Journal Issue

People’s Republic of China—Hong Kong Special Administrative Region Selected Issues1

International Monetary Fund. Asia and Pacific Dept
Published Date:
January 2017
  • ShareShare
Show Summary Details

Hong Kong Sar Banks’ Nonbank Mainland China Exposure1

A. Size of Exposure to Mainland China and Debt at Risk

1. Hong Kong SAR, as the international financial center serving China and a significant portal connecting China with the rest of the world, has sizable exposure to China. At the end of June 2016, Hong Kong SAR banks’ claims on nonbank China exposure—a broad measure that includes trade finance, lending to Mainland Chinese firms, overseas firms operating in Mainland China as well as some on- and off-balance sheet exposures—totaled HK$4.6 trillion (US$586 billion or 187 percent of GDP).2 Of these, claims on nonbank borrowers in China totaled HK$939 billion (US$121 billion or 39 percent of GDP). Meanwhile, claims on onshore Chinese banks totaled HK$1.5 trillion. Notably, this interbank exposure has fallen dramatically since 2015; the current exposure is only a little over half of its peak in mid-2014. Meanwhile, although current nonbank exposure remains near its peak, the rate of growth has slowed rapidly since 2015; over the first half of 2015, there was no lending growth in aggregate. The reduction in China exposure since 2015 may be due to a slowdown of the Mainland economy and some unwinding of currency exposure from Mainland corporates.

Exposure to China

(In billions of HKD)

Source: HKMA

Hong Kong SAR banks are exposed to potential losses from the Mainland

2. Given growing interconnectedness between Hong Kong SAR and the Mainland, Hong Kong SAR banks are exposed to credit risks from their Mainland lending. By one measure, potential credit loss for Hong Kong SAR banks could reach 2.33 percent of total loan portfolio if Mainland real growth rate falls to 4 percent in one year.3 To assess banks’ vulnerabilities from a granular perspective where the composition of lending portfolios is taken into consideration, staff combined data from the HKMA and private analysts to gauge, as a rough estimate, Hong Kong SAR banks’ Mainland exposure by borrower type–SOEs, private entities and non-Mainland entities—as well as the sectors where these loans were deployed. This more granular approach offers a baseline from which we could assess the varying degrees of vulnerabilities across different types of lending portfolios. In this analysis, we used current data—as of Q1:2016—to establish a baseline against which developments could take place; the analysis here focused on the composition of exposure by borrower type and economic sectors without the additional layer of downside risks. Additionally, we used the metric of ICR<1—the same as the approached adapted by the April GFSR—to signal potential corporate balance sheet stress. Although this approach is narrow and may understate debt at risk, it is intuitive—if operating cash flow is insufficient to cover interest expenses, it signals balance sheet stress.

3. Staff analysis suggests that relative to potential debt at risk for Mainland firms themselves, Hong Kong SAR banks, in aggregate, are less exposed to losses from Mainland firms—and more specifically, losses tied to Mainland-related lending. For reference, A-shares have a debt-at-risk ratio (relative to sample firms’ total liabilities) of 12.75 percent (based on 4-quarter moving averages), 4 or 1.42 percent of Hong Kong SAR’s total banking sector assets.5,6 By comparison, even under an extreme stress scenario where we assume no recovery—and using the same sectoral vulnerabilities as seen in A-shares—the estimated exposure to Hong Kong SAR banks is smaller, at 9.52 percent for loans used in Mainland (text chart below).7 This lower debt-at-risk ratio for Hong Kong SAR banks relative to A-shares, despite using the same debt-at-risk ratios for industry sectors, suggests that Hong Kong SAR banks in aggregate lend less to more at-risk firms on Mainland. Moreover, the assumption used—assuming that Mainland borrowers from Hong Kong SAR banks have the same balance sheet vulnerabilities as the average A-share companies in the same sector—suggests that estimates derived could represent the upper bound of potential bank losses from Mainland lending. That said, if downside risks—including a sizable growth contraction in China, and sharp increases in corporate insolvency—materialize, the knock-on effect on Hong Kong SAR banks could be sizable and exceed the estimates presented here, particularly given increasing interconnectedness between the Hong Kong SAR and Mainland economies.

4. Since 2014, guidance and close monitoring of bank lending activities, prompted in part by increased exposure to Mainland companies, in addition to existing strict credit approval processes and limited country and industry exposures have eliminated some of the risks of nonperforming loans.8 Therefore, it is reasonable to assume that Hong Kong SAR banks have minimal exposure to these worst-performing firms. If we trim the worst-performing borrowers—the bottom 10 percent—from our sample, Hong Kong SAR banks’ exposure falls to 5.8 percent or 1.07 percent of Hong Kong SAR’s total banking sector assets. Interestingly, for Mainland A-shares, the fall in ratio is larger—from 12.75 percent to 4.4 percent; this suggests that vulnerabilities in Mainland corporates are concentrated—not surprisingly—in the worst-performing firms. The more modest decline in Hong Kong SAR banks’ exposure suggests that Hong Kong SAR banks in general are less exposed to these riskier firms, though individual bank’s exposure could still be sizable. Indeed, staff analysis suggests a high degree of variation across banks.

