WP-2020-006

Center for Advanced Economic Study Fukuoka University

(CAES) 8-19-1 Nanakuma, Jonan-ku, Fukuoka,

JAPAN 814-0180 +81-92-871-6631

Junmin Wan and Qiqi Qiu

Fukuoka University, Japan

April 30, 2020

The Impact of Housing Bubbles on Industrial Investments in China

CAES Working Paper Series

The Impact of Housing Bubbles on Industrial Investments in China1

Junmin Wan and Qiqi Qiu2

Graduate School of Economics, Fukuoka University, Japan

April 30, 2020

1 This research was partially supported by the China National Natural Science Foundation - Peking University Data Center for

Management Science (research grant 2016KEY05) and a JSPS KAKENHI Grant (#16K03764). The first author gratefully

acknowledges the support of these funds.

2 The authors thank Hongwei Dai, Masayo Kani, Zhaoxin Niu, Ko Nishihara, Qian Sun, Mitsuo Takase, Takanori Tanaka, Konari

Uchida, Tong Wang, Wako Watanabe, Yuanwei Xu, Junren Yin, Zhongwen Zhang and the participants for their beneficial

comments when the paper was presented at Fukuoka University, Kansai University, Tsinghua University, Conference on “Applied

Finance, Macroeconomic Performance, and Economic Growth” at Zhejiang University of Economics and Finance, the 11th

International Conference on Financial Risk and Corporate Finance Management at Dalian University of Technology, and the 15th

Asia-Pacific Economic Association Annual Meeting at Fukuoka University, as well as the 6th International Conference on the

Chinese Economy: Past, Present and Future at Tsinghua University. The authors give special thanks to Tsutomu Miyagawa for the

constructive comments and to Kazuo Ogawa as well as four anonymous referees for their insightful suggestions. Any remaining

errors here are the authors’ responsibility. Correspondence: Nanakuma 8-19-1, Jonan Ward, Fukuoka City, Fukuoka 8140180,

Japan; (e-mail) Wan: wan@fukuoka-u.ac.jp; (tel) +81-92-871-6631(ext.4208); (fax) +81-92-864-2904.

1

Highlights

1) Housing bubbles in major cities have created a great deal of vacant new housing

(oversupply). We use the matrix of the direct input coefficients of an input-output

table to identify 13 of 36 Chinese industrial sectors as housing-related.

2) We estimate Marginal q values by sector for 2001–2016; sectoral investment is

explained by panel estimations based on Marginal q theory. The Marginal q

investment elasticities of 13 housing-related industries, the remaining 23 industries,

and all industries were 0.2412, 0.8539, and 0.6956 respectively.

3) The housing price Granger explains the producer price index (PPI), and the PPI

Granger the Marginal q values of the 36 industrial sectors. Overcapacity

(overinvestment) was evident in all 13 housing-related industrial sectors, including

metal and cement, during the housing bubble.

2

Abstract

Housing bubbles in major cities have created a great deal of vacant new

housing (oversupply). We use the matrix of the direct input coefficients of an

input-output table to identify 13 of 36 Chinese industrial sectors as housing-related. We

estimate Marginal q values by sector for 2001–2016; sectoral investment is explained

by panel estimations based on Marginal q theory. The Marginal q investment elasticities

of 13 housing-related industries, the remaining 23 industries, and all industries were

0.2412, 0.8539, and 0.6956 respectively. The housing price Granger explains the

producer price index (PPI), and the PPI Granger the Marginal q values of the 36

industrial sectors. Overcapacity (overinvestment) was evident in all 13 housing-related

industrial sectors, including metal and cement, during the housing bubble.

JEL: E22; E32

Keywords: China, housing bubble, industrial sector, input-output table, Marginal q

3

1 Introduction

Overcapacity of housing-related industries in China

Overcapacity (overinvestment) in housing-related industries in China has

created many non-performing loans (NPLs) at the individual bank level from the view

of financial sector (Wan 2018b), but we lack direct evidence of the housing-related

industrial sectors that are customers of commercial banks. Financial instability caused

by NPLs has become a top priority (Wan 2018b, Xi, April 26, 2017).3 The cause of

overcapacity must be clarified. Here, we develop a conceptual mechanism and collect

direct evidence of overcapacity within housing-related industries that produce far too

many vacant new buildings nationwide (Wan 2018b, p. 26: “building houses requires

steel and steel requires coal”) during the housing bubble as shown in Figure 1.4

The sectoral profit growths and the sectoral composite prices of outputs

[proxied by the producer price index (PPI)] of excess-capacity industries (such as steel

and coal; Figures 2–3) have fallen faster than those of other industries. Commencing in

2010, the government responded by raising the state capital ratio to address the

overcapacity issues (e.g., coal, Ren et al. 2019; steel and coal, Chen et al. 2018) and

3 For the details see: http://www.xinhuanet.com//politics/2017-04/26/c_1120879349.htm 4 Too many resources were held by housing-related industries; other industries found it difficult to obtain finance. Small- and medium-sized enterprises could not get loans, as described by former Chairman Zhou Xiaochuan of the People’s Bank of China (Wan 2015b, p. 5: “too difficult and expensive to obtain funds, Rongzinan, Rongzigui, in Chinese”).

4

prevent bankruptcies.5 For example, the state capital ratio of Coal Mining and Washing

was 0.86 in 2001, decreased steadily to 0.31 in 2010, but then increased again to attain

0.48 in 2016.6

China has removed 0.17 trillion tonnes of excess steel production capacity

(notably, China produced 0.83 trillion tonnes, 49.1% of all steel worldwide, in 2017),7

0.8 trillion tons of excess coal capacity (China produced 3.52 trillion tonnes, 45.5% of

all coal worldwide, in 2017, of which about 30% was used to make steel for housing

construction),8 and laid off 1.1 million employees between 2013–2017 (Li, March 6,

2018). These cuts significantly raised the prices of both steel and coal (Chen et al.

2018).9 The net profit of the steel industry rose 2.2-fold in 2017, with the top 15 steel

companies worldwide being based in China (Nikkei, May 25, 2018).

Overcapacity and international spillover

The overcapacity of housing-related industries in China significantly impacts

other countries because “overinvestments in housing-related businesses such as

5 President Xi Jinping has stated that China will reduce excess capacity (Xi, September 3, 2016) and that China should work to contain the housing bubble shown in Figure 1 (Xi, December 22, 2016). Additionally, efforts are being made to remove zombie firms contributing to excess capacity in identified industrial sectors (Xi, February 28, 2017; Dai et al. 2019; Shen and Chen 2017). Based on these comments by China’s leader, it is apparent that overcapacity in certain industrial sectors, and housing bubbles, have become key domestic concerns (Nikkei, November 21, 2018). 6 The state capital ratios are based on data of the National Bureau of Statistics of China. See http://data.stats.gov.cn/ 7 For the details, see: https://www.globalnote.jp/post-1402.html. Sourisseau (2018) reported that the steel production growth rate in China was 14.2% during 2000-2013 compared to 6% during 1980-2000, based on data from the World Steel Association. 8 For the details, see: http://www.coalchina.org.cn/detail//18/06/14/00000035/content.html We used Chinese input-output tables for 2012 and 2017 to estimate steel consumption. 9 Coal is also required by the cement industry. China produces over 50% of all cement worldwide, according to the World Cement Association. https://www.worldcementassociation.org/ Also see Selim and Salem (2010) and Hu et al. (2016).

5

construction materials’’ (Wan 2018b, p. 29) may trigger price collapses of materials

such as steel not only in China but also worldwide. In China, the export price of steel is

lower than that of foreign markets, significantly impacting international trade. The

China steel price Granger affects international steel prices (Guo et al. 2019). The

collapse of international steel prices is easily misinterpreted as “dumping”, for example

by the U.S. government.10

To date, no link has been established between collapses in the domestic and

international prices of housing-related materials, and housing bubbles. Here, we

conceptually and empirically examine this issue; we explore the overcapacities of

housing-related industries driven by housing bubbles in China.

Overcapacity caused by overinvestment

Overcapacity reflects an excess of capital stock (a stock variable) caused by

overinvestment (a flow variable). Here, we analyze overcapacity from the perspective of

overinvestment driven by excessive housing construction. A major housing bubble

increases both the demand- and supply-side capacities of steel, coal, and other industries.

