China’s Debt Hangover

China’s debt problem is mainly a local one. This product, updated as of May 2021, contains both a “Debt Stress Indicator” (the first map) and a “Debt Drag Indicator” (the second map).

The stress indicator measures the extent to which each region is facing pressure to deleverage and ranks each region according to the stress score. The drag indicator ranks each region based on its existing debt’s drag on local GDP. You can also click on the rankings to see a more detailed breakdown of how each province stacks up nationally in terms of debt.

For detailed explanation of this product and its data, see the Overview.

Debt Stress Indicator

 High Stress

 Medium Stress

 Low Stress

Debt Drag Indicator

 High Drag

 Medium Drag

 Low Drag


More than a decade after the global financial crisis, China is still reeling from the debt hangover that was largely caused by the explosion of local government financing vehicles (LGFVs). The expansion of off-budget borrowing through the LGFVs was officially tolerated because it was meant as an emergency stop-gap effort to stimulate the economy after 2009.

That lending was more regulated as late as 2014, when Beijing still formally kept a tight lid on borrowing and mandated local governments to run balanced budgets. But Beijing soon realized that the LGFV genie couldn’t be put back into the bottle.

Indeed, the LGFV debt/GDP ratio has grown through 2020, up more than 30 percentage points since 2008. Debt growth is not necessarily a problem in and of itself, so long as the borrower maintains the ability to service that debt.

What is troubling is the fact that many LGFVs’ debt-financed investments can no longer service that debt, indicating that LGFVs have not invested wisely. In fact, returns on LGFV investment are on average below 2%, far lower than the cost of borrowing.

These lenders’ poor returns would have persisted as long as Beijing was willing to kick the can down the road. That’s because dealing with debt was always going to be a political decision since it touched the third rail of local government fiscal health.

A high level of debt is not necessarily correlated with debt stress. Those above the line are generally fiscally stronger provinces that are more capable of servicing their debt, hence lower stress. Those below the line are fiscally weaker provinces, hence higher debt stress even with lower levels of debt.

But Beijing’s political calculus has changed. Having emerged from the pandemic, Beijing is doubling down on deleveraging, believing there exists a “window of opportunity” to tackle LGFV debt. We assume that in this political environment, regions and provinces must respond. In other words, it’s not a matter of whether, but how regions will deal their debt in coming years.

The central government’s newfound hawkishness on deleveraging will put tremendous stress on money-losing LGFVs that are essentially holding onto nonperforming assets. Given the scale of the LGFV debt challenge—totaling around $10 trillion—deleveraging will have major implications for regional growth.

How a region manages its deleveraging process will be one of the most important determinants of local economic performance. Fiscally strong regions will have the ability to control the pace of deleveraging, which will allow them to engineer an economic soft landing. In contrast, fiscally weak regions with plenty of money-losing LGFVs are going to have a rude awakening.

This is why we at MacroPolo sifted through 10 years of financial reports from more than 2,500 LGFVs and created both the Debt Stress Indicator and the Debt Drag Indicator. The first and new indicator aims to measure and rank the extent to which each region is facing LGFV deleveraging pressure. The second and updated indicator measures how much existing debt is dragging on the local economy.

What’s Different about This Product?

While other analyses of LGFV debt exist, we took a more refined approach to correct for bias and account for regional variation. We aim to present as realistic and accurate a picture of regional LGFV debt’s impact on the real economy as possible. Such province-by-province adjustments and extensive data imputations are important in ensuring the accuracy of the results and avoiding large errors due to bias in the data.

Our sample of 2,526 LGFVs accounts for 80% of total LGFV debt. To correct the problem of underrepresenting financially weaker LGFVs or those in economically weaker regions in our sample, we conducted substantial data imputation.

Some analyses simply assume that such under-representation of small and non-bond issuing LGFVs is evenly distributed across the country, when in fact strong evidence suggests that regional variation is substantial. For regions like Inner Mongolia or Liaoning, for example, bond-issuing LGFVs likely only account for less than 40% of total LGFV debt. In other municipalities like Beijing or Shanghai, bond-issuing LGFVs account for nearly all local LGFV debt. As a result, it is important to account for such variation and adjust regional LGFV debt accordingly.

For inquiries and suggestions, please contact Houze Song (

Learn more about the Debt Stress Indicator

The indicator is calculated by multiplying 1) the cost of borrowing with 2) the debt burden, as represented by this equation:

LGFV bond interest rate x (LGFV debt + local government debt)/tax revenue = Deleveraging Stress

The higher the borrowing cost and more severe the debt burden will lead to a higher stress score for the region. The score for each region, then, indicates the pressure to deleverage, which by implication means those regions with higher scores are more likely to experience an economic hard landing in the coming years.

Both variables are needed to construct a sound measurement of deleveraging stress. Simply relying on the cost of borrowing isn’t sufficient because many LGFVs don’t borrow from the bond market, making it difficult to compare across regions. It’s also difficult to get a sense of the potential scale of deleveraging from the cost of borrowing. In principle, one region could have higher borrowing cost but fewer LGFVs than another province, which means that the scale of its deleveraging will be smaller and have less of an impact on the local economy.

