The Risks Behind the Numbers
While such aggregate numbers make for good headlines, they are not particularly helpful for assessing the risks associated with shadow banking. There is much more to the story.
An implication of the term “shadow banking” is that diverse and distinct forms of nonbank credit can be grouped together under the same heading and thought of in the same way. In reality, each form of credit exhibits a dramatically different risk profile. Pointing to the riskiness of one form of shadow banking (usually trust products or wealth management products) and then citing the overall amount of shadow banking gives a misleading perception of the magnitude of the risks.
Broadly, there are two ways in which a shadow-banking sector can pose a risk. The first is direct: credit risk is increased by lending to financially weak borrowers. For this to become a serious issue, such lending must also be large in magnitude so that the potential losses are significant relative to the economy. The second form of risk comes when a shadow-banking sector is strongly interconnected with the rest of the financial system. This channel was key, for instance, to transforming relatively modest losses on mortgage portfolios in the United States into a systemic financial crisis in 2008.43 In China’s heavily bank-based financial system, the primary risk is of interconnectedness between shadow banking and the commercial banks.
Each individual form of shadow financing—entrust loans, bankers’ acceptances, trust companies, and wealth management products—should be analyzed in terms of these risks. And it should be analyzed on its own merits.
Entrust loans are a means by which nonfinancial corporations lend to each other, amounting to 14 percent of China’s GDP as of the end of 2012. They are often portrayed as riskier than traditional bank loans, but in reality, as much as 80 percent of entrust loans are between related companies and are low risk.44 As such, lending through entrust loans is arguably safer and more efficient than the average bank loan due to lower information asymmetries since the relevant companies are familiar with each other.
The underlying borrowers may be risky, but financial institutions bear no credit risk for entrust loans. Although banks play an important role in facilitating and overseeing such loans, all of the credit risk is borne by the company doing the lending. That means entrust loans pose virtually no direct threat to financial stability—even if the borrower is a high credit risk, the hit of a default will land largely on the company that did the lending, with minimal spillover to the wider financial system.
Bankers’ acceptances, which amounted to 12 percent of GDP as of the end of 2012, are a form of short-term credit used to facilitate business activity by having the bank guarantee payment, obviating the need to evaluate whether a business partner is creditworthy. Because of that, banks typically require proof of an underlying commercial transaction before issuing an acceptance, and they often require modest deposits as collateral on the acceptance draft, both of which act to reduce the risks of a credit loss.
Bankers’ acceptances are inherently interconnected with the financial health of banks, which bear the full credit risks of the loans. If losses mount, they could spill over to the broader financial system. However, the market for these products is small and most acceptances are at least partially collateralized by deposits and backed by underlying transactions, so the threat is modest.
In 2013, trust companies—a form of nonbank financial institution unique to China—surpassed insurance companies to become the second-largest type of financial institution in the country after commercial banks.45 Trust companies function in a way somewhat analogous to investment funds in the West and are authorized to manage assets for enterprises and individuals. But, unlike banks, they are prohibited from taking deposits. Instead of collecting deposits and issuing loans, trusts make loans, purchase securities, or invest in private equity, which they then sell to investors in the form of “trust products” that often offer interest rates as high as 9–11 percent.46 Because their investment activities are generally riskier, this aspect of shadow banking warrants more attention and regulatory oversight.
Trust products are frequently invested in a single project loan, but some involve packaging together multiple loans and investments in a particular sector. There is a wide range of restrictions on marketing and advertising trust products, and trusts rely on banks to sell their products to investors.