On the demand side, rising prices encourage speculative households to buy houses. On

10 The Organization for Economic Co-operation and Development (OECD) has required the Chinese government to monitor the negative impacts of China’s oversupply of iron, steel, coal, and other materials in the international market (Nikkei, May 28, 2016). The argument is that the low price of Chinese steel considerably reduces the price of steel in the U.S., a practice termed “dumping” by the U.S. government (U.S. Department of Commerce, Fact Sheet, November 13, 2017). Thus, the overcapacities of certain Chinese industrial sectors are hotly debated internationally (Liu and Woo 2018; Schnabl 2019).

6

the supply side, a housing bubble encourages developers to build more structures,

increasing the inputs of steel and other raw materials. Consequently, basic material

prices (as indicated by the PPI) rise and raw material industries are strongly incentivized

to expand. Overcapacities in the markets for housing-related materials [steel (Chen et al.

2018; Ren et al. 2018) and coal (Ren et al. 2019)] then develop.

Overinvestment in China

Not only a housing bubble but also other factors may induce overinvestment.

Lin et al. (2015) point out that state ownership of the Big Four banks means that most

loans are granted to inefficient state-owned enterprises (SOEs), implying that fund

misallocation could cause overinvestment in some firms (e.g., SOEs) and simultaneous

underinvestment in others. Chen et al. (2016) and Ding et al. (2019) show that listed

SOEs may enjoy “free cash flow’’ and inefficiently overinvest using only panel firm

data. Kou et al. (2017) quantified industrial policy and used a cost function method to

estimate utilization of industrial capacity; industrial policy created excess capacity

based on an empirical study of 33 industrial sectors of China from 1999 to 2014.

Overinvestment and bubbles in Japan and the U.S.

7

Both Japan and the U.S. have experienced stock and land (or housing) bubbles.

Chirinko and Shcaller (2001) show that the bubble in the Japanese stock market had a

significant positive effect on investment, and that the bubble boosted business fixed

investment by 6–9% from 1987 to 1989 by raising the Tobin Average q, as revealed by a

time series of aggregate data. Two other studies by listed firm-level data, Goyal and

Yamada (2004) on Japan and Chirinko and Shcaller (2011) on the U.S., reported similar

findings; the Tobin Average q rises were driven by stock bubbles. No research has yet

explored overcapacity issues of housing-related industries driven by housing bubbles in

China and elsewhere. This is our topic here.

Contribution of this research

Wan (2018b) reported that housing bubbles induced overinvestments in

manufacturing sectors, as revealed by NPLs on bank balance sheets, but we lack the

views of industrial sectors. Our first contribution is that we newly use an input-output

table to identify 13 housing-related industrial sectors among the 36 Chinese sectors, and

we find direct evidence on industrial overinvestments induced by the housing bubble.

Next, we estimate the Marginal q values of all 36 Chinese industrial sectors;

this is a significant contribution. Our approach is unlike the Tobin Average q works of

8

Chirinko and Shcaller (2001, 2011) and Goyal and Yamada (2004). That q value is

applicable only to listed companies, of which China has very few. We analyze both

listed and non-listed companies. Our study differs from that of Ogawa et al. (1994), who

used both sectoral Marginal q and sectoral Average q values to show that a fall in land

prices significantly reduced fixed investment via collateral channels after the land

bubble burst in the 1990s. To the best of our knowledge, no explicit study of

overinvestment by the industrial sector has used Marginal q values to analyze a housing

bubble. We find that the investments of all 36 sectors can be explained by Marginal q

theory, and that the elasticities of housing-related sectors are lower than those of others.

The investment behaviors of industrial sectors are thus well-explained by neoclassical

theory. All 36 sectors should be viewed as market-oriented, at least since the time China

joined the World Trade Organization (WTO) in 2001; neoclassical theory states that

corporate investment behavior is rational in that it is explained by the market.

Additionally, we analyze how a housing bubble affects housing-related

industries; we develop a conceptual framework and run some formal tests. The monthly

ratio of house price:rent, and the bubble test, are used to identify housing bubbles in

major Chinese cities. We find that the housing price bubble Granger explains the PPI,

and the PPI Granger explains the Marginal q.

9

Organization of the paper

The remainder of the paper is organized as follows. The research question and

the hypotheses are presented in Section 2. Section 3 identifies housing-related

industries; we use an input-output table to this end. Section 4 describes the data sources,

the empirical specifications, and the estimations. Section 5 summarizes the conclusions.

2 Research question and hypotheses

2.1 Ghost buildings, vacant housing, and the housing bubble

The ratio of investment to gross domestic product in China has increased

continuously since 1952, especially after 2000 (Horioka and Wan, 2007, 2008, Wan

2015a, 2019). Industrial policy has sought to promote industrial investment; however,

ultimately, this triggered overinvestment and/or resource misallocation (Kou et al. 2017).

In this context, “ghost cities’’ have appeared all around China. Zhang et al. (2016)

studied urban vacancies; Yin et al. (2019) analyzed vacant housing in Ordos (Inner

Mongolia), where a major housing bubble burst. In Chengdu (Sichuan province), vacant

houses caused by the ongoing housing bubble were termed “ghost buildings” by

Williams et al. (2019), who identified vacant buildings using the big data of social

10

media.

Wan (2018b, Figures 3 and 10) showed that housing investment correlated

positively with housing prices. Wan (2018b, p.39) identified housing bubbles in major

cities; the vacancies at the end of 2015 could accommodate 30 million persons if one

person was assumed to require 24 square meters. The housing bubble caused excess

demand that in turn triggered oversupply. We clarify the relationship between housing

bubbles and corporate overinvestment using an input-output table and panel data on all

36 industrial sectors and time series data on housing prices and the PPI.

2.2 Excess construction drives input oversupply

Given the observed excessive housing construction, the factor markets

(construction inputs) of steel and cement (about 30% of all steel and about 60% of all

cement are used for housing construction) should be oversupply driven by the factor

demand.11 Following Wan (2020),12 we assume that the housing bubble is an

exogenous shock experienced by factor markets (Figure 4). This perturbs the equilibria

of both quantities and prices (Figure 5); both corporate profits and the Tobin Marginal q

values should be (directly and indirectly) affected. To meet overdemand for steel or

11 We estimated the structure of cement consumption using the Chinese input-output tables for 2012 and 2017. 12 Transmission of factor input quantities and prices in the context of housing bubble or a preference shock is analyzed by adding the input-out table to a neoclassical theory of the general equilibrium framework of Wan (2020). We explicitly model one household sector, two final goods sectors (housing and health), and one intermediate goods sector (steel).

11

cement, increased production capacity, funded by bubble profit, is indispensable;

overinvestment occurs.

2.3 Investments of factor industries and the bubble Marginal q

To clarify the connection between a housing bubble and investment, we

initially identified housing bubbles following Wan (2015a, 2018b). We predicted

bubbles in major Chinese cities from 2004–2017, and then found Grange causality

between the housing price bubble, the PPI, and the Marginal q. To determine the

relationship between the PPI and the Marginal q, we used the PPI as a proxy of output

price when determining Marginal q values. Following Wan (2018a, b), a speculative

housing bubble would have at least some effect (potentially, the principal effect) on

factor industry profits, similar to what was observed in the steel and coal industries.

Thus, the Marginal q values would presumably be affected (even dominated) by the

bubble output prices of (especially) housing-related industries.

2.4 Key hypotheses

The standard theory of investments created by Jorgenson (1963), Tobin (1969),

and Hayashi (1982) indicates that corporate investments should be positively and

12

significantly correlated with the Marginal q values.

Hypothesis 1: Investment is guided by the Marginal q.

This generally means that corporate investment is rational, thus dictated by the

market; but this, in turn, implies other important meanings in China. Whether the

Chinese economy (the so-called “socialized market economy”) is market-oriented

remains hotly debated (see Yu 2019). In the time since China joined the WTO in 2001,

the domestic and global competitiveness of the industrial sector has increased annually.

If the corporate behavior of investors can be explained by the neoclassical theory of the

market economy, all 36 current Chinese industrial sectors should exhibit such behavior.

However, given the unique characteristics of housing factor industries, we explored

their investment behaviors in more detail.

Hypothesis 2: The Marginal q investment elasticities of housing-related sectors would

be lower than those of other sectors.

Housing-related industries such as steel always exhibit unique characteristics

13

(Wu 1998). These are very large-scale endeavors; fixed costs are very high, investment

shot size is very large and the industries must work in standard operating environments.

Such projects are associated with high adjustment costs, some imposed by

environmental assessments or regulations. The higher the adjustment cost, the lower the

Marginal q elasticity of investments (Abel 1980).

