Debt-Driven Property Boom, Land-Based Financing and Trends of Housing Financialization: Evidence from China

Abstract

To cope with the global financial crisis, China’s governments issued huge amount of debt to support public infrastructure projects. These financing mechanisms brought about rapid economic restoration, as well as large amounts of debt accumulation. Among other outcomes of this were increasing the leverage in property markets and advancing the extent of financialization in China’s local economy. Financialization level measures the proportion of the total volume of financing provided by the financial system to the real economy, which covers all the generated debts connecting within real economy and financial system. In this study, we outline the mechanism of housing-centered debt expansion process, land-based financing for local governments and trends of housing financialization in local China. Most importantly, the functions of land in such mechanism is highly emphasized. We run fixed-effect and random-effect models to testify the correlation between the kernel variables—financialization level and real estate investment and public infrastructure. To lower the endogenous problems in the estimation, we use instrumental variables (IV) methods and estimate by the two-stage OLS (2SLS) method. The results show that 1% increase (or decrease) of financialization level (measured by the indicator of Aggregate Financing to Real Economy as percentage of GDP) brings about a significant increase (or decrease) of 48% of real estate investment and 59% of public infrastructure investment nationally. Based on the results, we deduce an overview of debt-driven mechanism in China’s local economy named dual financing circulation, which contains two parallel financing circuits, governmental financing based on lands as collateral and market financing based on properties. Finally, the study reveals some new trends of financialization in property markets. Therefore, the major originality of paper is theoretically combining the governmental and private financing circuits as a whole framework for better understanding the financialized local economy of China and putting forward some policy implementations, such as reducing and setting ceilings on leverage of real estate developers in China.

1. Introduction

Financialization is defined as an occurring economic growth mode and redistribution pattern of capital all across the world. In this mode, varied financial tools are created and used to make financial profits through varied financial channels, extruding out proportion of real estate and taking larger position of economy. The traditional production modes of profit through trade and commodity are squeezed out and even gradually replaced [1]. The feature of financialization is the increasing role of financial motives, financial markets, financial actors and financial institutions in the operation of the domestic and international economies [2] and typically presented as an increase in the volume of debts [3]. Financialization of housing is a central aspect and a key objective of financialization [4]. It has been a nationwide and global phenomenon occurring in most countries, such as US, Germany, Singapore, and Brazil [5,6,7]. Moreover, it has become a prevalent feature of economies in the post-global financial crisis (post-GFC) era, originally for the 2008 and 2009 bailout of banks and provision of emergency liquidity and successfully stabilizing the system and stopping the run that threatened to bankrupt the system [3]. As a result, financialization process has reshaped the economy geography and has been engrained in the daily lives of personal individuals, capital usage, and financing circuits by financial institutions and even public infrastructure and service provided by public sectors all over the world, generating the tension between territorial and relational spatialities of geographic differentiation [8]. Even though the term is criticized for being an amorphous concept [9], it is inherent in the fact that the features of financialization linked to the housing sector are varied [10,11]. In China’s case, to cope with the negative impact of the global financial crisis, China’s governments leveraged large-scale public infrastructure investment by increasing public debt financing through an assortment of financial tools [12,13].

Aggregate Financing to Real Economy (AFRE) is a very crucial indicator to measure the total scale financing funds flowed from financial industry to real economy. According to the explanation from the People’s Bank of China in 2013, the definition of Aggregate Financing to Real Economy (AFRE) is the value measuring the total amount of financial funds pouring into the real economy, which aggregates RMB loans, foreign currency loans, trust loans, undiscounted bankers’ acceptance drafts, corporate bonds, etc. The People’s Bank of China started to publish this value for 31 provinces beginning in 2013. In this paper, we conceptualized AFRE into two separate parts, public financing for governmental or public use and social financing for the other use. During the years after global financial crisis, the ratio of AFRE-to-GDP soared up to 39.85%, an overall increase of 82.4%, which was higher than loans-to-GDP ratio in the total non-financial corporate sector (30.90%). To absorb the large amount of debt, China’s governments took actions in two parallel approaches by forming dual financing circulations. The first approach, the land-based financing circulation for local governmental, was mortgaging state-owned construction lands as collateral for financing public infrastructure, which drove land finance-led economic growth at the prefecture-level [14,15,16]. The other one approach, the property-based financing circulation for residents was using a market approach to encourage the private sector to invest in real estate projects. With the rapid yet uncoordinated development of population and land urbanization [17], real estate markets witnessed an unprecedented boom. Residential mortgages have been allowed and many financial products offered by different financial institutions (e.g., credit card, commercial trust, and micro-finance companies) have been widely used to make payments for purchasing property. Additionally, most real estate developers enjoyed low-level interest rates on development loans (kai fa dai). Based on these factors, two significant facts emerge. One is that increasing both demand and supply in the property markets stimulates the price of property. The other is that the total debt and ratio of debt-to-GDP is growing. The property boom in China is largely due to the influence of the debt accumulation that surged in China after the 2008 GFC [12,18]. Other research argues that the process of financialization, particularly featured as debt accumulation in China is driven by the increase of loans to both public and private sectors [19,20].

However, the actual relationship between debt expansion in China after the GFC and its two main financing approaches, investing in public infrastructure and investing in real estate sector, is not well understood and lacks a consistent analysis framework to explain this phenomenon. Additionally, existing mechanisms are limited in explaining the connection between the housing sector, financial sector, other related industries, and local governments. Spatial factors are taken into consideration, by examining the uneven allocation of funds among these entities, with respect to spatial relationships [21], or the relationship between territorial and relational spatialities with respect to competitiveness dynamics [8,22], friction of information flows [23], and globalization of financialization [24]. The uneven distribution of funds is deeply determined by varied distance of access to financial intermediaries or practices [25]. The outcome of this is that the economy is experiencing increasing movements in the prices of real estate and financial assets and by the burden of servicing financial obligations, such as debts [26], creating the new rentier class by channeling real sector savings to speculative short-term investments instead of long-term investment projects in developing countries, in particular [27]. Surely the capital income would more proportionately flow into bankers, homeowners, stakeholders of financial intermediaries, and local governments and less proportionately to people not owning houses. The generated debts and side effects of these are also disproportionately burdened within different sectors and individuals. However, how the proportion of loans for financing real economy affect real estate markets as well as public infrastructure investment is still not clear, and a working chart showing the operation of this mechanism has not been defined.

The first purpose and main achievement of the study is to propose a whole picture of mechanism of financialized economy in local China, deduced from stylized facts and empirical evidence. To fulfil the aim, we put descriptive analysis first, and do empirical studies ahead of theoretical parts as Lin et al. (2018) and Zhang et al. (2020) did [28,29]. This study is organized as follows. In Section 2, we review the main body of literature on the topic of housing financialization, and a concept of financialization of housing is defined in the end of this part. In Section 3, we introduce four stylized facts linking China’s debt-driven property boom by descriptive statistics and design empirical studies, including hypotheses set and models specification. In Section 4, the empirical results present the correlations between the financialization level and property and public infrastructure investment. In Section 5, we try to deduce the dual financing circulation mechanism based on the stylized facts in Section 3 and the empirical studies in Section 4. Trends in financialization of housing will be further discussed in this part too. Section 6 further discusses the significance of our empirical studies, developed mechanisms, and potential studies for the future. Section 7 concludes the study.

