Jonathan Anderson

jonathan@[Link]
May 3, 2023

How to Think About China (2023 Edition)

Part 5

• Part 1 – Lies, Damn Lies and Statistics

• Part 2 – A State Economy or a Market Economy?
• Part 3 – Explaining the Miracle
• Part 4 – The Most Important Sector in the Universe

• Part 5 – Debt Boom and Debt Bust

• Part 6 – Is There a New Growth Model?
• Part 7 – The Future of the Renminbi
• Part 8 – The Aging of China
• Part 9 – Commodities, Food and the Next Decade
• Part 10 – How to Follow Macro Policy
 Jonathan Anderson
May 3, 2023

Introduction and summary

In this installment of the How to Think About China series, we want to stress five key points:

1. A big debt doom. First, in the first half of the previous decade China had a massive debt explosion.
Between 2009 and 2016, overall financial debt levels rose by more than 100 percentage points of GDP, which
is not only epic but also a record-setting increase by emerging market standards. The debt party was heavily
driven by policy-related stimulus lending, but also by very lax regulatory environment where small/medium
banks and NBFIs were able to expand balance sheets for years with impunity.

2. But never a formal crisis. As big as the boom was, however, it never took China to the “tipping” point of
serious financial or macro crisis. A lot of factors helped here, including China’s closed capital account, lack of
external exposures, high domestic saving rate, a very safe starting point and the “virtual” nature of much of
the borrowing (about which more below). Things were definitely going in a bad direction, with rising risks –
but the party ended before outright disaster could strike.

3. Ended in 2016. Which brings us to the third point: the party did end, more than six years ago. This is
commonly misunderstood by a surprising number of investors, who assume that China is still in the midst of
an unabated expansion. In reality the government made a wrenching shift in policy at the end of 2016,
ending the quasi-fiscal credit frenzy and forcing small and medium banks to deleverage balance sheets.
Lending activity fell off sharply, with debt ratios peaking that year and generally stabilizing since.

4. Have already seen the main “bust”. Because debt ratios are still high, many investors also assume that
China is still on the verge of a looming debt “bust” – but the reality is that much of the post-boom shakeout is
already behind us as well. Soon after the policy turn, real growth momentum collapsed from double digits to
below 3% (and the downturn was even worse in nominal terms). The domestic property market stalled.
Dozens of banks went into de facto bankruptcy and restructuring; the basic materials sector suffered heavy
losses and a big cleanup as well; bond defaults and bad debt write-downs rose. All of these are ongoing
processes that will continue to weigh on growth prospects, but objective macro risks have fallen rather than
risen in China.

5. The one remaining hidden risk factor. With one exception: i.e., the long-term outlook for the renminbi
exchange rate, as growing tension between the ever-expanding stock of domestic monetary liquidity,
stagnant FX reserves, a quasi-pegged currency regime and a closed capital account threaten to destabilize the
renminbi before the current decade ends. We will discuss this further in the upcoming Part 7 installment of
this series.

2
 Jonathan Anderson
May 3, 2023

1. The debt boom

For a “one-stop” summary of the great Chinese debt boom, here’s the chart:

Chart 1. The debt boom

Share of GDP
350%
Total debt
300% Domestic financial claims

250%

200%

150%

100%

50%

0%
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Source: CEIC

The yellow line above is our broad estimate for overall debt as a share of GDP, and the blue line shows the
share of claims by the domestic financial system (see footnote below for details on both measures). 1

As shown, from an average level of around 170% of GDP in the 2000s, total debt outstanding first began to
jump sharply in 2009 and continued in an upward spiral for the next eight years, reaching 300% of GDP by
early 2017. And virtually the entire increase came from domestic lending; local financial system claims on the
rest of the economy rocketed from 140% to nearly 260% of GDP in the same time frame.

Are these big numbers?

Ahem, yes.

It’s not about debt levels. When we say that, mind you, we’re not really talking about debt levels. Overall
debt of 300% of GDP is certainly high by EM standards, nearly twice the broad level in the rest of the
emerging universe using the same comparable metric (Chart 2 below). But it would still be on the low side for
a DM country – and of course China has moved rapidly up the relative income scales over the last 30 years
and has one of the highest national saving rates in the world. As a result, we could argue back and forth all
day and still not reach a consensus as to what constitutes “too much” debt for the mainland economy.

1 The yellow line is our “apples-to-apples” comparative estimate for total debt outstanding used across all countries we follow,
defined as the sum of (i) domestic claims of the financial system on the non-budgetary economy, (ii) overall public debt, including
domestic and external liabilities, and (iii) private external debt. The blue line shows estimated domestic financial claims on the entire
economy, including government as well as households and corporates.

3
 Jonathan Anderson
May 3, 2023

Chart 2. Debt levels in comparative context

Total debt (estimate, % GDP)

500% DM
450% China
EM ex-China
400%
350%
300%
250%
200%
150%
100%
50%
0%
03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22

Source: CEIC

Sure enough, in Part 6 of the accompanying How to Think About Emerging Markets series (An Austrian
Primer: Savings, Debt and Default, 16 October 2022) we found that absolute debt levels don’t really tell you
much at all about underlying macro stress or risks in the EM universe. There are a host of factors that can
influence a country’s “equilibrium” level of debt: Higher-income economies, for example, have a larger
capital stock and more developed financial systems, and thus tend to have both higher levels of assets and
liabilities as a share of GDP. The same is true for smaller open trading economies, and also true for countries
with a higher level of structural savings, etc. etc. As a result, trying to compare headline debt ratios across EM
(or DM) turns out to be a rather useless exercise in terms of predictive power.

Rather, it’s about the delta. Instead, what really matters is the change in debt over time. Regardless of the
starting level, we consistently find that countries with a sharp run-up in debt ratios over a given time frame
run into subsequent trouble: big growth slowdowns, outright recession, protracted deleveraging pressures or
in some cases financial crises. Simply put, debt booms lead to debt busts in one form or another; this is both
an emerging and a global regularity. Which is why we use debt deltas rather than debt levels, for example, in
our regular EM macro risk indices.

And look where China sits on a delta basis. Chart 3 below shows comparative data for credit booms across
major emerging markets, defined as the largest trough-to-peak increase in the domestic credit/GDP ratio
over a 10-year period since 1970, using the World Bank’s WDI database:

4
 Jonathan Anderson
May 3, 2023

Chart 3. Comparative credit booms

Trough-peak increase in domestic credit/GDP, past 50 years
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
China

Turkey

Peru

Mexico
Latvia

Malaysia

Singapore

Saudi

Taiwan
Korea

Chile
Hong Kong

Israel
Ecuador
UAE

Lithuania

Slovenia
Estonia

Hungary

Brazil

Slovak

Colombia

Nigeria
S Africa
Indonesia
Czech
Bulgaria
Ukraine

Croatia

Romania

Poland
Thailand

Morocco

Russia

India
Vietnam

Kazakhstan

Philippines

Argentina
Egypt

Source: World Bank

You get the point: China’s boom was absolutely huge.

Not only is China at the very top of the list, but all the remaining countries that even come near to the same
mark are very small open economies like Latvia, Hong Kong, Bulgaria, or at most Thailand, Vietnam and
Malaysia. A credit boom of that magnitude is unheard of for large countries – just look at Brazil, India, Russia,
Indonesia or Mexico, with an average peak increase of 35% to 40% of GDP. However you measure, China
stands virtually alone.

How did this happen?

How did this happen? How did China wake up with one of the biggest headline debt booms in EM history?

There are two closely connected answers to this question:

• Falling growth vs. unrealistic targets. First, because the fading structural growth environment of the
2010s forced the government into repeated rounds of hurried policy stimulus in order to meet broad
macro targets.

• Lack of information on what was really going on. And second, because the government had no clear
idea what was happening on financial and quasi-fiscal balance sheets and thus no idea how big the
debt boom actually was.

Fading growth and policy stimulus. Let’s start with the first issue, i.e., the changing growth backdrop and
policy response over the past decade.

Remember from Part 3 of this series (Explaining the Miracle, 27 February 2023) that China’s world-beating
economic boom of the 1990s and 2000s was overwhelmingly driven by two sectors: exports, which began to
explode in the early 1990s and transformed the mainland into the world’s manufacturing workshop, and

5
 Jonathan Anderson
May 3, 2023

property, which burst onto the scene a decade later and quickly turned China into one massive construction
site at home.

But here’s the thing: By the early 2010s, it was over.

As discussed in the earlier report, China’s share of global industrial imports rose like a rocket from less than
5% in the 1990s to nearly 25% by the end of the 2000s … and then went absolutely flat from 2011-12
onwards, as high labor costs made low-end manufacturing increasingly uncompetitive and factories in
traditional consumer industries left for Vietnam, Cambodia and elsewhere in the region (see Chart 4). Market
share jumped again in 2020 due to Covid disruptions in developed market producers, but has since gone flat
once more.

In the property and construction sectors, as well, activity shot up dramatically throughout the 2000s, more
than doubling as a share of real GDP by the peak in 2013 … and then went into reverse. The absolute level of
sales, construction and materials usage more or less stagnated for the next seven years, and the implied
share in real activity posted a steady decline (Chart 5).

Chart 4. Export market share Chart 5. Property and materials intensity

Share of total industrial imports (sa 3mma) Intensity index to real GDP (2005=100 6mma)
35% 140
130
30%
120
25% 110
100
20%
90
15% 80
Starts/completions/sales
10% 70
Steel/cement usage
60
5%
50
0% 40
000102030405060708091011121314151617181920212223 0001020304050607080910111213141516171819202122

Source: CEIC Source: CEIC

Even if we do a very rough, simple estimate for the implied “all-in” contributions of exports and property to
real GDP growth (including direct and indirect linkages, see footnote for details) we get a chart like the
following: 2

2 For the property sector, we take the average two indices in Chart 2 above and use the resulting annual growth rate to calculate the
implied growth contribution, benchmarking the “broad” property sector (including indirect linkages to autos, appliances and
property-related infrastructure) at 25% of GDP in the peak years between 2006 and 2013, in line with our findings in the previous
Part 4 report of this series. For exports, we do the same thing using export volumes, benchmarking the peak share of direct and
indirect demand in the economy at 20%. These are both rough, “finger-in-the-air” estimates – but the key point here is that
regardless of the actual share, the resulting trend would look exactly the same as in the chart below.

6
 Jonathan Anderson
May 3, 2023

Chart 6. Export/property contributions to growth

Implied contribution to real growth (pp)

10%

8%

6%

4%

2%

0%

-2%

-4%
00 02 04 06 08 10 12 14 16 18 20 22

Source: CEIC

You see our point. Between 2003 and 2010 exports and property were massively important, together
contributing roughly 8pp to annual growth in the 2000s and a whopping 10pp in 2010. By 2011, however,
the export contribution had died out, falling to less than 1pp on average over the ensuing eight years – and
by 2013 the same was true for property and construction.

Sure enough, here are our best estimates for overall Chinese growth during the same period. The yellow line
in Chart 7 shows the path of official real GDP growth, while the blue line shows implied growth according to
our own China Activity index (introduced in detail in Part 1 of this series):

Chart 7. Real growth indicators Chart 8. Nominal growth indicators

Real growth (% y/y 3mma) Implied nominal growth (% y/y 3mma)
20% 30%

25%
15%

20%
10%
15%
5%
Official GDP 10%
Activity index
0%
5%

-5% 0%
95 00 05 10 15 20 95 00 05 10 15 20

Source: CEIC Source: CEIC

Notice anything? If you just look at the official real GDP path, the post-2010 slowdown doesn’t look too bad;
the economy was expanding at 10.5% annually in the 2000s prior to the 2008-09 crisis, and after the big 2010
recovery still managed to grow at 7.5% y/y between 2011 and 2019.

However, the Activity index shows a very different picture. For starters, the boom years were much bigger,
with average growth of 14% y/y in the years leading to 2008. And the subsequent slowdown was much more

7
 Jonathan Anderson
May 3, 2023

pronounced. By 2014-2015, for example, when the official figures were still showing real growth of more
than 7%, the Activity index had already dropped to under 3%.