3 Identification of housing-related industries

3.1 Direct housing construction inputs of the input-output table

To identify industries heavily involved in housing construction, following

Leontief (1941) and Morishima (1958), we used a matrix of the direct input coefficients

of 23 industrial sectors from 2002-2017 available from the National Bureau of Statistics

of China (NBSC). NBSC reported seven sets of coefficients for the general construction

sector and two sets for the housing construction sector.13 We matched the 23 sectors of

the input-output table to the 36 industrial sectors listed in Marginal q in this paper (not

including construction or housing construction) and estimated the direct input

coefficients to construction and housing construction (Table 1a and b).The coefficients

vary little over time; the housing construction coefficient is very similar to that of

construction.

13 “Construction” includes roads, bridges, and housing.

14

3.2 Direct vs. indirect impacts on factor input sectors

Direct impacts on 13 factor input sectors

Based on the average input coefficients of housing construction by sector

(Table 1a and b), we assume that if the value is greater than 1% [as for Manufacture of

Non-metallic Mineral Products (including cement) and Smelting and Pressing of

Ferrous Metals and Non-ferrous Metals (including iron and steel); 23 and 15%

respectively], those sectors are housing-related industries. Ten sectors (italicized in

Table 1a-b) were identified as housing-related industries. The classification of the 23

industrial sectors of the input-output table differs slightly from the 36 industrial sectors

with Marginal q values; we found that the 10 sectors of the input-output table

corresponded to 13 of the 36 sectors (Table 1c). We term the impact of housing

construction on these 13 sectors “direct”; they lie upstream of housing construction.

Indirect impacts on the remaining 23 sectors

When the direct inputs of steel or cement are examined using other

input-output tables, coal is the major input. Thus, we term the impact of housing

construction on coal “indirect”; the impact lies upstream of steel and cement, which are

15

driven by housing construction. The remaining 23 sectors of Table 1c could be

considered as indirect inputs to housing construction.

4 Marginal q values and investments of the 36 industrial sectors

4.1 Data

Ratio of house price to rent

We collected data on monthly housing prices and residential rental prices for

36 major cities from December 2004 to December 2017; we used the “China Monthly

Economic Indicators” of the NBSC.

Panel data on the 36 industrial sectors

We collected panel data from the National Data provided by NBSC

(http://data.stats.gov.cn/). The principal economic indicators of Industrial Enterprises

(above a designated size) by industry sector, thus 36 sectors, were downloaded. The

aggregation and statistical methods changed after 2000; therefore, we used only the

2000–2016 data.

4.2 Estimations of investments, depreciation rates, and interest payments

16

We lacked data on investments, and interest and depreciation rates; we

estimated these based on the original values of fixed assets and interest expenditures by

industry.

Estimation of investment

The investment and the book value of the fixed assets of industry m at time t

are given by , respectively; thus:

(1)

Estimation of depreciation rate

Following Qiu and Wan (2019), we obtained depreciation rates by industry.

The depreciation rate of industry m at time t is represented by as follows:

(2)

where the fixed asset total at t − 1 is represented by We used a depreciation

rate of 0.074 averaged across all industrial sectors, as estimated by Qiu and Wan (2019),

which is very close to the value of 0.077 used in the Japanese study of Ogawa et al.

(1994).

17

Estimation of interest payment

Wan (2019) found that, since 2011, outstanding corporate deposits on the

balance sheet of the banking sector have overtaken household deposits, leaving some

firms with negative interest payments even if the firm has debt, as the deposit may

exceed the debt. We had data only on final interest payments, thus interest paid on net

debts. Thus, we developed a coefficient to estimate interest paid on outstanding debt.

Briefly, the ratio of the sum of corporate and household deposits in the banking sector

to corporate deposits held by that sector, based on macro data from 2001 to 2016,

served as a multiplier of interest payment by industrial sector. In order, the annual

adjustment ratios from 2001 to 2016 were: 2.43 (2001), 2.45, 2.43, 2.41, 2.47, 2.43,

2.24, 2.38, 2.20, 2.24, 1.86, 1.90, 1.90, 2.21, 1.95, and 1.94 (2016). The annual interest

rate was:

. (3)

4.3 Estimation of Marginal q

Following Jorgenson (1963), Tobin (1969), Abel (1980), Hayashi (1982), and

Ogawa (2003), we considered the following investment framework:

(4)

18

, j= 1, 2, 3, 4…,

where Tobin’s Marginal q, interest rate, discount factor, price of investment,

depreciation rate, and rate of profit at time t are represented by , , , , and

, respectively. The investment equation is given an increasing function with q by

, (5)

where investment, capital stock, and Marginal q in industry m at time t are represented

by , , and , respectively.

Because this paper is the first to estimate Marginal q of industrial sector using

macro data for China, and the number of observations is also very limited, we prefer a

simple specification based on Ogawa (2003) rather than Abel and Blanchard (1986) to

obtain a simple result, while the approach of Abel and Blanchard (1986) for China is

left for next research. By adding assumptions on expectations (Ogawa 2003, Eq. (3),

(4)), i.e., the discount rate and the profit rate follow random walk independently,

(6)

, (7)

where and is stationary white noise, respectively. Then it is shown by

Ogawa (2003, Eq.(5)) that the Marginal q by industrial sector can be simply written

19

as,14

. (8)

The rate of profit of industry m at time t is defined as:

, (9)

where the total profit before tax and the fixed assets of industry m at time t are

represented by and , respectively. We estimate . To control for

potential multicollinearity and endogeneity issues, we use and to estimate

Marginal q by industrial sector. The estimated Marginal q values of the 36 sectors are

listed in Table 2a–d, and Marginal q values of some industrial sectors are illustrated in

Figure 6.

4.3 Empirical specifications

Bubble test

Following Mackinnon (1996), Phillips et al. (2015) and Wan (2015a, 2018b),

we performed unit root and bubble tests using the monthly series of the house price:rent

ratios of 36 major Chinese cities.

Granger causality among housing price, PPI, and Marginal q

14 The method of constructing the Marginal q here is also close to Gugler et al. (2004).

20

Housing price bubbles are regional issues; bubbles differ considerably among

regions and from one year to the next. Housing prices in a bubble should impact the

industrial PPI. However, we lacked housing price data by industrial sector. We only had

data on the housing price, PPI, and Marginal q, for all 36 sectors nationwide. Hence,

Granger causality testing of housing prices and PPI (Figure 7), and PPI and Marginal q

(Figure 8), were performed using the method of Toda and Yamamoto (1995).

Investment and Marginal q

We considered the following empirical investment function based on Abel

(1980), Chirinko (1993) and Ogawa et al. (1994, 2019):

, （10）

where the investment ratio and Marginal q of industry m at time t are represented by

and respectively. The coefficient of is , and , and

are a constant term, the industry-specific effect, the time effect, and the random error

term, respectively. Hypothesis 1 could be tested by specification of Eq. (10). Following

Mussa (1977), Abel (1980) and Chirinko (1993, Eq.(13)), the adjustment costs are

assumed to be,

(11)

21

where α and is parameter of quadratic function, respectively. Then the empirical

investment equation can be expressed by

, (12)

as presented in Chirinko (1993, Eq. (17)). Higher adjustment cost (e.g. α) will lower the

speed of investment response (coefficient of , ) by the adjustment cost

theory. Eq. (12) is the structural form of testing Hypothesis 2.

4.4 Investment by housing-related industries and other industries

We analyzed transmission from the housing market to its industrial factor

markets. We assumed that the investments for all industrial sectors were dependent on

the Marginal q values. Thus, as the Marginal q is somewhat affected by the PPI, and PPI

is induced by housing prices in a bubble, overcapacity or overinvestment issues will

arise in industrial sectors (i.e., Housing Price Bubble → PPI, PPI → Marginal q,

Marginal q → Investment) as shown in Figures 4 and 5.

4.5 Estimations

Bubble test

Tables 3a and b show the estimations of the unit root and bubble tests for the

22

monthly series of house price:rent ratios in 36 major Chinese cities. Both unit roots and

bubbles are apparent, confirming the housing bubble issue raised by Wan (2015a,

2018b).

Granger causality between housing price and the PPI

The estimations are summarized in Table 4. In terms of the hypothesis that

housing price does not influence the PPI in the sense of Granger causality, the null

hypothesis is rejected. In terms of the hypothesis that PPI does not exhibit Granger

causality in terms of housing price, the test is unable to reject the null hypothesis. Thus,

we conclude that the impact of housing price on the PPI is stronger than that of the PPI

on housing price. These results are consistent with the prediction of section 2.2 and the

transmission shown in Figures 4, 5, and 7.

