How serious is China’s housing market bubble?
- Housing prices in China have been rising rapidly since the beginning of 2015 and it is the question whether or not this is supported by economic fundamentals. China’s booming housing market as such is no direct cause for concern, but it can become a problem when a housing boom (locally) turns into a bubble, which has the characteristic to pop, once its elasticity limits are reached
- China’s housing market shows an exceptionally high affordability index which indicates that housing expenses are on average much higher than in other major cities, such as London. It should be noted, however, that funding of property in China also differs substantially from mainland Europe and the US, as it is based on a broader wealth basis and also includes substantial grey income
- Our analysis suggests that multiple Chinese cities are facing real estate bubbles, which are especially prominent Shenzhen, Suzhou, Wuxi, Hefei and even some tier 3 cities. Possible negative economic consequences might be severe, since our calculation indicate that 12% of Chinese urban inhabitants live in these ‘bubble-cities’
- A sharp house price correction in these bubble cities will weigh on fixed investment and private consumption, which will put downward pressure on real GDP growth. Residential construction activity would respond immediately, as this sector is lagging behind developments in the housing market. A negative scenario on the housing market will probably result in contracting year-on-year activity in the residential construction sector by the end of 2017. Additionally, a real estate market that is highly vulnerable to a severe price correction poses significant credit and liquidity risks to the sector itself and financial institutions
- The (local) government and the central bank have taken several measures to curb further rises in real estate prices, but it is questionable if these are sufficient to prevent the current situation from spiraling out of control, as fundamental land and fiscal problems are left unsolved. A controlled introduction of a property tax could help to orderly deflate the real estate market
There are increasing worries about China’s housing market. Since the beginning of 2015, house prices have been rising sharply, with a particular acceleration in H1 2016 in tier 1 cities such as Shanghai and Shenzhen (Figure 1). Indeed, in September 2016, property in Shanghai with an average floor size of 100 m2 will cost you USD675,000, and even amounts to a staggering USD825,000 in Shenzhen. These price levels are higher in absolute terms than the average sales price in other expensive urban areas in richer countries such as London (USD630,000) and New York City (USD670,000). Moreover, in Shanghai some land is even sold for more than three times the average price in Manhattan (see New York Times). Recently, in an exclusive interview with CNN, China’s real estate magnate Wang Jianlin warned that situation on the Chinese real estate market is “the biggest bubble in history” (see CNN Money).
One could also argue that the high prices rises are justified by economic fundamentals, such as the continuing rate of urbanisation, rising disposable income, and cultural values that favour property. Glaeser et al. (2016) recently argue that the strong level of demand for real estate in China might suggest that current prices are sustainable. Indeed, the Chinese housing market experienced surging house prices before, and opposed to predictions in the past, the market has never collapsed in recent history. In this report we try to assess if the current situation is healthy or not and, in case of local real estate bubbles, the potential risks to Chinese economy.
Why is the housing market booming?
There are a couple of reasons the housing market could be under so much stress. Below, we will separately discuss factors on the demand and supply side.
On the demand side, urbanisation could be driving up demand for property, especially in the tier 3 and 4 cities. Urbanisation has taken a flight since 1995, and in the 13th Five Year plan accepted in March 2016, the government aims to increase the urbanisation rate even further to 60% in 2020 (Figure 2). Migrants are largely attracted to cities, as income levels are three times as high as in rural areas, and migrants moving to the cities are allowed to buy real estate after three years of residence.
Secondly, policy by the People’s Bank of China (PBoC) could also have contributed to the rush on the real estate market. In February, the PBoC relaxed the down payment of first-home purchases from 25% to 20% and second homes from 40% to 30% to support the housing market. The relaxation did not apply to the tier 1 cities, such as Shanghai and Beijing. This was the first policy initiative of the government to stimulate house purchases in the tier 2, 3 and 4 cities and an attempt to soak up the large stock of empty houses in these cities. What possibly could also have contributed to the housing market boom is the loose monetary policy of the PBoC. It is striking that money supply M1 grew much faster than M2 (Figure 3). The deviation shows that loose policy isn’t boosting borrowing in other areas of the economy; so it indirectly shows why housing is booming, as there is nothing else seen worth investing in by the private sector.