Trusts have been growing rapidly—46 percent in 2013. But their growth has also been moderating rapidly, coming down from 71 percent in June 2013 and 60 percent in September 2013.47 Even with this decline, by the end of 2013, China’s trust companies had assets under management worth almost 20 percent of GDP.48
Credit extension through trusts amounts to about 9–13 percent of GDP or 3–5 percent of all banking assets. That is because less than half of trust assets are trust loans, which is the only component of trusts’ business that involves extending credit.49 The rest of trust assets are invested in bonds (whether for short-term trading or held to maturity investment), equity investment (typically private equity rather than publically traded stocks), bank deposits, and a hodgepodge of other assets. Some analysts have alleged that some of these nonloan activities have suspiciously loan-like structures and have argued that as much as 67 percent of trust assets should be considered loans.50
An outdated concern about this form of shadow banking is that trusts are just a way for banks to circumvent quantitative limits on their lending or restrictions on lending to particular sectors by moving loans off of their balance sheets as wealth management products (which leads to some double-counting in the size of shadow banking as discussed later) or into the interbank market to reduce their capital requirements. Such bank-trust cooperation, somewhat analogous to American banks’ use of “special investment vehicles” before the global financial crisis, accounted for two-thirds of all trust assets in 2010 and helped drive the early growth of the trust industry following the 2009 stimulus.51
Another concern is that trusts are disproportionately invested in risky sectors like real estate, raw materials, and infrastructure, and much of this is flowing to LGFVs. This is true, but trust assets are not as concentrated in these sectors as is often perceived: less than 40 percent of trust assets are invested in real estate, raw minerals, and infrastructure.52 Moreover, trusts’ investments in real estate grew more slowly than any other category in 2013, as trusts pulled back from the sector.53 This has continued into 2014, with issuance of property-based collective trusts falling by half in the first quarter.54
Even outside of those vulnerable sectors, credit risk is likely higher for trusts than banks because financially stronger businesses are more likely to have access to the cheap loans from the formal banking sector while more marginal firms are forced to borrow at higher rates from trusts. Thus defaults are a very real risk for investors in trust products, particularly with 4.5 trillion yuan (approximately $730 billion) worth of trust products maturing this year, up 77 percent on 2013.55
How much of a threat do defaults in the trust sector pose to financial stability? Defaults or other losses on the assets underlying trust products accrue to the investors in those products rather than the trusts themselves, which are forbidden to guarantee the principle or return on products they issue.56 This, combined with the lack of leverage in trusts, means that a systemic impact has to come from spillovers to the rest of the financial system, particularly the banks.
Spillover is unlikely to come through direct bank-trust cooperation, including partnerships in some lending arrangements. Since 2009, regulations restricting the scope and magnitude of bank-trust cooperation have cut such cooperation to just 20 percent of trust assets, a level that will continue to fall as old cooperation agreements mature.57 As such, exposure to defaults in the trust sector is increasingly shifting away from formerly linked banks to individual retail and corporate investors who are directly responsible for the trust products, reducing the risk of contagion from the trust sector to the financial system as a whole. Bank-trust cooperation appears to be on track to contract still further, particularly given the recent tightening of regulation on interbank transactions between banks and trusts.58
An oft-cited possibility is that spillover could occur through bailouts. That is, to protect their reputation, banks will bail out trust products that were marketed through their branches.
Yet, this does not seem to be what has happened in practice. For example, when a trust product tied to a coal mining project in Shanxi Province approached default in January 2014, the Industrial and Commercial Bank of China, which had marketed the product to investors, steadfastly refused to bail out investors.59 More recently the China Construction Bank refused to make investors whole on another coal-related trust product it had marketed and that experienced payment difficulties in February.60 Both banks, China’s largest, also appear to have temporarily stopped marketing trust products to their customers, further limiting their exposure.61
Such bailouts may yet materialize, but it is hard to imagine them doing serious damage to bank balance sheets. The entire trust sector amounts to less than 8 percent of bank assets, only a small portion of that will default, and only a small portion of those losses will be absorbed by the banks.62
What is more, even a loss of confidence in trusts is not particularly threatening to the banks. In the United States and Europe, banks were heavily reliant on wholesale funding from money market mutual funds, repurchase agreements, and other shadow-banking institutions and activities, so when the shadow-banking system froze up, they suffered a serious contraction of liquidity that spiraled into a systemic banking crisis.63 The story in China is quite different. As the bank UBS has repeatedly pointed out, if investors lose confidence in the trust market, their money will simply flow back into bank deposits because China’s capital controls prevent them from sending it abroad, and they have few alternative investment platforms.64 Thus a loss of confidence in trusts will, if anything, increase liquidity and reduce the cost of capital for banks in China.