This is already a tremendous deceleration in its own right – but that still doesn’t compare with the drop in
nominal momentum in the economy, as shown in Chart 8 above. And here it really doesn’t matter that much
if you use official GDP numbers or our own proxy index. Either way, “China Inc.” was growing at eye-popping
pace of nearly 20% y/y in the pre-crisis 2000s and again in the early 2010s. And either way, nominal growth
collapsed to the mid-single digits by the middle of the decade.

This was an unexpected and profound shock for the Beijing leadership, which had gotten used to years of
high, stable growth during the full-blown “miracle” export- and property-driven period. Or, better put, a
series of shocks, as outlined in the next two charts below.

The first came in 2008, which the combination of big earlier policy tightening at home and the sudden
explosion of the US financial crisis led to an outright collapse of growth momentum. On our index, both real
and nominal growth dropped to zero. Export volumes tanked by 20% to 30%. For the first time since the
housing boom began, domestic property sales contracted outright (see Chart 9), and for the first time in
more than a decade corporate profit growth turned negative as well (Chart 10).

Chart 9. Property sales Chart 10. Corporate profits

Property sales growth (% y/y 3mma) Profit growth (SOE/industrial avg, % y/y, 3mma)
80% 80%

60% 60%

40% 40%

20% 20%

0% 0%

-20% -20%

-40% -40%
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19

Source: CEIC Source: CEIC

In short, this was a state of macro emergency, with no clear sense of when and how the global economy
would recover or how long the downturn at home might last. In response, the Chinese government pulled
out all the stops and pushed through its first, massive policy stimulus package at the end of 2008 to get things
moving again. (We’ll spend a lot more time talking about the nature and specifics of that stimulus shortly.)

By 2010, of course, it was clear that the global economy had roared back faster than expected, and also clear
that the government had hugely overshot the mark at home. Real growth was in the low teens officially but
20% y/y on our index, and nominal growth was a stunning 25% y/y at the peak.

This led to the next round of sharp domestic policy retrenchment … at the same time as the global recovery
petered out, with exports contracting once again and the EU back in recession. Looking back at the above
charts, by the end of 2011 Chinese growth had tanked again, losing 10-15pp of nominal momentum virtually

8
 Jonathan Anderson
May 3, 2023

overnight. The local property market was back in recession, and corporate profits were falling in absolute
terms.

Once again, the government got scared, and once again they eased in a hurry, pushing credit and spending
through the system.

This worked, with a rebound in property demand and corporate profits at home. However, as shown above,
it didn’t take China anywhere close to where it had been before. Global demand was still muted, exports
were flat and commodity and other upstream prices were falling. In this environment, renewed stimulus
allowed China to meet official growth targets … but barely, with real growth of 6% to 7% y/y and nominal
growth that was no higher.

And then, in 2014-15, another sudden downturn. On the external front, the oil market collapsed, leading to
big declines across the commodity sector; global trade was also contracting in volume terms, putting
pressure on China’s export sector. And the domestic construction and property market had gone into
recession once again, taking corporate profits with it.

There wasn’t much impact in the official Chinese GDP data, but boy, you could see the impact in the
alternative figures. Both real and nominal growth in our Activity index tanked right back down to 2008 levels,
which were the weakest on record in the past 25 years (Charts 7 and 8 above). By all bottom-up accounts, the
economy was flatlining, and the local rumor mill was rife with speculation about how bad the underlying
situation really was.

This was also, importantly the first downturn on President Xi Jinping’s watch; who had taken over the reins
shortly after the end of the previous cycle and was staking the success of his leadership tenure on a doubling
of real incomes by 2020. Sure enough, by mid-2015 Xi had seen enough, and hastily called for another round
of aggressive pro-growth policies.

What we talk about when we talk about “stimulus”

In short, this is the first part of the story, i.e., “why” policy stimulus. Chinese leadership was reacting to a
combination of adverse global conditions and the end of the country’s high-growth years at home. As long as
the economy was riding a massive export- and property-driven expansion wave there was never a need to
provide an extra policy push during the 2000s. But from 2008 onwards, the government repeatedly found
itself in situations where it had to play the debt card in order to meet growth expectations and keep things
moving.

But this is only half of the picture. To fully understand how China ended up with so much debt, we have to
talk about the “how” of stimulus. And this involves more detailed discussion of who was borrowing, who was
lending … and how the authorities kept track (or, in this case, failed to keep track) of what was happening.

Who was borrowing? So again, returning to the very first chart above, overall Chinese debt levels jumped by
more than 100% of GDP from the late 2000s through the mid-2010s. Where did debt this go? Who was doing
all the borrowing?

9
 Jonathan Anderson
May 3, 2023

Ah. Let’s have a look. Here is a copy of the domestic claims line in that first chart, but now broken down by
borrower category.:

Chart 11. Credit outstanding by borrower

Total domestic financial claims (% of GDP)
300%
Government
250% "Quasi-fiscal"/other
Household
Indentified corporate
200%

150%

100%

50%

0%
03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Source: CEIC

The blue bars at the bottom show estimated credit to the “formal” enterprise sector, estimated using
reported balance sheet data on debt liabilities for (i) all SOEs, (ii) industrial, trade and construction
enterprises, and (iii) all listed enterprises, i.e., a best guess on debt/GDP trends for both state and private
firms. As you can see, there simply wasn’t that much happening; corporate debt penetration expanded in the
first half of the 2010s, but this was a moderate fraction of the overall picture, rising only 20pp of GDP from
the 2007-08 trough and even less when measured from the early 2000s.

There was clearly more going on with the household sector (in green), but as shown – but as shown, much of
the real action with household borrowing occurred after 2016, with consumer debt levels rising over the past
seven years even as the total debt ratio has flattened out. If we take household borrowing during the 2009-
16 boom itself, we find that this segment also contributed around 20pp to the overall increase.

What about the remaining 65% to 70% of GDP? Well, some 15pp came from the yellow bars, i.e., direct
borrowing by central and local governments. But the rest – more than 50pp of GDP, or roughly half the total
debt increase during the period – came from the unprecedented explosion in the red bars, defined as the
unexplained residual once all household, traditional corporate and formal government claims are accounted
for.

What is this “unexplained residual”? As we discussed in Part 2 of this series, this is where China’s “quasi-
state” sits: a bewildering hodge-podge of generally local government-oriented administrative and corporate
entities such as local finance vehicles, development funds, infrastructure and construction firms, public-
private partnerships (PPPs), etc.

All of these bodies have a few things in common. First off, in contrast to their traditional SOE counterparts,
most of whom have been operating for decades, they were generally not pre-existing firms. Instead, most
were newly created out of thin air specifically for policy stimulus purposes. (And we should stress that they
are rarely recorded in any available corporate database, nor are they counted in the standard register of

10
 Jonathan Anderson
May 3, 2023

state-owned or controlled firms maintained by SASAC or the NBS; they are almost completely opaque,
existing outside of the sectoral and macro metrics we follow.)

Second, as a result, they had no revenue or profit streams, and only the most tenuous asset structure; local
governments or SOE parents usually inject some state land or other commercial assets in order to facilitate
bank borrowing.

Third, when they could, these entities did borrow … a lot. With the aid of a bit of asset collateral, implicit
guarantees from local governments and a free-for-all banking environment (about which more below), some
of these borrowers historically took on debt loads 10 or 20 times their notional equity base. And again, with
no current revenues or business income to speak of and often only the sketchiest of plans for future
infrastructure projects.

And finally, they were big. Each individual entity tends to be small, mind you, far smaller than any “serious”
traditional SOE, but with many tens of thousands of them cropping up all over China during boom periods,
they collectively accounted for a large share of overall credit activity.

Who was lending? That’s the borrowing side. Now let’s turn to lenders. Our next chart, reprinted from
financial analyst Crystal Cheng’s work, show the size of China’s financial system between 2008 and 2019,
aggregated by gross assets as a share of GDP.

Chart 12. Credit outstanding by borrower

% GDP
600% NBFIs
Bank's off-b/s
Small/medium and other banks' on-b/s
500% Big 5 banks' on-b/s
PBC

400%

300%

200%

100%

0%
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019

Source: CEIC

The dark blue bars show the PBC balance sheet; the light blue are assets of China’s “big five” state
commercial banks that historically dominated the system, and the remaining small and medium commercial
banks are in green. We have also added banks’ off-balance sheet asset positions in grey, and relevant assets
of non-bank lenders (NBFIs) such as trusts, securities and asset management firms in yellow.

What is the chart telling us? Two things.

First, the magnitude of the credit and debt party was even bigger when measured this way. Gross financial
balance sheets were 300% of GDP in 2008, on the eve of the boom; by end-2016, they had exceeded 500% of

11
 Jonathan Anderson
May 3, 2023

GDP. That’s a quadrupling in absolute nominal terms and a near-doubling as a share of the economy … all
within the space of eight years.

Second, and equally if not more important, look where all the lending came from. Not from the “traditional”
banking sector, i.e., the PBC and the big five state banks; together their GDP shares actually fell over the
same period. Rather, the entire boom came from an absolutely madcap explosion in the “periphery” of the
system: small and medium banks, off-balance sheet activity and NBFIs. Taken as a group, their balance sheets
didn’t just quadruple; they increased seven-fold in nominal terms from the late 2000s onward.

You can see this vividly by comparing the two pie charts below; the one on the left showing the asset
composition of the financial system in 2008 and that on the right in 2017, as a share of GDP:

Chart 12. China financial structure 2008 Chart 13. China financial structure 2017

Central
bank,
Big five
44.2%
Central banks,
bank, 113.1%
64.87% Big five NBFIs,
banks, 103.1%
108.34%
NBFIs, 26.71%

Bank's off-
b/s, 22.78% Small/medium
Small/medium Bank's off- and other banks,
and other b/s, 65.1% 179.4%
banks, 92.12%

Source: Various official, EM Advisors Source: Various official, EM Advisors

Again, the story is all about the ballooning shares in the green, yellow and grey segments.

Did Beijing even know what was going on? Allow us to pause and summarize what we have so far. To wit,
massive borrowing by opaque, unregistered “quasi-state” entities at the local level … overwhelmingly funded
by an explosion of small banks and “shadow” non-bank lenders in the system.

If this sounds like unstructured chaos, well, yes, that’s exactly what it was.

And it leads us to two crucially important but all-too-often misunderstood facets of Chinese macro policy
over the past 10-15 years.

First, “official stimulus” is not about planners in Beijing carefully doing math and allocating funds to a handful
of big national projects, and it’s not about expanding the size of central or even local government budgets.
Rather, the way things worked in 2009, and again in 2012 and once again in 2015, was that Chinese
leadership sent the word down to localities to “get things done”, i.e., regional and local governments should
find ways to boost growth at all costs, banks should support whatever lending necessary to facilitate local
initiatives – and all on the tacit understanding that there wouldn’t be too many questions asked.

12
 Jonathan Anderson
May 3, 2023

In other words, a wrenching removal of normal regulatory and policy restraints in the system.

Which brings us to the second point: The authorities in Beijing clearly didn’t have a firm idea what was going
on, neither in the details of borrowing and lending activity nor the overall magnitude of the resulting debt
boom.

This was true almost by definition from a bottom-up perspective. Again, the fact that stimulus was carried
out not through formal budgets but rather a hidden mass of unreported quasi-commercial structures was
specifically aimed at obfuscating the truth of what was happening on the ground. Indeed, in the aftermath of
the first 2009-10 policy package, it wasn’t until much later, in 2012, that the central government was able to
get even the most rudimentary picture via its first nationwide audit of LGFVs and other off-balance sheet
entities. And each successive stimulus round has led to a similar process of ex-post “catch-up”.

But that’s not all. It was also true for the very macro-level financial data policymakers were using at the time
to track was happening in the overall economy. We discussed this issue in Part 1 of this series, but with
apologies to regular readers we need to repeat for emphasis here.