A third contributor to the fast increase in house prices are the limited investment opportunities. As capital movements are under heavy scrutiny in China, there are only three options for money to generate revenue: savings deposits, the stock market and the property market. The equity market is considered to be unattractive ever since the collapse in September 2015. Since interest rates have been declining, investment in property is currently regarded as the only profitable option, which puts additional upward pressure from a household demand perspective. Recently, the government is selectively and cautiously lowering capital controls. If Chinese households are able to move more of their funds abroad, this would alternatively, all else equal, lead to lower demand for domestic property investment which downwardly affects housing prices.
Due to slowing demand from 2010 onward and increasing vacant property outside the tier 1 cities, overall real estate development has been very subdued. Growth of investment and building activities have been on a downward trajectory, only to show a slight increase at the beginning of 2016. This illustrates that there is already massive oversupply in the market, and investment has responded to that fundamental. However, real estate development always tend to lag behind an increase in demand and we also receive indications that developers are rushing in to profit from the booming housing market. Property developers have been using the corporate market extensively to acquire capital in order to purchase land. The benchmark rate cuts by the PBoC from 5.4% in January 2015 to 4.4% has induced many developers to use relatively cheap bonds to fund their activity. In addition, banks are not allowed to provide funding to the real estate market, but have circumvented these policies by setting up trust funds and asset management funds to provide real estate developers with additional source of funds (see BMI). All these abundant funds have driven up land prices to a significant degree, but also leads to increased credit risks all across the board in case of declining property prices. We will return later to this issue more extensively, when we will discuss the exposure of the financial system to the property market (see Section 5).
The fact that the housing market in China is booming is no direct cause for concern. It can, however, become a problem when a housing boom in certain cities turns into a bubble, which has the characteristic to pop, once its elasticity limits are reached.
What is a housing market bubble and what consequences can it have?
To answer the question if China has a housing market bubble, we first need to define what a bubble actually is. An asset bubble arises when asset prices (in our case house prices) start to deviate from the fundamental price component. This means that:
where p is the asset price, f the fundamental component and b the bubble component. In practice, however, it is hard to determine the real actual value of assets. In case of house prices, the conventional method is to relate the price to disposable income per capita. This way, it is possible also assess the affordability of housing in a certain urban area. An alternative measure is to compare the growth of house prices to rental prices, as renting is a suitable substitute for buying a house. Especially the second rules of thumb presents a problem when looking at the Chinese housing market. The picture is muddied by the fact that the rental market in China is largely underdeveloped, as the institutional rights of renters are not safeguarded. In addition, renting does not bode well with the culture of China, where buying property is practically a social necessity for an adult (see Forbes).
Figure 5 shows the UBS real estate bubble index between various expensive cities, putting real estate price developments in perspective against the fundamentals. The index shows that London, Stockholm and Hong Kong are in bubble territory, whereas the first-mentioned two cities are also on a negative trajectory. Amsterdam is interesting, as it is overvalued according to the UBS index, and heads to bubble territory fast. Unfortunately the UBS bubble index is not available for Chinese cities. We will use the outcome of the index, however, to define London as our international benchmark city.
Explanation: the index is a weighted average of five standardized city sub-indices: price-to-income, price-to-rent, change in mortgage-to-GDP ratio, change in construction-to-GDP ratio and the relative price-city-to-country indicator. Undervalued: -1.5 to -0.5; fair valued: -0.5 to 0.5; overvalued: 1.0 to 1.5; real estate bubble: > 1.5.
What happens when a real estate bubble bursts?
To illustrate what happens when a real estate bubble bursts we take on a very simple model (see Figure 6), where AD represents aggregate demand and AS aggregate supply. AD is represented by:
Where A are the expenditure components (i.e. private and public investment, private and public consumption, net exports), M is money supply, P the price level and t a time index. AS is defined as:
Where Y*represents potential output and Pe the expected price level. The general idea is that wages and prices respond with a certain time lag to economic shocks. Wages and prices keep adjusting until Y=Y*.