Granger causality between PPI and Marginal q

The results are shown in Table 5. In terms of the hypothesis that PPI is not the

Granger cause of Marginal q, the test rejects the null hypothesis. In terms of the

hypothesis that Marginal q is not the Granger cause of PPI, the test does not reject the

null hypothesis. These results are consistent with the prediction of section 2.3 as shown

23

in Figure 8.

Investments of 36 industries, 13 housing-related and 23 others

Figure 9 shows the transition of Marginal q and the ratio of investment to real

capital stock. The principal statistics are summarized in Table 6 and the results are

shown in Table 7. In the first and the second columns of Table 7, the Marginal q

significantly impacts investment regardless of sector size (proxied by the ratio of total

assets to the total value of fixed assets). Thus, investment behavior of 36 industries is

explained by the Tobin Marginal q theory. The investment elasticity of Marginal q is

0.6956. These results support the prediction of Hypothesis 1.

In the third and fourth columns of Table 7, the Marginal q coefficients are

significantly positive. The investment elasticity of Marginal q is 0.2412 for the 13

housing-related industries. The asset size coefficients are also significantly positive.

Thirteen factors including steel and cement are identified as direct inputs to housing

construction (Smelting and Processing of Non-ferrous Metals, Manufacture of

Non-metallic Mineral Products, the 13 italicized items in Table 1a-b); all exhibit

overcapacity driven by overinvestment, because their downstream industries (housing

construction) are overproduced. These are “direct effects” driven by housing oversupply.

24

In the fifth and sixth columns of Table 7, the Marginal q coefficients of the remaining

23 sectors are also significantly positive; the elasticity is 0.8539.

In the seventh, eighth and ninth columns of Table 7, all the coefficients of

are significantly positive, and the coefficient for 13 housing-related

industries is significantly smaller than that of the remaining 23 industries. These results

are based on structural model of adjustment cost theory in Eq. (12), and are similar to

the reduced form in Eq. (10). The elasticities of Marginal q on investment based on Eq.

(12) are almost the same as the ones based on Eq. (10). We used the estimated

coefficient based on Eq. (12) to draw the adjustment cost function of Eq. (11) as

illustrated in Figure 10. It is obvious that adjustment cost increases with investment for

the 13 housing-related industries more sharply than that for the remaining 23 ones.

Hence, these results support the prediction of Hypothesis 2.

Elasticity of Marginal q on investment in the U.S., Japan, and China

We calculated the elasticity of Marginal q on investment of manufacturing

companies for both the U.S. and Japan. For the U.S. during 1977-1996 based on Gugler

et al. (2004) and for Japan during 1972-1990 based on Ogawa et al. (2019), the

elasticity is 0.2556 and 0.3544, respectively. The elasticity (0.6956) in China is larger

25

than those of both the U.S. and Japan. At least two factors may contribute to this

difference. The first one would be that our Marginal q is based on profit before tax

(because of lacking data on tax), and the second one would be that our estimation is

based on macro data while Gugler et al. (2004) and Ogawa et al. (2019) are all from

micro data.

Overinvestments, Marginal q, and their implications

The Marginal q values of the coal and other industries were abnormally high in

the housing bubble era, as shown in Table 2a–d, but decreased sharply (to below 1 for

some industrial sectors such as coal) after the government shrank the bubble (Figure 6).

In the bubble era, a high Marginal q reflected a small capital stock; the firm required

more investment. If the Marginal q was lower than 1, this implies that that the industry

engaged in past overinvestment. The PPI created an abnormally high Marginal q during

the housing bubble. Thus, a Marginal q lower than 1 after bubble control policies were

in place would robustly identify overinvestment in industrial sectors. Thus, housing

construction exerted ``indirect impact’’ on coal driven by steel and cement (coal is the

upstream industry of steel and cement; steel and cement are upstream industries of

housing construction; i.e., coal→steel and cement→housing construction).

26

These results directly support the principal finding of Wan (2018b), who

argued that overinvestments in housing-related industries induced by housing bubbles

significantly increased the numbers of NPLs held by Chinese banks. The results are

consistent with those of Chen et al. (2016) and Ding et al. (2019) who used a cash flow

approach to identify bubbles caused by “free cash flow’’. Our results are not only

consistent with those of Kou et al. (2017) in terms of the cost function and industrial

policy approach but also with the policy of “Five Measures to Exclude Excessive

Industrial Capacity’’ proposed by the National Development and Reform Commission

of China (NDRC, Jan. 13, 2016) and the 2015 Annual Report of the NBSC, which

revealed that housing oversupply/overinventory reduced investments in steel, coal, and

other housing-related industries.15

Given that physical and human capital are limited, overinvestment in housing

and related sectors reflects underinvestment in other sectors such as health or education.

Our findings are consistent with the non-linear impact of housing price on private

investment described by Li et al. (2018) using provincial panel data. Housing prices

below a threshold encourage private investment but high prices inhibit investment.

5 Conclusions

15 See Xinhuanet (Jan. 20, 2016) for details.

27

We identified housing bubbles in several major Chinese cities via the bubble

test and by calculating the ratio of house prices to rent; the housing bubble has created a

massive amount of vacant new housing, thus housing oversupply. To identify the

material inputs of housing-related industries, we used a matrix of the direct input

coefficients of the input-output table and found that 13 of the 36 industrial sectors were

housing-related. We then estimated Marginal q values by sector from 2001–2016 and

performed panel estimations. The investments of all 36 industrial sectors are explained

by the Tobin Marginal q theory. The Marginal q investment elasticities of the 13

housing-related industries, the remaining 23 industries, and all 36 industries were

0.2412, 0.8539, and 0.6956 respectively. Finally, we found that the housing price

Granger explains the PPI and the PPI Granger the Marginal q values of all 36 industrial

sectors. Hence, overcapacity caused by overinvestment is evident in at least 13

housing-related industries (including metal and cement), caused by the housing bubble.

The implications of our findings follow. The Tobin Marginal q theory is the

dominant neoclassical framework used to analyze real corporate investment behavior.

The investments of the 36 manufacturing sectors (excluding housing construction) can

be explained by Marginal q theory; all sectors seem to have invested by reference to the

market. When the upstream demand market of housing construction is driven by a price

28

bubble, the overcapacity and overinvestment issues of housing-related industries such as

steel, cement, and coal directly or indirectly reflect the distorted housing market.

Therefore, rectifying the housing market by soft landing or shrinking bubbles [Wan

(2018a, b)] is urgent; this would help solve the overcapacity issues of housing-related

industries. Another implication is that most of China’s economy has gradually changed

to a market basis.

We will build a theoretical decision model of investment within bubble

scenarios and perform more detailed analyses using firm- or loan-level data with a focus

on activities upstream of the housing market.

6 References

Abel, Andrew B (1980) Empirical Investment Functions: An Integrative Framework,

Carnegie Rochester Conference Series on Public Policy, Vol.12, pp. 39-91.

Abel, Andrew B. and Olivier Blanchard (1986) The Present Value of Profits and

Cyclical Movements in Investment, Econometrica, Vol.54, pp. 239-273.

Chen, Linxi, Ding Ding and Rui Mano (2018) China’s Capacity Reduction Reform and

its Impact on Producer Prices, IMF Working Paper WP/18/217.

Chen, Xin, Yong Sun, Xiaodong Xu (2016) Free Cash Flow, Over-investment and

29

Corporate Governance in China, Pacific-Basin Finance Journal, Vol.37, pp. 81-103.

Chirinko, Robert S. (1993) Business Fixed Investment Spending: Modeling Strategies,

Empirical Results, and Policy Implications, Journal of Economic Literature, Vol. 31,

pp. 1875-1911.

Chirinko, Robert S. and Huntley Schaller (2001) Business Fixed Investment and

Bubbles: The Japanese Case, American Economic Review, Vol. 91, pp. 663-679.

Chirinko, Robert S. and Huntley Schaller (2011) Fundamentals, Misvaluation, and

Business Investment, Journal of Money, Credit and Banking, Vol.43(7), pp.

1423-1442.

Dai, Xiaoyong, Xiaole Qiao and Lin Song (2019) Zombie Firms in China’s Coal

Mining Sector: Identification, Transition Determinants and Policy Implications,

Resources Policy, Vol.62, pp. 664-673.

