Innovation investment of MEs is substantial and costly, and their innovation activities are easily affected by external influences. As for the influencing factor of real estate price, HP can affect the expectations of market entities, thereby affecting IME. Market expectations integrate multidimensional considerations from multiple parties, such as enterprises and individuals, and are an important factor affecting investment management and consumption (Beracha et al, 2022). The fluctuation of HP is closely linked to market expectations. Therefore, fluctuations in HP profoundly impact society and the economy (Li and Guo, 2020). A rise in HP can stimulate an increase in investment in the real estate industry and change related enterprises’ investment management strategies, which can have a significant driving effect on the development of related industries (Teng et al, 2021). HP can directly or indirectly drive the development of related fields such as decoration, construction, and even retail, resulting in a prosperous economy in the short term (Chen et al, 2020). People generally have optimistic expectations about the future, and this sentiment is reflected in the behavior of market participants. The positive economic outlook and increased demand for high-tech products driven by rising HP can lead to higher returns on innovative products, prompting MEs to allocate more resources and energy into innovation to increase investment in independent research and development (Yu and Zhang, 2017). In addition, a moderate increase in HP may lead to an increase in production costs for ME, but it may also encourage them to change management strategies and focus on efficiency improvements and technological innovation to offset the cost increase. Furthermore, because real estate can serve as collateral, and MEs generally have substantial housing assets, the rise of HP will increase the value of the housing assets owned by ME, which helps alleviate the FC faced by ME, thereby enhancing their willingness to innovate and promoting innovation research and development expenditures. Based on these arguments, hypothesis H1 is formed.

H1: In the short term, the rise of HP is beneficial for IME.

In recent years, the profit margins of ME in China have continued to decline. In contrast, the profit margins of the real estate industry have steadily increased, with the continuous rise in HP directly driving the increase in investment returns in the real estate industry. In the long run, the continuous rise of HP has two main effects on IME. On the one hand, when the return on investment in industries such as manufacturing falls below that of real estate, the speculative arbitrage motive of enterprise management prompts them to shift capital toward real estate to obtain higher profits. The profit-driven nature of capital has led to a large influx of capital into the real estate sector, further exacerbating the misallocation of resources. The rise of HP drives up land and rental prices, which increases the production costs of enterprises. The rise of HP also diverts funds needed for production, operations, and innovation toward the real estate market, weakening innovation capacity, and hindering the sustainable development of the manufacturing industry (Miles, 2009). On the other hand, during an economic downturn, the excess returns on general bank loans often fail to compensate for their associated risks. As a result, banks tend to shift their asset allocation strategies and direct loans toward sectors with higher returns and lower risk. In the context of rising HP, the real estate industry is considered a typical high-profit, low-risk, and high-quality lending sector. The high investment return rate of the real estate industry leads credit resources to tilt toward real estate enterprises (Chaney et al, 2012). Attracted by these high profit levels, banks adjust their fund management strategies, increase credit allocation to real estate enterprises, and reduce credit availability to other industries, resulting in a decline in research and development (R&D) lending for enterprises. Furthermore, social capital as a whole becomes increasingly inclined toward the real estate sector in terms of investment willingness, reducing investment in the manufacturing industry and creating a crowding-out effect on innovation in the manufacturing economy (IME). Based on these arguments, Hypothesis H2 is proposed.

H2: In the long term, there is an inverted U-shaped relationship between HP and IME.

The rise in HP typically attracts capital into the real estate sector, thereby reducing the availability of capital for the real economy, especially the manufacturing industry. In the context of segmented capital markets, the siphoning effect of high HP may lead enterprises to divert funds into speculative ventures, limiting investment in research and development (R&D) and weakening innovation capacity. This resource misallocation and imbalance in fund management not only reduce the capital utilization efficiency of the manufacturing sector but may also further hinder the growth of total factor productivity (Chen et al, 2011). Moreover, rising HP is often accompanied by increased land costs—one of the key production factors for MEs. High land prices may undermine the competitiveness of MEs in resource allocation, prompting some firms to shift management strategies away from innovation activities requiring substantial R&D investment toward production activities with higher short-term returns, thereby obstructing innovation-driven development (Zhu, 2018). Finally, the impact of high HP on the labor market is also significant. Increased HP often drives up living costs, especially in areas where housing expenses comprise a large share of household expenditure. This may compel firms to adjust management strategies by raising wages to attract high-skilled labor (Tao and Xue, 2019). For small and medium-sized MEs with limited resources, such cost pressures may reduce innovation budgets, leading to inadequate R&D investment. In addition, high HP may push skilled workers to migrate from manufacturing to real estate-related industries or regions, further exacerbating labor resource mismatches. The impact of HP on innovation in manufacturing enterprises (IME) through the RA mechanism is reflected in the reallocation of key production factors such as capital, land, and labor. This mechanism not only affects the internal optimization of enterprise resources but also influences industrial upgrading and economic growth through broader macroeconomic structural adjustments (Wang and Wen, 2019). Based on these arguments, hypothesis H3 is formed.

