom. That ratio is the same now as then. But if you add the $4 trillion of debt the Fed owns that ratio is higher today.
I don’t agree that the Debt/GDP ratio itself broke the economy.
It broke when debt service costs outstripped disposable income. For example, if I have $100 in debt but it only costs me $1/year to service it, I can keep the “party” going for longer than if my debt service costs $5/year. Once the economy as a whole reaches a point where debt service costs are higher than disposable income, then people have no choice but to deleverage through defaults or bankruptcy.
Sadly FRED only has data on household debt service and government interest payments, but nothing on corporate debt interest payments. If I were able to calculate that somehow, I’d be willing to bet that the ratio is much below 2008 since interest rates dropped to 0% for a prolonged period of time.
This is really excellent information.
It looks to me like some sort of combination of the Tech bubble and Personal debt bubble are happening at the same time. The market capitalization of stocks to GDP is very high like 2001 but broader than it was back then so stock pickers won’t have as much room to hide. Most people I talk to say the market is just fine because PE ratios look alright (They laugh at me!). If you take a look at revenue, you can see that companies are converting revenue to earnings at abnormal rates. My worry is that this trend may continue because of the tax cuts pushing up PE ratios (a one time shock that may reverse if the laws expire) and new revenue recognition standards that would increase revenues while increasing earnings by a larger percentage. Those two items combined would convince all of the very focused bull market investors to engage in more buying.
I also see housing prices in a bubble compared to current income levels. You can see that reflected in the monthly inventory change data before year end. That inventory was affected by the spike in 30yr rates in November. 50 basis points had an enormous effect (added about 3 month of housing inventory). I suspect that 100 basis points would all but stop home sales at current levels based on the debt service required.
Based on prior recessions we can see some of the best indicators are an increase in unemployment and an increase in housing inventory by month. I like to use those because they focus on personal earnings and that is the weakest link for the working class. Based on those two indicators, I would pay very close attention to the 30 mortgage rates and how long this government shutdown lasts. Even though the government workers get back pay some of them don’t have enough saved to pay bills.
I wish I had a crystal ball but if I had to guess I would say that in the absence of a rising trend in unemployment or a sharp rise in housing inventory the probability of the market making a sideways move is the most probable thing that could happen for the next year or two. I believe something has to break before the bull market people cave. At this point the market cap is really skewed toward wealthy people, that implies that it can stay irrational for longer than we expect because that demographic doesn’t need money for the basics!
Thanks for your valuable contributions!
Narayana,
Great work, appreciate what you have pulled together.
Thanks
Sweet blog! I found it while surfing around on Yahoo News.
Do you have any suggestions on how to get listed in Yahoo News?
I’ve been trying for a while but I never seem to get there!
Appreciate it
There’s nothing “beautiful” about this deleverage. This is just a small dip over whole 10 years. It’s not getting us anywhere. And when the next recession hits, and it will hit sooner or later, all this progress will be wiped out in an instance.
The deleverage in the 30s-40s may have been brutal, but it was actually effective. When you run such reckless debt policies, there can’t be no half-measures.
Wonderful story, reckoned we could combine a couple of unrelated data, nevertheless seriously worth taking a search, whoa did 1 learn about Mid East has got a lot more problerms at the same time.
Hey ! Thanks for putting this together!
For Corporate debt, what data series did you use? Also, did you include municipalities in your government debt category?
Any guidance would be greatly appreciated!
Hi David,
If you look at the dashboard, go to the second page (see page selector at the bottom). You can see all the series and FRED IDs. The government debt category includes Federal debt as well as State/Local, though note that the latter is combined into one series.
Wonderful story, reckoned we could combine a few unrelated data, nonetheless really really worth taking a appear, whoa did one particular learn about Mid East has got far more problerms also.
Great work, I was trying to look at the Total Debt to GDP ratio but encounter the same problem as you. Thanks for putting all these together
Debt service payment over GDP would be an interesting gauge. Any ideas to find the reliable source for that?
That’s a tough one. Last I checked, FRED only has a series on interest payments on US government debt. I am thinking of a way to create an estimate of debt servicing costs for Corporate/Household debt where I use prevailing interest rates and estimated average time to maturity, but it would be a rough guess at best.
Hi Guys,
I found this website by accident while searching for the debt/gdp levels. It was cool to find some people who read Dalio’s book as well and who spend some thoughts on it. I was not able to figure out one graph though on page 13. It shows the Debt/GDP ratio, amortisation, debt service/gdp & interest payments/gdp.
I d suspect that debt service includes interest payments, but nevertheless debt service minus interest rate payments would equal 200% of GDP (for 2015 or so). Does this mean that 200% debt of GDP is rolled over every year? that seems quite a large number in contrast to the total 330% Debt of GDP burden.
Another question relates to the interest rate burden. It amounts to about 50% of GDP and this would give a yield of 15% on the Debt of 330% of GDP. This number seems also to be quite inadequate.
Maybe somebody got this better and could elaborate on this.
Regards
Georg
Hi Narayana,
How did you get the Debt to Income data for each of the sectors? I see how you split out the debt but the income part doesn’t make sense to me.
Thanks
Income is GDP so I divided total debt by GDP.
Any Corona-Updates so far? I guess debt/gdp ratio must have gone up over 400% again because of it.
Yes, it looks like the Q1 data finally updated on the Fed’s web site. We jumped from 364% to 377%! Should be even more of a jump when Q2 data comes out in three months.
Hi Narayana,
Ray’s book have chart mention “Total Debt (%GDP) which is similar to your chart. What about the “Debt Services (%GDP). Just wondering whether you know / research any about that.
Hi Toshi, great question. See my comment from December 18, 2018 for more info – unfortunately it’s not easy to estimate debt service as % of GDP, though I am thinking to try soon!
Super interesting. Can I subscribe for automatic updates?
Great research,
On the book, there’s also
1. change in Debt-GDP Ratio (ann) and
2. Debt Growth (%GDP, Ann)
Did you figure out if Debt-GDP Ratio means Debt Minus GDP? or Debt/GDP?
Debt-GDP means Debt/GDP in my analysis. Thanks for stopping by!
Use to be an annual publication of “The Statistical Abstract of the United States” which contained the combined government (at all levels less trusts funds), corporate (all businesses less equity), and private (household) debt. The details could be found in the “Banking, Finance, and Insurance Section (Section 25 of the last publication in 2012 based on 2011 data). GDP could be found in the “Income, Expenditures, Poverty, and Wealth” section (Section 13 of the last publication in 2012). The publications made it easy to get the debt as a % of GDP. There are other publication that continued to the present date; but, not in convenient form. The site where all the information resides at is https://www.census.gov/library/publications/time-series/statistical_abstracts.html.