2. Literature Review: Housing Financialization in China’s Context

2.1. Definition

Housing financialization is a process that housing markets tightly connect to financial markets via varied financial channels. The concept origins from the financialization of economy. In urban and housing studies, the term financialization is defined as a phenomenon, which there is an increasing dominance of financial actors, markets, practices, measurements, and narratives at various scales across the world. The fact results in a structural transformation of economies, firms (including financial institutions), state and households [30]. According to the previous literature review and China’s specific features in the financialization process, we define housing financialization as follows.

Housing financialization refers to a process that real estate markets and financial markets get more tightly connected. Capital pours into the housing sector through the financial market, resulting in the boom in real estate investment and property purchasing in the way of increasing leverage, finally resulting in property values continuously appreciating. Its stylized features are as follows: (a) First, the value of housing assets is more affected by changes in debt in the economy from the financial system; (b) second, real estate developers use varied financing channels to expand their real estate development and construction, and jointly produce housing-related financial products with financial institutions, resulting in an extremely high-level debt expansion; (c) third, through the financial market and new financial instruments, the residential sector has seen an increase in the leverage ratio of the proportion of residents’ total debts.

There are multiple formats of housing financialization due to the different commercial, social, and cultural background in different countries [30]. For example, in China’s context, housing financialization is not featured as mortgage securitization boom but as a debt expansion due to the creation of owner-occupied housing as a financialized asset to propel a systematic borrowing mechanism in economy [31]. There are two typical facts of financialization in China. First, there is a rapid increase in the volume of speculative real estate (SRE), over 5.4 times averagely from 2006 to 2019. Most of SRE is held by public companies, averagely occupying over 20% of total fixed assets of companies. Second, there underwent a leap of leverage ratio of real estate sectors. The leverage ratio of real estate developers rose from 74.5% in 2011, to 80.4% in 2020. Meanwhile the leverage ration of residents rose from 14.7% in 2011, to 30.1% in 2020. Moreover, the percentage of net new loans (domestic only) of the real estate sectors in total banking loans has increased from 16.9% in 2012 to 44.8% in 2017 (at the peak time). On this basis, China’s housing financialization goes into the center of financialization of China’s economy.

In more detail, the term financialization contains four implications in this study. First, financialization is a pattern of accumulation in which profits accrue primarily through financial channels rather than through trade and commodity production [1]. Second, financialization implies the increasing role of financial motives, markets, actors, and institutions in the operation of domestic and international economics [2]. As a result, the private and non-financial sectors participate in the financialization process through multiple financial instruments such as banking or shadow banking systems and the housing sector is notably affected. Third, in China’s case, the financialization process was reinforced by state-owned financial institutions and the real estate corporations through their activities of maximizing stakeholders’ interests, alongside with land owned by local governments and used by multiple state-owned market players [12,32]. Some research argues that the housing market or “land regime” is treated as an asset of speculative investment. It has been highly associated with the financialization process and exposes the vulnerability to speculation and cyclical risk over the last few decades [30]. Fourth, whether in developed countries or in China, the financialization process and its outcomes are unevenly distributed among different investors, local authorities, and residents who may or may not own properties likely heightening the existing inequalities in income, housing affordability, and stability [33,34].

There is also a growing number of literature on China’s land reform, China’s land, and housing financialization [17,31,32]. These literature present some of the features of China’s financialization process that distinguishes it from other countries. First, the housing financialization in China is not featured in assertization of assets such as stocks, bonds, mortgage, and so on, but featured in the housing-centered mortgage boom in residential property markets due to the state-controlled financial environment [31]. In this case, debt expansion rather than an equity securities boom is the most typical feature of China’s financialization. Second, the process of financialization is highly connected to state macroeconomic policies, which propels the property market boom and expanding scale of public infrastructure investment. It is featured as spatially uncoordinated development of population urbanization and land urbanization across the nation, since China’s governments dominate the allocation of funding, land supply, and resources either directly via enormous fiscal spending, large-scale government-backed borrowing (i.e., state-owned companies have some financing convenience to borrow from banks in lower interest rate than average level of that in markets, all due to the higher-quality credit raised by their actual owners, such as state-owned Assets Supervision and Administration Commission, SASAC), local governments land transferring or indirectly via the state-owned banking system or local government financing platforms (LGFPs). In China, LGFPs refer those companies established by local governments with function of financing as the main business purpose, including different types of urban construction investment, development and so forth, which operate mainly by charging public facilities fees, fiscal funds, and other operating income as source of repayment of debts. Third, China’s financialization operates in the context of promoting economic recovery after the global financial crisis in 2008. Hence, it forms main parts of the picture of the debt expansion process.

Moreover, the topic of housing financialization tightly lies in housing finance system. It is an increase in research on housing finance and relationships between housing and finance sectors. There is also a body of literature on topics above. However, in China’s context, financing practice mainly relies on banking financing rather than non-banking financing (i.e., stock markets and bonds markets), which makes it different. On the basis, real estate markets (including properties and land) take critical roles as collateral channel for either corporate or governmental investment [35]. Moreover, as a kind of asset, properties held by residents provide an approach to expand their credit constraints and share property value appreciation process [36]. Unlike some other countries, where there is deeper mortgage markets as well as property-derivatives markets [37], where there is more open access to global investors and thus more risk exposure in real estate markets via relevant stock markets, subprime financial product markets, and so on [38] and where market failures and policy failures may commonly occur [39], China’s real estate market functions as a collateral for debts financing, which could dramatically improve the response of aggregate demand to housing price shocks [40], so as to stabilize the fluctuation of economy as one of the policy tools, and meanwhile as share stock of local economic growth [41], and therefore to propel incentives all over societies. In this study, we go further and would like to briefly outline a whole image of housing financialization in China’s modes, jumping out of traditional discussion of institution or efficiency of housing finance and connection between debts or financing behaviors and housing markets. An image of housing-centered debt expansion process of China is illustrated in Section 5 of the paper.

2.2. Housing-Centred Debt Expansion Globally and Property Financialization in China

The phenomenon of financialization is occurring across nations all over the world [6,10,11,42]. A typical feature is the entwinement of financial markets, the urban built environment, and housing sector in particular [34,43]. It occurs not only in property markets, but also transfers to rental markets in many cities of the world [11,44,45,46]. By examining different combinations between home ownership ratios and mortgage-to-GDP ratios, there are four trajectories in housing-centered financialization for developed countries globally [47]. Financialization in most developed countries develop around debt accumulation, in particular residential mortgage loans and their derivatives [48]. However, China’s housing-centered financialization has developed in a different fashion from these four trajectories, as it is not simply mortgage-oriented. The process is featured as a high-level home ownership ratio as well as a high-level mortgage-to-GDP ratio. Moreover, such a financialization process is not defined as a typical property mortgage securitization process but more of a creation process of owner-occupied housing as a financialized asset to propel the wider financialization of the economy in the context of a state-controlled financial environment [31]. Moreover, this process operates by expanding opportunities in the urban infrastructure investment process for restoring the economy after the global financial crisis [17,32].

A central issue for governments across the world is how to finance urban infrastructure to contribute to economic recovery, especially following the time of global financial crisis [13,20,49]. Local governments act more as entrepreneurial governors whether in the UK, Eastern Europe or in East Asia [50]. Local governments attempt to treat infrastructure as part of their investment portfolio using land as collateral to provide long-term and stable revenue streams that are isolated from business and financial cycles [51]. Different from most of other countries, China’s local governments monopolize the supply of urban lands according to the Constitutional Law and Land Management Laws 1988, which allows them to bundle packages at low transactional costs. More importantly, to finance public infrastructure projects, they pledge the land as collateral to borrow money from state-owned banking or non-banking financial institutions, such as trust companies, by issuing city investment bonds (CIBs or cheng tou zhai) or trust plans. However, most of these financial institutions are owned by state or local governments. When the global financial crisis happened, changes in these processes and the financial structure were instituted for governments to propel economic growth in a quick way when handling a crisis situation.