Ding, Sai, John Knight and Xiao Zhang (2019) Does China Overinvest? Evidence from

a Panel of Chinese Firms, The European Journal of Finance, Vol.25, pp. 489-507.

Goyal, Vidhan K. and Takeshi Yamada (2004) Asset Price Shocks, Financial

Constraints, and Investment: Evidence from Japan, The Journal of Business, Vol.

77(1), pp. 175-199.

Gugler, Klaus, Dennis C. Mueller and B. Burcin Yurtoglu (2004) Marginal q, Tobin's q,

30

Cash Flow, and Investment, Southern Economic Journal, Vol. 70(3), pp. 512-531.

Guo, Sui, Huajiao Li, Haizhong An, Qingru Sun, Xiaoqing Hao, Yanxin Liu (2019)

Steel Product Prices Transmission Activities in the Midstream Industrial Chain and

Global Markets, Resources Policy, Vol.50, pp. 56-71.

Hayashi, Fumio (1982) Tobin's Marginal q and Average q: A Neoclassical

Interpretation, Econometrica, Vol. 50(1), pp. 213-224.

Horioka, Charles Yuji and Junmin Wan (2007) The Determinants of Household Saving

in China: A Dynamic Panel Analysis of Provincial Data, Journal of Money, Credit

and Banking, Vol.39(8), pp. 2077-2096.

Horioka, Charles Yuji and Junmin Wan (2008) Why Does China Save so Much? China,

Asia, and the New World Economy, edited by Barry Eichengreen, Yung Chul Park

and Charles Wyplosz, Oxford University Press, pp. 371-391.

Hu, Hui, Xiang Li, Fuxia Yang and Jesmin Islam (2016) Total Factor Productivity and

Energy Intensity: An Empirical study of China’s Cement Industry, Emerging

Markets Finance and Trade, Vol.52, pp. 1405-1413.

Jorgenson, Dale W. (1963) Capital Theory and Investment Behavior, The American

Economic Review, Vol. 53, pp. 247-259.

Kou, Zonglai, Xueyue Liu, and Jin Liu (2017) Did Industrial Policy Induce

31

Overcapacity? Evidence from the Industrial Sector in China. Fudan Journal

(Social Sciences), No.5, pp. 148-160. (in Chinese)

Leontief, Wassily (1941) The Structure of American Economy, 1919–1929, Cambridge:

Harvard University Press.

Li, Jiangtao, Jianyue Ji, Huiwen Guo and Lei Chen (2018) Research on the Influence of

Real Estate Development on Private Investment: A Case Study of China,

Sustainability, Vol.10(8), pp. 1-17.

Li, Keqiang (March 6, 2018) Government Work Report. http://paper.people.com.cn/

rmrb/html/2018-03/06/nw.D110000renmrb_20180306_1-02.htm (in Chinese)

Lin, Justin Yifu, Xifang Sun and Harry X. Wu (2015) Banking Structure and Industrial

Growth: Evidence from China, Journal of Banking &Finance, Vol. 58, pp.

132-143.

Liu, Tao and Wing Thye Woo (2018) Understanding the U.S.-China Trade War, China

Economic Journal, Vol.11, pp. 319-340.

Mackinnon, James G. (1996) Numerical Distribution Functions for Unit Root and

Cointegration Tests, Journal of Applied Econometrics, Vol.11(6), pp. 601-618.

Morishima, Michio (1958) Prices, Interest and Profits in a Dynamic Leontief System,

Econometrica, Vol.26(3), pp. 358-380.

32

Mussa, Michael L. (1977) External and Internal Adjustment Costs and the Theory of

Aggregate and Firm Investment, Economica, Vol.44, pp. 163-178.

National Development and Reform Commission (Jan. 13, 2016) Five Measures to

Exclude Excessive Industrial Capacity.

http://www.gov.cn/zhengce/2016-01/13/content_5032457.htm (in Chinese)

Nikkei Shimbun (May 28, 2016) Iron and Steel Excess Production Requested to China,

Morning Edition, p.7. (in Japanese)

Nikkei Shimbun (May 25, 2018) Net Profit of World Steel Industry Rose by 2.2 Times:

Top 15 Companies in China Overcame Overproduction in 2017, Morning Edition,

p.17. (in Japanese)

Nikkei Shimbun (November 21, 2018) Miscalculation of Debt Reduction Policy in

China, Morning Edition, p.2. (in Japanese)

Ogawa, Kazuo (2003) Financial Distress and Corporate Investment: The Japanese Case

in the 90s, ISER Discussion Paper No.584, Osaka University.

https://www.iser.osaka-u.ac.jp/library/dp/2003/DP0584.pdf Included in Fumio

Hayashi ed., Malfunctioning of Financial Systems, (Empirical Analysis and Design

of Economic Systems, Vol.2, in Japanese), Chap.2, pp. 35-63, Keiso Shobo

Publishing Co. Ltd., 2007.

33

Ogawa, Kazuo, Shin-ichi Kitasaka, Toshio Watanabe, Tatsuya Maruyama, Hiroshi

Yamaoka, and Yasuharu Iwata (1994) Asset Markets and Business Fluctuations in

Japan, The Economic Analysis, Economic Research Institute, Economic Planning

Agency, No.138, pp. 17-97.

Ogawa, Kazuo, Elmer Sterken and Ichiro Tokutsu (2019) Why Is Investment So Weak

Despite High Profitability? A panel study of Japanese manufacturing firms, RIETI

Discussion Paper Series 19-E-009.

https://www.rieti.go.jp/jp/publications/dp/19e009.pdf

Phillips, Peter C. B., Shu-Ping Shi and Jun Yu (2015) Testing for Multiple Bubbles:

Limit Theory of Real Time Detectors, International Economic Review, Vol.56(4),

pp. 1079-1134.

Qiu, Qiqi and Junmin Wan (2019) Depreciation Rate by Perpetual Inventory Method

and Depreciation Expense as Accounting Item, CAES Working Papers Series

WP-2019-012, Fukuoka University.

http://www.econ.fukuoka-u.ac.jp/researchcenter/workingpapers/WP-2019-012.pdf

Ren, Mengjia, Lee G. Branstetter, Brian K. Kovak, Daniel E. Armanios and Jiahai Yuan

(2019) Why Has China Overinvested in Coal Power? NBER Working Paper No.

25437.

34

Schnabl, Gunther (2019) China’s Overinvestment and International Trade Conflicts,

China and World Economy, Vol.27, pp. 37-62.

https://mpra.ub.uni-muenchen.de/24464/

Selim, Tarek and Ahmed Salem (2010) Global Cement Industry: Competitive and

Institutional Frameworks, MPRA Paper No. 24464.

Shen, Guangjun and Binkai Chen (2017) Zombie Firms and Over-capacity in Chinese

Manufacturing, China Economic Review, Vol.44, pp. 327-342.

Sourisseau, Sylvain (2018) The Global Iron and Steel Industry: From a Bilateral

Oligopoly to a Thwarted Monopsony, Australian Economic Review, Vol.51(2), pp.

232-243.

Tobin, James (1969) A General Equilibrium Approach to Monetary Theory, Journal of

Money, Credit and Banking, Vol. 1(1), pp. 15-29.

Toda, Hiro Y. and Taku Yamamoto (1995) Statistical Inference in Vector

Autoregressions with Possibly Integrated Processes, Journal of Econometrics,

Vol.66 pp. 225-250.

U.S. Department of Commerce (November 13, 2017) Fact Sheet.

https://www.trade.gov/enforcement/factsheets/factsheet-prc-hardwood-plywood-pro

ducts-ad-cvd-final-111317.pdf

35

Wan, Junmin (2015a) Household Savings and Housing Prices in China, The World

Economy, Vol.38, pp. 172-192.

Wan, Junmin (2015b) Non-performing Loans in Housing Bubbles, CAES Working

Papers Series WP-2015-006, Fukuoka University.

http://www.econ.fukuoka-u.ac.jp/researchcenter/workingpapers/WP-2015-006.pdf

Wan, Junmin (2018a) Prevention and Landing of Bubble, International Review of

Economics & Finance, Vol. 56, pp. 190-204.

Wan, Junmin (2018b) Non-performing Loans and Housing Prices in China,

International Review of Economics & Finance, Vol.57, pp. 26-42.