Now let’s see what happens in case of, for example, a sudden increase of interest rates or the announcement of a property tax. The increased costs of finance for houses will not only make new house purchases less attractive, but will also confront home owners with lower net wealth due to declining house prices and higher (future) interest payment. Firms and households which are exposed to declining house prices will try to repair their balance sheets and postpone investment and consumption. If the real estate market takes a turn and bubbles would burst, the subsequent reduction of property sales will additionally weigh on the real economy via downward pressures on fixed (real estate) investment and private consumption. And as real estate fixed investment makes up for a quarter of overall fixed investment, an additional lowering of economic growth would presumably be caused by slowing construction and other real estate related activities such as steel, cement, and construction machinery. In addition, real estate fixed investment could suffer more via slowing fiscal revenues of Chinese local governments given their dependency on these revenues for investment in property and land.
In our model, a bust in house prices will lead to a shift in aggregate demand from AD to AD’ and the economy moves from point 1 to 2. Under price flexibility and perfect working capital markets a demand shock will not cause permanent damage and AS moves from AS to AS’ and the economy will move from point 2 to point 3, until full production potential is reached.
However, capital markets are not functioning perfect. The declining house prices will cause a rise in the non-performing loans (NPL’s) of both households, as well as real estate developers which have been speculating on price increases. In the deteriorating market, banks will increasingly face moral hazard and adverse selection problems. Indeed real estate developers which are involved in relatively risky projects which can generate high profits will pursue in their demand for funding, while the less risky projects will be terminated due to the higher funding costs. To counteract the increase in moral hazard and adverse selection, banks will scrutinize on the finance conditions by asking higher risk premiums from borrowers and increase screening activity. This will consequently raise the finance costs, leading to the more safe project to back out. What more, property often functions as collateral, which will decrease in value due to the decline house prices, increasing uncertainty and further credit constraints. In short, if the banking sector is largely exposed to the real estate sector, a bust will deteriorate the financial intermediation function of the economy. This is not much different than a (technological) supply shock (see King, 1994), which will lead AS to shift from AS’’. Ultimately, the economy end up in point 4 and the potential production is permanently damaged (from Y1*to Y2*). Mishkin (1996) shows that it is ultimately a change on the supply side of the economy which results in permanent damage of a financial crisis.
Does China have a housing market bubble?
As China’s renting market is very immature, we need to look at the price-to-income level. In order to do so, we calculate an affordability index, which links monthly house expenses to disposable income. We arrives at ridiculously high figures for Chinese cities (Figure 7). In our calculations, the housing expenses in Shenzhen are 6 times higher than average disposable income, whereas in London this percentage is about 70%.
The wide discrepancy between Chinese cities and London is not so much caused by the absolute deviation in house prices, but mainly due to the fact that average lending interest rates are higher and, above all, average disposable income per capita is much lower. In 2015, annual disposable income in London is more or less USD55,000per capita, whereas in Shanghai and Beijing this was roughly somewhat more than USD8,000, a difference by a factor 7. Admittedly, data on disposable income to cover for home expenses do not tell the whole story in China, as it does not include a large proportion of grey income (income in the informal economy), which share is still substantial in China. Secondly, in contrast to mainland Europe or the US, houses are actually funded by a much broader basis than merely household disposable income. In China, buying a house is a very social matter. Especially parents, but also family and sometimes even friends might contribute to the purchase. In addition, people in China buying property use their own capital to finance (part) of the property. Being a home owner is virtually a necessity from a social point of view and often parents (as related home buyers) tend to move into their children’s homes when getting older, letting house ownership be a multi-generational affair (Forbes, 2016).
Nonetheless, the exceptionally high affordability index does show us that the population base that is able to make a move on the Chinese housing market will likely be much smaller than total urban population figures are showing, especially in the tier 1 and tier 2 cities. In fact, migrants from rural areas that have been moving to rural areas in large number often have a very low disposable income and will be never able to own a piece of property in tier 1 and 2 cities. This implies that people that are active on the housing market earn substantially more than the average disposable income and threat property as an investment object. Indeed, income inequality in China has grown quite rapidly (see Xu and Jin, 2015); and the speculative component on the Chinese housing market must also be quite substantial.