Therefore, to cope with the economic recession, China’s central and local governments chose to expand public debt by establishing local government financial vehicles (LGFVs) or local government financial platforms (LGFPs) and issuing city investment bonds (CIB, chen tou zhai) to promote local economic growth [32], which created a “develop-by-borrowing” economy [16]. Moreover, the Chinese government issued a four trillion RMB (approximately USD 586 billion) stimulus package at the end of 2008, in which most of the public investment funds went into public infrastructure construction (see

Table 1. The approximate amount and investment areas of the 4 trillion RMB stimulus package from the fourth quarter of 2008 to the end of 2010.

Public Investment Area	Amount: Billions
Transportation infrastructure (e.g., railway, highways and airport)	1500
Post-disaster reconstruction	1000
Public housing (e.g., low-rent housing and shantytowns transformation)	400
Infrastructure in rural area (e.g., water, electricity and gasoline network)	370
Innovations and structural adjustment	370
Energy conservation and emission reduction	210
Technology, education, cultural and health care	150

Source: China’s National Development Reform Commission (NDRC) website.

).

Public financing due to its reliance on local financing platforms (LFPs) and increasing issuance of city investment bonds (CIBs) contributed to the debt-driven economy in China [16]. By financing through local government financial vehicles (LGFPs), sufficient public funds were available to promote transportation and other public infrastructure projects (e.g., high-speed rail), which increased the ratio of governmental loans-to-GDP by 26.8 times over the period of 2008 to 2015. The number of city investment bonds (CIBs) increased by 788% and the balance increased by 470% from 2012 to 2020 (see Figure 1).

In the meantime, real estate markets and public infrastructure investment witnessed a continuous boom. Property prices experienced an almost four-fold increase from 2003 to 2019. Moreover, enlarging real estate investment undertook a very large proportion of China’s GDP and investment, accounting for around 25% of investment and over 13% of GDP per year (see Figure 2). However, the logic connecting debt accumulation -featured financialization process, property boom (characterized as a rapid increase in real estate investment and skyrocketing prices), and increased infrastructure investment has not yet been outlined and discussed. This will be covered in Section 3, where we outline four stylized facts linking financialization, property investment, and public infrastructure expansion.

3. Empirical Studies: Stylized Facts of China’s Debt-Driven Property Boom

3.1. Data and Methods

As demonstrated in introduction part, we select the indicator Aggregate Financing to Real Economy (AFRE) as the presented variable of financialization in economy. AFRE-to-GDP ratio represents the proportion of marginal financing funds into real economy in economic productivity (GDP). AFRE-to-GDP ratio is first used to measure the extent of debts funded into economy from financial system [12]. We use it as financialization level (fl) in this study. In order to test hypotheses a and b, we estimate the correlations between the variables real estate investment (rei) and financialization level (fl) variations, as well as the connections between the variables public infrastructure investment (pii) and financialization level (fl) variations. Three dummy variables, east, middle, and west are set to make a comparative analysis based on spatial factors. According to China’s National Statistical Bureau, the variable public infrastructure investment (pii) aggregates the investment of public transportation, storage, post, management of water conservancy, environment, and public facilities.

At the end of 2022, the Notice on Stopping Illegal Financing Activities by Local Governments (Document No.463) is issued. The document is to restrict the varied illegal financing practices for rapid credits expansion by local governments and financial corporations. The issuance of the document is treated as an influencing exogenous shock to local economy. The financing practice of local governments and financial corporations stepped into a new era of legalization. Considering to eliminate the exogenous impact on economy system in local China, we selected data for the years of 2013 to 2019 to remove the effect of such exogenous policy working on financing infrastructure projects. In order to further test the relationship between the two kernel variables, a fixed or random effects model needs to be run based upon panel data that we collected from 30 provinces.

The data for property markets were collected from the China Real Estate Yearbook from 2013 to 2020, which is published by the Ministry of Housing and Urban-Rural Development. The key explanatory variables, the data of AFRE are collected from the People’s Bank of China and summarized by province. The controlled variables regarding economic growth of local areas are from China’s Statistical Yearbook from 2013 to 2020, including GDP (gdp), population (pop), population density (popdensity), land revenue (landrev), governmental expenditures (ge), taxation (tax), per capita disposable income (pcdi), real estate price (reprice), urbanization rates (urbanrates), interest rates (rates), and housing mortgaged rates (housingrates). The sample size of provinces is 30, excluding Tibet, because there are no data available for the period of the study due to the real estate markets of Tibet not yet being considered mature. Moreover, for all data concerning prices, the inflation effects are removed and re-counted using the CPI index and the property investment are revised with the fixed assets investment price index (FAIPI). Both CPI and FAIPI were collected from China’s Statistic Yearbook for 30 provinces. Considering the spatial imbalance between population growth, land use, and housing supply is the central issue for urban China [52], test of spatial difference of the correlation between kernel variables are needed. Using panel data for 30 provinces over the seven year period of 2013 to 2019, we test the spatial difference of those correlations, grouped into three regions (east, middle, and west) in China. We set a series of D_i,t as dummy variables for the estimation. When the variables are in the eastern, middle, western, or northeast areas, the D_i,t−1 = 1, otherwise 0. All statistics of variables at the nationwide level are described in

Table 2. Aggregate financing to real economy, financialization level and housing investment-to-GDP changes from 2013 to 2017 in the eastern, middle and western regions in China.

	2013	2014	2015	2016	2017	2018	2019
Aggregate Financing to Real Economy	17,316.9	15,876.1	15,406.3	17,799.9	26,153.6	22,492	25,673.5
Financialization level	29.20%	24.67%	22.36%	23.85%	31.43%	24.47%	25.91%
Real Estate Investment (billion RMB)	8601.34	9503.56	9597.89	10,258.06	10,979.85	12,016.48	13,219.43
Public Infrastructure Investment (billion RMB)	7445.4	8944.07	10,487.96	12,253.8	14,355.6	14,866.20	15,329.25
GDP (billion RMB)	59,296.32	64,356.31	68,885.82	74,639.51	83,203.59	91,928.11	98,651.52
Population (million persons)	1355.16	1362.46	1370.88	1379.84	1388.34	1396.53	1403.85
Density (persons per sqr km)	2817.61	2850.03	2775.55	2809.45	2796.58	2894.29	3008.10
Land Revenue (billionRMB)	43,745.30	34,377.37	31,220.65	36,461.68	51,984.48
Reprice (RMB per sqr meters)	6237	6324	6793	7476	7892	8726	9310
Government expenditures	11,974.03	12,921.55	15,033.56	16,035.14	17,322.83	18,819.63	20,374.32
Taxation (billion RMB)	53,890.88	59,139.91	62,661.93	64,691.69	68,672.72	75,954.79	76,980.13
Pcdi	26,467	28,844	31,195	33,616	36,396	39,251	42,359
rates	6.43%	6.66%	5.93%	5.50%	5.18%	5.06%	5.29%
Urban rates	54.45%	55.55%	56.64%	57.85%	58.98%	59.99%	60.85%

Source: China’s National Statistics Bureau.