H3: HP has an impact on IME through RA.

The rise of HP will significantly increase the market value of the real estate held by enterprises, thereby enhancing their asset collateral capacity. This process creates more opportunities for enterprises to obtain external funds by alleviating FC and improving their credit rating in the capital market. For innovative activities, due to their high risk and long-term characteristics, financing needs are more urgent. Therefore, the increase in mortgage capacity caused by the rise of HP may have a positive incentive effect on innovation investment (Sasidharan et al, 2015). Furthermore, the rise of HP improves the financing conditions of enterprises by enhancing their balance sheet. On the one hand, the increase in real estate value enhances the financial stability of enterprises, making it easier for them to obtain low-cost funds from banks or other financial institutions. On the other hand, asset appreciation may reduce a company’s cost of capital, making it more feasible to conduct high risk and high investment innovation activities. In addition, from the perspective of corporate behavior decisions, the rise of HP may indirectly affect innovation activities by changing corporate investment preferences (Liu et al, 2023). On the one hand, high HP may encourage companies to invest more resources in investment areas related to real estate, thereby diverting funds that should be used for research and technological innovation, forming a “crowding out effect” on innovation activities. On the other hand, the wealth effect released by high HP may prompt companies to increase their R&D investment to cope with a more competitive market environment. Finally, from a macroeconomic perspective, fluctuations in HP may indirectly affect corporate innovation activities through the risk transmission mechanism of financial markets. On the one hand, the rise of HP may trigger excessive credit expansion in the financial market, lead to a decrease in the efficiency of capital allocation, and suppress innovation activities (Wang and Zhang, 2016). On the other hand, if HP fluctuate too much, companies may face increasing uncertainty, which may further worsen their investment decisions, especially in supporting long-term innovation projects. Based on these arguments, hypothesis H4 is formed.

H4: HP have an impact on IME through FC.

High HP raise expectations for real estate appreciation—particularly for assets such as land—thereby encouraging MEs to divert funds originally allocated for research and technological innovation into the real estate sector. This capital reallocation or shift in management strategy may lead to internal resource mismatches, weakening both the ability and willingness of firms to invest in long-term innovation and ultimately suppressing their innovation efficiency. Specifically, firms may become more inclined to pursue asset appreciation through short-term, high-return real estate investments rather than engage in high-risk, long-term R&D initiatives (Cheng and Xia, 2020). Furthermore, rising HP may alter the behavioral incentives of ME management and ownership. In an environment of rapid HP growth, high returns from real estate investments may prompt managers to prioritize short-term profit maximization over long-term innovation strategies (Wu et al, 2021). Such short-termism can further exacerbate firms’ neglect of innovation. In addition, under persistently high HP, enterprises may face rising pressures on employee wages, rents, and overall operating costs, further constraining the financial space available for innovation-related activities. While high HP can potentially enhance a firm’s financing capacity by increasing the value of its property holdings, this benefit is likely to be more pronounced for small and micro enterprises with severe financial constraints and may have limited relevance for larger firms with abundant capital (Karahan, 2015). More importantly, even when financing capacity improves, firms may still channel these funds into real estate investments due to preferences shaped by high HP, rather than allocating them to innovation—thereby weakening the intended pro-innovation effect of increased financing. Based on these considerations, Hypothesis H5 is proposed.

H5: HP have an impact on IME through REI.

In summary, the path of HP acting on IME is shown in Fig. 1.

Due to uneven regional development in China, there are significant differences in economic levels and industrial structures among different regions. Therefore, the impact of HP on IME in different regions is also not the same. When HP continue to rise, the impact of high HP on IME is reflected in two aspects. On the one hand, local governments monopolize land resources and use them as collateral to obtain loans from banks (Rong et al, 2016). However, government financing will significantly inhibit IME, and this mismatch of land resources will lead to premature “deindustrialization” in China, significantly reducing urban innovation. This impact is more pronounced in developed regions, such as the eastern region of China (Feng et al, 2019). On the other hand, due to the close connection between the real estate market and the land market, the rise of HP will directly push up the land use costs of ME.