5. Applying a loss ratio of 60 percent—the same ratio applied to the April GFSR analysis of Chinese corporates—to the analysis, debt at risk falls for all categories. For Hong Kong SAR banks, the ratio falls from 9.5 percent to 5.7 percent; for A-shares, 12.7 percent to 7.7 percent. Applying a less stringent loss ratio of 45 percent—also the Basel standard—lends an even lower debt at risk of 4.3 percent and 5.7 percent for Hong Kong SAR banks and A-shares, respectively.

Comparison of Debt-at-Risk Ratio and Potential Losses

(Percent of loans, 4-quarter moving average)

Sources: HKMA; WIND; and IMF staff estimates.

Exposure is concentrated in real estate, large SOEs and multinationals

6. In aggregate, Hong Kong SAR banks’ exposures are not very heavily concentrated in particular industries. Exposures to “old” industries such as mining and textiles are small. However, lending to property developers and the construction industry account for a relatively high portion of total Mainland-related lending. There is also sizable exposure to the manufacturing sector as well as the wholesale and retail sectors.

7. Meanwhile, about 40 percent of Mainland-related lending in Hong Kong SAR went to SOEs, with another 40 percent to multinational companies operating in China and the rest to top-tier listed Mainland private firms. In general, these borrowers tend to have strong credit fundamentals. Multinational companies, in particular, are international corporations and Hong Kong SAR conglomerates operating in Mainland who have diversified income stream and are less reliant on revenues from Mainland operations. As such, Hong Kong SAR banks’ exposure to Mainland could be well below that suggested by aggregate data. Meanwhile, collaterals posted—a common practice with Chinese loans, both onshore and in Hong Kong SAR—serve as a risk buffer, helping to insulate Hong Kong SAR lenders.

Potential risks are higher among SOEs

8. Using the composition of A-shares in each major borrower type for SOEs and private entities as a baseline, and assume that this composition is representative of Hong Kong SAR banks’ lending portfolio, potential risks could be higher in loans to SOEs. In line with findings elsewhere (including the April GFSR), SOEs in the A-share market have a debt-at-risk ratio of 14.2 percent, higher than the ratio seen for other borrowers—9 percent for private firms in the A share universe, for example. Given the large exposure to SOEs, potential losses from these loans could be sizable, particularly if the assumed implicit government guarantee were to weaken. That said, data on the actual composition of lending to SOEs—by types of economic sectors, for example – for use in Mainland is not available; debt at risk for corporate that borrowed from Hong Kong SAR banks could well be different from the average SOE in Mainland, given stringent lending standards in Hong Kong SAR. Anecdotal reports from analysts and supervisors suggest that most of these SOE borrowers are large SOEs that receive strong government support; risks presented by these borrowers could be lower than the general SOE population.

9. That said, the risk buffer could be inadequate in a downside risk scenario, particularly in which collateral recovery may be sluggish and collateral valuation is heavily discounted. Not surprisingly, vulnerabilities are particularly high in the overcapacity sectors, with debt-at-risk ratios reaching as high as 39 percent for metal products, rubber and plastic manufacturers and the mining sector.9 By contrast, risks appear limited in sectors tied to the more service-oriented “new” industries such as electricity and gas production as well as printing and publishing; debt-at-risk ratios for these sectors were as low as 1.2 percent. Against this context, HKMA’s macroprudential and microprudential measures are important and remain the first line of defense.

B. Large Divergence in Individual Bank Exposures

10. There is a high level of divergence in bank exposures to Mainland-related lending. Of the four banks studied,10 potential debt at risk ranges from 9.2 percent of their respective Mainland exposure to 11.8 percent.11 Notably, while Bank A is particularly exposed to its SOE lending (16.3 percent of total assets), Bank B has outsized exposure to Mainland private firms, with 21 percent of total assets in Mainland lending. This underscores the need to look at banks individually.

Table 1.Selected Banks’ Nonbank Mainland China Exposure in Total Assets in 2015(In percent:)
Bank ABank BBank CBank D
In percent of total assets
Total Non-bank mainland China exposure22.9034.5016.4015.20
Mainland non-private entities16.306.806.706.30
Mainland private entities3.7021.104.905.50
Non-Mainland entities2.601.203.901.80
Others for PRC0.305.400.901.50
Source: Citi Research.
Source: Citi Research.

11. Analysis using other metrics offer qualitatively the same conclusions and underscores heterogeneous risk profiles of different banks’ lending portfolios. Relative to total loans (that is, banks’ entire loan portfolios, inclusive of Mainland-related lending), Bank A remains particularly exposed in its SOE lending, at a sizable 42 percent, while Bank B is particularly exposed to private firms, which accounted for 36 percent of total loans. Meanwhile, looking at risk-weighted assets to gauge potential recapitalization needs from these lending activities, Banks A and B, as expected, are particularly vulnerable.

Lending Share Mainland China Corporates in Risk-Weighted Assets

Source: Citi Research.