.

For the empirical study methods, we will outline overall characteristics of infrastructure, property markets, and lands first using descriptive statistics. Then, we will estimate the correlations between the financialization level (fl) and property investments based on the panel data by running fixed-effect (FE) or random-effect (RE) regression models to verify hypothesis a. Then, dummy variables representing different regions of China are introduced to see if the correlations hold true on a regional basis. Afterwards the focus will be on the correlations between financialization level (fl) and infrastructure with regards to figure out what correlations more significantly affect each other and whether the increase in AFRE impacts development in local areas by financing both public infrastructure and the housing sector or if it only contributes to one area.

3.2. Descriptive Statistics and Hypotheses

This section documents four stylized facts about China’s property financialization. The first and second facts are simply described. The first fact is that the amount of AFRE at the nationwide level increased very quickly from 2011.2 billion RMB in 2002 up to 25,673.5 billion RMB in 2019, an over ten-fold expansion. The indicators of AFRE and its data point to the debt expansion trends in China’s economy. Moreover, the ratio of AFRE-to-GDP, the measurement of financialization level, remains around 25% after shooting up to nearly 40% in 2009 (see Figure 3). In this context, the debt in China’s economy expands accordingly.

The second fact is that both public infrastructure investment and real estate investment experienced an increase in order to rehabilitate the economy (see Figure 4). As we can see in Table 1, most financed loans or other common funds described above were poured into these two sectors.

However, the connection between either of the two kinds of investment and the debt expansion process is ambiguous. Since economic development varies in different regions of China, it can significantly affect the spatial differentiation of financializaton levels, public infrastructure, and real estate investments and the interconnecting factors between them. These connections need to be tested to better understand the nature of their relationships. In order to test the relationships between the real estate investment boom and debt expansion (the third fact) and simultaneously between public infrastructure increase and debt expansion (the fourth fact), we build an econometric model based on the panel data of 30 provinces from 2013 to 2019 to test the two hypotheses as follows:

Hypothesis a.

The property boom is positively affected by financialization level.

Hypothesis b.

The public infrastructure boom is positively affected by financialization level.

3.3. Model Specifications

To test Hypotheses a and b, we specify and estimate the basic econometric model (1) and (2) as follows:

$$\begin{gathered} \ln {\mathrm{rei}}_{\mathrm{i},\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1\ln {\mathrm{fl}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_2\ln {\mathrm{gdp}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_3\ln {\mathrm{popd}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_4\ln {\mathrm{landrev}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_5\ln {\mathrm{rep}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_6\ln {\mathrm{ge}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_7\ln {\mathrm{pcdi}}_{\mathrm{i},\mathrm{t}}+ \\ {\mathsf{\beta }}_8\ln {\mathrm{rates}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\eta }}_1{\mathrm{east}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\eta }}_2{\mathrm{middle}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\eta }}_3{\mathrm{west}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\gamma }}_{\mathrm{i}}+{\mathsf{\eta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}, \end{gathered}$$

(1)

$$\begin{gathered} \ln {\mathrm{pii}}_{\mathrm{i},\mathrm{t}}={\mathsf{\chi }}_0+{\mathsf{\chi }}_1\ln {\mathrm{fl}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\chi }}_2\ln {\mathrm{gdp}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\chi }}_3\ln {\mathrm{pop}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\chi }}_4\ln {\mathrm{landrev}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\chi }}_5\ln {\mathrm{rep}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\chi }}_6\ln {\mathrm{tax}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\chi }}_7\ln {\mathrm{pcdi}}_{\mathrm{i},\mathrm{t}} \\ +{\mathsf{\chi }}_8\ln {\mathrm{rates}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\theta }}_1{\mathrm{east}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\theta }}_2{\mathrm{middle}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\theta }}_3{\mathrm{west}}_{\mathrm{i},\mathrm{t}}+{\varnothing }_{\mathrm{i}}+{\mathsf{\phi }}_{\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i},\mathrm{t}}. \end{gathered}$$

(2)

In model (1), rei_i,t is the property investment in urban areas for the 30 provinces from 2013 to 2017. i is for the number of province, while t is for the number of year. fl_i,t is the measurement of the financialization level. We also select the GDP (gdp_i,t), population density (popd_i,t), land revenue (landrev_i,t), governmental expenditures (ge_i,t), per capita disposable income (pcdi_i,t), real estate price (reprice), and housing mortgaged rates (housingrates_i,t) for controlling the effects of economic growth [16], public finance, households’ income, real estate price [36], and interest rates, which might have significant impact on property investment. We also set financialization level (fl_i,t) as lagged (t − 1) phases for model (1–3) to (1–4) to estimate the effects of lagged financialization level (fl_i,t). To consider the might-be effects of spatial difference, we set dummy variables, east_i,t, middle_i,t, and west_i,t, to do regression of basic models by groups and check the difference within groups. To test the dynamic changes between the variables, we code all variables into ln format.

To test hypothesis b, public infrastructure investment (named pii_i,t) is the total amount of public infrastructure investment in 30 provinces during the time between 2013 and 2017. We select the controlled variable GDP (gdp_i,t), representing economic growth rate, population (pop_i,t) at i province at year t, land revenue (landrev_i,t), and taxation (tax_i,t), representing land revenue and fiscal revenue for local governments, at i province at year t, variables representing features of real estate markets at i province at year t, like real estate price (reprice_i,t) and interest rates (rates) and other variables presenting life condition of residents at i province at year t, like per capita disposable income (pcdi_i,t) and urbanization rates (urbanrates_i,t). All variables are coded into ln format as well as in model (1).

4. Estimation Results

4.1. Debt-Driven Housing Investment Model Results

Table 3. Estimation results (1).