So an additional variable of the local debt burden is needed for better cross-regional comparisons and a better estimate of the true scale of deleveraging. To do so, we included debt of those LGFVs that did not release financial statements and were not included in our dataset. We first imputed the total size of LGFV debt that didn’t issue financial statements for each region, and then estimated the portion of that debt that is nonperforming and will have difficulty getting refinanced.

Using the final stress scores we ranked China’s 31 provinces and regions and grouped them into three categories: high stress, medium stress, and low stress.

Another way to interpret deleveraging stress is to think of it as the annual debt servicing burden as a percentage of provincial tax revenue. For example, a 0.1 stress score means the region needs to spend 10% of its tax revenue to service debt. That’s because we assume all money-losing LGFVs need fiscal support to pay back their debt, which in turn puts stress on regional fiscal coffers (actual debt servicing burden will likely be smaller in reality).

The more tax revenue a province needs to cover its LGFV debt implies that its fiscal condition will deteriorate, signaling to creditors that the LGFV supported by that local government is more likely to default. As such, creditors will put more pressure on these provinces to force deleveraging rather than continue to roll over their debt.

Stress Indicator Methodology

Data source

The LGFV data is accessed through Wind, which contains the financial reports of all LGFVs that have ever issued a bond. Any bond-issuing LGFV that is the subsidiary of other bond-issuing LGFVs has been removed to avoid double counting.

LGFV cost of borrowing

The average LGFV cost of borrowing is calculated as the average credit premium plus the 10-year central government bond yield. In particular, we used the credit premium of AA+ rated LGFVs because entities with this credit rating most closely represent the typical LGFV. To derive the average LGFV cost of borrowing, we added the credit premium to the average 10-year central government bond yield in the past decade.

Although bonds account for less than 20% of total LGFV debt, we believe the bond yield is a more accurate measurement of LGFV default risk. This is because the majority of LGFV financing is provided by state banks, and they have incentives to provide artificially cheap financing to LGFVs in spite of default risks.

Regional Debt Burden

The regional debt burden is composed of the sum of 1) all money-losing LGFV debt and 2) total on-budget local government debt. The reason for including on-budget debt is that local governments with high on-budget debt are less capable of supporting troubled LGFVs.

That sum is then divided by the regional tax revenue, which is estimated based on the total local tax revenue minus the mineral tax revenue. We chose tax revenue instead of total fiscal income because tax revenue is a more stable and accurate measurement of local fiscal revenue and does not include other sources of revenue such as the sale of state assets.

One such state asset is land, which has frequently been used to repay LGFV debt. We decided to exclude land from local fiscal revenue, in part because such revenue is highly concentrated—the top 30 cities receive around 40% of national land sales revenue. In addition, since land sales and tax revenue are highly correlated, for the sake of cross-regional comparison, looking at tax revenue alone is a sufficient gauge of fiscal capability.

The mineral tax revenue is subtracted because it is highly pro-cyclical, meaning that it’s at its lowest level when the economy is bad and LGFV default risk at its highest. Going forward, the mineral tax revenue will likely grow slower as the Chinese economy becomes less resource intensive. In 2019, the total national mineral tax revenue was 10% lower than its previous peak in 2011. As a result, regions that are highly reliant on the mineral tax will likely experience revenue stagnation in the coming years.

Finally, it is worth noting that we make stringent assumptions on debt and fiscal revenue to both improve the accuracy of the stress indicator and to make it more comparable across regions and time.

Learn more about the Debt Drag Indicator

The indicator is calculated by multiplying 1) the difference between LGFV return and LGFV cost of capital and 2) LGFV debt (percentage of GDP), as represented by this equation:

(LGFVc – LGFVr) x LGFV debt (percentage of GDP) = Debt Drag Indicator

The result for each province/region can be interpreted as the annual financial loss (as a percentage of local GDP) from LGFV debt. For example, a 0.01 score means the region suffered an annual loss equal to 1% of GDP. In some provinces and municipalities, such as Guizhou and Beijing, the loss may not fall entirely on the local economy since much of the lending is from national state banks.

Based on this indicator, we ranked China’s 31 provinces and regions and grouped them into three categories: high drag, medium drag, and low drag.

This indicator is not a measure of the overall financial vulnerability of a province. For those interested in financial risk, they should instead use the Debt Stress Indicator.

Drag Indicator Methodology

Data source

The LGFV data is accessed from Wind, which contains the financial reports of all LGFVs that have ever issued a bond. Our sample contains more than 2,500 LGFVs with data covering 2009 to 2020. Any bond-issuing LGFV that is the subsidiary of other bond-issuing LGFVs has been removed to avoid double counting.

LGFV return

LGFV return is calculated by dividing net cash from operating activities by total assets. This is because, compared to profit, net cash from operating activities is less prone to manipulation and represents a more accurate picture of LGFV performance.

LGFV cost of capital

LGFV cost of capital is calculated by dividing annual interest payment (based on cash flow statements) by the total amount of interest-bearing debt.