Specifically, HP in the eastern region of China is generally high, which may lead to an increase in operating costs (such as labor costs and rent) for enterprises, thereby affecting innovation investment. In addition, the eastern region has more research institutions, universities, and talent and has great potential for innovation in the manufacturing industry. However, high HP may inhibit the innovation capabilities of small and medium-sized enterprises. The policies in the eastern region tend to support high-end manufacturing and innovative industries, which may partially offset the negative impact of rising HP. On the contrary, HP in the central and western regions are relatively low, and there is less pressure on business operating costs. However, low HP may reflect insufficient market demand and limit the motivation for companies to expand production. There are relatively few scientific research institutions and talents in the central and western regions, and the innovation foundation of the manufacturing industry is weak. Moreover, the impact of HP on innovation may be more direct. The central and western regions attract investment and industrial transfer through national policies, but the improvement of innovation capabilities still requires time.

Compared to the central and western regions, HP in the eastern region of China is generally higher, and land use costs are also higher. This situation will encourage ME to consider the production cost differences caused by the price differences of land factors when innovating and laying out production. In addition, when HP is at a high level, it will also increase the cost of living for residents, who will demand an increase in wages, thereby affecting the labor costs of enterprises (Liu and Shi, 2019). Based on these arguments, hypothesis H6 is formed.

H6: There are regional differences in the inhibitory effect of HP on IME, with the eastern cities being more significantly affected compared to the central and western cities in China.

This study selected ME listed on the Shanghai and Shenzhen A-shares in 275 cities in China from 2013 to 2022 as the research sample. The data about the input and output of IME were from the China Research Data Service Platform (CNRDS). Other data related to listed companies were sourced from the China Stock Market & Accounting Research (CSMAR) database, Wind database, and official websites of listed companies. The HP data came from housing transaction websites such as Anjuke, Fangtianxia, and Housing Price Market Network. Other data related to cities were sourced from the National Bureau of Statistics of China, the China Regional Economic Statistical Yearbook and the China Urban Statistical Yearbook for relevant years.

In terms of research observations, on the one hand, the securities code of listed companies was used as a matching identifier, and missing data such as the establishment date and ownership form of individual companies were searched and filled in through the “Tonghuashun” software (4.6.0, Zhejiang Hexin Flush Network Services Ltd, Hangzhou, Zhejiang, China). On the other hand, companies that were treated by Special Treatment (ST), Special Treatment (delisting warning) (*ST), or Particular Transfer (PT) due to financial system abnormalities during the sample period and companies in the real estate industry were deleted. In addition, considering the differences in financial data requirements for companies in the B-share market, we also excluded companies listed on the B-share market during the sample period.

Based on the above data, this study matched the information of the city where the listed company is located with the city’s HP data, and screened the sample data as follows. Firstly, we listed companies registered in county-level cities were excluded. Secondly, because of missing data, the samples of Qinghai, Xizang, Inner Mongolia and Xinjiang were excluded. Furthermore, Hong Kong, Macau, and Taiwan were not included. Finally, a matching data set was formed, which includes 441 listed ME and 275 cities in China. Through the above methods, we selected 1760 valid samples. To reduce the interference of extreme values on the results, winsorization was applied to the 1% and 99% quantiles of continuous variables.

According to the benchmark regression results in Table 5, HP, as the core explanatory variable, has a significant negative impact on IME. After controlling for individual and time fixed effects, it can be seen from column (1) that the estimated coefficient of HP is –0.454, which is statistically significant at the 1% level. This indicates that the rise of HP is not conducive to the improvement of manufacturing innovation capabilities. Subsequently, after controlling for variables such as EDL, GSI, FDL, and OPL, we again conducted an estimated regression analysis on the model. From column (2), it can be seen that the estimated coefficient of HP is –0.428, which is significant at the 1% level. According to the estimated regression results in column (2), after considering control variables, high HP still has a significant negative impact on IME. This study preliminarily suggests that as the real estate industry develops excessively, high HP tends to suppress innovation in the manufacturing industry. According to the bubble theory, the rapid rise of HP leads to an increase in speculation in the real estate industry. High returns lead ME to focus on short-term interests, reduce innovative research and development, diminish patent output, and further weaken innovation capability.