Yi,t-1	LnREIi,t,	Nationwide (1-1)	Naitionwide (1-2)-fe	Nationwide (1-3)	Naitionwide (1-4)-fe	Nationwide Year 2013–2016 (1-5)	Nationwide Year 2013–2016 (1-6)-fe	Nationwide Year 2017–2019 (1-7)	Nationwide Year 2017–2019 (1-8)
Xi,t-1	LnAFRE/GDPi,t	0.112 *** (3.78)	0.088 ***(2.98)			0.081 *** (2.46)	0.072 *** (2.31)
	L.LnAFRE/GDPi,t-1			0.059 *** (2.16)	0.055 ** (2.08)			0.055 * (1.45)	0.067 * (1.83)
Controlled variables	Ln pricei,t	−0.187 * (−1.52)	−0.349 *** (−2.44)	−0.049 (−0.34)	−0.282 ** (−1.77)	−0.059 (−0.35)	−0.328 (−1.24)	0.067 (0.35)	−0.304 (−1.10)
	LnGDPi,t	0.256 *** (3.43)	0.207 ** (2.59)	0.209 *** (2.71)	0.172 ** (2.11)	0.340 *** (2.85)	0.0367 (0.26)	0.273 *** (2.22)	−0.011 (−0.08)
	Ln densityi,t	−0.170 *** (−2.84)	−0.136 ** (−1.89)	−0.217 *** (−3.28)	−0.172 *** (−2.24)	−0.145 **(−1.91)	−0.246 *** (−2.57)	−0.190 ** (−2.12)	−0.310 *** (−2.90)
	LnLand Revenuei,t	0.300 *** (10.34)	0.268 ** (9.09)	0.290 *** (9.25)	0.259 *** (8.29)	0.347 *** (8.44)	0.256 *** (5.73)	0.367 *** (8.28)	0.263 *** (5.55)
	Lngovernmentexpendituresi,t	0.463 *** (4.07)	0.600 *** (3.34)	0.556 *** (4.87)	0.690 *** (3.73)	0.273 * (1.53)	−0.154 (−0.49)	0.365 ** (2.00)	−0.188 (−0.55)
	lnpcdii,t	−0.697*** (−4.04)	−0.609*** (−2.70)	−0.847 *** (−4.55)	−0.709 *** (−2.89)	−0.893 *** (−2.91)	0.365 (0.76)	−0.963 *** (−3.01)	0.506 (0.93)
	lnratesi,t	−0.308 *** (−3.79)	−0.248 *** (−2.70)	−0.291 *** (−3.36)	−0.232 *** (−2.66)	−0.793 *** (−3.25)	−0.494 * (−1.72)	−1.028 *** (−4.06)	−0.594 ** (−1.99)
Dummy variables	Easti	0.594 *** (4.24)		0.586 *** (4.18)		0.671 *** (3.97)		0.621 *** (3.73)
	Middlei	0.257 ** (1.84)		0.287 ** (2.08)		0.349 *** (2.18) **		0.369 *** (2.36)
	Westi	0.144 (1.12)		0.150 (1.17)		0.278 (1.93) **		0.285 ** (2.02)
	Constants	9.672 *** (9.09)	9.741 *** (8.94)	10.259 *** (9.11)	10.160 *** (8.87)	11.674 *** (5.51)	8.893 *** (3.23)	11.862 *** (5.18)	8.609 *** (2.79)
	R2	0.92	0.56	0.91	0.55	0.95	0.58	0.95	0.56
	rho	0.68	0.90	0.65	0.90	0.86	0.987	0.83	0.99
	Sample	210	210	203	203	90	90	87	87
Hausman test	Chi-square		193.93		20.44		28.94		29.30
	p-value		0.000		0.0088		0.0003		0.0003
	FE or RE		FE		FE		FE		FE

t or z value are in parentheses. ***, p = 0.01; **, p = 0.05; *, p = 0.1.

shows the results of model (1) and its related variants. The main results are as follows: (a) according to the estimation results of model (1-1) and (1-3), at the nationwide level, the financialization level (fl) changes present a positive correlation to real estate investment (rei) changes at the 1% significance level. As running the model (1) without controlling its fix or random-effects model, the coefficient of β₁ is 0.112, which indicates that every 1% of financialization level increase (or decrease) will bring an 11.2% (increase or decrease) of real estate investment (rei), and 1-period lag fl is also positively correlated to real estate investment (rei) at 1% significant level (the coefficient is 0.059). As the indicator, financialization level, refers to an aggregate credits scale financed into economy. It shows a tiny change of proportion of debts to economy will bring about a big leap of real estate investment. The controlled indicators GDP, population density (popdensity), land revenue (landrev), real estate price (reprice), governmental expenditures (ge), per capita disposable income (pcdi), and interest rates (rates) also significantly affect real estate investment (rei). Moreover, spatial factors do matter. Both the east and middle regions as spatial parameters have a positive correlation to real estate investment (rei) as well. (b) According to the estimation results of model (1-2) and (1-4), if we control the factors in the estimation formula so that it does not change along with timing, say, estimating the fixed-effect format of model (1), the results are still robust. Compared to econometric estimation based on time series data of some certain real estate markets on single nation or city as research sample [38], we choose panel data in this study to check out some unobservable individual characteristics at the provincial level of China. We also carried out the Hausman test to figure out whether or not fixed-effects model is more significant than random-effects model on the condition that the estimation of two models are both unbiased and consistent. The results of the Hausman-test found that the fixed effects models are all consistent to be selected. According to the estimated model (1-2), the value of financialization level (fl) is also significantly positive to real estate investment (rei) changes on a national level. 1% of the financialization level (fl) increase (or decrease) contributes to 8.8% of real estate investment (rei) growth statistically significant at the 1% level. Moreover, the controlled variables GDP, population density (popdensity), land revenue (landrev,), at the 5% significance level, governmental expenditures (ge), per capita disposable income (pcdi), interest rates (rates), at the 1% significance level, present positive correlations to the real estate investment (rei) variable. (c) Moreover, the positive correlation between land revenue (landrev) and real estate investment (rei) are consistent in all models (the coefficient is about 0.30), which indicates that land, as a major proportion of cost to real estate investment and as collateral to a vast proportion of financed loans from government and corporations to the banking system, contribute largely to the property boom.

4.2. Debt-Driven Infrastructure Model Results

Table 4. Estimation results (2).

Yi,t-1	LnPIInvesti,t,	Nationwide (2-1)	Nationwide (2-2)-fe	Nationwide (2-3)-iv	East (2-4)-fe	Middle (2-5)	West (2-6)-fe
Xi,t-1	LnAFRE/GDPi,t,t	0.179 *** (3.19)	0.177 *** (3.23)	0.590 **(1.81)	0.192 ***(3.15)	0.424 * (1.73)	0.094 * (1.59)
Controlled variables	LnGDPi,t	1.393 *** (7.84)	1.254 *** (6.61)	1.411 *** (5.63)	0.838 *** (3.63)	2.770 *** (2.87)	1.716 *** (4.57)
	Lnpopulationi,t	−0.116 (−0.48)	−0.087 (−0.05)	1.424 (0.58)	0.606 (0.30)	−30.278 *** (−2.89)	0.945 (0.39)
	LnLand Revenuei,t	0.038 (0.72)	−0.024 (−0.44)	−0.036 (−0.56)	−0.122 * (−1.69)	0.373 (1.47)	−0.107 * (−1.59)
	Ln ppricei,t	−0.608 *** (−2.67)	−0.606** (−2.25)	−0.387 (−1.09)	−0.790 *** (−3.15)	−2.875 *** (−2.68)	0.983 *** (2.59)
	Lntaxationsi,t	−0.544 *** (−3.21)	0.049 (0.23)	−0.271 (−0.77)	0.262 (0.81)	−1.329 ** (−2.09)	−0.619 ** (−1.84)
	lnpcdii,t	1.935 *** (7.35)	1.419 *** (4.07)	1.132 *** (2.46)	2.039 *** (3.47)	3.909 ** (2.01)	0.725 (1.43)
	lnurbanratesi,t	−0.883 * (−1.68)	1.229 * (1.74)	1.720 ** (1.91)	0.786 (0.84)	0.261 (0.261)	−1.951 (−1.55)
	lnratesi,t	−0.143 (−0.61)	0.030 (0.13)	−0.256 (−0.74)	0.042 (0.19)	−0.593 (−0.82)	−1.052 *** (−3.08)
	Constants	−12.570 *** (−3.72)	−18.757 (−1.29)	−32.005 * (−1.63)	−25.029 * (−1.66)	233.922 *** (2.79)	−17.397 (−17.397)
Dummy variables	East	−0.015 (−0.06)
	Middle	0.238 (0.97)
	West	0.502 ** (2.11)
IV ratio	lndebts			0.719 ** (2.15)
	lnrdinvest			0.013 (0.32)
	lnrdratio			−0.075 (−0.94)
	R2	0.80	0.80	0.73	0.90	0.85	0.93
	F-test		12.38	9.34	13.66	2.66	7.66
	rho	0.65	0.94	0.97	0.98	0.999	0.97
	Sample	210	210	210	70	42	77
Hausman test	Chi-square		39.14
	p-value		0.0000
	FE or RE		FE

t or z value are in parentheses. ***, p = 0.01; **, p = 0.05; *, p = 0.1.

reports the estimation results of model (2) and its variants. Results show that the correlations between the financialization level and infrastructure investment are significantly positive. The main results are as follows: (a) 1% financialization level (fl) increase will bring about 17.9% (17.7% in FE model) of public infrastructure investment (pii) increase at the 1% significant level. (b) The spatial factors have less significant impact on public infrastructure investment (pii) as in model (1), only in dummy variable west_i,t, is it found to have positively correlated to pii (the coefficient is 0.502) at the 5% significance level. (c) The controlled variables, GDP (gdp), real estate price (reprice), taxation (tax), per capita disposable income (pcdi) and urban rates (urbanrates) are positively correlated to public infrastructure (pii), but the impact of taxation (tax) is not significant in the FE estimation. (d) The variables reprice in both models (1) and (2) have negative correlation to either real estate investment (rei) or public infrastructure investment (pii) at a significant level.