From the 1980s, when China’s modern commercial banking system first came into existence, through the end
of the 2000s, the PBC and the NBS basically tracked one financial time series: standard bank loans (see the
yellow line in Chart 14 below). In an environment where there were only a few large banks, with limited
financial diversification and a firm regulatory hand, monitoring loan activity and loan growth was a
reasonable and effective tool at the policy level, and the authorities relied heavily on annual lending quotas
in order to keep things under control.

When the first big 2009-10 stimulus round hit, however, things fell apart. In the face of heavy pressure for
“growth at all costs”, banks found very creative ways to expand credit without exceeding regulatory loan
caps. They began accepting short-term corporate bills for discount, which were held off-balance sheet; the
same was true for intra-corporate lending, where one company lent to another at the bank’s request, with
the bank acting as custodian. And NBFIs first exploded on to the scene here as well, as trusts and securities
firms stepped in to provide credit in large amounts.

As a result, if you were just following formal bank loans, there didn’t seem to be that much going on. From
the yellow line in Chart 14, for example, loans outstanding only increased from 108% of GDP at end-2007 to
128% of GDP by the end of 2013, a meager rise of only 20pp.

13
 Jonathan Anderson
May 3, 2023

Chart 14. Credit aggregates

Oustanding credit as a share of GDP (%)

300%

260% "New" TSF

Aggregate domestic claims
PBC TSF
220% Loans

180%

140%

100%

60%
02030405060708091011121314151617181920212223

Source: CEIC

It wasn’t until well after the first stimulus round, however, in 2011-12, that the PBC began to systematically
include the above credit items (acceptances, intra-firm and trust loans, as well as corporate bonds) in a new,
broader credit measure called “total social finance” or TSF for short. And when they looked at the historical
figures … lo and behold, the debt boom was suddenly much bigger. As it turned out, aggregate TSF credit
increased not by 20pp of GDP but rather a whopping 60pp of GDP over the same six-year span (the green line
in the chart). The government had completely “missed” a full two-thirds of overall debt creation at the time it
was happening.

And the same thing occurred once again, starting in 2014 and continuing through the last 2015-16 stimulus
round. By this time the PBC and regulators were actively monitoring all components of TSF – so when banks
came under heavy renewed pressure to lend out, they were forced to be even more entrepreneurial in their
approach to expansion. As it happened, they turned to yet another “new” class of instruments, structured
credit products whereby NFBIs extended corporate loans and then sold the income rights to banks; the end
product sat on banks’ balance sheets as an investment rather than credit, and was reported as an interbank
claim rather than an asset claim on the rest of the economy.

You can see the results in Chart 14 above. From end-2013 to end-2016 the green TSF line only rose by around
20pp of GDP, a fairly moderate increase – but the blue line, which includes all NBFI claims on bank balance
sheets (as well as local government bond issuance), rocketed upwards by 55pp of GDP.

Sure enough, the government had missed nearly two-thirds of the actual debt spiral at the time. And once
again, it wasn’t until a few years later that the PBC came out with a new expanded TSF measure including
DAMP and TBR instruments as well as local government bonds that the true magnitude of the problem
became apparent (see the red line in the chart).

This, in sum, is how you get a missive debt boom in the economy: almost without realizing it until after the
fact.

14
 Jonathan Anderson
May 3, 2023

2. Long since over

With that lengthy introduction to the boom behind us, we can now quickly turn to the next point on our list:

The Chinese debt party ended by the beginning of 2017 – already seven years ago.

You can see this in the circled portion of Chart 15. Whether we measure using overall liabilities or domestic
claims, the debt/GDP ratio stopped rising precipitously at the end of 2016 and has basically gone flat since.
And you can see it even more clearly in Chart 16, which shows monthly net lending on a seasonally-adjusted
basis (defined as an index to GDP). New credit flows dropped sharply in 2017 and (with the exception of a
brief stimulus spike in Q2 2020) have remained below the 20-year average pace from 2018 onwards.

Chart 15. Debt estimates Chart 16. New monthly lending flow index
Share of GDP New domestic claims relative to GDP (index 2005=100, 3mma)
340% 350
Total debt Monthly
300% Domestic financial claims 300 3mma

260% 250

220% 200

180% 150

140% 100

100% 50

60% 0
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Source: CEIC Source: CEIC

And you can see it in the chart on aggregate financial system balance sheets (Chart 17). Here as well, total
estimated assets peaked in 2016 as a share of GDP, fell gradually through 2019 and have simply stabilized
since 2020.

Chart 17. Aggregate financial balance sheets

% GDP
600%
NBFIs
Bank's off-b/s
500% Small/medium and other banks' on-b/s
Big 5 banks' on-b/s
Central Bank
400%

300%

200%

100%

0%
08 09 10 11 12 13 14 15 16 17 18 19 20 21 22

Source: Various national, EM Advisors

15
 Jonathan Anderson
May 3, 2023

What happened?

What happened is that after seven years of more or less unabated debt expansion, a madcap proliferation of
all kinds of marginal banks and “shadow” NBFIs, visibly worsening asset quality and rising bad loans, better
financial statistics that more accurately highlighted the size of the debt problem – and, we should add, a
chorus of economists and analysts, ourselves included, who were alarmed at growing financial risks in the
system – the Chinese leadership changed its focus, turning away from an incessant push for high growth in
favor of a much stronger emphasis on deleveraging balance sheets and reducing systemic risk.

Other broader factors also played a role. A sudden, dramatic spike in capital outflows that caused the country
to lose nearly US$1 trillion in FX reserves between mid-2015 and mid-2016 showed that China’s external
“walls” were not as impregnable as many thought. Having sidelined much of his internal political competition
via anti-corruption purges in the first five years of his administration, President Xi was more willing to
consider the necessary trade-offs between growth and macro stability. And the worsening geopolitical
environment that begin with the US-China trade war in late 2017, further deteriorated during the Covid
pandemic and then completely tanked since the onset of the Russia-Ukraine war has arguably also helped
heighten awareness of potential macro risk factors at home.

Is the government serious? Fair enough, but can we really say it’s “over”? After all, Chinese debt ratios have
only stabilized, not fallen. If this were a “real” deleveraging, wouldn’t we see the ratio decline outright? And
couldn’t this just be a pause in the next inevitable wave of debt-led growth?

There’s no guarantee, of course, but the short answer is that on balance we do see this as a serious structural
policy turn.

To begin with, remember the long-standing point we reiterated above: Debt levels don’t matter; instead, it’s
all about the rate of change. In this sense, to shift from years of steady dependence on ever-rising ratios to a
zero delta is already a meaningful “deleveraging”.

Second, again, as we write today it has already been seven years, with a headline debt ratio that is not
meaningfully higher than it was at end-2016. And the government has broadly held the line through
tremendous economic shocks and volatility.

Which brings us to the third point. The COVID shock of 2020 was intense, and although short-lived, it was
much sharper than the 2008 global crisis. At the trough in early 2009, implied growth according to our China
Activity index fell to zero; by comparison, during the Q1 2020 lockdowns it was at -15% y/y and contracted
sharply again during the 2022 “zero Covid” closures.

However, whereas in 2009 the government opened the debt floodgates and kept them open for nearly 18
months, long after a booming recovery had already occurred, before moving to seriously rein them in, this
time around the government allowed for exactly three months of stimulus lending in 2020 … full stop. It
rolled back policy at the mere initial onset of recovery at home, while activity in the rest of the world was still
collapsing. This is very different behavior than we would have seen a decade ago.

Finally, and crucially, overall debt ratios may be flat – but look at the composition. Here’s a reprint of Chart
11 above without household and corporate debt exposures, i.e., zoomed in just on fiscal and “quasi-fiscal”
borrowing:

16
 Jonathan Anderson
May 3, 2023

Chart 18. Fiscal and “quasi-fiscal” claims

Total domestic financial claims (% of GDP)
80%

70% Government
"Quasi-fiscal"/other
60%

50%

40%

30%

20%

10%

0%
03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Source: CEIC

You see the point: there has been a tremendous outright decline in the “quasi-state”, with an almost
complete stop in new lending and ongoing write-offs of existing debts. By any definition, this is deleveraging
where it matters most. (Albeit offset to some degree by an expansion in formal fiscal debt outstanding as a
share of GDP.)

The same is true within banking system balance sheets. Chart 19 below shows gross interbank exposures of
commercial banks (i.e., the sum of interbank claims and liabilities, which as we will argue shortly below are a
significant indicator of potential stress in the financial system); starting in 2017, one of the first things
regulators did was to curtail and then reverse these kinds of funding and asset operations, and from the chart
gross exposures have subsequently fallen to roughly half of the original buildup.

Chart 19. Gross interbank position

Interbank/NBFI exposures as a share of total assets (%)
30%

25%

20%

15%

10%

5%

0%
06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22

Source: CEIC

As above, this also shows a commitment to outright deleveraging where it really counts. Which is why, again,
we say that the Chinese debt boom has been over now for a long while.

17
 Jonathan Anderson
May 3, 2023

3. And then the bust

And after the boom came the bust.

At face value this may sound a bit silly to those who follow China. The economy did suffer significantly under
Covid, of course, with lockdowns and restrictions taking down services activity as well as the property market
– but there’s been no sign of the financial stress that so often follows credit bubbles in other cases. Major
banks didn’t fail. There’s been no currency crisis, no mass corporate stress, and no social unrest.

But that’s not what we mean by “bust”. We’ll talk more about outright crisis (and why China didn’t have one)
in the next section below.

What China did have, however, is much lower growth. Which, again, started nearly eight years ago.

Allow us to stress this point for emphasis: The Chinese economy is not facing a “new”, looming debt
hangover today. It has been in relative hangover already for most of the past decade.

An EM digression

In order to frame the discussion, we need to reprint a few arguments and charts from Part 6 of the
companion How to Think About Emerging Markets series (see An Austrian Primer, 16 October 2022).

As noted above, the first major conclusion from that report was that debt levels don’t really matter much in
determining macro outcomes.

In order to show this, the first thing we did was to revisit the experience of 2008-09. Why 2008-09? Because
the onset of the global financial crisis brought on the biggest worldwide “sudden stop” in capital flows in
more than half a century, one that devastated the most exposed countries and markets. Simply put,
everyone’s party ended at the same time.

However, if you look at economic performance across EM countries during and after the crisis, you are forced
to conclude that absolute debt levels played no role at all in determining subsequent trends. Chart 20, for
example, shows the end-2007 debt levels (defined as domestic household and corporate credit outstanding
as a share of GDP) plotted against the cumulative change in real GDP levels over the subsequent five years for
all major and frontier EM countries we follow, and Chart 21 shows a similar scatter plot for the five-year
change in real growth rates following the crisis.

18
 Jonathan Anderson
May 3, 2023

Chart 20. Debt levels vs. growth performance (i) Chart 21. Debt levels vs. growth performance (ii)
Change in 2012 real GDP vs. end-2007 level Change in 2009-12 real GDP growth vs. 2003-08 pace
80% 4%
70% 2%
60%
0%
50%
40% -2%

30% -4%
20% -6%
10%
-8%
0%
-10% -10%

-20% -12%
0% 30% 60% 90% 120% 150% 180% 0% 30% 60% 90% 120% 150% 180%
Private credit/GDP ratio, end-2007 Private credit/GDP ratio, end-2007

Source: IMF Source: IMF

In either case, the most striking feature is the visible lack of correlation between debt/GDP levels and the
ensuing macro trends. Indeed, the worst-performing economies in the aftermath of the crisis were medium-
debt countries: the broad Central and Eastern European bloc, including the Baltic states, Hungary, Ukraine,
Bulgaria, Romania and the former Yugoslav region – all of which had unexceptional debt ratios in the 50% to
90% of GDP range. Whereas there wasn’t a single country with “high” private debt/GDP levels above 100%
that fell below the emerging post-crisis average in terms of GDP and output momentum.

So looking at absolute debt/GDP ratios clearly not a useful exercise.

What does work, then?

It’s in the deltas. The answer is the change in the debt level. As it turns out, a strong, sustained increase in
overall credit and debt/GDP ratios has been one of the very best predictors of subsequent trouble in
emerging markets.