Wan, Junmin (2019) Corporate Saving, Household Saving, and Gross National Savings:

Analysis by Provincial Panel and Time Series Data in China, CAES Working Papers

Series WP-2019-017, Fukuoka University.

http://www.econ.fukuoka-u.ac.jp/researchcenter/workingpapers/WP-2019-017.pdf

Wan, Junmin (2020) Transmission of Housing Bubble among Industrial Sectors,

Fukuoka University, mimeograph.

Williams, Sarah, Wenfei Xu, Shin Bin Tan, Michael J. Foster and Changping Chen

(2019) Ghost Cities of China: Identifying Urban Vacancy Through Social Media

Data, Cities, Vol.94, pp. 275-285.

36

Wu, Yanrui (1998) The Economics of the East Asia Steel Industries: Production,

Consumption and Trade, Taylor & Francis Ltd.

Xi, Jinping (September 3, 2016) China is Trying to Exclude Excess Capacity.

http://politics.people.com.cn/n1/2016/0903/c1001-28689050.html (in Chinese)

Xi, Jinping (December 22, 2016) Restraining Housing Bubble.

http://henan.people.com.cn/n2/2016/1222/c351638-29497758.html (in Chinese)

Xi, Jinping (February 28, 2017) Exclude Excess Capacity and Zombie Firms.

http://finance.china.com.cn/news/20170228/4117271.shtml (in Chinese)

Xi, Jinping (April 26, 2017) Stability of Financial System and the Economy.

http://www.xinhuanet.com//politics/2017-04/26/c_1120879349.htm (in Chinese)

Xinhuanet (December 21, 2015) From New Normal to Reform of Supply Side.

http://www.xinhuanet.com/politics/2015-12/21/c_128551111.htm (in Chinese)

Xinhuanet (January 20, 2016) Economic Efficiency Overwhelmed by Overcapacity.

http://www.xinhuanet.com/finance/2016-01/20/c_128646462.htm (in Chinese)

Yin, Duo, Junxi Qian and Hong Zhu (2017) Living in the “Ghost City”: Media

Discourses and the Negotiation of Home in Ordos, Inner Mongolia, China,

Sustainability, Vol.9(11), pp. 1-14.

Yu, Miaojie (2019) The Status of China's Market Economy and Structural Reforms: The

37

Issues Behind the US–China Trade War, Asian Economic Papers, Vol.18(3), pp.

34-51.

Zhang, Chuanchuan, Shen Jia and Rudai Yang (2016) Housing Affordability and

Housing Vacancy in China: The Role of Income Inequality, Journal of Housing

Economics, Vol.33, pp. 4-14.

38

Table 1a: Matrix of direct input coefficents of industrial sectors, 2002-2017

Input OutputConstruction Housing Construction

2002 2005 2007 2010 2012 2015 2017 Avg. 2012 2017 Avg.

Mining and Washing of Coal 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%Extraction of Petroleum and Natural Gas 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%Mining and Processing of Ferrous Metal Ores and Non-Ferrous Metal Ores 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

Mining and Processing of Non-metal Ores 2% 2% 1% 1% 1% 1% 1% 1% 1% 1% 1%Manufacture of Food from Agricultural Products, Foods, Liquor Beverages, Refined Tea and Tobacco

0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

Manufacture of Textile 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%Manufacture of Textile Fabrics Wearing Apparel , Accessories, Leather Fur Feather , Related Products and Footwear

0% 0% 0% 0% 0% 1% 0% 0% 0% 0% 0%

Processing of Timber Manufacture of Wood Bamboo Rattan Palm, Straw Products and Furniture

3% 2% 2% 2% 2% 3% 2% 2% 1% 2% 1%

Manufacture of Paper, Paper Products, Printing Reproduction of Recording Media, Articles for Culture Education Art and Carfts, Sport and Entertainment Activities

0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

Processing of Petroleum and Coking of Nuclear Fuel 3% 2% 2% 1% 1% 1% 1% 2% 1% 1% 1%

Table 1b: Matrix of direct input coefficents of industrial sectors, 2002-2017 (cont.)

Manufacture of Raw Chemical Materials and Chemical Products 4% 3% 4% 3% 4% 1% 3% 4% 2% 3% 3%

Manufacture of Non-metallic Mineral Products 11% 21% 21% 23% 19% 19% 19% 19% 24% 22% 23%

Smelting and Pressing of Ferrous Metals and Non-ferrous Metals 12% 7% 16% 12% 16% 12% 11% 12% 17% 12% 15%

Manufacture of Metal Products 5% 4% 4% 3% 4% 5% 5% 4% 4% 5% 4%

Manufacture of General Purpose Machinery and Special Purpose Machinery 4% 3% 3% 3% 1% 1% 1% 2% 1% 1% 1%

Manufacture of Automobiles, Railway Vessel Aerospaceand Other Transport Equipments 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

Manufacture of Electrical Machinery and Equipment 3% 3% 4% 4% 4% 4% 3% 4% 4% 4% 4%

Manufacture of Communication Equipment Computers and Other Electronic Equipment

0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

Manufacture of Measuring Instruments and Machinery 1% 1% 0% 0% 0% 0% 0% 0% 0% 0% 0%

Utiliztion of Waste Resources 0% 0% 0% 0% 0% 0% 1% 0% 0% 0% 0%Production and Supply of Electric Power and Heat Power 1% 2% 1% 1% 1% 1% 1% 1% 1% 1% 1%

Production and Supply of Gas 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

Production and Supply of Water 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Table 1c: Matching 23 industrial sectors of input-outpt table with 36 industrial sectors

23 industrial sectors of input-output table 36 industrial sectorsMining and Washing of Coal Mining and Washing of CoalExtraction of Petroleum and Natural Gas Extraction of Petroleum and Natural GasMining and Processing of Ferrous Metal Ores and Non-Ferrous Metal Ores

Mining and Processing of Ferrous Metal Ores Mining and Processing of Non-Ferrous Metal Ores

Mining and Processing of Non-metal Ores Mining and Processing of Non-metal Ores

Manufacture of Food from Agricultural Products, Foods, Liquor Beverages, Refined Tea and Tobacco

Processing of Food from Agricultural ProductsManufacture of Foods Manufacture of Liquor Beverages and Refined TeaManufacture of Tobacco

Manufacture of Textile Manufacture of Textile

Manufacture of Textile Fabrics Wearing Apparel , Accessories, Leather Fur Feather , Related Products and Footwear

Manufacture of Textile Fabrics Wearing Apparel and AccessoriesManufacture of Leather Fur Feather and Related Products and Footwear

Processing of Timber Manufacture of Wood Bamboo Rattan Palm, Straw Products and Furniture

Processing of Timber Manufacture of Wood Bamboo Rattan Palm and Straw ProductsManufacture of Furniture

Manufacture of Paper, Paper Products, Printing Reproduction of Recording Media, Articles for Culture Education Art and Carfts, Sport and Entertainment Activities

Manufacture of Paper and Paper Products Printing Reproduction of Recording MediaManufacture of Articles for Culture Education Art and Carfts, Sport and Entertainment Activities

Processing of Petroleum and Coking of Nuclear Fuel Processing of Petroleum and Coking of Nuclear Fuel

Manufacture of Raw Chemical Materials and Chemical Products

Manufacture of Raw Chemical Materials and Chemical Products Manufacture of Medicines Manufacture of Chemical Fibers Manufacture of Rubber and Plastics Products

Manufacture of Non-metallic Mineral Products Manufacture of Non-metallic Mineral ProductsSmelting and Pressing of Ferrous Metals and Non-ferrous Metals

Smelting and Pressing of Ferrous Metals Smelting and Pressing of Non-ferrous Metals

Manufacture of Metal Products Manufacture of Metal ProductsManufacture of General Purpose Machinery and Special Purpose Machinery

Manufacture of General Purpose Machinery Manufacture of Special Purpose Machinery

Manufacture of Automobiles, Railway Vessel Aerospaceand Other Transport Equipments

Manufacture of Automobiles, Railway Vessel Aerospaceand Other Transport Equipments

Manufacture of Electrical Machinery and Equipment Manufacture of Electrical Machinery and Equipment

Manufacture of Communication Equipment Computers and Other Electronic Equipment

Manufacture of Communication Equipment Computers and Other Electronic Equipment

Manufacture of Measuring Instruments and Machinery Manufacture of Measuring Instruments and Machinery

Utiliztion of Waste Resources Utiliztion of Waste ResourcesProduction and Supply of Electric Power and Heat Power

Production and Supply of Electric Power and Heat Power

Production and Supply of Gas Production and Supply of GasProduction and Supply of Water Production and Supply of Water