China’s urban infrastructure financing is heavily dependent on government loans and that is yet to change [53]. Even though some types of infrastructure have not been well managed, the rate of return on such investments will decrease in the next two decades. The infrastructure-led economic growth model will be maintained as it contributed to an increase in firms’ productivity in local areas of China, e.g., high-speed roads [54]. If taking spatial factors into consideration, the lower levels of governments with smaller fiscal budgets will always have to borrow larger amounts to support local economic growth [53]. These loans are an essential part of AFRE from the banking system to local governments. Therefore, the financialization level shows a highly positive correlation to infrastructure investment in time and space.

4.3. Endogenous Problems and the Solution

For both basic estimation model (1) and model (2), the kernel independent variable financialization level (fl = AFRE/GDP) might be correlated to error term ε_i,t, or be affected by dependent variables real estate investment (rei) or public infrastructure investment (pii) resulting in a bias of estimation. To solve the endogenous problem, we use instrumental variables (IV) to estimate by the two-stage OLS (2SLS) method. Table 4 and

Table 5. First-stage OLS estimation results.

Yi,t-1	LnREIi,t,	Nationwide (1-9)	Naitionwide (1-10)	Nationwide (1-11)	Naitionwide (1-12)
Xi,t-1	LnAFRE/GDPi,t,t	0.466 ** (1.85)	0.480 ** (1.95)	0.487 ** (1.99)	0.487 ** (1.99)
	L.LnAFRE/GDPi,t-1
IV- First stage	lnrfavgwage	0.426 ** (1.98)	0.422 ** (1.96)	0.411 ** (1.89)	0.411 ** (1.89)
	numebersfinstitute	5.85 × 10−6 (0.47)	6.44 × 10−6 (0.51)	6.55× 10−6 (0.52)	6.55 × 10−6 (0.52)
	fpopulation	0.002 (0.35)	0.002 (0.30)	0.002 (0.26)	0.002 (0.26)
	lnrdinvest		−0.012 (−0.61)	0.002 (0.04)	0.002 (0.04)
	lnrdratio			−0.032 (−0.38)	−0.032 (−0.38)
Controlled variables	Ln pricei,t	−0.260 (−1.25)	−0.256 (−1.21)	−0.255 (−1.20)	−0.255 (−1.20)
	LnGDPi,t	0.642 ** (2.09)	0.659 ** (2.18)	0.667 ** (2.21)	0.667 ** (2.22)
	Ln densityi,t	−0.172 * (−1.67)	−0.173 * (−1.66)	−0.174 * (−1.65)	−0.174 * (−1.65)
	LnLand Revenuei,t	0.297 *** (6.54)	0.298 *** (6.49)	0.299 *** (6.46)	0.299 *** (6.46)
	Lngovernmentexpendituresi,t	−0.058 (−0.12)	−0.083 (−0.17)	−0.095 (−0.19)	−0.095 (−0.19)
	lnpcdii,t	−0.409 (−1.20)	−0.401 (−1.16)	−0.397 (−1.14)	−0.397 (−1.14)
	lnratesi,t	−0.595 ** (−2.32)	−0.609 *** (−2.40)	−0.615 *** (−2.43)	−0.615 *** (−2.43)
	Constants	7.650 *** (3.73)	7.568 *** (3.69)	7.531 *** (3.66)	7.531 *** (3.66)
	R2	0.14	0.11	0.10	0.10
	rho	0.5	0.63	0.62	0.62
	Sample	210	210	210	210
F test	F test	6.98	6.73	6.61	6.61

z values are in parentheses. ***, p = 0.01; **, p = 0.05; *, p = 0.1.

present the results of the estimation by IV methods in model (1-x) and model (2-y).

Theoretically, more financial institutions, larger financial population, and higher wages of the sector would bring about the higher financialization level. But on the other hand, the variables are not significantly correlated to real estate sectors. There is no evidence to show that the staff of financial sector have preference and do take actions to purchase house or do real estate developers so as to propel the real estate investment. In fact, according to National Statistics Bureau, the proportion of population employed in financial sector in China, was about 5% in 2020, ranked No. 6 of all 19 industries. It takes a very small proportion compared to other industry, which means there is very little impact on property markets by people in financial markets. In models (1-9) to (1-12), we select the indicator average wage in finance industry (avgfwage), numbers of financial institutes (nfi), and finance industry population (fip) as instrumental variables, which have direct correlation to financialization level but little impact on real estate markets, and it is nearly uncorrelated to error term ε_i,t (in model 1). Additionally, in China’s up-to-down innovation planning system, the R&D ratios are set by governments so generally the indicator is treated as an exogenous variables set in economy system [55], thus it has little endogenous impact on some industries, such as real estate. Moreover, the connection between innovation and real estate industry is traditionally not tight all across the world, even though there is an occurrence for further development of sustainability in urban real estate sectors [56]. Especially in China, real estate is not classified into industry with high technology and innovation [57]. However, the innovation industry requires large amounts of financing that could be a lucrative investment, and so we add variables r&d investment and r&d ratio (=r&d investment/gdp) in the first-stage model. The estimations show a consistent result to model (2). (a) The correlation between financialization level (fl) and real estate investment (rei) is still positive at a 5% significant level, but the coefficient increase is over 0.48. It means that 1% of financialization level (fl) increase (or decrease) will contribute to over 48% of real estate investment (rei) increase (or decrease), much higher than that in the basic model (1-2). (b) For the coefficient of variables GDP (gdp), population density (popdensity), land revenue (landrev), and interest rates (rates), the estimation is also consistent in model (1-2), which shows a development and robustness of the specified models.

We use the same methods to develop the basic model (2-2) by selecting real estate developers’ loans (rreedebts) as instrumental variables, which have significant impacts on the financialization level but have little correlation to public infrastructure investment. R&D investment and r&d ratio are also added into the first-stage estimation. Because the innovation industry requires financing demand in higher quantity, and it proves that innovation practice brings about the boom of financial sector [58]. On the indirect approach, the developed financial approach would bring about the convenience of financing of public infrastructure so as to propel the volume of its investment. But on the other hand, there is no evidence to show that R&D investment and R&D ratio has any impacts on infrastructure investment. The results of model (2-3) shows consistency to model (2-1) and (2-2): (a) the coefficient of financialization level (fl) is still positive at 1% significant level and increases to 0.59, which indicates that 1% of financialization level (fl) increase (or decrease) will bring a 59% increase (or decrease) in public infrastructure investment (pii). (b) The estimations of coefficient of the controlled variables GDP (gdp), per capita disposable income (pcdi), and urban rates (urbanrates) also show consistency to the basic model (see Table 4).