Going back to the broad 2008 crisis sample, you can see this in Chart 22 below. If you list the EM countries
that saw a cumulative increase in credit outstanding of, say, more than 35% of GDP in the five years leading
to 2008, almost all of them had a collapse of output during the crisis. By contrast, relatively view of the
countries where credit growth was more moderate suffered a similar fate.

19
 Jonathan Anderson
May 3, 2023

Chart 22. Changes in debt vs. growth Chart 23. Change in debt/funding vs. growth
Change in 2009-12 real GDP growth vs. 2003-08 pace Change in 2009-12 real GDP growth vs. 2003-08 pace
4% 4%

2% 2%

0% 0%

-2% -2%

-4% -4%

-6% -6%

-8% -8%

-10% -10%

-12% -12%
-40% -20% 0% 20% 40% 60% -40% -20% 0% 20% 40% 60% 80%
Change in private credit/GDP ratio, 2008 vs. 2003 Credit and L/D delta, 2003-08 (average)

Source: IMF Source: IMF

Indeed, there’s only one other indicator that works as well as or better than the change in credit in predicting
trouble: the change in net funding exposure of that credit, as measured by the aggregate financial system
loan/deposit ratio. In Chart 22 above we combine the two metrics (i.e., the five-year delta in credit/GDP and
the five-year delta of the loan/deposit ratio) ... and as you can see, it’s an exceptionally strong directional fit.
The combination of rapidly rising credit penetration and a rapid increase in net non-deposit funding was a
very reliable indicator of subsequent stress, and their absence was a reliable guarantor of stability.

Not just about 2008. And again, while the 2008-09 crisis experience was exceptional in many ways, the same
results hold at other times as well. In Charts 24 we repeat the exercise for the 40 major EM countries under
coverage between 1997 and 2012, using the five-year change in credit outstanding vs. the subsequent
change in five-year average growth rates.

Chart 24. Change in debt vs. growth

Change in overall growth, next five years
15% 2012
2007
10% 2002
1997
5%

0%

-5%

-10%

-15%
-100% -50% 0% 50% 100%
Five-year change in private credit/GDP

Source: IMF, BIS, CEIC

You see the point: the results are far from perfect, but there is nonetheless a big directional correlation – and
especially strong at the ends of the spectrum. Countries with the most rapid run-up in credit and debt

20
 Jonathan Anderson
May 3, 2023

penetration almost uniformly saw the worst growth outcomes, while those with well-behaved balance sheets
fared far better.

Lessons learned:

First, if you have a debt boom, you get a subsequent debt bust – regardless of whether you started from low
or high levels of penetration.

And second, by “bust” we don’t necessarily mean crisis. Rather, we mean a sustained drop in growth.

China is no exception

And China is no exception. Here again is the picture for Chinese growth momentum over the past couple of
decades. As we highlighted earlier, you don’t see the trend so clearly in the official GDP figures … but the
path of the China Activity index could not be more obvious. During the peak debt buildup period in the first
half of the decade, the index grew at 9.5% y/y on average in real terms. By the second half, however, real
growth had fallen to nearly 4% y/y, and was running under 3% in 2019 on the eve of the pandemic (Chart 25).
And while it’s too early to tell where structural growth is coming out in 2024-25 after the successive Covid-
era shocks, the poor performance over the past three years to date is evident from the chart as well.

Chart 25. China growth indicators

Real growth (% y/y 3mma)

20%

15%

10%

5%
Official GDP
Activity index
0%

-5%
95 00 05 10 15 20

Source: CEIC

Needless to say, there were a lot of other things happening in China in the 2010s besides just the debt story.
We already saw that growth was on a significant downward trajectory regardless, as export market share
peaked and the property boom matured – and indeed, that debt stimulus was in large part a reaction to that
very trend. One could easily argue that the causality is completely reversed in China’s case, i.e., that the
country had a credit boom because the economy was slowing, rather than the other way around.

Fair enough. But look what happens when we put the Chinese figures for the peak five-year expansion in
private credit/GDP in 2019-14 and the subsequent five-year change in real growth from 2015-19 into the
above-citied EM-wide scatter plot from Chart 24:

21
 Jonathan Anderson
May 3, 2023

Chart 26. Where China fits in

Change in overall growth, next five years
15%

10%

5%

0%

China
-5%

-10%

-15%
-100% -50% 0% 50% 100%
Five-year change in credit/GDP

Source: IMF, BIS, CEIC

You see the point. Prior to Covid, at least, China was right on track. Irrespective of causality and other
incidental factors, its economic performance in the second half of the 2010s was almost exactly what one
would have expected it to be, given the magnitude of the debt boom in the first half.

That’s not all. Since 2016 China has exhibited other obvious manifestations of “post-boom” malaise as well,
first and foremost in the banking system. Remember that the credit explosion was overwhelmingly driven by
small/medium banks and their NBFI counterparts, lending heavily to “quasi-state’ commercial firms that
often had no existing sources of revenue and very limited asset collateral. As you might imagine, once the
credit cycle stopped and regulators were forcing banks into balance sheet retrenchment, these quasi-fiscal
players were the first to go bankrupt in droves. One rarely hears about them, as they are not generally bond
issuers and are almost wholly beholden to banks (and banks in turn had no option but to quietly shelve the
resulting bad loan claims on their balance sheets pending subsequent write-downs). But they are there. And
the numbers are huge, with non-performing debts that could easily amount to 20% or 30% of GDP.

Sure enough, from 2015 onwards – after more than a decade-long hiatus since the completion of the state
bank cleanup in the early 2000s – China has seen dozens of smaller bank restructurings involving significant
asset write-offs and forced mergers and/or recapitalizations. These have been done quietly, without fanfare
and mostly without any defaults on outstanding liability obligations, but the biggest cases have involved
banks with more than RMB1 trillion in assets – and there are more still waiting in the queue (for further
details see our 7 March 2019 and 2 July 2019 reports on the topic).

Similarly for the property market. Sales and construction activity were already slowing in the first half of the
decade – but went almost completely flat in the second half, and as shown in Chart 5 above declined sharply
as a share of real GDP. Consumption of property-related materials like steel and cement rose by more than
70% in between 2010 and 2014 … but only a meager 11% over the next five years.

This, in short, is a bust.

From a macro perspective you can see the impact of slowdown and deleveraging in our EM Macro risk index.
As a reminder, the index is compiled semi-annually and measures risk across a number of areas, including

22
 Jonathan Anderson
May 3, 2023

banks and finance, fiscal and external exposures (see the latest 9 April 2023 installment for full details). Here
are the end-year scores for China over the past 30 years:

Chart 27. Macro risk index scores

Macro risk scores, China
8

2
1990 1995 2000 2005 2010 2015 2020

Source: IMF, CEIC

Between 2008 and 2016 there was a clear, dramatic “risking up” as debt ratios rose, external surpluses fell
and quasi-fiscal imbalances widened. Over the past five years, however, risk scores have gradually but
steadily decreased as debt levels stabilized and other imbalances narrowed.

The de-risking process is still going on, of course, but it is important to recognize that is already well
underway.

23
 Jonathan Anderson
May 3, 2023

4. Why no crisis?

Which leads us to the next question: Why didn’t China have a crisis?

I.e., if the debt expansion itself was really so epic, why didn’t things end in a more “epic” manner as well?
And here we’re not just talking about a long post-boom period of writing down bad debts and cleaning up
balance sheets; as discussed above, every country goes through that, and China was no exception. Rather, by
crisis we mean something “breaking”: a currency failure; a Lehman-style “Minsky moment” at home, with
outright bank failures and a sudden stop in intermediation; abrupt, widespread debt defaults; collapsing
property and/or other asset markets. And an economy that not only slows but actually falls into sharp
contraction.

Other EM countries have been laid low for seemingly much smaller debt bubbles. Why not China?

For readers in a hurry, the short answers are as follows:

• An extremely safe starting position, with clean balance sheets and high external surpluses
• Sky-high domestic saving rates
• A very closed capital account
• The “virtual” nature of quasi-state credit boom
• Lack of leverage in the property market

For the longer version, we need to start by setting the EM-wide background, with an examination of the
historical relationship between debt and crises.

Debt and crises in EM

Or, better put, the lack of a relationship.

Chart 28 below shows peak debt levels, defined as the maximum ratio of gross domestic credit of the banking
system to GDP across the major EM countries we follow for the past three decades. In the chart, we color-
code each country according to whether it had a financial crisis in the same period, with blue representing a
crisis and yellow for no-crisis cases (and by “financial crisis” we mean either an external currency collapse, a
domestic banking collapse, or both; note that we are being very careful with our definitions, so please do
refer to the footnote for details). 3

3We need to be careful about what we mean by banking or currency “crises”. In standard academic literature on the topic, for
example, a bank crisis is normally defined as a period of bank losses and capital insolvency. I.e., if the government is forced to step in
and recapitalize or restructure banks, it’s a crisis. For our purposes, however, this is not a very useful working definition. Any period
of rapid credit expansion will, by definition, result in higher-than-usual bad loans and other impaired assets, and generally requires
subsequent workout or recapitalization of at least some lenders. By this standard, China has already been in “crisis” for half a decade.
Rather, we want to know if the problems are severe enough to (i) impact funding for major, systemically important institutions, (ii)
affect liquidity and confidence across the financial system, and (iii) impair credit intermediation. I.e., severe enough to cause a
“break” in credit flows and availability. As we discuss below, none of these have been the case in the mainland economy.
Turning to the external side, by “currency crisis” we mean either (i) a forced, disorderly break from an existing peg or quasi-peg
arrangement, or (ii) a sudden, dramatic depreciation caused by country-specific imbalances and capital outflows. Just to take a couple
of representative counterexamples, when a freely floating unit like the South African rand sells off alongside the entire basket of EM

24
 Jonathan Anderson
May 3, 2023

Chart 28. Debt levels vs. crises

Maximum domestic credit/GDP ratio, 1995-2020
250%

200% No crisis
Crisis

150%

100%

50%

0%
China

Turkey

Peru

Mexico
Taiwan
Malaysia
Singapore

Latvia

Saudi
Korea

Chile
Hong Kong

Israel

Ecuador
Lithuania
Brazil

UAE

Slovenia

S Africa
Estonia
Slovak

Hungary

Colombia

Nigeria
Czech

Indonesia
Bulgaria
Ukraine

Croatia

Poland

Romania
Thailand

Russia
India
Vietnam

Philippines

Kazakhstan

Venezuela

Argentina
Egypt

Source: IMF

Hmm. Clearly high-debt countries are no more prone to financial crises than their low-debt counterparts.
Thailand and Korea blew up with headline domestic debt levels that were approaching 200% of GDP … but so
did Indonesia and Argentina, with debt/GDP ratios that were closer to 60% at the time.

In fact, if anything countries with low headline debt/GDP ratios were more likely to have a crisis than those
with high debt levels. Of the ten countries at the far right end of the chart, eight recorded a financial crisis.
Whereas of the ten countries at the far left, only four did.

This comes as no surprise. Once again, absolute debt levels don’t really tell us anything about underlying
macro risk or growth in the emerging world.

Ok, then, what about debt growth?

The picture is a good bit better here, to be sure. In Chart 29 we replace peak debt/GDP ratios with the
maximum trough-to-peak swing in the debt ratio as a share of GDP within a ten-year period, and as you can
see, most crisis cases fall into the left-hand side of the chart, in countries that had big debt booms.

resource-exporting currencies due to a drop in global commodity prices, or when the Russian ruble depreciates in responses to falling
crude oil markets, this is not a “crisis”. Unless there are other extenuating circumstances, it’s just a depreciation.
And finally, we also need to point out that for the purposes of Chart 29 and the next few that follow, we only count crises that occur
in direct response to the debt boom in question. Brazil and Egypt, for example had crises in the late 1990s and the early 2010s
respectively, but in both cases these occurred well before the peak debt period referenced in the chart.