Magnitude of the Chinese bubbles
As the Chinese funding of property deviates from the situation in other expensive cities across the globe, we also need to consider China’s housing market fundamentals by its intertemporal standards in order to assess if there are housing market bubbles growing at this moment. We use the models discussed in Box 1 to forecast the price level per square meter (M2) based on several fundamentals, i.e. the growth of the average lending rate, the second derivation of the mortgage lending rate, and the house price-to income ratios in the short and medium term. The house price based on the fundamentals is compared to the actual rise in house prices to identify which part of the house price development is in speculative bubble territory. A bubble is defined as the situation where the deviation between actual and fundamental price growth exceeds the standard deviation of the entire observation period by a factor 2.5.
Table 2 shows the extent of the bubbles in Chinese tier 1 and 2 cities. According to our estimates, real estate bubbles are present in Shenzhen (Figure 8) and some tier 2 cities, such as Wuxi (Figure 9), Suzhou, Nanjing, Wuhan and Hefei. In some cases, the price level exceeds the price level predicted by the assumed underlying fundamentals by as much as 25%. But according to our definition, Shanghai is also in bubble territory, since the 9% price discrepancy in 2016Q3 in rarely seen in the period of observation and prices have been much in line with fundamentals.
Box 1: Estimation of a house price model for Chinese cities
We use the following house price model inspired by the work of Garretsen et al. (1999) for a panel of cities:
where H is the average house price per m2 (data from CREIS) and π is the inflation rate (data from NBS). This implies that our dependent variable is growth of real house prices per m2. Furthermore, r reflects the average mortgage lending rate (data from PBoC). We also add the first derivation of the mortgage lending growth L (data from CEIC) and the growth of the population P (data from UN). Y is disposable income per capita (data from CEIC). Term a4 and a5 captures the nominal house price to disposable income ratio in the short and longer term, respectively. Furthermore, i represents city, t month, d is a dummy for each city to tackle non-observed heterogeneity between cities which is not captured by our variables. Finally, D indicates that a variable is expressed in y-o-y mutations on a quarterly basis. The model is estimated for panels of tier 1, tier 2 and tier 3 cities separately to examine if there is a bubble, given the impact of the development of the underlying institutions. We estimate the model for the sample period 2010Q2-2015Q4 and forecast the house prices based on these assumed fundamentals. By comparing the actual development of house prices with the forecasted values of the model in the period 2015Q1-2016Q2, we are able to identify if house prices deviate from their fundamentals and, consequently, the implied magnitude of the housing market bubble.
In Table 1, the estimation results of our models are presented. The average lending rate (a1) has a significant negative effect on the growth of real house prices in the types of cities: tier 1, 2 and 3. Lower lending rates stimulate house prices, as affordability increases. The mortgage lending growth (a2) as a second indicator of funding possibility shows a significant positive sign in the tier 1 and 2 cities. The higher possible mortgage lending growth, the higher the demand for property and the higher prices will be. In the tier 3 cities, the effect is insignificant, possibly as the additional mortgage loan growth mainly applies to the tier 1 an 2 cities. Population growth (a3) does not seem to have a significant effect on growth of property prices in the tier 1 and 2 cities. This means that the continuing urbanisation does not seem to affect house price increases, as rural migrants are unable to buy expensive property in the tier 1 and 2 cities. This is in line with the already mentioned conclusion that a relatively small and wealthy group of insiders in the tier 1 and 2 cities are active on the housing market and outsiders moving to these cities are not able to participate. In the tier 3 cities, however, population growth has a very prominent effect on house price development. This indicates that chances to buy property in tier 3 cities is easier for new residents due to lower price levels. Meetings of Rabobank with large banks in China in October confirm these observations. In the tier 1 and 2 cities, people usually only finance the top up price in order to move to a more expensive house within the city. Especially the older generation which already owns an apartment in the tier 1 and 2 cities are able to move up the housing market ladder. In the end, we do not expect the continuing urbanisation to support continuing demand for property in the tier 1 and 2 cities, but we do expect that the ongoing urbanisation will continue to put upward pressure for housing demand in the tier 3’s. In the tier 3 cities where oversupply is low, this will continue to put upward stress on prices. In the cities where oversupply is high, urbanisation will continue to soak up the oversupply in housing. Finally, the house-to-income ratio lagged one quarter (a4) shows a significant positive effect in the tier 1 and 2 cities, whereas the same variable lagged five quarters (a5) shows a negative effect in all cities. If in the medium run house prices exceed income to a substantial degree, this will put a brake on further increases of house prices, as affordability becomes increasingly difficult. In the short run, however, a rise in house prices will fuel speculation in anticipation of further increases, which will cause house prices to increase even faster. This speculation mechanism seems to be absent in the tier 3 cities, as these cities still cope with a large stock of empty houses.