4.4. Robustness Test and Summary

To test the robustness of our running models, we use different samples to run models by excluding some exogenous factors such as public policies or spatial differences. To test the robustness of model (1-2), we select the year of 2017 as a treatment and separate the sample into two parts. The year of 2017 was chosen, because at the end of 2016, China’s central government proposed the principles of regulation on property markets, “the houses are for accommodation, not for speculation”, which symbolized the start of strict regulation on the markets and burst of the property boom in China. The results show that in both the periods (year 2013 to 2016 and year 2017 to 2019), the correlation between financialization (fl) and real estate investment (rei) are significantly positive and estimations for coefficient of some other variables are consistent as well (see Table 3). Such results show a robustness for model (1-2).

For model (2-2), we divide samples according to four dummy regional variables, east, middle, and west, because the regional differentiation of economic growth, fiscal capacity, and so forth may illustrate the impacts associated with difficulties in public infrastructure investment. The results show that either in east, middle, or west regions, the coefficient of variable financialization level (fl) is positively correlated to public infrastructure investment (pii), 0.192 in east, 0,424 in middle, and 0.094 in west, which represent a robustness of model (2-2).

To summarize the estimation results, we find a steady and consistent relationship between the financialization level and either property investment or public infrastructure investment. Moreover, GDP (gdp), land revenue (landrev), and other indicators show robust correlation to real estate investment (rei) and public infrastructure investment (pii) at a significant level. The results above are the third and fourth stylized facts of China’s debt-driven property boom, which provide evidence that most of the funds borrowed from the financial system to real economy were actually input into the property markets and public infrastructure sectors. In Section 5, we try to map out the financing mechanism and provide an analysis of the property boom through a new approach.

5. Dual Financing Circulation Mechanism: A New Approach from a Property Financialization Perspective

The crash of the world economy during the global financial crisis (GFC) of 2008 deeply affected debt accumulation in both private and public sectors across the world. The sizable debts in China were produced by two main sectors: infrastructure and housing. As land is controlled by local governments and used as collateral for financing, it bridged the two sectors creating a closed-loop circulation between infrastructure-led public debts and property-led private or social debts.

5.1. Land-Based Dual Financing Circulation and Expanded Debts: Essential Feature of the Financialized Economy

5.1.1. Land-Based Financing Circulation for Local Governmental Financing

Based on the four stylized facts of China’s debt-driven property boom in Section 3, we are going to deduce and create a whole picture of the mechanism of the debt-driven property boom. In Figure 5, the debt-driven mechanism for economic recovery and growth after the global financial crisis (GFC) is outlined in terms of a dual financing circulation, which is comprised of public financing and social financing. For the government financing part, local governments invest assets they actually own into the local financing platforms (LFPs) as capital funds to meet the minimum legal qualifications of finance companies. Local financing platforms (LFPs) are legally eligible to receive financing from financial institutions, which can be used for funding public infrastructure, and in the long term can generate stable cash flow as revenue. The invested assets are mainly those lands that local governments have accumulated but have not yet been sold by guapai (two-stage auction), paimai (British-mode auction), and zhaobiao (sealed bid) approaches [59]. These lands are then mortgaged to banking and non-banking institutions (e.g., trusts, securities and other funds) as collateral. With the release of Document No.463 in 2013, the financing approaches via local financing platforms (LFPs) from banks, commercial trusts and other lenders were severely curtailed and stricter supervision was implemented. This led to some special financing options, such as zheng xin he zuo (cooperation between commercial trust companies and local governments) and banking financial products packaged within public loans at high interest rates along with financial guarantees offered by local governments, being prohibited. The result of these restrictions was the decrease of governmental financing and financing products from multiple financial institutions (the shares of infrastructure trusts decreased approximately from 62.8% to 35.17% in 2013 and to 13.10% in 2015), as well as the rapid switching of social financing funds to property investment where more reliable, stable, and high return of investment (ROI) can be maintained.

5.1.2. Property-Based Financing Circulation for Residents, Developers and Financial Institutions

Compared to governmental financing, the market financing used in residential properties is infinitely more complex. In Figure 5, residential properties are associated with financial systems on both the demand and supply sides of the housing market. On the demand side of the market, residents can get individual mortgage loans from commercial banks. Investors and residents are also eligible to purchase banking products in which the real estate development loans are packaged with other bank financing products as an investment portfolio. In China, due to the restrictions and legal framework in the financial industry, there are still no real financial products packaging individual mortgage loans (IMLs) as one financial product (e.g., MBS or REITs), and at this time they cannot be sold to the public by banking or non-banking systems. The products containing IMIs can only be traded and transferred in interbank markets. For some high-net-worth individuals (HNWI), commercial trust plans are another option. The funds are mostly provided to real estate developers as loans with a fixed return. As housing market was booming and capital for developing real estate projects was competitive in need, the fixed return rate was much higher than saving rate or expected rate of bank financing products. On the supply side of the market, real estate developers get funds as real estate development loans (kai fa dai) from banking or non-banking financial institutions. In order to secure the expected return, the real estate developers mortgage the lands and build properties they owned as collateral to the banks or other financial institutions (e.g., commercial investment trusts corporations).

Land acts as a bridge to connect market players and financial systems. Land is not only a financial resource for local governments under the traditional fiscal framework, but also an expanded approach to financing both public investment and real estate development to increase returns on lands in the possession of local governments [60]. For individuals with residential mortgage loans (RMI), they may mortgage their purchased properties to banks that lend money in return for the residential land beneath their home. For China’s residents, they own 70-year land use rights for their properties after which the rights revert back to the government. In this case, the traditional model of land financing in local areas has turned into a model of land financialization [61].

The nature of the funds flowing in the closed-loop dual circulation system are actually debts [18]. This system depends on local governments being able to repay the debts accrued from the public financing side and is highly reliant upon whether or not the debts connected to the property boom will continue to be repaid by real estate developers with development loans and commercial trusts and residents and investors with mortgage loans. Additionally, developers and property owners maintaining this circulation of debt rely on the appreciation of these properties that is driven by the investment in public infrastructure led by local governments with large amounts of governmental financing and the property boom supported by market financing. The two major paths for local economic growth, property-led growth and infrastructure-led growth, form a morphed financialization process in China’s specific context.

5.2. Infrastructure as an Economic Growth Engine, Properties as Asset Appreciation and Lands as Collateral in China’s Case

5.2.1. Governmental Financing Based on Infrastructure and Market Financing Based on Properties

According to statistics, we can see that the indicator of Aggregate Financing to Real Economy (AFRE) in 2009 rocketed to 13,910.4 billion yuan, approximately double the number from 2008 with the financialization level (AFRE-to-GDP) ratio rising to about 39.8% after Article 10 of the Packages removed bank credit limits. Article 10 of the Packages increased the support for economic growth by multiple financial tools, including removal of restrictions for the amount of commercial banking credits and expansion of credit, especially the credit extended to aid priority projects related to agriculture, rural areas, and rural residents (san nong issues), etc. The real estate or housing sector and infrastructure investment are still the two leading factors of the fixed assets investment (FAI), which contributed around 80% to the GDP (e.g., the FAI-to-GDP ratio was 56.85% in 2019, see Figure 6). It was also designed to reduce debts generated by this process. The local governments also have strong incentives to carry out the packages issued by the State Council and involve themselves in the two-paths debt expansion process in terms of financialization.

