Sovereign-Bank Default Correlation
A review of the IMF's April 2022 Report on Global Financial Stability
The International Monetary Fund (IMF) recently published another issue of their Global Financial Stability Report in April 2022. The report is broken down into 2 chapters, the first focusing on Commodity effects from Russia's invasion of Ukraine - mostly forgettable with a lot leaning on unsound causality. The second chapter is much more interesting, and contains a wealth of data irrespective of any author's biases. This chapter is titled "The Sovereign-Bank Nexus in Emerging Markets: A Risk Embrace". The link to this work can be found here (and given the importance of this chapter, I would recommend a read): https://www.imf.org/en/Publications/GFSR/Issues/2022/04/19/global-financial-stability-report-april-2022. This article is a summary of key figures in an order that I believe presents the purpose of this analytical work in a more direct approach. The raw summation of this work boils down to a few key points:
Banks have increased their holdings of sovereign debt greatly since the beginning of the COVID pandemic in 2020.
In environments of financial stress, banks and sovereign default likelihood increases in a correlating manner.
Bank and Sovereign default likelihood increases as public debt level increases and Bank-Sovereign exposure increases, presenting a negative feedback loop for Banks holding sovereign debt as that debt increases in ratio and size.
Sovereign debt rating decreases correlate with an increase in non-performing loans over time as bank-sovereign exposure increases, showing a 2-5 year lag in metric response.
The most recent stress test for banks was done at above-historical average values for stressful conditions, not simulating a real stress event, and that a stress event according to these norms would be potentially catastrophic for exposed banks.
Figure 1 is fairly simple, a theoretical framework for how the IMF staff view the Sovereign, Banks, Corporate Sector, and Foreign Investors in their working models. The figure works pretty well. I am curious to how the IMF truly view Foreign Investors in the grand schema, as they are shown out of the system yet the major cause of problems in tightening financial conditions
Figure 2 is a simple chart analyzing historical causes of Crises in Emerging Markets vs Advanced Economies. Emerging Markets have frequent Currency related stress events, where as Advanced economies tend to have Banking-sourced crises. Obviously government stability plays a huge role as well, with a significant frequency of related crises.
Figure 3 is the IMF's quantitative proof that banks have increased Government debt holdings. While I agree with the assertion, the raw data is fairly obvious in showing a mandated-increase in government debt holdings across the board. This was one of the big changes pushed post 2007-08 Financial crisis, the Basel III accords are supposed to control bank leverage ratios via forcing collateral held in government debt. Aside from there, the sheer volume of government debt pushed post-COVID in an effort to keep the entire system running offers some weight, not just to the figure, but the arguments being made inside the text. But this figure isn't swell as the keystone of the entire work.
Figure 4A shows an increase in correlation between Bank and Sovereign defaults as Bank Stress increases. 4B is cited by the authors as proof that Sovereign and Bank default risk moves together via historical trends. Specific numbers of defaults are fairly low, but this is a raw chart. I would love to see the IMF authors use some alternative models to pull apart who best follows that trend, and not.
F 5 is where things get sweaty. This one makes sense: Sovereign governments with high public debt ratios are going to be more sensitive to tightening financial conditions, and so banks holding high levels of this debt are, too. However, this is stratified by High public debt levels and Average public debt levels, only of Emerging markets. This author would like to remind readers of the increasing Public debt ratio for the United States and Eurozone economies since 2020, and that AE have reduced frequency of financial crises, but they do have them. The second half of Figure 5 is the corresponding bank default risk trends, showing high bank-sovereign exposure level increases Bank default risk following stress shocks. Figure 6 specifically looks at Sovereign downgrade effects in Investment quality. The first half is logical, the greater the risk the greater the effects post-downgrade. Most of the figures popped out by IMF for this work are logical, that is generally how data works. However, I want to stress the lagging effect on Nonperforming loan ratios. While it may feel like Ratings changes are always ex-ante, post-event and irrelevant, the nonperforming loans triple via historical trends 2 years on. In a pressing environment of Sovereign debt stress leading to Ratings changes, there are countries that have been burned in FX markets over the last year that may not be done feeling the pain.
Figure 7 shows the effects on Bank growth over 5 years post Sovereign distress events. Normal times shows a standard ~1-4% growth, where sovereign distress events only cause major growth effects >3 years from event. Furthermore, the figure shows negative growth to all banks regardless of capitalization standards. The second half of Figure 7 is more relevant to the accompanying discussion of this data, but nonperforming loans dramatically increase for only poorly capitalized banks post stress event, but data from the 2007-08 GFC shows the average non-performing loan rate for US banks reached 4-5% in 2009-2011. While the IMF are clearly suggesting modern megabanks should be immune to these effects, historical data disagrees. And that's how I lead into Figure 8; on a general basis, either a 5% decrease in equity and loans following monetary inflation events, or a 15% decrease when there is stress in economies with higher expected maturing debt.