The tier 3 cities (see Table A.2 in the appendix) show bubbles in Nantong, Wuhu, Dongguan, Zhuhai. In these cities property prices all exceed fundamentally-based price levels by at least 25%. Next to these cities, Zhengzhou, Tangshan and Shijiazhuang are experiencing price bubble according to our definition.
Taking all examined cities into consideration, the population in Chinese cities which are in bubble territory approximates 92 million inhabitants, covering roughly 12% of China’s total urban population.
What could happen if a bubble bursts
Although some cities are coping with quite extreme real estate bubbles, it is impossible to predict if the housing market will bust in the near future. The Chinese housing market has seen surging house prices before (in 2008-2010 and 2012-2013) and although experts have been predicting busts, the market has never collapsed in recent history. What we can do, however, is examine the exposure of key actors in the economy in case of a hypothetical drop in house prices. We will also pay special attention consequences of a hypothetical bust on the construction sector.
Households as borrowing and spending actors
In general, household debt as a percentage of GDP in China is still relatively low (40%), but mortgage lending has been rising (Figure 10). Households are increasingly exposed to the property market, as they have scaled down on deposits (even though money supply has been growing) and put their money in the real estate market. Surging house prices in major Chinese cities have made housing assets more and more important in terms of household wealth. As a result, housing assets on average now account for more than 70 percent of household wealth, and even 80 percent in large cities like Beijing and Shanghai (Xi and Jin, 2015).
What increases the exposure of households is the fact that mortgage lending as a % of total loans have been growing in a very short time from 13% early 2015 to almost 15½% now. Even though various provincial governments have introduced additional measures to curb house price increases since October this year (BMI, Eurasia, 2016), households have circumvented these measures by even willing to divorce in order to obtain property (e.g. FT, BBC).
The worrying thing is that residential sales in general show a very distinctive cycle and even are leading vis-à-vis private consumption by approximately 8 quarters (Figure 11). However, the boom cycle that has started in 2015 seems to be fading already, whereas private consumption is still on a downward trend. In fact, the downward trend in consumption expenditure could be even speeding up if the property market would turn the corner, and this is a realistic scenario at the current point in time. Decreasing housing prices would deter household consumptions in two ways. First, Chinese consumers would – via a deteriorating wealth effect – perceive themselves as less wealthy than before, as their housing assets are valued lower than before. Second, housing transactions do generate spinoffs in private consumption as increasing housing sales have a positive effect on the consumption of household durables, such as furnishings and electrical appliances. As such, decreasing housing prices would negatively affect private consumption levels.
Financial institutions exposed to higher risk
As discussed, for a housing market bust to lead to permanent damage to the real economy, the financial sector somehow needs to be exposed to a drop of real estate prices (see Figure 6). Exposure of the financial sector runs directly via mortgage lending, but also indirectly by using special financing structures in order to circumvent prohibitions for banks to invest in real estate projects directly.