Wealth Management Products
The final component of shadow banking is wealth management products, which are investment products created by banks but often confused for a type of deposit.65 They are somewhat analogous to a money market fund offered by a bank. While they may not have introduced additional credit risk, they have generated a variety of other vulnerabilities. What sets WMPs apart is their intrinsic connection to the banks, which creates large spillover risks. A loss of confidence in WMPs could pose significant systemic risk to the banks and the economy.
Wealth management products are quite distinct from the other forms of shadow banking because they do not involve any direct credit extension—WMPs are simply a distribution channel for existing assets. Rather than making loans, as trusts do with a portion of the money they collect through trust products, WMPs invest in a range of underlying assets, including bonds, interbank assets, securitized loans, trust products, discounted bankers’ acceptances, and in some cases equity. This simply shifts the ownership of existing debt rather than creating new debt.66 In principle, banks just manage WMP assets for their customers, who bear the risk of defaults. However, unlike trusts, banks are allowed to guarantee the principal or return on a WMP, although they do so for less than 40 percent of products.67
Wealth management products have been one of the most rapidly expanding elements of the financial system, growing almost sixfold between 2009 and 2013 at more than 50 percent a year.68 The interest rate on WMPs tends to be about 1–2 percentage points higher than that on bank deposits. This has drawn investors into the WMP market, which did not exist before 2005 but now accounts for more than 10 percent of all deposits.69
Roughly half of the funds that have flowed into WMPs are invested in relatively safe interbank assets and financial bonds while the other half are invested in riskier equity stakes (some of which may be disguised loans), corporate bonds (including LGFVs), and “nontraditional credit assets,” primarily trust products and simple securitized bank loans.70 Nontraditional credit assets and LGFV bonds are the primary sources of credit risk for WMPs.
WMP credit risks are significantly lower than those of many other financial products. It is widely assumed that corporates that have issued bonds are less risky than those that have not and that the government is well situated to guarantee the debts of LGFVs when necessary. The risks associated with WMP loans are likely to be lower than trust products. Standard Chartered Bank has argued that the credit standards for WMP “loans” (that is, holdings of trust products and securitized bank loans) are not significantly looser than for traditional on-balance-sheet bank loans, although they are still concentrated in riskier sectors.71 The ratings agency Standard and Poor’s and Goldman Sachs have gone so far as to argue that WMPs have lower credit risk than most traditional bank loan portfolios, though that assessment probably goes too far.72
As with trust products, many analysts of China’s financial system allege that banks are likely to bail out failing WMPs even if they are not guaranteed. The case is much more compelling with WMPs because banks have stronger incentives to preserve confidence in the WMP market. For one thing, the failure of a bank-crafted product, such as a WMP, is a more direct reputational hit to the bank than the failure of a product they simply sold. Banks also rely on WMPs as an integral part of their business,73 whereas sales of trust products simply raise some fee revenue on the side.
It might initially seem that a bailout of and loss of confidence in WMPs would work much as a loss of confidence in trust products: it would simply lead to a flood of cash out of WMPs and into deposits and thus banks would not be adversely affected. However, if non-principal-guaranteed products are bailed out, both deposits and the troubled assets themselves would be brought back onto banks’ balance sheets. This could lead banks’ loan-to-deposit ratios to rise or fall, but their capital adequacy ratios and reserve requirements would be unambiguously eroded, leaving them more vulnerable.
Such bailouts would also create an accelerator effect where banks are forced not only to absorb the losses but also to put aside capital and reserves for the remaining assets and deposits that have been shifted onto their balance sheet unexpectedly, leaving them less flexible than before the bailout.74 Thus problems with some of the assets underpinning WMPs could have negative spillover consequences for the banking sector more generally.
Two aspects of WMPs make them particularly vulnerable to a loss of confidence: maturity mismatches that encourage investors to exit early and a lack of transparency about the underlying assets that in part stems from that mismatch.
At the end of 2012, fully 60 percent of WMPs in China had maturities of three months or less while most of their underlying assets were multiyear loans and securities.75 As a result, investors in maturing WMPs are often paid out with the funds from new WMP issuance, leading the head of the China Securities Regulatory Commission to call WMPs a “Ponzi scheme.”76 If there is a sudden loss of confidence in the WMP sector and new investors cannot be found, existing investors in WMPs may find it impossible to withdraw their capital until the underlying assets mature. This gives investors a strong incentive to be the “first out the door,” a structure inherently prone to runs. This is less of a vulnerability for the majority of WMPs invested in liquid assets that can be sold off to pay out investors. But it is a serious issue for the minority of WMPs invested in nontraditional credit assets, which would be hard-pressed to pay out investors if cash stopped flowing into the sector regardless of underlying asset quality.