25
 Jonathan Anderson
May 3, 2023

Chart 29. Debt growth vs. crises

Maximum trough-peak change in domestic credit/GDP ratio, 1990-2020
100%
90%
80% No crisis
Crisis
70%
60%
50%
40%
30%
20%
10%
0%
China

Turkey

Peru

Mexico
Latvia

Malaysia

Singapore

Saudi

Taiwan
Korea

Chile
Hong Kong

Israel
Ecuador
UAE

Lithuania

Slovenia
Estonia

Hungary

Brazil

Slovak

Colombia

Nigeria
S Africa
Indonesia
Czech
Bulgaria
Ukraine

Croatia

Romania

Poland
Thailand

Russia

India
Vietnam

Kazakhstan

Philippines
Venezuela

Argentina
Egypt

Source: IMF

But far from all. A full half of the economies with the lowest change in debt ratios over the past three
decades (Argentina, Mexico, Nigeria, Colombia and Ecuador) recorded financial crises. And while seven of the
top ten in terms of debt dynamics had crises, the top three (Hong Kong, China and Vietnam) didn’t.

(China has been quietly fixing smaller banks in the system, and Vietnam in the early 2010s went through a
period of significant write-downs in larger institutions similar to China’s clean-up in the late 1990s. Hong
Kong’s credit boom, meanwhile, is still ongoing, so the aftereffects are still unclear. But none of these
economies suffered major financial shocks or instability at home, and none of them had a currency shakeout
or collapse.)

Here again, there’s no hard and fast rule.

Simply put, it’s not just about debt. As we saw above, changes in debt ratios over time are an excellent
predictor of future growth momentum and prospects – but when it comes to outright crises, they aren’t very
reliable.

What does predict crises?

Here’s a big hint. Or two. We call them the “two margin calls”.

The “domestic margin call”. On the home front, if we were to summarize the issue in a single sentence it
would be this: It’s not about debt … but rather the funding of the debt.

As we’ve learned in pretty much every EM and DM banking crisis, it’s not asset quality per se that brings
down banking systems; we have a number of examples (including China itself in the 1990s, about which more
below) where banks have happily continued to operate and lend in the face of crushing bad debt ratios and
de facto insolvency.

Instead, what determines a true “breakdown” scenario is the nature of banks’ and other financial institutions’
liability exposures. The lower the share of domestic retail deposits, and the more dependent the system is on

26
 Jonathan Anderson
May 3, 2023

non-traditional/external sources of funding, the more likely a sudden funding pullout when things go wrong –
i.e., a crisis.

Here’s a scatter chart that shows what we mean. The vertical axis of Chart 30 is a macro-level “loan/deposit
ratio”, defined as total domestic credit of banking institutions divided by broad money (deposit plus cash)
liabilities, using standardized IMF data from its annual IFS database. Specifically, for each of the 40 major EM
countries we cover the axis shows the maximum ratio recorded since 1995.

Meanwhile, the horizontal axis shows the associated change in the ratio in percentage points, for the five
years leading to the peak level.

In short, this is a very simple chart on overall banking system non-deposit funding exposure, in terms of both
levels and rates of change:

Chart 30. Domestic funding exposures

Max domestic claims/deposit ratio, 1995-2019
225%
LAT
No crisis
200% v Crisis
THA RUS
v
ECU EST
175% UKR
KAZ
KOR NIG LIT
150% TUR HUN SLO
S AFR ARG MAL UAE
125% CRO INDO
CZE COL ROM
BRA SLK POL
ISR EGY SIN
CHN BUL
100% MEX CHIVIE
TAI IND HKG
PER
SAU
75%

50%
-20% 0% 20% 40% 60% 80% 100%
Five-year change in ratio

Source: IMF

But despite its crude simplicity, it correlates extremely well with actual crises. Countries with both a rapid
run-up in credit relative to deposits and a high resulting headline ratio almost always ended up in financial
turmoil, while countries with lower, more stable credit/deposit indicators broadly avoided worst-case
outcomes.

There are exceptions, of course. Mexico and Indonesia, for example, went through crises with seemingly
“safe” funding exposures by relative emerging standards, and countries like Bulgaria in the late 2000s
avoided banking and currency collapses despite more elevated readings. And needless to say there are more
precise and nuanced ways to measure actual exposures in each individual country case.

Again, however, the striking thing about the chart above is just how well it does in capturing the essence of
crisis risks on an EM-wide basis.

Two “non-crisis” examples. A couple of good “non-crisis” examples should help focus minds on this point. In
the interest of choosing cases close to home, we want to discuss (i) China’s own past in the 1990s, and (ii) a
more recent example from neighboring Vietnam.

27
 Jonathan Anderson
May 3, 2023

Long-time China watchers will remember vividly the credit and investment bubble of 1991-95. The fledgling
post-liberalization economy exploded in a frenzy of bubble activity, with huge numbers of ill-advised,
indiscriminate property development and industrial investment projects; asset prices went skyrocketing and
retail pyramid schemes were rampant.

By the time the authorities finally turned off the credit taps in the middle of the decade, thousands of firms
were put out of business, millions of workers were sent home – and banks were saddled with one of the
largest bad asset hangovers in major EM history, with a system-wide NPL ratio at anywhere from 40% to 60%.
Again, by any standard the entire Chinese banking system was massively insolvent.

However, the one thing the banking system didn’t have in the 1990s was meaningful funding exposure. There
was a hodge-podge of tiny alternative investment trusts and other financial institutions that had cropped up
during the bubble, but the handful of large state banks completely dominated the landscape at the time, with
a liability base that was almost completely made up of traditional deposits.

As a result, there were no defaults on deposit or other financial obligations, and not a single depositary
institution of any meaningful size was shut down. The government didn’t even bother to do a serious
recapitalization; they took a small portion of bad loans off commercial bank balance sheets and simply
instructed banks to evergreen the rest. And savings continued to pile into the banking system and aggregate
credit continued to grow at a double-digit pace every single year from 1995 onwards through the following
decade.

Vietnam in the 2000s was very much the same story. As shown in Chart 8, between 2000 and 2010 Vietnam
posted one of the largest debt booms the emerging world has ever seen, with an increase in aggregate
domestic debt levels of close to 90% of GDP over the decade. As in China in the 1990s, most of that increase
went into a massive property and construction bubble, with wildcat developers springing up across the
economy, as well as white-elephant industrial SOE investment.

Sure enough, after years of wasteful spending and wild overdevelopment, the bubble burst; corporate
borrowers went bankrupt in droves, asset quality plummeted and Vietnamese banks ended up with
tremendous holes in their balance sheets.

But here as well, there was never an actual “crisis” – not as we define it. Banks were steadily supported by
the government; funding remained ample through the ensuing downturn, and the ensuing workout has been
relatively slow and quiet. Lending growth slowed visibly from 2010 onwards, but remained meaningfully
positive; there was never a collapse of liquidity, and the economy never fell below 4% real growth on an
official basis.

How did Vietnam manage this, when neighboring countries like Thailand, Malaysia and Korea went into sharp
recession and more serious currency and financial system stress despite having debt booms that were
objectively smaller than Vietnam’s? Part of the answer is in Chart 9 above: Vietnam’s boom was funded
almost entirely through retail deposits funneled through a limited number of state banking institutions, and
the aggregate credit/deposit ratio never once peaked above 100%.

28
 Jonathan Anderson
May 3, 2023

The “external margin call”. But that’s only part of the answer, of course. The other main difference between
Vietnam and neighboring Thailand or Korea was that Vietnam had much more limited overseas funding
exposures. And this brings us to the second, extremely important “margin call”: external imbalances.

There are a few ways to think about the external side of the equation. One is to simply look at the current
account or the “basic” balance of payments as a measure of dependence on external funding. And as shown
in Chart 31, it’s certainly true that nearly every country with “massive” external deficits eventually blew up
(the vertical axis in the chart indicates the historical minimum current account balance as a share of GDP over
the past 25 years, and the horizontal axis shows the preceding five-year change in the balance).

However, looking at the chart, it’s also true that most crisis and non-crisis cases are jumbled together in the
middle, with more moderate deficits and moderate deterioration, and no easy way to choose between them.
And of course you also have crisis outliers on the other end like Russia and UAE, which had banking and/or
currency collapse while effectively in surplus.

Chart 31. The role of the current account Chart 32. The role of banks’ external funding
Min current account balance, 1995-2019 Min bank NFA/asset ratio, 1995-2019
10% 10%
No crisis No crisis
v Crisis 5% v Crisis SAU
5% CHN
v TAI
v NIG VIE VEN TAI
CHN UAE 0%
0% HKG KOR MAL
EGY
RUS -5% PHI INDO IND
VEN
-5% RUS SIN CZE ISR BRA CHI
-10% S AFR
POL PER
HKG UAE BUL ECU
CRO
-10% -15% COL
ROM ARG HUN
CRO TUR SLO SLK
-15% -20%
EST THA
LIT EST -25% UKR
KAZ
MEX
LAT
-20%
-30% ROM
-25% LIT
BUL -35%
LAT
-30% -40%
-20% -15% -10% -5% 0% 5% -40% -30% -20% -10% 0% 10%
FIve-year change in balance FIve-year change in ratio

Source: IMF Source: IMF

Another way to come at the issue – and one that is more directly related to credit and debt booms – is to
focus on external funding of the financial system. Thus, in Chart 32 we plot the relationship between
commercials banks’ net foreign asset/liability exposure (as a ratio to total assets) and the five-year change in
that ratio.

This works somewhat better. There are still plenty of outliers, but the broad bulk of crisis cases are in the
lower left quadrant, with significant external banking system liabilities, while most non-crisis cases are in the
upper right, with less overseas funding and a lower pace of expansion. And suddenly cases like Russia and
UAE, which were very surprising on a current account basis, are more easily explainable.

Finally, if we put the two indicators together, using the simple overage of the current account balance and
the banking system NFA ratio, we get a relationship that works very well indeed, with decently high
predictive power (Chart 33):

29
 Jonathan Anderson
May 3, 2023

Chart 33. Current account/NFA average

Min current account/NFA avg, 1995-2019
10%
No crisis
5% v Crisis
v CHN TAI
SIN
0% NIG SAU
HKG KOR
-5% RUS VIE INDOIND BRA
ISR CHI
PHI CZE MAL S AFR
UAE POL ARG PER COL
-10% SLO ECU
TUR CRO
-15% KAZ SLK HUN
THA
BUL UKR MEX
-20%
EST
ROM
-25%
LIT
-30% LAT

-35%
-30% -20% -10% 0% 10%
FIve-year change avg

Source: IMF

The chart is far from perfect; it still leaves crisis outliers like Korea, Indonesia and Nigeria, for example, which
had relatively low foreign exposures by emerging standards but still fell apart in an idiosyncratic manner. But
on an EM-wide basis, it still does highlight the crucial role of external funding in explaining emerging crises.

Back to China

With that in mind, let’s return to China. What have learned?

Again, when it comes to outright crises it’s not about debt – it’s about the funding of the debt.

And in China’s case, it may have had epic amounts of debt, but it simply didn’t have the kind of funding
exposures necessary to generate a financial breakdown or collapse. As we show below, the situation was
clearly deteriorating on the domestic and external “margin call” front. But in both cases exposures still fell
well short of “red zones” when China put a stop to the debt bubble in 2017.

The domestic margin call. Let’s start with the domestic funding side. And here it’s not just that aggregate
Chinese debt levels increased; much more important is how that debt increased. Sure enough, as we showed
above, the mainland credit boom was accompanied by a veritable explosion of financial innovation and
differentiation: a massive rise in the share of small and medium banks in overall lending, and a dramatic
increase in the role of trust, securities and finance companies, off-balance sheet wealth management
products and other “shadow” financial institutions.

I.e., the big story of the past decade was not just the frenetic growth of balance sheets, it was also the
increased fragmentation of the financial system. At late as the mid-2000s, Chinese households and firms
mostly put their savings into one of the large state banks, which in turn provided most of the loans going
back out into the economy. By the latter part of the 2010s those savings were equally likely to go to a non-
bank/off-balance sheet fund management or wealth management product, which would then provide
funding to smaller banking institutions, which then provide credit back to businesses and households.

In other words, we’re talking about the rise of wholesale funding intermediation in China.