Low interest rates and loosened monetary policy has induced real estate developers to issue corporate bonds as the main source of debt financing for acquiring land. In order to lower the risk of insolvency, real estate developers have also resorted to off-balance-sheet financing vehicles, such as wealth management products. Due to such financial engineering, the actual leverage of these companies can be higher than presented (BMI, 2016). In that case, property firms and banks are exposed to significant credit and liquidity risks in the event that housing prices fall substantially. Meanwhile, banks have issued trust loans or asset management plans to real estate developers as they are not allowed to invest directly in the real estate sector. By using this financing structure, banks are investing investor’s funds into trust plans or asset management plans from trust companies or asset management companies, who in turn set up trust loan schemes or buy corporate bonds or equity from real estate investors.This financing structure, where Chinese banks are indirectly exposed to real estate price developments, have led to a surge in corporate bonds as a financing option for Chinese domestic real estate developers. By issuing property bonds, the real estate developers are able to acquire land at a fast pace which in turn upwardly affected real estate prices. The significant inflows of funds from banks' wealth management products poses credit and liquidity risks to the real estate and banking industry in the event of a drop in prices. Especially banks who are exposed to risky real estate related sectors such as steel, cement and construction would suffer from a step up in credit risks in relation to higher NPL’s. Overall, the negative consequences of a housing bubble burst can be severe.
Higher than estimated debt levels of real estate developers could lead a further rise in NPL’s and banking sector stress in the event of a sharp and sustained property price correction. China’s NPL’s are clearly on the rise, both on the level in billions of RMB’s and the NPL ratio. Total registered NPL’s amount to almost 2000 billion RMB’s, making up for a NPL percentage of total outstanding loans of 1.75%. Underling NPL exposure to the housing market (via mortgages and real estate loans) is rising, which pose additional risks to the banks as lenders for real estate (Figure 12). However, there are indications that these figures might be understated. BMI, for instance, argues that China’s total NPL ratio could be as high as 20%, but also admits that it is difficult to accurately point out the NPL ratio as a result of lack of data (BMI, 2016).
In the event that house price gains lose steam, property developers would likely struggle to pay a portion of their loans, which in turn result in a further increase in banks' NPL’s and would also force them to record investment losses. At the same time, banks have become increasingly exposed to property developers through trust companies and directional asset management plans, on top of the usual mortgage loans, which are also increasing, creating an additional layer of risk and uncertainty to bank’s balance sheet risks which are already downwardly pressured by lower profitability (Figure 13).
Government policies to curb price surges
The government is also worried that the property market is overheating and has been actively signaling this to the public. In recent months, several articles published in the People’s Daily defend measures which have been installed this year curbing the rise in real estate prices, such as home purchases restrictions (for second and third homes) and higher mortgage down payments. Additional restraining measures for real estate developers sources of financing were introduced in order to cut rising corporate debt, including encouraging mergers and acquisitions, swapping debt for equity and facilitating firms’ bankruptcy. Next to this, an 8 year old article by Xi Jinping’s most important economic advisor, Liu He, has been spreading rapidly on social media. The article explains the importance of maintaining affordability of houses for low-income groups. In the short term, however, housing prices are only driven up, because buyers are rushing in before new restrictions by local governments are put in place (see BMI).
More than 20 Chinese cities have introduced restrictions on home purchasing this year, including Shanghai, Beijing, Tianjin, Chengdu, Wuxi, Wuhan and Zhengzhou. The emphasis of restrictions seemed to be more related to home buyers than real estate developers, as they do not prevent property developers from gaining access to cheap credit, which has encouraged them to acquire land at record-breaking values and has contributed to surging housing prices. However, China’s State Council has announced additional plans to reduce elevated corporate debt levels. The plans include encouraging mergers and acquisitions, allowing a greater number of bankruptcies, and promoting debt securitization. Additionally, the PBoC will take into account off-balance sheet financing at commercial banks to assess their overall financial health, requiring commercial banks to count high-yielding wealth-management products as part of their overall credit (e.g. Reuters, WSJ). As a result, banks have to hold more capital for their off-balance sheet exposures. Next to the PBoC, the Shenzhen Stock Exchange tightened rules for listed real estate firms seeking to issue bonds, joining its Shanghai counterpart in efforts to curb excessive capital flows into the property market (see BMI).