One approach the banks have developed to cope with this is to pool funds from multiple WMPs together, which allows the excess returns on some assets to subsidize losses on others. More than half of WMPs are of this mixed form.
This fix reduces transparency of the underlying risks and introduces ambiguity as to what assets are actually backing a particular WMP. It also makes it more difficult for the banks to resist guaranteeing the WMPs—it is much easier to allow a default when they can point to a specific failed investment than when those funds have been mixed into a pool that includes successful and failed projects. This also makes it more likely that investors will lose confidence in these products, forcing banks to bail them out.
There have been positive steps toward curbing these risks. In March 2013, the China Banking Regulatory Commission issued new regulations that required banks to restrict the share of nontraditional credit assets in their WMP portfolio to 35 percent of WMP assets and 4 percent of bank assets. This improved WMP liquidity and reduced, though did not eliminate, the vulnerabilities derived from their maturity mismatches given the short-term nature of the WMPs relative to the multiyear nature of their underlying assets. The commission also required that banks link all WMPs to an underlying asset rather than drawing on an amorphous pool of mixed assets, improving transparency and allowing investors to better assess the risks they are exposed to through their WMP investment. While the first requirement appears to have been successfully implemented and enforced at most banks, the practice of pooling WMP funds still appears to be fairly widespread. More rigorous enforcement of the restriction on pooling is needed to truly address these vulnerabilities.
Yet, these regulations are still insufficient to curb the risks WMPs could pose to financial stability if they continue to grow at their current pace. The China Banking Regulatory Commission must find a way of clearly distinguishing WMPs from deposits to put an end to implicit guarantees, or it should recognize the inherently deposit-like structure of WMPs and begin to regulate them as such. Creating a clear divide seems implausible given banks’ incentives to maintain investor confidence in the WMP market and the government’s concerns that isolated protests over failed WMPs might get out of hand. Thus it would be more practical for the China Banking Regulatory Commission to treat WMPs as a form of deposit subject to capital and reserve requirements similar to those on traditional bank deposits and lending.
All in All, Not a Serious Threat
Ultimately, the exposure of the traditional banking system to risky shadow-banking activities is not that significant. It would take truly disastrous performance over a short period of time for these assets to pose a serious threat to the stability of the banking system or the economy.
Alongside the high-risk, high-return products that draw most of the attention are a vast number of relatively low-risk investments that play healthy and productive roles in facilitating economic activity. Some shadow-banking activities pose no extraordinary risks: entrust loans cannot easily spill over to impair banks’ balance sheets, and bankers’ acceptances are not acutely risky. The real concerns are more narrowly concentrated in the activities of trust companies and banks’ wealth management products, although even here the risks are often overstated.
In reality, the much-talked-about, high-risk components of the shadow-banking system are overhyped. All forms of shadow banking amount to a total of 84 percent of GDP. By the numbers, the shares of trust companies and WMPs in the broader system seem large, amounting to 36 percent of GDP or about 15 percent of bank assets. But those figures are overestimated, because WMPs’ investment in trust products are double counted, and they could account for as much as 30 percent of WMP assets. Further, over half of WMP investments are in safe interbank assets and government and financial bonds, while less than half of trust assets are in the form of loans or exposed to risky sectors. So ultimately the high-risk components of shadow banking, concentrated among WMPs and trust products and accounting for about 18 percent of the total, are unlikely to exceed 15 percent of GDP or about 6 percent of bank assets (see figure 10).
Granted, there are second-order financial risks associated with the shadow-banking system, and potentially dangerous situations could arise, such as a vicious cycle in which the unraveling of some portion of shadow banking slows economic growth. The ensuing economic slowdown could then damage the fiscal health of marginal borrowers, potentially leading to wider systemic consequences.
Three unlikely things must happen for such a vicious cycle to arise. Each assumption is questionable in its own right, and together they are implausible.