Figure 9 shows that historical averages of stressful conditions are more severe than the conditions used for the recent stress test, and that an actual test using appropriate numbers would result in failures of key institutions.
Right off the bat, I want to extend my professional and personal gratitude to the IMF staff that wrote and published this work. While the next segment of this article might be critical, I argue that the authors work is relevant and necessary for the present. My base problem with this work isn't so much as what is in it, but what's not. The authors have a difficult task of comparing global economies through time on general quantitative variables, a task that is more like trying to compare potatoes to tactical nuclear missiles rather than apples to oranges. Comparing America to Azerbaijan is difficult enough, but comparing America in 2020 to America in 2006 is a stretch in itself. My main issue with the work is inappropriate stratification and scaling of appropriate subjects. Primary to point, the authors' focus on post-COVID is relevant, but limiting back data to 2005 with variable party representation is faulty. If 2005 is the earliest time point with relevant data, then so be it, but in assumption of future monetary and fiscal policy, limiting trends to the same phase period is where faulty models lead to massive failures rather than their designed operating margins. Sourcing is peculiar, utilizing a wide array of public and private institutions is fine, but quality checking is absent. Citing failures by rating agencies to diagnose system issues during the Great Financial Collapse is enough to warrant refining data sources to ensure quality of model. The final issue is benchmarking, resulting in a failure for results to scale. I accept that Macroeconomics involves using a lot of inputs and outputs that don't mesh perfectly, but the axes of the figures throughout this work are in question, leading to the only sure take away as a quantitative up or down without scale.
The second half of this article has figures and data that illustrate what the IMF have propositioned as risk for emerging markets, is relevant to advanced economies as well. Defining variables makes it hard to reason out why Sovereign debt from advanced economies and emerging markets should be so naively separated. I'm unaware of a different set of rules for many of the Advanced Economies on the IMF's list that specifically bars financial crises, given that many of them are undergoing massive rates of inflation currently. The results can be summarized as "Banks exposed to high amounts of sovereign debt for nations with high public debt ratios exhibit stress events and defaults at a higher frequency, and so does that sovereign". The effects on those banks are an increase in non-performing loans, a decrease in equity and loan value, and an increasing likelihood of default. Post-2008 GFC US banks saw an increasing Non-performing loan rate similar to the trend presented by the IMF here. Continuing, their losses in equity and total loans were greater than the average. Furthermore, the US exhibited the described increasing public debt to GDP ratios as seen below in Figure 10.
Figure 11 expands on this by showing some of the major Eurozone economies and their public debt ratios compared to the IMF's average for AEs and EMs. The authors include an altered version of this in their work, but theirs mutes the effect. Figure 12, from the IMF text, shows AE/EM total amount in dollars increases relatively similar through time, both ~50%. According to IMF classifications, there are 39 AEs and 161 EMs (China being one of them). I do not respect the conditional placed on the data that this is an Emerging Market property rather than a mechanical one.
Take an environment of financial tightening conditions, i.e. one where interest rates are increasing, liquidity decreasing, thus cost to borrow increases as difficulty to borrow increases. As interest rates increase, so will sovereign debt yields. Figure 13 (below) is the US 2/5/10/20 Year yield and Federal Funds rate. The Federal Funds rate typically serves as the floor for yield, which makes sense. That means that in periods of financial tightening conditions, the cost to borrow will increase and any sovereign debt will have to deal with that. Does this mean that the Federal Reserve creates a problem for the Treasury? Possibly, given that Treasury Secretary Yellen has recently spoken against the Interest rate increases by Federal Reserve Chairman Jay Powell. But either way, any Bank with high exposure to a sovereign dealing with increasing costs to borrow on increasing volume to borrow (high public debt) is not immune to the symptoms defined. The part most interesting to me is that even after the crisis has begun, the full weight won't be felt for 2-5 years, meaning crises can stack very quickly.
Figures 14 and 15 review current Nonperforming loans and historical NPLs in the EU, where Figure 16 illustrates the US trend since 1995. There are a lot of key differences in rules and regulations that drive the nonperforming loan rate delta from the Dotcom bubble and the GFC. However, it does suggest that there is some key modifier to this behaviour that needs to be studied and appropriated for usage again if the need arises. Given the continuing environment, it is likely to be needed again. In summary, the IMF report a trend correlating Sovereign debt, Bank exposure, tightening financial conditions, and increasing risk of default, increasing quantity of nonperforming loans, and recession of bank assets. This trend is hyper-relevant to current conditions. While the IMF suggests this is an Emerging Market specific phenomena, I believe the presented historical data argues this is a mechanical property of the current system architecture.