Critics argue that China’s government leaves fundamental land and fiscal problems unsolved, as the curbing rules are viewed as top-ups of restrictions just mentioned. Accordingly, they fail to address structural issues as dependence of local governments on property and land deals for their income as well as a lack of investment opportunities. One way to solve this could be the installment of a property tax, putting pressure on speculation on the property market. Some cities imposed a property tax scheme before since 2011 (e.g. Shanghai and Chongqing), but the programme was not implemented thoroughly, due to the lack of a real estate and property registration systems. Without such systems, it is almost impossible to get knowledge about property owners, making it unmanageable to tax property. It is however noteworthy that China is currently inspecting to launching a property registration system by the end of the year (e.g. Reuters).
Implications for the construction sector
The developments in the housing market also directly affects adjacent sectors, such as the steel sector, residential construction and the cement sector. Also from a political point of view it is important to keep construction activity going, as many rural migrants have jobs in these sectors. We can make an assessment of the potential fallout in the residential construction sector in case of a hypothetical housing market bust by using a model (Box 2) and conduct scenario analyses.
In our model, we have linked growth in residential floor space under construction to mutations in residential house prices, sales and the stock of empty houses. Next, we make assumptions on these three determinants (see Figure 15) and use the model to forecast how the construction sector is affected in a baseline scenario, a positive scenario and a negative one. The scenarios should not be interpreted as forecasts of what will actually happen on the Chinese housing market. The scenarios solely provide a bandwidth of the potential impact on the residential construction sector under different hypothetical circumstances.
Box 2: The impact of the housing market on the real estate construction sector
We estimate the following model to translate developments in the housing market into real estate construction sector activity:
where R isresidential floor space under construction (data from NBS), H is an real estate residential price index (data from BIS), S is the amount of residential sales (data from NBS) and V is the number of vacant houses (data from NBS). Furthermore, t is quarter and D indicates the year-on-year mutation of the variable. In short, our model captures all dynamics of the residential real estate market: the residential sales component, the stock of empty residential houses and the price development (which also contains a potential speculative component). We have also included a lagged dependent variable to tackle possible omitted variable bias. We use the two and three-year lagged difference in floor space construction as instruments for the lagged dependent variable. General Methods of Moments (GMM) estimation techniques are used to estimate the model for the sample period 2007Q1-2016Q3. Real estate prices are lagged by one quarter and vacant home by two quarters.
In Table 3, the estimation results of our models are presented. All included variables have a significant effect on growth in construction activity and show the expected sign. A rise in sales or house prices will result in higher residential construction activity: a2 and a3. Moreover, destocking of empty houses will also urge real estate developers to prop up their activity, as shown by the significant negative impact of a4. The model explains 90% of the variance in growth of residential floor space under construction (see Figure 14).
The impact of the housing market on the construction sector
Next, we will discuss the outcome of three scenarios. In our baseline scenario (orange dotted lines), the decline in growth of residential sales will continue (Figure 15b), prices will reach their maximum growth of 7% in the first quarter of 2017 (Figure 15c) and will subsequently follow a downward trajectory and, finally, the stock of empty houses will stabilize (Figure 15d). In our positive scenario (light blue dotted lines), sales will continue to grow which will help to destock the housing market at a fast pace (Figure 15b and 15d) and price growth will max out on 11% (y-o-y) in Q2 of 2017 (Figure 15c). In our negative scenario (dark blue lines), panic sales are triggered in Q4 of 2016 (Figure 15b) and will even build up to -30% in 2017 (y-o-y growth). This will result in a sharp correction of residential house prices (Figure 15c) and empty house stocks to build up again during the second half of 2017 (Figure 15d).
The impact of these three scenarios on residential real estate activity is plotted in Figure 15a. In all three scenarios construction will follow an upward growth trajectory (y-o-y) up till the first half of 2017. This is due to the fact that the construction sector is lagging behind the developments in the housing market. Even in our negative scenario, the panic sales will not directly trigger a decline in construction activity, as many ongoing real estate project cannot be terminated immediately. This is also in line with indications that real estate developers are still rushing in to profit from the surging house prices. After the second quarter of 2017, the effects of the different scenarios will unfold itself. In the baseline scenario, the increased construction activity will peak at 22% growth later in 2017. In our negative scenario, construction activity is on a negative trajectory and will even result in declining activity (compared to activity one year ago). In our positive scenario, developers keep rushing in and construction growth more or less resembles surges that we have witnessed in 2008 and 2011.