The first two assumptions relate to a collapse in lending. First, there must be a crisis of confidence that leads to a collapse in lending through one or more shadow-banking channels. Second, that collapse must be of sufficient magnitude to have a noticeable impact on the economy.
To reach sufficient magnitude to be of concern, the collapse would have to occur across multiple channels simultaneously—and that is unlikely. A collapse in the bankers’ acceptance market is unlikely given its structure. But a run is possible in the entrust loan, trust product, or WMP markets that are driven by individual and corporate investors because of their spillover effects in the banking system. Even so, only the 30–50 percent of WMPs and trust products that are involved in credit extension would be involved. However, even an implausibly severe collapse of 30 percent in one of these markets would only amount to a few percentage points of GDP. Further, a significant portion of this credit was likely used to buy land or roll over existing debt, activities that do not contribute to GDP, so the true impact could be even smaller.
The third assumption is that the government would not respond to the unraveling of the shadow-banking market and allow bank lending to expand to compensate for the reduction in shadow lending. Under this scenario, the government would simply allow credit, and thus growth, to collapse. This is unlikely because regulators could compensate for much of the resulting contraction in shadow credit by raising banks’ loan quotas and cutting the reserve requirement ratio to allow greater credit growth through traditional bank lending. There would be some damage in the interim, but growth would not be allowed to collapse.
All in all, then, the much-hyped, high risk parts of shadow banking are unlikely to trigger a financial crisis.
Of course, there is always the fallacy of the heap: each individual component of the financial system does not look overly menacing, but the combination of the various risks together could pose a threat. To assess whether the heap is truly more menacing than its individual components, there are stress tests and scenarios that should be explored.
A number of scenarios come from the past. After all, this is not the first time China has faced serious debt strains. Throughout the 1990s, nonperforming loans at China’s banks were steadily rising as a result of excessive lending to poorly managed SOEs. The official nonperforming loans of state-owned commercial banks reached their peak in 1999, when they accounted for more than 30 percent of loans, about 29 percent of GDP.77 Unofficial estimates ranged up to as much as 40–50 percent of total loans.78
China’s current challenges may be more complex, but the financial situation in the late 1990s was much more severe, and even then difficulties proved manageable. The government created four asset management companies to buy up banks’ nonperforming loans at face value. The central government also injected substantial amounts of capital into the biggest state-owned banks so they could write off many of the remaining nonperforming loans. By 2005, the total face cost of these and other efforts to restructure the banks had reached nearly 4 trillion yuan, or over 40 percent of GDP in 1999.79
Although the direct cost of the bailout was immense, it allowed the banks to return to lending and allowed economic growth to accelerate over the coming decade.80 As the banks’ balance sheets expanded with new and better loans, nonperforming loans as a share of total loans steadily declined to under 1 percent in recent years (see figure 11).81 All of this occurred in the context of a modest deleveraging between 2003 and 2008, when nominal GDP growth was outpacing credit growth and gradually pushing down China’s debt-to-GDP ratio.82
So what are the prospects of a repeat of this successful, if costly, cleanup of the financial sector? Quite good, actually. Even the most pessimistic current estimates of credit losses fall far short of what was observed in the late 1990s and early 2000s, with no serious macroeconomic fallout. Goldman Sachs’s comprehensive report on China’s “credit conundrum” estimates the maximum potential credit loss incurred during a bailout would be between 10 and 21 percent of GDP, depending on how aggressively the government responds to the situation.83 A report from RBS lays out moderate and severe stress scenarios and estimates credit losses between 10 and 20 percent of GDP as well.84 And a more recent exercise by the consultancy Oxford Economics also concludes that nonperforming assets in the financial system are likely to reach between 10 and 20 percent of GDP.85
These levels of bad debt would not be an unreasonable burden on public finances. Between 30 and 60 percent of the losses in these scenarios are from LGFVs, which are already accounted for in China’s total public debt of 56 percent of GDP. In fact, the estimate that China’s expected debt is only 44 percent of GDP uses more pessimistic loss assumptions about LGFV debt than any of these scenarios. Thus, adding the non-LGFV losses of between 2 and 15 percentage points of GDP to the expected debt still leaves China’s total public debt well below the accepted threshold of 60 percent of GDP.