Source: Information from National Bureau of Statistics of China. http://data.stats.gov.cn/

Table 2a: Marginal q of 36 industrial sectors, 2001-2016

Year National Total

Mining and Washing of

Coal

Extraction of Petroleum

and Natural Gas

Mining and Processing of

Ferrous Metal Ores

Mining and Processing of Non-Ferrous Metal Ores

Mining and Processing

of Non-metal Ores

Processing of Food from Agricultural

Products

Manufacture of Foods

Manufacture of Liquor

Beverages and Refined Tea

Manufacture of Tobacco

2001 0.7268 0.1746 2.8868 0.6566 0.9264 0.3516 0.6680 0.8203 0.7942 2.29132002 0.9450 0.3645 2.8787 0.8354 1.1374 0.4514 0.9049 1.0992 1.0379 3.05892003 1.2068 0.5071 3.4715 1.7095 1.7427 0.6060 1.2204 1.2654 1.1755 3.82932004 1.7282 1.2875 5.2922 7.0221 4.0397 0.9165 1.7351 1.6289 1.3123 5.69032005 1.9399 1.8702 7.9303 5.8690 7.7674 2.1105 2.5305 2.2556 2.1131 6.76152006 1.7763 1.5364 7.0963 4.3460 9.0975 2.4391 2.4991 2.0898 2.0800 5.98442007 1.9332 1.6518 5.2536 6.1421 7.8010 2.4407 3.0824 2.4610 2.5590 7.26292008 1.3402 2.2658 4.1733 6.8459 4.2923 2.5420 2.4830 1.9003 2.0095 4.52182009 2.9779 4.2563 3.6040 6.6183 5.9316 4.9867 5.5608 5.3547 4.8488 11.58432010 1.5628 2.1477 1.6467 4.3196 3.4620 2.4975 2.8160 2.5991 2.4960 4.57552011 1.2956 1.9132 2.0154 3.0485 3.1599 2.3392 2.3178 2.2621 2.3329 3.38432012 2.1370 2.3808 2.7810 4.5547 4.9309 4.1102 4.1224 4.0820 4.3995 6.61502013 1.5577 1.0797 2.0450 3.0362 2.7197 2.7949 2.8273 3.0549 3.0848 6.66352014 1.6926 0.6177 2.0900 2.3585 2.5548 2.7870 2.7115 3.3923 3.1247 7.75722015 2.1187 0.2344 0.5244 1.9045 2.4372 3.6059 3.5453 4.6259 4.3694 10.60722016 1.5144 0.4581 -0.2929 1.0177 1.6488 2.2254 2.4774 3.1521 2.9368 5.5997Avg. 1.6533 1.4216 3.3373 3.7678 3.9781 2.3253 2.5939 2.6277 2.5422 6.0117

Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Table 2b: Marginal q of 36 industrial sectors, 2001-2016 (cont.)

Year Manufacture of Textile

Manufacture of Textile Fabrics

Wearing Apparel and Accessories

Manufacture of Leather Fur

Feather and Related Products and

Footwear

Processing of Timber

Manufacture of Wood Bamboo Rattan Palm and Straw Products

Manufacture of Furniture

Manufacture of Paper and

Paper Products

Printing Reproduction of Recording

Media

Manufacture of Articles for

Culture Education Art

and Carfts, Sport and

Entertainment Activities

2001 0.4770 1.7116 1.3115 0.5449 1.2391 0.4597 1.1291 1.42322002 0.7303 1.9794 1.9587 0.6751 1.4613 0.7411 1.1705 1.88262003 0.8687 2.0332 2.3763 0.8730 1.7801 0.7854 1.4223 1.77992004 0.9197 2.3391 2.8512 1.3984 2.5776 1.0405 1.6575 2.00052005 1.3925 3.0132 3.6518 1.9009 2.6182 1.1721 1.6471 2.25012006 1.3153 2.7733 3.2025 1.9586 2.4984 1.0604 1.4721 1.75142007 1.4755 2.8675 3.7170 2.5735 2.2189 1.2811 1.7053 1.86132008 1.1426 2.3585 3.0223 2.3064 1.6331 0.9706 1.4272 1.25812009 2.8197 5.5614 7.3891 4.7114 4.4540 2.3003 3.3974 3.62832010 1.6821 2.9439 4.0789 2.5135 2.4859 1.1917 1.6795 1.95412011 1.4260 2.4090 3.3829 2.1057 2.1154 0.9066 1.4281 1.53172012 2.3190 5.0656 6.2437 4.0040 3.7301 1.4952 3.1518 9.62392013 1.8746 3.0008 3.8700 3.1028 2.5446 1.0532 2.4119 3.63752014 2.1459 3.3779 4.2435 3.2730 2.8848 1.1121 2.7451 4.02572015 2.9540 4.4821 5.8379 4.0523 3.9681 1.6960 3.6444 5.33822016 2.1014 2.9855 3.9223 2.7811 2.8329 1.2627 2.3705 3.6359Avg. 1.6028 3.0564 3.8162 2.4234 2.5652 1.1580 2.0287 2.9739Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Table 2c: Marginal q of 36 industrial sectors, 2001-2016 (cont.)

Year

Processing of

Petroleum and Coking of Nuclear

Fuel

Manufacture of Raw

Chemical Materials

and Chemical Products

Manufacture of Medicines

Manufacture of Chemical Fibers

Manufacture of Rubber and

Plastics Products

Manufacture of Non-metallic Mineral

Products

Smelting and

Pressing of Ferrous Metals

Smelting and

Pressing of Non-

ferrous Metals

Manufacture of Metal Products

Manufacture of General

Purpose Machinery

2001 -0.0485 0.3377 1.4439 0.1855 0.8225 0.3736 0.3863 0.4484 0.9768 0.69662002 0.2179 0.6460 1.7134 0.3519 1.1228 0.5237 0.6125 0.5640 1.2970 1.21672003 0.5205 0.9889 1.7855 0.6703 1.2162 0.8896 1.1695 0.9308 1.6216 1.71562004 1.3748 1.9245 1.7963 0.5820 1.5987 1.2955 2.0334 1.7976 2.4301 2.70752005 -0.5424 2.1461 2.0307 0.5213 1.7195 1.1823 1.7621 2.1701 2.9902 3.22592006 -0.9616 1.6214 1.4858 0.5541 1.5378 1.2387 1.4841 2.9910 2.5591 3.04092007 0.5093 1.9730 1.8963 1.1124 1.8960 1.7096 1.6918 3.1322 2.7011 3.32172008 -1.5052 1.2755 1.6283 0.3667 1.4029 1.5130 0.8004 1.2711 2.2337 2.56742009 2.8790 2.8580 4.1140 1.8562 3.7650 3.5991 1.4195 2.5346 4.4395 5.07102010 1.1590 1.5574 1.9590 1.5831 1.9831 1.8386 0.7414 1.5067 2.4036 2.60122011 0.2852 1.3358 1.6558 1.2235 1.5551 1.5652 0.5695 1.3364 1.8785 1.92492012 0.3271 1.9743 2.8965 1.3331 2.7041 2.3316 0.7540 1.8599 3.6035 2.96232013 0.4624 1.3825 2.1266 0.8857 2.2206 1.7745 0.5764 1.2175 2.2224 2.46582014 0.0687 1.4231 2.2766 1.0823 2.2889 1.9333 0.5766 1.1929 2.4203 2.68792015 0.8380 1.9118 3.1940 1.5497 3.1557 2.3330 0.2584 1.3312 2.7635 3.49432016 1.4674 1.3554 2.2448 1.3053 2.2019 1.7043 0.5293 1.2030 2.3165 2.3381Avg. 0.4407 1.5445 2.1405 0.9477 1.9494 1.6128 0.9603 1.5930 2.4286 2.6274

Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Table 2d: Marginal q of 36 industrial sectors, 2001-2016 (cont.)