There are increasing worries about China’s housing market. Since the beginning of 2015, prices have been rising sharply. One could also argue that the high prices rises are justified by economic fundamentals, such as the continuing rate of urbanisation, rising disposable income levels, and cultural values that favour property holdings. In this Special we have assessed whether the current situation is healthy or not and, in case of a real estate bubble, the potential risks to Chinese economy. There are a couple of reasons the housing market could be under so much stress. On the demand side, loose monetary policy and limited options for households to invest capital could be factors driving up demand for property. On the supply side, real estate development is lagging behind due to slowing demand from 2010 onward and increasing vacant property.
A simple affordability index, linking monthly house expenses to disposable income for Chinese cities, arrives at ridiculously high figures. In our calculations, the housing expenses in Shenzhen are 6 times higher than average disposable income, whereas in London this percentage is about 70%. However, we have to take into account that substantial grey income is not included in the data on disposable income. Furthermore, funding of property in China is a multi-generation social affair, involving substantial wealth which is put into the real estate market. Nonetheless, the exceptionally high affordability index does show us that the population base that is able to make a move on the Chinese housing market will likely be much smaller than total urban population figures are showing, especially in the tier 1 and tier 2 cities.
As the Chinese funding of property deviates from the situation in other expensive cities across the globe, we choose to consider China’s housing market fundamentals by its own intertemporal standards in order to assess if there are housing market bubbles growing at this moment. We use a Chinese house price model to forecast the price level per square meter based on several fundamentals, i.e. the growth of the average lending rate, the mortgage lending rate, and the house price-to income ratios in the short and medium term. The house price based on the fundamentals is compared to the actual rise in house prices to identify which part of the house price development is in speculative bubble territory. The outcome suggests that real estate bubbles are present in the tier 1 cities Shenzhen and Shanghai, the tier 2 cities Wuxi, Suzhou, Nanjing, Wuhan and Hefei, and even a couple of tier 3 cities (Nantong, Wuhu, Dongguan and Zhuhai). These cities together have an estimated amount of 92 million inhabitants, covering roughly 12% of China’s urban population.
If the real estate market would collapse in the bubble cities, GDP growth would face increasing downward pressure as the reduction of property sales will additionally weigh on economic growth via lower fixed (real estate) investment and housing sales related to private consumption. Residential construction activity would not immediate respond, as this sector is always lagging behind developments in the housing market. A negative scenario on the housing market will probably result in contracting year-on-year growth in the residential construction sector by the end of 2017.
The introduction of measurements to curb further rise in real estate prices, such as home purchases restrictions and higher mortgage down payments and restraining measures for real estate developers sources of financing by the Chinese government and central bank should put downward pressure on price developments in order to prevent the negative consequences of a real estate bubble environment. It is, however, hard to tell if the current set of measures is sufficient to prevent a future housing market crisis or more macro prudential policies are needed. There is critique that fundamental land and fiscal problems are left unsolved and a controlled introduction of a property tax could help to orderly deflate the real estate market. The government is currently working on such an initiative.
 China’s cities are classified according to a so-called tier system, which involves different criteria, such as population size to rank cities. In the appendix we provide an overview of China’s tier system according to Sinostep.
 Average sale price of a family home in December 2015.
 In contrast to the measurement of money supply, M1 in China only covers cash in circulation and corporate and government demand deposits. M2 encompasses M1 plus household deposits and other deposits.
There is no official division of China’s cities into tiers, although the classification is used widely and extensively by analysts and even policymakers. We use the classification of Sinostep, based on four criteria: their population size, development of services, infrastructure, cosmopolitan nature. In some classifications Hangzhou is regarded as a tier 2 city instead of a tier 1. The exclusion of Hangzhou from our tier 1 group does not significantly alter the estimation results.