30
 Jonathan Anderson
May 3, 2023

Here’s how it looks at the level of the aggregate banking system. Chart 34 shows the nominal amount of
gross bank assets and liabilities in 2008, as and you can see the numbers were both (i) small, around RMB 50
trillion, and (ii) dominated by “traditional” household and corporate deposits on the liability side and
standard term loans on the asset side.

Chart 34. Banking system balance sheet 2008 Chart 35. Banking system balance sheet 2017
Commercial banks, end-2007 (RMB tn) Commercial banks, January 2017 (RMB tn)
250 250
Interbank Foreign assets
NBFI PBC reserves
200 200 Bonds
Other Other
Equity claims
150 150

Generic Loans
100 100 deposits

50 50
Generic Domestic
deposits assets
0 0
Liabilities Assets Liabilities Assets

Source: CEIC Source: CEIC

Fast forward to early 2017, at the end of the debt boom, and things were much more complicated (Chart 35).
To begin with, generic loans shrank visibly as a share of balance sheets, from nearly two-thirds of total assets
back in 2007 to less than half. Instead, banks were been providing credit by buying government bonds or
corporate paper, or more recently in the form of so-called “investment receivables”, essentially hidden
lending structured through NBFI intermediaries and sold to banks as investment products. As a result,
“other” non-loan claims on the rest of the economy had ballooned to nearly 40% of the balance sheet.

And on the liability side traditional deposits shrank at an equally visible pace, from 75% of total bank
liabilities in the late 2000s years ago to only 60% a decade later. This decline reflected a sharp trend increase
in the share of “other” funding items: banks’ own bond and paper issuance, deposit and NCD liabilities to
NBFIs and gross interbank borrowing.

What do these latter items all have in common? The answer is that they are all liabilities to other financial
institutions. I.e., they are all wholesale rather than retail.

That’s not all. Those exposures are even more visible if the large state commercial banks and just focus on
the small and medium banking sector. As shown in the circled portion of Chart 36, by 2017 small and medium
banks had an even larger share of non-loan credit assets, an even smaller share of traditional deposits
relative to total liabilities, and a correspondingly larger proportion of wholesale funding items on the balance
sheet.

31
 Jonathan Anderson
May 3, 2023

Chart 36. Small/medium bank balance sheet 2017

Smaller commercial banks, January 2017 (RMB tn)
140
Foreign assets
120 Interbank PBC reserves
NBFI
100 Bonds
Other Other claims
80 PBC
Equity
60
Generic
deposits
40
Loans
20

0
Liabilities Assets

Source: CEIC

You can see the trend in the following chart, which shows traditional household and corporate deposits as a
share of overall liabilities for the past decade. Again, for the banking system as a whole the ratio fell from
75% in the mid-2000s to around 62% by 2017 – and for small/medium banks the ratio declined faster still to
less than 55% (the blue line in Chart 37).

Chart 37. Standard deposits to total liabilities

Traditional deposits/total liabilities
80%

75%

70%

65%

60%

55% All banks

Large banks
50%
Small/medium banks
45%

40%
06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21

Source: CEIC

Since the boom ended in 2017, however, the trend has reversed. As we saw above, over the past five years
regulators not only aggressively tightened access to credit and slowed the pace of balance sheet growth; they
also specifically targeted the most risky types of asset and liability operations, i.e., interbank funding and
structured interbank claims (including claims and liabilities vis-à-vis NBFIs). As a result, again, gross interbank
positions have fallen steadily as a share of balance sheets (Chart 38) – and the share of “standard” deposits in
overall liabilities has stabilized for the system as a whole and risen outright for smaller and medium banks
(Chart 37 above).

32
 Jonathan Anderson
May 3, 2023

Chart 38. Gross interbank position

Interbank/NBFI exposures as a share of total assets (%)
30%

25%

20%

15%

10%

5%

0%
06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22

Source: CEIC

China in EM-wide context. If we translate all of this into the headline credit/deposit ratios that we used as
our standard exposure gauge across EM economies in the discussion above, we get the following picture:

Chart 39. China credit/deposit ratios

Credit/deposit ratio
140%
Using total augmented claims
130% Using domestic credit from banks
120% Using loans
110%
100%
90%
80%
70%
60%
50%
40%
02030405060708091011121314151617181920212223

Source: CEIC

To begin with, if we just look at the “old-school” loan/deposit ratio (the yellow line) … lo and behold, it barely
budged all through the cycle; it was around 70% two decades ago and was still only 75% at the end of the
boom in 2016 (indeed, it only began to rise after the onset of deleveraging, as regulators cracked down on
other forms of lending).

But as discussed above, that’s only because banks diversified aggressively away from standard loans; once we
include other on-balance sheet credit claims in the numerator, the overall domestic credit/deposit ratio –
which is our “apples to apples” comparator on an EM-wide basis – actually rose sharply in the first half of the
2010s reaching 110% in 2017 before stabilizing during the post-boom era (the green line in the chart).

33
 Jonathan Anderson
May 3, 2023

(That’s not all, of course. As we also showed, the other notable hallmark of the stimulus-led credit years was
the move off formal bank balance sheets altogether, in the form of off-balance sheet acceptances, intra-
corporate lending or NBFI loans. If we include these items in the numerator as well, the ratio of overall
augmented claims to deposits rose to more than 120% in 2017 before subsequently going flat, see the blue
line.)

And where does China fit in when we put those ratios in the domestic funding/crisis scatter plot from Chart
30 above?

It fits in here:

Chart 40. Where China fits in – domestic exposure

Max domestic claims/deposit ratio, 1995-2019
225%
No crisis
200% v Crisis
v
175%

150%

125%
China 2021
100% China 2017

75%
China 2008
50%
-20% 0% 20% 40% 60% 80% 100%
Five-year change in ratio

Source: IMF

In 2008, China was as “safe as houses”, with an extremely low domestic credit/deposit ratio that had been
falling over time, putting it at the far lower left corner of the chart.

By the end of the debt boom in 2017, again, the ratio had jumped to 110%, for a 25pp increase over the
preceding five years. This was a clear risking up, of course … but still put China very much in the middle of the
“no crisis” EM counterpart cases (and the same is true even if we were to use the slightly higher augmented
credit/deposit ratio from the previous chart). As we stressed at the time, it would have taken another five
years of unabated debt expansion to really put China at meaningful risk of domestic financial system stress.

Needless to say, this didn’t happen. Instead, Chinese leadership reversed course, tightened policy and began
to de-lever the economy. As a result, China today has shifted all the way back to the left-hand axis, putting
the financial system in a much safer place.

The external margin call. And as for the external front, well, China was never even remotely close to trouble
during the boom years in the first place.

The economy did post a sharp drop in the current account balance, mind you, from more than 10% of GDP at
the 2008 peak to less than 1% by 2018 (Chart 41) – but nonetheless remained in surplus throughout, and
throughout subsequent years as well. Even more important, the Chinese banking system has always been a
net external creditor, with a steady relatively steady net claims position as a share of total assets (Chart 42).

34
 Jonathan Anderson
May 3, 2023

(Indeed, even if we strip out the “big five” banks we find that the remaining small/medium banking sector
may have had rising wholesale funding exposures at home, but almost no foreign assets or liabilities to speak
of. So again, this was never a meaningful point of concern.)

Chart 41. Current account and basic balance Chart 42. Banking system NFA/assets
"Basic" balance (% GDP 12m cum) Commercial bank NFA (% of overall net assets)
18% 10%
16% 9%
14% Net FDI
8%
Current account
12% 7%
10% 6%
8% 5%
6% 4%

4% 3%

2% 2%

0% 1%

-2% 0%
01020304050607080910111213141516171819202122 02030405060708091011121314151617181920212223

Source: IMF Source: IMF

And when we put China in the earlier EM-wide scatter chart on external exposures, we get the following
picture:

Chart 43. Where China fits in – external exposure.

Min current account/NFA avg, 1995-2019
10% China 2008
No crisis
5% v Crisis
China 2017
v
0%
China 2021
-5%

-10%

-15%

-20%

-25%

-30%

-35%
-25% -20% -15% -10% -5% 0% 5%
FIve-year change avg

Source: IMF

Once again, China began the boom in 2008 in the safest possible position, with (ii) high current account
surpluses and a net creditor position in the banking system and (ii) both metrics improving over the previous
half-decade. The situation did deteriorate in the 2010s during the long debt expansion, but even at the 2017
peak China was still above the horizontal axis and just barely over the vertical axis, i.e., still very far indeed
from any meaningful risk on a comparable basis. And since the end of the party, China’s position has once
again moved in a safer direction. Simply put, there’s not much to talk about here.

35
 Jonathan Anderson
May 3, 2023

[At least not on these metrics. In the last section below we will mention another analysis that highlights the
“hidden” long-term risk to the renminbi regime, to be covered in full in Part 7 – but more on that shortly.]

Why?

Summing up, China in the 2010s may have seen one of the great EM debt booms, but even towards the end it
was never at serious risk of falling apart.

Why? What allowed China to print money and credit with impunity for nearly a decade without having to
worry about domestic or external instability, while so many other EM countries that tried the same thing
broke down much earlier along the way?

The answer is a combination of the following elements:

1. The best possible starting position. Looking back at Charts 40 and 43 above, the striking thing in both
cases is just how protected China was at the start of the boom: at the far lower left quadrant of the domestic
scatter and the far upper right quadrant of the external scatter. Most EM crisis countries did not have the
luxury of starting from an external current account surplus of 10% of GDP and a “basic” balance of payments
surplus of nearly 16% of GDP when they began their leveraging process. They were also not necessarily just
coming off a decade of record-high growth that allowed the government to undergo a massive cleanup of
large banks, writing off large amounts of outstanding debt in the process and delivering the lowest historical
financial system credit/deposit ratios in Chinese history.

Simply put, China got very lucky. As it turns out, 2008 was the best possible starting point for a subsequent
credit boom; you could not have asked for better conditions. And this allowed for a lot of misbehavior and
deterioration along the way without taking the economy into danger zones.

2. Sky-high saving rates. Needless to say, another thing that sets China heavily apart from most EM
economies is its enormous national saving rate – a whopping 47% of GDP on average over the past decade on
a headline gross domestic saving basis, or 20pp above the EM-wide average. This helped deliver China’s
external surpluses, lowering the spending impact on the current account, and allowed credit to “recycle”
back into domestic deposits, thus slowing the pace of deterioration in banks’ outside funding ratios.

3. China’s “virtual” credit economy. It’s easy to talk about “savings” in aggregate … but in China, we also
have to ask: whose savings? The generic story repeated ad nauseum by observers and pundits is one of
Chinese households putting huge amounts of money in the bank, to be used in turn by corporates and
government for spending and investment – but as we showed in the previous Part 4 installment of this series,
that story is a complete fiction. Households do have gross saving rates, but their net financial saving rate has
actually dropped to near zero over the course of the past decade. Households are not putting money in the
bank; instead, they are spending it directly on housing.

I.e., in China we’re really talking about corporate savings. What does this mean? Have traditional firms been
saving more to allow for a mass borrowing binge by the state and especially the “quasi-state”?

No. As it turns out, a big share of the savings during the debt boom came from the quasi-state itself; the
same entities that were bingeing on credit were also turning around and pocketing the money rather than

36
 Jonathan Anderson
May 3, 2023

spending on real goods and services. This is what slowed the pace of financial system funding deterioration,
and this is what slowed the pace of real demand and external pressures.

We call it the “virtual credit boom”: you get massive borrowing, but much less spending. In our view it’s hard
to fully comprehend China during the 2010s without understanding this facet of the credit cycle. We will
devote the next section of the report to this topic, so stay tuned.

4. No property bubble. It’s also hard to understand China’s “soft” post-2016 landing without looking deeper
into the single largest and most important sector in the economy, i.e., real estate and construction. After all,
if you go back to Chart 3 above showing the magnitude of debt booms across EM, almost every single high-
debt country on the left-hand side was driven by aggressive property bubbles: Hong Kong, Vietnam, Thailand,
the Baltics and Balkans, UAE, Malaysia, etc., etc., all of whom had (i) sharply rising property prices in income-
adjusted terms, (ii) a dramatic increase in mortgage leverage as a share of sales, and (iii) a big jump in
construction activity as a share of GDP as well. As a result, when the turnaround finally came and capital was
suddenly pulled out, property markets plummeted and the sector collapsed.

