Discussion of “The Great Wall of Debt: Real Estate, Political Risk, and Chinese Local - PowerPoint Presentation

Slide1

Discussion of “The Great Wall of Debt: Real Estate, Political Risk, and Chinese Local Government Credit Spreads”Andrew Ang, Jennie Bai, and Hao Zhou

Jasper HoekBoard of Governors of the Federal Reserve System2016 Pacific Basin Research ConferenceCenter for Pacific Basin Studies, Federal Reserve Bank of San FranciscoNovember 18, 2016Disclaimer: The views expressed in this presentation are those of the presenter and do not necessarily represent those of the Board of Governors or the Federal Reserve System.Slide2

BackgroundLively policy debate on role local governments play in China’s debt expansion, but little researchProblem: lousy dataNational AuditLocal government financing vehicle (LGFV) bondsAng, Bai, and Zhou (ABZ) is one of the first papers to make use of LGFV bond dataSlide3

This paper’s storyLocal government financing vehicles’ (LGFV) credit spreads are determined by the creditworthiness of the local governments backing them.Use of land-use rights as collateral for LGFV bonds links them to local real estate markets.Local governments have much control over local real estate markets via land sales, which drives economic development. In principle, depending on how astutely local governments exert this control, real-estate-driven development may either increase or reduce their creditworthiness, and, in turn, the creditworthiness of the LGFVs they implicitly guarantee. In practice, it appears to increase local governments’ creditworthiness.Local governments’ control over land-use rights also opens the door to rent-seeking and hence political corruption.In principle, corruption could facilitate more deal-making, thus promoting economic development, but in practice it appears to hinder economic development and thus impair local governments’ creditworthiness.Moreover, corruption mitigates the beneficial impact of real-estate-driven development on local governments’ creditworthiness. Slide4

The evidenceDespite the existence of implicit government guarantees, LGFV credit spreads vary across provinces and over time.Regression of LGFV excess bond yields on province-level real estate and corruption proxies, and their interaction, shows that credit spreads are:Negatively correlated with importance of real estate in a provincePositively correlated with presence of political corruptionPositively correlated with the interaction of corruption and real estate

developmentSlide5

The implicit guaranteeOne of the most interesting contributions of the paper is to add nuance to the notion of the implicit guarantee in China.Common perception is that the implicit guarantee is very strong because bonds are rarely allowed to default.

But the fact that: (a) collateral is required at all, and (b) bond prices vary with the value of collateral suggests that, contrary to popular perceptions, there is some market discipline on local government finances.

But doesn’t tell us how strong investors perceive the implicit guarantee to

be.

For that, we need to know the counterfactual of yields on similar debt that doesn’t benefit from a

guarantee.

The

story

ABZ

tell is focused on the strength of the local governments ability or willingness to back the debt. But there is still the question of why the central government guarantee of local government debts doesn’t eliminate variation in spreads.Slide6

The real estate effectFocus on explaining cross-sectional variation in bond yields exposes estimates to potentially serious omitted variable bias.Example

: Real estate GDP strongly correlated with GDP per capita across provinces, which presumably affects credit risk but is not controlled for.

Would

be more convincing if controlled for province fixed

effects. Unfortunately, not

much within-province

variation over time.Slide7

The real estate effectReverse causality—low yields may drive investment in real estate, rather than vice versa. Could use lags of real estate GDP but, again, not enough time variation in this measure.

Province-level real estate GDP is a pretty noisy measure of the local real estate market conditions that may be affecting the credit risk of a given LGFV.

Most

LGFV debt is at the municipal and county level

.

The precise connection between real estate GDP and credit spreads is murky.

ABZ’s “growth engine” story: real estate development -> stronger tax revenues -> stronger fiscal position of backing local government -> implicit guarantee worth more -> lower credit spreads

Alternative real estate-driven story: real estate market strong -> collateral worth more -> lower credit spreadsSlide8

The corruption effectSimilar issues apply to the estimates of the effect of corruption. Ideally, would control for province fixed effects, but corruption proxy does not vary over timeProvinces that are poorer and have weaker institutions may be more susceptible to corruption and also have higher credit risk. But this does not imply that corruption is driving up credit spreads by “destroying value”.What to make of the fact that corruption appears to explain some cross-sectional variation in credit spreads but credit spreads don’t generally rise after corruption announcements?Taken at face value, suggests that graft probes don’t in and of themselves reveal any information and/or affect the likelihood of default.Another interpretation is that corruption itself does not have much of an effect on credit spreads, but is correlated with unobserved variables in the cross section.Slide9

Some suggestionsTie municipal LGFV credit spreads to real estate market conditions in that municipality, e.g., property prices, sales, land sales? More time variation, closer correspondence with LGFV riskUse non-LGFV SOEs and/or private firms in the same municipalities as “control” groups? Still problematic, but could give more insight into strength of implicit guarantee on LGFV debt.E.g., Compare LGFV and non-LGFV reaction to regulatory announcements cracking down on implicit guarantee.Tie corruption probes more closely to specific municipalities or firms. Include LGFV-level controls in the analysis (e.g., leverage, profitability).Slide10

Policy implicationsBond swap currently underway appears to be imposing a haircut on holders of debt -> Investors were right to price in credit riskBut there appear to be few repercussions for local governments Highlights importance of the fiscal reform to address local government incentives.Proving very hard to implement because local government have significant financing needs and central government relies on local governments to prop up growth.