Year

Manufacture of Special Purpose

Machinery

Manufacture of Automobiles， Railway Vessel Aerospaceand

Other Transport Equipments

Manufacture of Electrical

Machinery and Equipment

Manufacture of Communication Equipment Computers and Other

Electronic Equipment

Manufacture of Measuring Instruments

and Machinery

Utiliztion of Waste

Resources

Production and Supply of Electric

Power and Heat

Power

Production and Supply

of Gas

Production and Supply of Water

2001 0.6511 0.9671 1.3737 2.2023 1.5494 0.3914 -0.0184 0.0790 2002 1.2158 1.7025 1.7906 2.1732 1.8133 0.4278 -0.0286 0.0432 2003 1.5686 2.4353 2.1381 2.3296 2.6390 0.4442 0.1536 0.0114 2004 1.9326 2.4396 2.9217 3.0442 2.8594 14.0942 0.5385 0.3075 0.0472 2005 2.4322 1.8925 3.3505 2.5520 3.8961 5.5056 0.6196 0.3184 -0.0094 2006 2.5149 1.9678 2.9856 2.2390 3.4523 5.1757 0.6328 0.4320 0.1226 2007 3.1932 2.5691 3.5592 2.3075 3.7512 3.5779 0.5737 0.8112 0.1231 2008 2.4793 1.9415 3.1626 1.4889 2.7648 4.2889 0.0915 0.8888 0.0734 2009 4.9537 5.1436 6.7285 3.4329 6.3418 6.5255 0.4677 2.6358 0.1437 2010 2.7041 2.7717 3.1783 2.1534 3.2600 3.4808 0.2502 1.2348 0.1211 2011 2.1523 2.1893 2.2566 1.2776 2.4075 3.6615 0.1859 1.0893 0.1104 2012 3.3239 3.2996 3.6290 2.8435 3.8251 4.3289 0.4487 1.7644 0.2028 2013 2.3479 2.7327 2.5985 2.1604 3.3636 3.4129 0.4342 1.2157 0.1993 2014 2.3331 3.5123 3.0764 2.6201 3.9144 4.0532 0.5215 1.5782 0.3154 2015 2.8624 4.4805 4.4945 3.7967 5.1927 5.1116 0.7804 2.0361 0.5385 2016 1.9758 3.1564 3.3701 2.8407 3.7291 2.8101 0.4157 1.1725 0.3664 Avg. 2.4151 2.7001 3.1634 2.4664 3.4225 5.0790 0.4515 0.9744 0.1555

Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Table 3: Bubble test for monthly ratio of housing price to rent in 36 major cities

A: Unit root testPrice-renting ratio (Dec. 2004 – Dec. 2017, 154 observations after adjustments)

Null Hypothesis: The series has a unit root

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -1.328 0.616

*MacKinnon (1996) one-sided p-values.

B: The SADF test and the GSADF test of the price-renting ratio

Null hypothesis: The series has a unit root

Price-renting ratio (Dec. 2004 – Dec. 2017, 157 observations after adjustments)

SADF GSADF

Test statistics 6.648 6.660

p-value 0.000 0.000Right-tailed test.

Note: Critical values of tests are obtained by Monte Carlo simulation with 1,000 replications.The smallest window has 24 observations. The author’s calculations.

Table 4: Granger causality tests for housing pice and PPI for industrial sectors, 2001-2016

H0: Housing price does not Granger cause PPI H0: PPI does not Granger cause Housing price

statistical value p-value statistical value p-value16.62 0.0000 *** 1.73 0.1882

Note:***, ** and * denote rejection of the null hypothesis at the 1%, 5% and 10% significance level, repectively.

Table 5: Granger causality tests for PPI and Marginal q for industrial sectors, 2001-2016

H0: PPI does not Granger cause Marginal q H0: Margimal q does not Granger cause PPIstatistical value p-value statistical value p-value

5.52 0.0189 *** 2.02 0.1544

Note:***, ** and * denote rejection of the null hypothesis at the 1%, 5% and 10% significance level, repectively.

Table 6: Summary statistics of 36 industrial sectors，2001-2016

Variable Obs Median Mean Std. Dev. Min Max

（36 industrial sectors）

Investment(t) / Capital Stock(t-1) 573 0.1962 0.2215 0.2303 -0.2047 3.5904

Marginal q (t) 573 2.1105 2.3553 1.7535 -1.5052 14.0942

Total Assets(t-1) / Total Value of Fixed Assets(t-1) 573 2.7857 2.8811 0.8695 1.2734 6.8588

Year 573 2009 2008.534 4.6013 2001 2016

（13 industrial sectors）

Investment(t) / Capital Stock(t-1) 208 0.2025 0.2094 0.1220 -0.1495 0.5679

Marginal q (t) 208 1.8897 1.8885 1.2211 -1.5052 6.7285

Total Assets(t-1) / Total Value of Fixed Assets(t-1) 208 2.5041 2.7267 0.7752 1.4206 4.9187

（23 industrial sectors）

Investment(t) / Capital Stock(t-1) 365 0.1929 0.2283 0.2734 -0.2047 3.5904

Marginal q (t) 365 2.2448 2.6213 1.9463 -0.2929 14.0942

Total Assets(t-1) / Total Value of Fixed Assets(t-1) 365 2.8421 2.9566 0.9093 1.2734 6.8588

Source: See the text.

Table 7: Determinants of investments in 36 industrial sectors, 2001-2016

(36) (13) (23)

Marginal q (t) 0.0724 ** 0.0654 ** 0.0283 *** 0.0267 *** 0.0812 ** 0.0744 **

(0.0312) (0.0265) (0.0077) (0.0069) (0.0370) (0.0319)

0.0697 ** 0.0302 *** 0.0782 **

(0.0288) (0.0076) (0.0345)

0.0982 0.1390 *** 0.0826

(0.0725) (0.0445) (0.0743)

Year -0.0060 -0.0101 -0.0008 -0.0062 ** -0.0078 -0.0116

(0.0041) (0.0061) (0.0011) (0.0024) (0.0056) (0.0077)

Constant 12.1943 20.1502 1.6966 12.2343 ** 15.7673 22.9968 0.1256 *** 0.1822 *** 0.0996 **

(8.1930) (12.0737) (2.1803) (4.7484) (11.1386) (15.1936) (0.0395) (0.0068) (0.0567)

Observations 573 573 208 208 365 365 573 208 365R-squared 0.1867 0.2006 0.0413 0.0718 0.2244 0.2338 0.1909 0.0511 0.2259

36 13 23

Note: Robust standard errors in parentheses (FE), *** p<0.01, ** p<0.05, * p<0.1.

36 13 13 23 23

[Marginal q (t) - 1]*Price Index for Investment inFixed Assets

Total Assets(t-1) / Total Valueof Fixed Assets(t-1)

Number of industiral sector 36

(36) (13) (23)

reduced form of adjustment cost model (Eq. (10))(industrial sectors)

(Panel estimation with fixed effect and robust standard errors (FE) )

Dependent Variable = Investment(t)/Capital Stock (t-1)

structural formof adjustment cost model (Eq. (12))

(industrial sectors )

Independent Variables

Figure 1: Monthly housing prices and ratio of price to rent during Dec. 2004-Dec. 2017

Source: The data are obtained from the China Monthly Economic Indicators by National Bureau of Statistics from Jan. 2005 to Jan.2018.

Figure 2: PPI by industrial sector during 2004-2016 (previous year=100)

Source: The data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Figure 3: Ratio of total profit to fixed asset by industrial sector during 2000-2016 (%)

Source: The data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Housing market

Housing related industries

vs. the left ones

Bubble?

Price bubble induces over supply of housing

Producer price commoving with

change of housing related factor

market

H Impact ?

Factor demand of housing construction by

input-output table

Profit, Tobin’s q of housing related factor market and the left industries

q and investment equation?

Difference may exist between housing related industries and

the left ones

H Impact ?

Producer price may affect profit of factor market

Figure 4: Transmission from housing bubble to investments of housing-related industries

Source: Drawn by the authors.

Figure 5: Demand and supply of housing-related factor market (input i of housing construction )

Source: Drawn by the authors.

Quantity of input i of housing construction

S: Supply of input i Price of input i

of housing construction S’

D: Demand of input i D’

E

E’

p

p’

Quantity 1 Quantity 2

Figure 6: Marginal q by industrial sector during 2001-2016

Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Figure 7: Growth rate of ratio of housing price to rent vs. PPI during 2001-2016

Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Figure 8: PPI vs. Marginal q during 2001-2016

Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

Figure 9: Marginal q vs. ratio of investment to real capital stock during 2001-2016

Source: Authors' estimations based on data from the National Data by National Bureau of Statistics of China. http://data.stats.gov.cn/

0

2

4

6

8

10

12

14

16

18

20

-0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3

G(I, K, τ) of 36 industries

G(I, K, τ) of the 13 housing‐related industries

G(I, K, τ) of the remaining 23 industries

I/K

Figure 10: Implicit adjustment cost function of the 13 housing-related industries and the remaining 23 industries in China, 2001-2016

Source: See the text.