AS Global Debt Chartbook, Q4-2024 - How long is a piece of string?
I’ve been sitting on this project for a while, but I’m finally ready to bring it above the fold. I’ve long wanted a straightforward overview of global debt levels—both public and private—and an easy way to compare them across countries, alongside their respective external dynamics. This is essential material for macro investors and researchers, yet it’s rare to find all the relevant information compiled in one place. The AS global debt chartbook is a first attempt at this. Like the LEI Chartbook, this project runs on Python code generated and compiled with the help of my trusty OpenAI assistant, with a few manual adjustments along the way. At the moment, it draws data from an Excel spreadsheet, but integrating APIs should be relatively straightforward down the line.
The chartbook uses data from the BIS for debt levels and debt-service ratios—expressed as shares of GDP—for both the private and public sectors. It also incorporates OECD data on net international investment positions (NIIPs), nominal GDP, and current accounts. I’m using market value indices from the BIS, which typically result in higher reported debt levels compared to nominal or par values, as it adjusts down the price of debt outstanding with higher interest rates. However, this methodological choice should not affect cross-country comparisons, as long as it's applied consistently. That said, using market values significantly reduces the country sample in the total debt calculations, primarily because market value data for government debt is often missing outside the largest developed economies.
By their nature, these kinds of data are long-lagged. BIS debt figures are usually delayed by two to three quarters, while NIIP and GDP data lag by one to two quarters. One obvious extension to the methodology would then be to estimate these variables more frequently, but since most of them evolve gradually, I’m not convinced it would be worth the additional work. Moreover, the more detail you include, the greater the risk of shrinking the country sample due to data gaps. Other potential enhancements could include variables such as the ownership composition of debt, rollover schedules, currency mismatches, and the asset breakdown of net foreign positions. Each of these, however, increases the likelihood of sample attrition—though some, like rollover schedules, should be feasible. I am open to all suggestions on how to improve data timeliness and add more variables.
With that out of the way, what do the data tell us at the end of 2024, and what can we infer about the trajectory so far this year?
levels vs changes
Asking whether a given debt level is “high” is a bit like asking how long a piece of string is. But in this case, it’s fair to say the string is quite long, for a lot of countries. The first two charts in the chartbook highlight its core framework: they plot the twin deficit on the Y-axis—the sum of the budget and current account balances as a percentage of GDP—and private as well as total debt (also as shares of GDP) on the X-axis. Adding to the NIIP to the total debt calculation introduces double-counting, but also adds a perspective on debt ownership. In future versions of this chartbook, I will attempt to plot the NIIP on a third dimension, or perhaps enhance the visual element to show the size of the NIIP.
Twin deficits everywhere!
Higher term premia needed?
The first chart shows that twin deficits are now widespread across the G20, here defined as trailing current account deficits and projected budget deficits through 2026 in the IMF WEO. Many countries are running twin deficits around—or well above—5% of GDP. Only the usual thrifty suspects, such as Japan, South Korea, and Germany, remain in the black, mainly due to still-solid external positions.
When we match these numbers to debt levels, the U.S., France, and the U.K. clearly stand out as points of vulnerability in the developed world. The second chart zooms in on the largest developed market economies, by virtue of the shrinking sample for government debt at market value, using a broader measure of total debt. Consider the U.S. position: as of the end of 2024, Uncle Sam was running a twin deficit of nearly 10% of GDP, with total debt at a staggering 500% of GDP, which includes a negative NIIP just under 90%. Talk about “your dollar and your problem.” France and the U.K. also show precariously high twin deficits combined with elevated domestic debt levels, while Italy, Canada, and Australia occupy an intermediate position.
Looking ahead, there’s little reason to expect meaningful improvement in the near term. In the U.S., Mr. Trump is attempting to reduce the trade deficit through tariffs—essentially a form of domestic tightening via taxation. But since the administration seems unwilling to accept any accompanying economic slowdown, fiscal policy is loosening, likely far more than tariffs will generate. To complicate matters, the Fed may soon be led by someone who favors aggressively looser policy, due principally to accommodate the rising federal interest bill. In this context, I'm not optimistic about any near-term improvement in the U.S. debt-growth nexus.
In France and the U.K., domestic politics also seem unlikely to deliver near-term relief. Even within the so-called Club Thrifty, cracks are emerging. In Japan, rising inflation—and concerns about long-term debt sustainability—have pushed up bond yields. Meanwhile, Germany is beginning to price in a sustained fiscal loosening to support defense and infrastructure spending.
Judging by the numbers in the charts above, a crisis of confidence in DM government debt no longer seems far-fetched. One could reasonably expect persistently higher risk premiums on long-dated debt, steeper yield curves, and, in short, pain and suffering for duration.
However, a closer look at changes in debt ratios presents a somewhat more sanguine picture. Specifically, strong nominal growth and inflation have masked a number of fiscal vulnerabilities—a dynamic that would be even clearer if the charts were constructed using nominal par values. In fact, private sector debt levels have been falling steadily in many developed markets and have reached multi-year lows in some cases. Even in the public sector, which bore the brunt of Covid-related fiscal expansion, debt as a share of GDP is off its highs in many countries, though it is now rising again in many countries, and will continues to do on the current fiscal trajectories in key developed markets.
This leads me to two overarching conclusions:
High debt levels and large macro-deficits are back, but unlike the run-up to the global financial crisis, the public sector is now the culprit, not the private sector. This suggests a scenario of gradual erosion rather than sudden collapse—a death by a thousand cuts, unless bond vigilantes target one or more high-profile countries. Yes, I’m looking at you, France and Germany. Japan represents a special case, where “enemy-from-within” vigilantes could eventually trigger a spectacular reversal if and when things spin out of control.
Governments and central banks now desperately need Goldilocks to persist: inflation and nominal growth must remain high enough to erode debt levels and stave off a debt-deflation trap, yet low enough to maintain the illusion that monetary policy is committed to an inflation target, and that it would never abandon such a target to allow significantly above-target inflation to burn away the debt. Any deviation—whether too hot or too cold—could shatter that delicate balance. Maintaining this equilibrium could prove exceedingly difficult in certain regions and countries.