Table 2.Selected Banks’ Nonbank Mainland China Exposure in Total Loans and Risk-Weighted Assets in 2015(In percent)
Bank ABank BBank CBank D
In percent of total loans
Total non-bank mainland China exposure58.7358.4731.8641.16
Mainland non-p rivate entities41.8111.5213.0217.06
Mainland private entities9.4935.769.5214.89
Non-Mainland entities6.672.037.584.87
Others for PRC0.779.151.754.06
In percent to total risk-weighted assets
Total nonbank mainland China exposure57.4553.5242.9844.32
Mainland non-p rivate entities40.8910.5517.5618.37
Mainland private entities9.2832.7312.8416.04
Non-Mainland entities6.521.8610.225.25
Others for PRC0.758.382.364.37
CET1 capital ratio (latest) 1/18.5812.2016.8016.10
Sources: Citi Research, Bank earning reports, Bloomberg; IMF staff calculations.

2016H1 for Banks A, C, & D; 2015 for Bank B.

Sources: Citi Research, Bank earning reports, Bloomberg; IMF staff calculations.

2016H1 for Banks A, C, & D; 2015 for Bank B.

C. Continued Monitoring Crucial

Staff’s analysis, combining information on A-shares, HKMA data as well as methodologies used in the April 2016 GFSR, suggests Hong Kong SAR banks are generally less exposed to riskier Mainland corporates. By applying the sectoral composition of A-shares and their debt-at-risk ratios to Hong Kong SAR banks’ Mainland portfolios, we derived a rough estimate of potential losses for Hong Kong SAR banks. These estimates represent the upper bound of potential losses under current circumstances. These relatively low estimates compared to average A-shares suggests that Hong Kong SAR banks’ exposure to more vulnerable Mainland corporates may be limited.

That said, a sharp deterioration in the balance sheet of Mainland corporates, as well as a sharper-than-expected downturn in the Mainland economy could negatively affect Hong Kong SAR banks, raising debt at risk well above estimates suggested here. As Hong Kong SAR banks generally have sizable buffers against downside risks, the best approach to such a scenario is vigilance, including maintaining high origination and underwriting standards. Staff would also support continued close cooperation with Mainland supervisors to ensure information sharing and risk monitoring. Additionally, existing policies, including stable funding requirements imposed in October 2013, could help further insulate banks from negative shocks. Specifically, stable funding requirements discourage rapid loan growth, especially those used outside of Hong Kong SAR; they require banks to ensure adequate stable funding to support their lending business from 2014 onwards. As evidence of the effectiveness of this measure, loan growth has decelerated gradually since then.


    Hong Kong Monetary Authority2016Half-Yearly Monetary and Financial Stability Report (March).

    Hong Kong Monetary Authority2015Practice Note on Credit Risk Management to the Corporate Sector,Ref #: B10/1C (Hong KongSAR).

    • Search Google Scholar
    • Export Citation

    Citi Research2016Hong Kong Banks, “NBMCE”—An Effective Tool for Measuring China Exposure Risks.

1These summary notes provide background documentation for the 2016 Article IV consultation discussions with Hong Kong SAR.
1Prepared by Sally Chen and Daniel Law (Resident Representatives Office Hong Kong SAR).
2The total tally of this broad category could include double counting of some transactions. For example, some trading financing transactions could also be included as a part of on-balance sheet items.
4Results are qualitatively the same with 8-quarter moving averages.
5A similar exercise by HKMA—using ICR<1 to indicate balance sheet stress and relying on listed firms’ data in Hong Kong SAR from 1995 to 2015—compared to staff’s sample of listed firms in Mainland—suggests a debt-at-risk ratio of about 23 percent for nonlocal corporates. The difference in debt-at-risk ratios stems primarily from different datasets used (i.e., different samples of companies) and different definitions of ICR. Indeed, the debt-at-risk estimates are found to be sensitive to definitions of income, interest expenses and samples. For example, the debt-at-risk estimate for nonlocal corporates listed in Hong Kong is found to be lower at 13 percent if EBITDA is used instead of EBIT, according to HKMA estimates.
7As data on sectoral exposure for Hong Kong SAR banks are not available, using the debt-at-risk ratios for A shares allows us to estimate an upper bound of potential losses for banks’ Mainland loan portfolios. The debt-at-risk ratio of Chinese A-shares is used as a rough gauge the credit quality for firms operating in China; for Hong Kong banks, these ratios are weighted according to Hong Kong banks’ sectoral exposure for the aggregate debt-at-risk ratio.
8Hong Kong Monetary Authority (2015). Practice note on credit risk management to the corporate sector, Ref #: B10/1C.
9This exercise assumes that the composition of Hong Kong SAR bank lending to Mainland is the same as the sectoral composition of A-shares.
11We apply system-wide debt-at-risk ratios by borrower type (based on A-shares) to each banks’ lending portfolio to estimate the exposure relative to the total balance sheet, as proxied by total nonbank Mainland exposure, total lending and total risk-weighted assets. While the ratios we used are derived from Mainland exposures, total loans and assets used here are based on each bank’s aggregate balance sheets, and thus include lending activities elsewhere.

Other Resources Citing This Publication