The Setup:

China watching, even from a data and charts angle as we do in this newsletter, often feels akin to the blind men and the elephant parable — you feel your way around in the darkness, piecing together a picture, wondering how close or off the mark you really are. But in the land of the blind, the one-eyed man is king, so fingers crossed that this week’s charts will spark some valuable connections for our readers.

We build on the local government financing vehicles Fragility Index from the last post — “LONG VIEW: LGFV Fragility” — to construct an aggregated national index. We make some interesting inferences from additional real estate charts and adjust the index using a proxy indicator. Data files attached.

Contents:

1. Some Context: Real Estate Woes Finally Bite

2. National LGFV Fragility Index

3. Inferences & Adjusted Fragility Index

4. Data Files (.xlsx)

5. Bonus Chart

1. Some Context: Real Estate Woes Finally Bite

December data came out to close out the year, which was picked up by the WSJ in this chart. The year 2022 clearly trumps 2020-21 in severity.

2. National LGFV Fragility Index

As a quick reminder, using publicly available macro data, we built a Fragility Index for 33 cities in China to be used as a backdrop proxy for Local Government Financing Vehicle (LGFV) fragility. See the previous post for a breakdown of index components and equations.

Previously, we only provided a point-in-time cross-section of the index. After sourcing city-level year-by-year population data, we built a time series version of the index. Note that some cities have somewhat different index readings (e.g. Zhengzhou) from the last post on account of this — city-level populations are given for metro areas, which are smaller than the broader population numbers we found previously. Also, please note that time series data for land sale dependence at the city level is sparse and hard to come by. As such, the 2021 point-in-time readings from the Bloomberg chart shown in the previous post were used throughout, meaning that that particular index component does not have an impact across time in the calculations.

And here’s the elephant. By aggregating the 33 cities, we build a national LGFV Fragility Index. There are two aggregation methods — a simple average of all the cities (cyan) and city-level population-weighted (red). Together, the 33 cities sum up to 187 million people, i.e. 13.24% of China’s total population. So a decent sample but not entirely comprehensive. Remember that city-level data is only ever available for the larger cities. We suspect many smaller cities are doing worse than those we have data for.

So we’ve put numbers on the elephant. The question to ask now is: judging by the chart, are we still below the 2015-2016 highs in fragility? As mentioned, without a city-level land sales dependence time series, we can't know for sure. But we can make inferences from another real estate indicator and use it as a proxy for land sales dependence. We can then adjust the Fragility Index to reflect this. More on this below.

3. Inferences & Adjusted Fragility Index