Not in China, however. When the debt boom ended in 2016 and the authorities stopped the party the
following year, the property sector went … just flat. As discussed above, sales simply stabilized. Housing
prices have continued to rise. Steel and cement demand slowed, but continued to increase very gradually in
the ensuing years. The construction sector sank as a share of the economy, but not in absolute terms.

Why? Because as we showed in Part 4, the Chinese property market was not a bubble. You did have the third
element mentioned above, i.e., a sharp rise in the construction share of GDP in the 2000s and early 2010s –
but not the first two. Income-adjusted housing prices were flat all through the 2000s and have actually fallen
all through the past decade. Mortgage leverage, as well, has been extremely stable on trend, never rising
above a very moderate 30% of new sales on a structural basis. This is why the sector didn’t fall apart when
China tightened five years ago. And while we do expect the Chinese property sector to continue to shrink in
relative terms and eventually in absolute terms as well going forward, there’s no pending collapse in the
wings.

5. Tightly closed capital account. We can’t make a list of contributing factors without mentioning capital
controls. After all, China’s capital account is not just “closed” – it’s almost sealed, with the most binding
controls regime of any EM country we follow (or know of) outside of perhaps Iran and North Korea. Which
also helps explain China’s ability to print debt at home without having to worry too much about external
capital movements.

(At least to date. We’ll have more to say about “hidden” renminbi risks shortly, and will discuss in detail in
Part 7.)

6. The “PBC put”. Finally, we need to mention the role of the pervasive “PBC put”.

Here’s what we mean. The two lines in Chart 44 show the closest thing we have to benchmark short-term
rates in China: the weighted average interbank bond repo rate (which falls between overnight and 7-day
rates in terms of average tenure), and the 7-day SHIBOR rate, which is the unsecured lending rate between
large “market center” banks.

37
 Jonathan Anderson
May 3, 2023

As you can see, in previous policy tightening rounds – of which there we many, even during the secular boom
period, e.g., 2007-08, 2011, 2013-14 – the PBC and regulators were not only restricting the volume of credit
via macro-prudential directives; they were also aggressively pulling liquidity out of the financial system. As a
result, even the shortest-term interbank money market rates jumped visibly in terms of average levels, with
regular and massive upward spikes as well.

Chart 44. Benchmark money market rates

Interbank rates (% per annum)

10%
9% Weighted bond repo
7-day SHIBOR
8%
7%
6%
5%
4%
3%
2%
1%
0%
03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Source: CEIC

But since the debt boom ended? As shown in the circled portion of the chart, rate levels barely rose at all
between 2016 and 2018, with only a meager pickup in volatility as well, with none of the dramatic spikes that
characterized the last few cycles. And since 2019, rates have been on a steady decline back towards historic
lows.

In short, the PBC may have brought down headline credit flows – sharply – since 2017 but they never really
tightened liquidity. Quite the opposite: every player in the Chinese financial system has enjoyed ample access
to funding throughout the process. You can see this in the news headlines as well. Back in 2011 and again in
2014 the press was in a frenzy over the first wave of defaults by financial lenders in the form of trust and
other investment products; informal “sidewalk” lending rates had shot up to high double-digit levels, and
small/medium corporate borrowers were in significant stress.

Whereas in the past seven years … nothing. Investment products have not exploded. Sidewalk rates are very
stable. And not a single shadow lender of any meaningful has been forced into illiquidity or default. Yes,
markets have been up in arms over selected defaults by Chinese conglomerates and developers, of which
Evergrande is the latest, but despite repeated rounds of small and medium bank restructuring and
consolidation there are no financial defaults.

Why hasn’t the PBC been tightening funding liquidity since the end of the boom? Precisely because it is now
so focused on deleveraging and financial cleanup. In an environment where all but the largest banks and
NBFIs are still struggling with heavy bad asset loads, and are not able to expand balance sheets, the PBC has
no other choice but to keep rates low and funding liquidity as plentiful as possible, to make sure that there
are no “accidents” along the way à la Lehman in the US.

38
 Jonathan Anderson
May 3, 2023

In short, the authorities have played it very well at home in recent years, having learned copiously from the
lessons of the US financial crisis as well as other EM crisis cases. They did not try to liberalize the capital
account, and did not try to impose liquidity shocks or “market discipline” on financial institutions. Instead
they have gone about quietly writing down exposures, accompanied by whatever liquidity injections needed
to keep things stable. And this has allowed the process to go on with a minimum of contagion risk.

39
 Jonathan Anderson
May 3, 2023

5. China’s “virtual credit” economy

[Note: The following text was originally published in Part 1 of this series; we are reprinting it here verbatim
from that report.]

A credit boom … and a growth bust? The issue is as follows: Between 2009 and 2017 China embarked on one
of the greatest credit bubbles in recorded history. You can use any lending aggregate you want – official
“total social finance” (or TSF), or our own measure of total financial claims in the economy; in all cases,
outstanding credit exploded upwards by more than 100pp of GDP during the period (Chart 45).

And yes, that’s a lot. In all of EM, only Hong Kong has surpassed the mainland Chinese mark in terms of
trough-to-peak credit expansion within any given ten-year period. Ever.

At the same time, however, the pace of real growth slowed. Dramatically. Again, between 2002 and 2007
China was posting average GDP growth of nearly 12% y/y, and more like 14% y/y according to our own
China Activity Index (Chart 46). By contrast, after a brief recovery spurt in 2009-10, by 2012 official GDP
had dropped below 8% and continued to fall to 6% by the end of the decade. And our Activity Index
averaged a meager 4% y/y between 2014 and the onset of the Covid pandemic in 2020.

Chart 45. The China debt boom Chart 46. The China growth bust
Total domestic claims oustanding (share of GDP %) Real growth (% y/y 3mma)
280% 20%
260%
240% 15%
220%
200% 10%
180%
160% 5%
Official GDP
140%
Activity index
120% 0%
100%
80% -5%
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 95 00 05 10 15 20

Source: CEIC Source: CEIC

That’s a trend growth deceleration of anywhere from 5pp to 9pp … during the biggest credit boom in EM
history. And no, this is most emphatically not how it works in the rest of the emerging world. Almost every
other recorded case of sharply rising private credit penetration in EM was accompanied by a significant
increase in the pace of domestic demand and growth.

Where the money went. Where did all that lending in China go? As most readers know, a large part of the
headline answer is stimulus spending in the form of new construction and infrastructure investment.

Sure enough, when we look at reported construction and infrastructure activity there was indeed a
tremendous notional explosion in China. The yellow bars in the charts below show the real volume of
reported activity, defined as the average of:

• New facilities construction and installation (in the monthly fixed asset investment accounts)

40
 Jonathan Anderson
May 3, 2023

• Real estate and infrastructure investment (in the monthly fixed asset investment accounts)
• Aggregate gross construction output
• Current property floor space under construction

In every case these figures rocketed upwards, rising nearly four-fold in real terms between 2008 and 2017 –
almost exactly commensurate with the four-fold increase in real domestic debt outstanding (Chart 47) as well
as a four-fold increase in new flow credit in real terms (Chart 48).

Chart 47. Debt stock vs. investment Chart 48. Debt flows vs. investment
Real index (2008=100 sa 3mma) Real index (2008=100 sa 3mma)
600 450
Construction/infrastructure investment Construction/infrastructure investment
400
TSF/govt outstanding TSF/govt flow (12mma)
500
Total claims outstanding 350 Total debt flow (12mma)

400 300

250
300
200

200 150

100
100
50

0 0
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22

Source: CEIC Source: CEIC

I.e., China had a massive credit boom and – according to these figures – an equally massive investment and
construction boom. So far the numbers add up just fine. What’s the problem?

Here’s the problem. Start with Chart 49 below, which shows the volume of reported construction volume in
China (the average of gross construction output deflated by investment prices and property floor space under
construction) plotted against actual domestic Chinese materials usage, as measured by electricity, steel,
cement and non-ferrous metals.

Wow. The materials consumption data tell a completely different story. As you can see, materials use did not
rise four-fold over the past decade. Indeed, it barely rose two-fold. Usage did jump visibly in the very first
years of the credit boom … but by 2012 the party was over, and numbers barely increased from 2013
onwards, even as the debt bubble continued to rage.

41
 Jonathan Anderson
May 3, 2023

Chart 49. Construction vs. materials use Chart 50. Investment vs. production
Real index (2008=100 sa 3mma) Real index (2008=100 sa 3mma)
450 450
Construction/infrastructure investment
400 Electricity usage 400 Construction/infrastructure investment
Steel product usage Domestic equipment usage
350 Cement usage 350
Overall goods production
300 300

250 250

200 200

150 150

100 100

50 50

0 0
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22

Source: CEIC Source: CEIC

The same is true for goods and equipment production in Chart 50. The yellow bars in the chart are the
average of reported new facilities investment and reported real estate and infrastructure investment in real
terms, while the two lines show (i) average real production across all industrial goods and (ii) net domestic
usage of machinery and equipment, defined as domestic production of general and special equipment less
net exports.

Again, actual equipment usage was far less than the reported investment figures would suggest – and again,
most of the jump in usage came in the first years of the boom, with growth fizzling out over the most recent
half-decade.

Or take domestic listed company capex in Chart 51. China’s listed sector is not necessarily representative of
the economy as a whole (there are only around 2,400 listed firms compared to hundreds of thousands of
enterprises covered by to the National Bureau of Statistics) – but they have the advantage of providing
audited income statements showing actual ex post financial flows, rather than just ad hoc numbers reported
up to the NBS.

Chart 51. Investment vs. listed capex

Real index (2008=100 sa 3mma)
450

400 Construction/infrastructure investment

350 Listed capex spending

300

250

200

150

100

50

0
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22

Source: CEIC, Bloomberg

42
 Jonathan Anderson
May 3, 2023

And, ahem, look at the difference. The yellow bars are again reported investment spending in real terms,
while the blue line is aggregate real capex volume according to listed company data:

You get the picture. Starting in 2008 China had a stupendous eight-year jump in borrowing and debt creation.
It had an equally stupendous jump in reported spending and activity. But when we look at actual spending
and physical activity in the economy, by any available metric … we see much, much less.

In short, something “disappeared” along the way. From borrowing and reported activity to actual spending,
there’s a huge black hole.

How big of a hole? It’s difficult to say with precision, but looking back at the charts above it appears that
roughly half of all increased infrastructure, real estate investment and construction-related borrowing since
2008 somehow fell by the wayside. By our estimates, the cumulative increase in debt related to these
categories exceeded 40% of GDP by 2017, which implies a cumulative net loss of perhaps 20% to 25% of GDP
… or more than US$2 trillion of “missing” funds. This is a wildly rough estimate, but it doesn’t matter. The
point is that in any event, it’s enormous.

Which naturally raises the question: How does one “lose” a few trillion dollars?

In macroeconomic terms, what we’re really saying is that the money was saved, not spent. In other words, it
was just recycled directly back to domestic deposits or other financial asset holdings without having an
impact on real demand and real activity. This is why the record Chinese credit bubble did not have a net
positive impact on overall growth, why the economy slowed rather than accelerated during the past decade.

The virtual credit economy. Ok, but why would anyone borrow trillions of dollars and then … just stick it back
the bank?

Here’s a hint. Please turn to the next chart:

Chart 52. Financial claims by borrower

Total domestic financial claims (% of GDP)
300%
Government
250% "Quasi-fiscal"/other
Household
Indentified corporate
200%

150%

100%

50%

0%
03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Source: CEIC

43
 Jonathan Anderson
May 3, 2023

Chart 52 shows total outstanding domestic financial claims as a share of GDP with an estimated breakdown
by borrower. The blue bars show estimated lending to the “traditional” corporate sector (defined using
available enterprise balance sheet data, please see footnote for details). 4 The green bars are for borrowing
by households. The yellow bars show direct borrowing by central and local governments … and the red bars
indicate implied credit to other borrowers.