The financial institutions that own and manage those funds attempt to put the money into two main sectors, public infrastructure and real estate (see Figure 7). Real estate continuously led the top shares of total trust funds across China after the GFC, especially in 2009 and 2010 (33.79% and 22.75% respectively, see Figure 8).

These trends have made properties especially valuable and oftentimes are the only asset households hold onto for appreciation purposes [62]. Moreover, unlike commercial banking and other financial institutions in other countries, China’s banking or shadow banking systems play a role in lending quasi-fiscal subsidies to the industrial sectors with reliable, high rate of investment return and low risks [44] and in return, banks and those financial institutions acquire assets with high-value as mortgages. As a result, residential mortgage loans (responding to the household sector) and real estate development loans (responding to developers) have increased simultaneously since 2009, while the number of domestic loans (DL) to real estate developers increased by 122% and individual mortgage loans (IMD) from residents almost doubled (179%) from 2009 to 2017. Moreover, the proportion of IMD-to-funds-to-real-estate-developers has continuously grown from 9.81% in 2008 to 16.92% in 2016 (see Figure 9).

In the meantime, the total revenue and supply of state-owned construction lands witnessed a dramatic increase in the years following 2008 (see Figure 10). The numbers almost quadrupled and doubled by 2017, 406.68% (from 10,259.799 in 2008 to 51,984.475 billion RMB in 2013) and 164.85% (from 234.185 in 2008 to 620.246 billion square meters in 2017 from the peak 750.835 billion square meters in 2013) respectively.

Overall, stated-owned lands serve as collateral and public infrastructure serves as economic recovery engines for local governments [63,64]. Real estate serves as asset appreciation tools for private residents.

5.2.2. Tendency of Financialization in Property Markets

Within the functioning of the dual financing mechanism and its associated stylized facts, there are some trends of financialization in the property market that draw our attention. First, the securitization process rose quickly beginning in 2015. The number and balance of asset backed securities (ABS) products increased by nearly 53 times and 45 times respectively. More importantly, over 76.1% of ABS are properties. This indicates that the trend of securitization of property markets has driven the last few years and brings forth a brand new and prospective stage of financialization of housing (see Figure 11 and Figure 12).

With the guideline “houses are for accommodation, not for speculation” by China’s central government, the housing financialization process started to be strictly restricted and evidently slowed down, since the year 2017. Since the year of 2021, peering into the abyss of debts default across the industry, real estate developers started lowering their leverage, or in the other word, de-financialization process [64]. The average leverage ratio dropped from over 80% at the peak time in 2019 to 78.9% in 2022. Moreover, the proportion of real estate sector in new RMB loans in financial institutions has reduced from 44.8% at the peak time in 2017 to 26.3% in 2021. Developers reject borrowing to expand their reinvestment and narrow their debts due to their balance sheet deflation. Affected by this, land markets turns into recession and expectation of buyers and investors in real estate markets turns into lower level. The revenue of local governments apparently decreased. All in all, the flow of dual financing mechanism shrank and the directions reversed.

6. Discussion

There are also some policy implications toward the tendency. First, the tendency of financialization of housing increased the connections between property markets and debts expansion of the macro economy and thus might have lowered the policy effects of lowering the deleveraging of macro economy. To cope with that, China’s governments should continue lowering the speculative attributes of properties by increasing the total number of public housing construction and rental housing markets, joint-ownership housing and long-term-tenure public rental housing for example. Second, some taxation legislation should be put forward, such as property tax, in the way of increasing the cost of holding the properties and thus lowering the speculative demands. Third, infrastructure, real estate, and manufacture sectors take the three of the largest proportions of investment sectors of China in the past decades. However, in recent years, the growth rate of manufacture went down rapidly compared to the other two sectors as the financialization process of the two goes further. To change the addiction symptoms, some research and policy agenda should be set for de-financialization of housing process [65]. In such agenda, the emphasis on the innovation of industries of cities and supply of public goods and social governance should be important in getting rid of the dependence on the modes of economic growth of local led by debt-driven property boom and the financialization process with it. For example, local governments enlarge the public investment into innovation fields by using governmental innovative investment funds supporting the fundamental science. More institutional arrangements are also made for tracing the path of such R&D proceedings and reclaiming the revenue of innovative industry in the fiscal or finance ways.

For further study, we are going to specify and figure out the interactions of Aggregate Financing to Real Economy (AFRE) funds within provinces or regions and test how such interactions would affect the spatial effects (e.g., the spillover of property prices) of property markets in local China. Moreover, we did not construct an indicator to measure the housing financialization in this paper due to the topic and length of the paper. We are trying to accomplish it in our following studies. The measurement of financialization in more micro level, at the corporate level, and its effects toward macro economic growth is also one of our research topics in the future.

7. Conclusions

In response to the global financial crisis, the Chinese local governments used the public debt expansion process to stimulate and drive economic recovery. In this study, we provide an illustration and explanation of economic recovery in China mode. Compared to the approach of Quantitative Easing (QE), by the USA and some other countries, China underwent a land-based financing way connecting governments, enterprise, and residents, as a whole, named as dual financing mechanism. Different from feature of long-term interest rate of financial markets, high-level inflation level, and export USD liquidity to countries all over the world, one of the features of the China mode is enlarging the volume of infrastructure investment and real estate investment, leading to a property boom with rapid increase in housing price, and large quantity of debt in domestic economy. The 4-trillion-yuan stimulus package released funding for public infrastructure and kicked off the expanded borrowing by local governments using public financing methods (such as LFPs, MBs, etc.,). This process brought about a series of outcomes. On one hand, the accumulated public debt created high demand for innovative financing tools from assorted financial institutions. As a result, banking or non-banking financial corporations and the entire financial industry witnessed a rapid growth of debt accumulation. On the other hand, it created large amounts of public debt held by local governments, which introduces increased risk into the economy. In this case, the governmental financing approach was essentially prohibited in 2013 and property markets became a prime investment for the funds resulting in the property boom. With AFRE supporting both the buyer and seller sides of the real estate market, financial institutions prioritize the return of investment (ROI) and maximizing stakeholders’ interests, lending the money to residential households as well as real estate developers.

According to the descriptive and empirical studies in Section 3 and Section 4, we find out four stylized facts linking the debt-driven property boom. The first is the rapid increase in AFRE and the second is that both property and infrastructure investment increased during the same period. But the connection between financialization level (=AFRE/GDP) and property and infrastructure investment is not clear. By estimating the econometric models, we find out both the property boom and infrastructure expansion are positively affected by financialization level at a very significant level, which are the stylized facts 3 and 4.

Moreover, there are strong connections within the two financing circulations around the lands monopolized by local governments. Whether or not local governments will be able to repay the debts accrued from governmental financing is reliant on two things. One is, whether or not, the debts related to the property boom will be repaid by real estate developers in the form of varied loans, such as development loans and commercial trusts loans. The other thing is, whether or not, debts will be repaid by residents and investors in the form of mortgage loans. Such debt chains also rely on the appreciation of properties and lands, which are driven by these two major paths, public infrastructure investment led by governments with large amounts of public financing and the property boom supported by a varied and tremendous number of market financing tools. After 2013, most volume of Aggregate Financing to Real Economy (AFRE) were from the market financing circulation part and mainly poured into the real estate markets in China.