Further, the actual cost to the government is likely to be significantly lower than the estimated credit losses under these scenarios because the government will force the financial industry to absorb a share of the losses. The RBS estimates assume that the financial industry will absorb losses on the first 2.3 percent of assets to go bad, based on the fact that banks currently have provisions for losses on 2.8 percent of their loan books. What is more, as was seen in the previous bank restructuring from 1999 to 2005, the costs are likely to be spread over a number of years, which will allow time for bank earnings to absorb a larger share of the losses.
All these scenarios are meant to be extreme stress tests and do not represent expected outcomes, so the credit loss estimates themselves are likely overestimates. China’s financial and fiscal metrics simply do not point to a looming “Lehman moment.”
Short-Term Outlook: Property Correction
Although financial woes are unlikely to bring China to its knees, the property market may pose a serious obstacle to short-term growth.
Over the past decade, land prices in China have surged fivefold according to the Wharton/NUS/Tsinghua Chinese Residential Land Price Index (see figure 12), while the construction sector has expanded from 10 to 13 percent of GDP.86 With such a dramatic increase in land and property prices and a potential glut of housing in the cards, this has understandably led to concerns that China may be experiencing a property bubble. These concerns have been brought to a boil by deteriorating sales figures and media accounts of large price cuts during the first few months of 2014. In April 2014, Nomura Securities declared that it is no longer a question of “if” or “when” but “how severe” the property correction will be.87
A correction does indeed appear to be coming, but the severity of the correction may be limited. For a number of reasons, China’s situation is not particularly worrisome.
Unlike other bubble-afflicted economies, the growth in China’s land prices coincided with the emergence of a real property market that did not exist until housing was privatized a decade ago. Because of this, the sharp rise in property prices may be indicative of the market trying to establish appropriate prices for an asset whose value had previously been hidden by the socialist system, a pattern also observed in Russia after the fall of the Soviet Union.88 Although this does not rule out the possibility of a bubble, it does suggest that much of the rise could reflect the true underlying value of land. Further, it means that rapidly rising land prices alone cannot be interpreted as evidence of a bubble.
And there are strong fundamental drivers of urban housing demand that suggest China has suffered from undersupply until quite recently, indicating that the expansion of the sector is driven by real demand. Over the past decade China has been experiencing the largest population movement in history with an average of 21 million migrants a year moving into urban areas, driving exceptionally strong demand for new housing.89 Similarly explosive growth of urban wages, which have doubled since 2009, has increased the affordability of housing despite rising prices and has driven massive upgrading demand as households seek out more spacious and modern apartments.90 Even more demand comes from the need to replace old communist-era housing stock and other shoddy construction; the average Chinese building only lasts twenty-five to thirty years, compared to an average of seventy to seventy-five years in the United States.91 Gavekal estimates that these three factors together added up to underlying demand for over 1.1 billion square meters of floor space a year between 2000 and 2010, compared to an average supply of less than 700 million square meters a year.92
Thanks to easing urbanization pressures and a construction boom following the global financial crisis, Gavekal now believes that China is entering a period of modest oversupply, but it still expects strong fundamental demand of around 10 million units a year (see figure 13).93 It is fairly clear that China’s housing supply has overshot the market, but it is not clear that it has done so by a dramatic margin.
Widespread distortions could also be artificially inflating demand for property. For one, Chinese households have limited investment options; the closed capital account leaves few avenues for savers seeking returns higher than savings deposits can provide. Real estate is treated as an investment vehicle in China to fill the gap. According to the China Household Finance Survey, 18 percent of households own more than one home, suggesting that demand for property as an investment vehicle is substantial.94 Moreover, the fact that China does not tax property, aside from pilot programs in Shanghai and Chongqing, means that there are lower costs to purchasing and holding unused property. Much of this activity has been curbed in recent years as the government has gradually tightened restrictions on multiple home ownership and mortgage standards.95 The dip in housing prices in 2011–2012 is generally attributed to the tightening of these regulations.
A correction is likely in the short term. As China moves into a period of structural oversupply, especially in the second and third tier cities rather than major cities like Beijing and Shanghai, prices will begin to fall and the investment incentives that fed the bubble will evaporate and cause a severe collapse in both prices and construction activity. It is difficult to evaluate the balance between the fundamental drivers of the expansion of the property market and irrational distortions, making a confident assessment of how severe the correction might be impossible. Still, there are a number of reasons to believe that the correction will not be destabilizing.