What do we mean by “other borrowers”? Essentially we are talking about “non-traditional” corporate
entities, most often affiliated with local governments in one form or another: LGFVs, local government-run
infrastructure and development funds, ad hoc stimulus-related construction firms, PPPs and others, including
both direct lending from banks and a huge jump in “hidden” exposures in the form of DAMPs, TBRs and other
credit instruments intermediated through shadow NBFIs. We don’t have direct data on these entities’
balance sheets or debt liabilities – but we can see their impact in the residual once all other borrowers are
accounted for.

What is the chart telling us? It’s telling us that China’s big debt boom was heavily quasi-fiscal in nature. Yes,
household and traditional corporate borrowing picked up steadily as well, but they only account for half of
the cumulative increase. If we measure from 2008 to the debt peak at the beginning of 2017, the real story is
in the red and yellow bars above: an eruption of implied borrowing by local governments and related
corporate and quasi-corporate entities, which rocketed from less than 10% of GDP in 2008 to nearly 70% of
GDP over the period.

This stands in dramatic contrast to the other EM credit bubble countries we discussed above. In virtually
every other case the lending explosion was overwhelmingly private in nature (with the sole exception again
of Vietnam, which saw a big boom in state enterprise and other quasi-official borrowing).

What’s the difference between a private credit boom and a quasi-fiscal credit boom?

Two things. First, in a private credit boom, households and firms borrow because they have something they
want to purchase or build. In a Chinese “policy-led” credit boom, local agencies, infrastructure and
development vehicles load up on debt because they are required to – and then cast around for stimulus
projects that they can perhaps spend on.

And second … well, as anyone in the developing world can tell you, public funds just have a certain way of
getting mislaid.

Fair enough. But can these two factors really account for trillions of dollars worth of funds in China?

Judge for yourself. On the first front, we already showed the increase in credit claims outstanding as a share
of GDP above. Now let’s show the corresponding trend for banking system deposits:

4 Chinese banks do not break out overall credit claims by corporate segments. Instead, we have used directly reported data on overall

debt liabilities of (i) all industrial enterprises, (ii) all non-financial SOEs and (iii) all listed firms to estimate the growth of credit
outstanding to the “standard” corporate economy.

44
 Jonathan Anderson
May 3, 2023

Chart 53. Deposits by category

Banking system deposits (% GDP 12mma)
220%
Other govt/quasi-fiscal
200% Fiscal
Other
180%

160%

140%

120%

100%

80%
07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Source: CEIC

As shown, from the trough in 2008 to the 2017 peak, total system deposits rose by nearly 60% of GDP. And
more than half of that increase, or 30pp of GDP, came from an explosion of deposits of non-budgetary
government and government-affiliated organizations (the yellow bars in the chart) – not traditional state-
owned enterprises, which are counted under normal corporate deposits, and not formal central or local
government deposits, which are shown in the green bars, but rather “other” government and quasi-
government players. Deposits of these entities were almost nothing in 2008, but by 2017 they had reached
RMB27 trillion, or nearly US$4 trillion dollars, before subsequently stabilizing and fading as a share of GDP.

Needless to say, this government/quasi-government “savings binge” corresponds exactly in terms of timing
with the fiscal/quasi-fiscal “debt binge” we saw above. Keep in mind that the time series above is not
perfectly reliable; it appears that some of these government-affiliated deposits were previously classified as
normal enterprise funds and gradually moved over to the new category, so we don’t want to overstate the
trend. Nonetheless, however, the point is clear: the same class of entities that did most of the borrowing in
China account for most of the deposit growth as well.

But why take all that money and then just hold it? When we first noticed this very odd behavior in 2009 and
2010, we assumed it was simply “hoarding”, i.e., local governments taking advantage of Beijing’s sudden
emergency lifting of all credit restrictions at end-2008 and the accompanying strident marching orders to
borrow and invest by any means possible by taking on far more debt than they could possibly spend in one
year – i.e., simply so that the money would be available through future periods of tightening, if and when
Beijing changed its mind.

However, as time wore on, with an ever-widening gap between reporting and reality in real activity and an
ever-increasing pile of quasi-fiscal deposits, we began to add an alternative explanation: stealing.

Think about it. Again, every local government wakes up one morning in 2009, and again in 2012, and again in
2015, to find that the central authorities have completely lifted all forms of credit restrictions, and that
regions are now mandated as an urgent priority to lever up and find ways to spend on social housing,
infrastructure, environmental improvement and urban redevelopment in order to promote growth at home
and counter the effects of a weak global economy abroad. Moreover, the preferred vehicles for doing all of

45
 Jonathan Anderson
May 3, 2023

this are off-budget, non-official commercial entities that are almost completely opaque: no regular reporting
or statistical coverage, completely beholden to the provincial, city and county officials who oversee them …
and maybe once every couple of years an auditor might show up to see what’s going on.

What would you do? Given the incentive structure in China, of course you would borrow and take on projects
on a grand scale, quickly reporting starts and the maximum possible amount of activity underway in order to
show compliance and meet quotas. But with no one watching the till, it would be awfully hard to resist the
temptation to side-track the funds, put them away in related official accounts or pay them out through
padded contracts to other connected suppliers and friends, who themselves then hide the money in deposits
as well. (And keep in mind that – in contrast to, say, Russia, where excess earnings are routinely sent offshore
as a matter of course – China has an extremely closed capital account, making it very difficult for funds of any
meaningful size to leave the country, with the result that pilfered funds are forced into “captive” savings at
home almost by definition)

Which, according to the data, is essentially what’s going on: an absolute maximum of reported activity, an
absolute minimum of actual spending and implementation, and an absolute maximum of embezzlement and
waste along the way.

Summing up. This, in sum, is the core of the Chinese “virtual” credit hypothesis. During the 2008-17 lending
boom, the policy-led nature of the upturn incentive structures that were skewed to produce:

• a maximum amount of debt and leverage creation, and

• a maximum amount of “reported” economic activity, but
• a minimum amount of actual spending and real demand in the system.

Instead, borrowed funds are side-tracked directly back into financial savings, whether through pilfering,
misappropriation or sheer hoarding of resources.

This is how you get a big credit bubble with only little growth. And, since the 2017 peak and subsequent
deleveraging, how you get a credit bust with no economic collapse.

46
 Jonathan Anderson
May 3, 2023

6. Debt, the impossible trinity, and the end of the renminbi peg

Which brings us to the final, short section of this report, where we want to return briefly to question of the
credit boom, the renminbi exchange rate and currency risks.

In the section on crises above we talked about external risks in terms of the current account position and
external liabilities, and again, China ranks as extremely safe on both metrics. The country continues to run
current and “basic” BOP surpluses; the financial system is a net foreign creditor, and economy-wide gross
external debt is very low by EM standards. Simply put, there’s nothing here that would flag any troubles on
the exchange rate front, and sure enough, the renminbi has been one of the best-performing units in EM
over the past decade.

There is, however, another way of looking at currency risks – one that is directly related to the debt, money
and credit creation. And this shows a very different, more alarming picture.

In order to set the stage we need to review the “impossible trinity”, i.e., the well-known adage in
international macroeconomics that a country cannot do the following three things at the same time:

• Control monetary and financial conditions at home

• Control the value of the exchange rate
• Maintain an open capital account

And the point is this: the “trinity” is not just an adage. It’s very real, and operates across the entire emerging
universe.

Consider first the two EM-wide charts below. Chart 54 on the left shows the average path of (i) FX reserves
and (ii) broad money, both expressed in US dollar terms, for major EM economies that successfully peg or
heavily manage their currencies, a sample that includes Hong Kong, Malaysia, Singapore, Korea, Taiwan,
Thailand, Saudi Arabia, UAE, Bulgaria, Czech, and others.

Chart 54. EM fixers Chart 55. EM floaters

USD index 2010=100 USD index 2010=100
350 350

300 300

FX reserves FX reserves
250 250 Broad money
Broad money
200 200

150 150

100 100

50 50

0 0
00 02 04 06 08 10 12 14 16 18 20 22 00 02 04 06 08 10 12 14 16 18 20 22

Source: CEIC Source: CEIC

47
 Jonathan Anderson
May 3, 2023

What is the chart telling us? Simple. If you want to fix your currency you have to follow the well-known “peg
rule”: i.e., you have to let the domestic money stock expand or contract in line with FX reserves. In essence,
you give up domestic autonomy in order to achieve exchange rate stability.

So far so good, and this relationship is well understood.

What many readers don’t realize, however, is that it’s exactly the same for floating countries. Just look back
at Chart 57 on the right, which shows the path of FX reserves and dollar-adjusted broad money for major EM
“floaters”: Argentina, Brazil, Chile, Colombia, Hungary, India, Indonesia, Israel, Mexico, Peru, Philippines,
South Africa, Turkey, etc.

As you can plainly see, it’s no different at all – except that the causality runs in the other direction. If
domestic broad money aggregates get too far “out of whack” with the overall external balance (i.e., the path
of FX reserves), currencies inevitably adjust in an offsetting manner to ensure that the dollar-adjusted money
stock falls right back into line. The end result is that peg “discipline” still holds ex post, just through the
exchange rate rather than domestic adjustment.

In short, this is just the impossible trinity at work once again: If you want to control your currency you have to
give up monetary autonomy, and if you want domestic monetary independence you have to let the currency
adjust freely.

And this is the case almost everywhere in the world; the above charts hold for pretty much the entire EM
universe ….

… except, as it turns out, for China. Here’s the corresponding chart for the mainland economy:

Chart 56. China

USD index 2010=100
500
450
400 FX reserves
Broad money
350
300
250
200
150
100
50
0
00 02 04 06 08 10 12 14 16 18 20 22

Source: CEIC

Whoa. The two lines moved together in the 2000s, but once the debt boom got underway, they began to
diverge sharply. FX reserves effectively stopped rising ten years ago and have been flat or falling since –
whereas the broad money stock has gone up … and up, and up, more than quadrupling in dollar terms over
the past decade alone. (And it’s important to note that while the process slowed after China tightened policy
in 2017, it has not stopped. Indeed, the biggest divergence of all has occurred over the past 18 months, as the

48
 Jonathan Anderson
May 3, 2023

renminbi and other global currencies strengthened sharply against the US dollar, pushing up the dollar value
of the local money stock, which now stands at nearly US$40 trillion in terms of M2 and an astounding US$55
trillion for overall banking system liabilities.)

What’s going on?

What’s going on is that China is trying to “have its cake and eat it too”, by balancing a heavily managed
“basically stable” exchange rate and the world’s most colossal expansion in domestic liquidity at the same
time. How does that that work?

Duh. By cutting off the third leg of the impossible trinity. Over the past half-decade China has abandoned its
earlier renminbi liberalization policies and aggressively tightened capital account controls … particularly,
needless to say, as they relate to currency outflows.

And as we will argue in the upcoming Part 7 installment, this can’t go on forever. At some point even minor
shocks and leakages in the capital account regime would be sufficient to overwhelm existing foreign
exchange reserves and force a destabilizing break in the exchange rate. Simply put, it won’t be external
deficits or funding liabilities that eventually take down the renminbi; it will be the sheer massive accumulated
weight of local liquidity that is steadily building up behind capital walls.

But again, we’ll go through the full details in that report. Stay tuned.

© 2023 Emerging Advisors Group Limited

49
 Jonathan Anderson
May 3, 2023

Emerging Advisors Group is a China-based macroeconomic research firm covering emerging markets in Asia, Latin
America, Europe, the Middle East and Africa.

EM macro research
Jonathan Anderson jonathan@[Link]
Mandy Ong mandy@[Link]
(Website) [Link]

We also work in partnership with Applied Global Macro Research, an independent research advisory focused on
developed/global macro and market trends with offices in Colorado, USA and Copenhagen, Denmark. For
information on AGM Research product offerings and subscription terms, please refer to their website and contact
details below.

Carsten Valgreen [Link]@[Link]

Niels-Henrik Bjorn [Link]@[Link]
Mary Allyn [Link]@[Link]
(Website) [Link]

50