First, the trends in land and housing prices in China are inconsistent with the typical pattern of a bubble. In a typical bubble, investors buy in not because they see value in what they are buying but because they expect the price to continue rising. Thus, when prices show signs of falling (or even just slowing growth), investors rush for the door and cause a collapse.
Prices in China did show signs of falling. Despite a general trend toward higher prices, over the course of 2011, land prices fell by 10 percent, and a portion of that decline was passed along to homebuyers as the price of newly constructed housing fell in late 2011 and early 2012 (see figure 14). It is difficult to find any precedent of a bubble in which prices declined significantly and persistently before continuing to inflate; it is certainly not the pattern observed with U.S. housing before the global financial crisis or in Japan before its lost decade (see figure 15).
What is more, falling prices in China did not precipitate a collapse of the market in 2011–2012, which suggests that the high prices up to 2012 are supported by something more real and persistent than “irrational exuberance.” This also suggests that 2012 levels may be a reasonable floor for prices during the coming correction, implying that prices will fall by less than 20 percent from early 2013 levels. Such a fall would have serious economic consequences, but it is dwarfed by 30–50 percent collapses in the United States and Japan.
The other reason to be relatively sanguine about the severity of a price correction is that China’s property market is not highly dependent on leverage since down payments on property as a share of the purchase price are much higher in China than in the United States, making it less likely that banks will be drawn into a foreclosure process in the event of a major downturn. The results of the China Household Finance Survey indicate that even after a 30 percent decline in prices, only 2.9 percent of households would be underwater thanks to high down payments and the equity that accumulated over the course of their ownership.96 The robustness of household balance sheets helps to insulate the banks from a property downturn because two-thirds of their direct exposure to the sector is in the form of mortgages.97
The highly leveraged property developers that account for the other third of banks’ direct exposure are more concerning. Although those developers have been building up large liquidity buffers, they will certainly see high default rates during the correction.98
Even with this result, the direct impact on banks’ balance sheets would be limited. The strength of the household balance sheets, combined with the fact that only 20 percent of bank credit goes to the property sector, has led the Bank of China to estimate that even a 30 percent drop in real estate prices would have a negligible impact on banks’ nonperforming loan ratios.99
The indirect impact of a correction would likely be more severe but still manageable. The falling price of land would erode the value of collateral that is backing other loans, particularly those to LGFVs, and there would be knock-on effects in steel, cement, and other industries closely tied to construction.100 While these indirect effects would increase the damage done by a correction, the workout scenarios discussed already assumed losses of between 10 and 35 percent on the LGFV and shadow bank lending. A 20 percent price correction in the property sector would be unlikely to result in financial scenarios any more severe than those already considered—and China remains relatively stable under those scenarios.
While the likely price correction may not be a major issue, correction in the volume of construction will have a negative impact on growth. Construction and real estate have been gradually rising as a share of GDP over the past three decades (see figure 16).101 As with property prices, these sectors have not exhibited the pattern common to other bubble countries of a sharp increase in share over a short period. However, amounting to 13 percent of GDP as of 2013, construction and real estate have become important components of China’s economy. And accounting for related industries, such as steel, cement, and construction equipment, would show that construction’s contribution approaches 20 percent of GDP. This makes China’s economy quite vulnerable to a construction downturn.
That downturn is likely to happen soon. Given that underlying demand totals roughly 10 million units a year, and 11 million units came onto the market last year, it seems clear that construction volume must slow by roughly 10 percent. UBS estimates that such a decline could subtract 2.5 percentage points from GDP growth.102
But the actual impact of the correction on GDP growth is likely to be relatively muted because the entire correction need not occur in a single year and because an anticipated increase in infrastructure investment will compensate for a portion of the decline in housing construction. Demand is also likely to rise as China’s urbanization plan is further implemented, which will reduce the magnitude of the adjustment that must take place.103
As a result of these factors, it would be reasonable to expect growth to decline to 6–6.5 percent in the short term as the property sector works through the oversupply of housing. But the darker scenarios in which growth collapses to 5 percent or lower are unlikely to come to pass.