In addition, considering the possible non-linear relationship between HP and IME, this study added a quadratic term of HP for estimation regression. According to column (3) and column (4) in Table 5, it can be seen that the coefficient of HP and its quadratic terms are statistically significant. Regardless of whether control variables are added or not, the coefficient of HP remains significantly positive, indicating that HP has a positive impact on IME. This validates hypothesis 1 that HP is beneficial for IME in the short term. The coefficient of HP² is significantly negative, indicating that HP may have an inverted U-shaped relationship with IME. It should be noted that to verify the existence of the inverted U-shaped relationship, researchers usually add the square term of the independent variable to the regression model and determine the inverted U-shaped relationship by whether the quadratic coefficient is significant and whether the quadratic coefficient is different from the first-order coefficient. This method is generally feasible. However, this method has certain limitations because inverted U-shaped relationships can appear everywhere, even in places where inverted U-shaped relationships do not exist. Therefore, the inverted U-shaped relationship can lead to the conclusion that the quadratic coefficient of the independent variable is significant, but the significant quadratic coefficient of the independent variable cannot lead to the conclusion that there is an inverted U-shaped relationship (Simonsohn and Nelson, 2014).

To avoid the above errors, we need to adopt the approach of Aghion et al (2005), and the specific steps are as follows. Firstly, we should find the maximum point of the inverted U-shaped curve. The regression coefficients of HP and HP² are 1.745 and –3.753, respectively. We assume that the coefficient of HP² is a and the coefficient of HP is b. According to the formula for finding the maximum value of a quadratic function, when HP = b/–2a, IME has the maximum value. Note that the HP at this time is HP_max. We will continue to generate three new variables: HP_low, HP_high, and high. The specific definitions of variables are shown in Eqns. 8,9, and 10:

(8)$${\mathrm{HP}}_{\text{low}}=\{\begin{array}{c}\hfill \mathrm{HP}-{\mathrm{HP}}_{\max },\text{if} \mathrm{HP}\le {\mathrm{HP}}_{\max }\hfill \\ \hfill 0,\text{otherwise}\hfill \end{array}$$

(9)$${\mathrm{HP}}_{\text{high}}=\{\begin{array}{c}\hfill \mathrm{HP}-{\mathrm{HP}}_{\max },\text{if} \mathrm{HP}\ge {\mathrm{HP}}_{\max }\hfill \\ \hfill 0,\text{otherwise}\hfill \end{array}$$

(10)

Finally, we need to perform breakpoint regression, as shown in Eqn. 11.

(11)$$MEI=c*H{P}_{low}+d*H{P}_{high}+e*high$$

If the signs of coefficients c and d are opposite and both are significant, it can be determined that there is indeed an inverted U-shaped relationship between IME and HP (Aghion et al, 2005). Through mathematical formulas, we can calculate that the turning point of HP (HP_max) is 0.232. By using Stata15.0, we conducted breakpoint regression and obtained coefficients of 2.173 for HP_low, –3.296 for HP_high, and 0.104* for high. From the results, it can be seen that the coefficient signs of HP_low and HP_high are opposite, and both are statistically significant at the 1% level. Therefore, we can conclude that there is an inverted U-shaped relationship between HP and IME, which verifies hypothesis 2. On the one hand, in the early stages of the rise of HP, HP can increase the financing scale of enterprises, increase their willingness to engage in technological innovation, and promote patent inventions by enterprises. On the other hand, when HP continues to rise, the positive effect of HP on IME weakens and gradually produces significant negative effects. This is mainly due to the high-yield nature of real estate, which leads to an increase in speculative demand and triggers more manufacturing enterprises to turn to real estate investment, reducing investment in IME.

In terms of control variables, the coefficients for EDL, GSI and ISD are significantly positive, indicating that they are conducive to IME to a certain extent. The good development of the social economy, government’s support and reasonable industrial structure can provide a favorable environment and relevant policy support for ME to carry out innovation activities. The coefficient of FDL is positive, passes the 10% significance test, and indicates that financial development can provide financial support for innovative activities of ME. Although the coefficient of OPL is positive, it is not significant enough, indicating that the promoting effect of OPL on IME is not significant enough in the research sample of this study.

We have analyzed the mediating effect of HP on IME through stepwise regression. Table 6 shows the empirical results. Column (2) and (3) present the regression results with the mediating effect of RA added, column (4) and (5) show the results with the mediating effect of FC added, and column (6) and (7) display the regression results with REI added.

From the perspective of RA, column (1) shows that HP has a positive impact on RA at a significance level of 1%, with an impact coefficient of 0.317. Column (2) shows that RA effectively inhibits IME at a significance level of 5%. HP can effectively hinder IME through RA. Therefore, it can be concluded that RA plays a mediating role and has a significant impact. With the misallocation of resources, HP can inhibit IME, and hypothesis 3 has been verified.

From the perspective of FC, model (3) shows that the impact coefficient of HP on FC is –0.804, which passes the significance test at the 1% level. This indicates that the higher the HP, the less FC MEs face. Column (4) shows that FC is still significant, and the coefficient of HP is significant after introducing FC. The coefficient of HP is positive, while the coefficient of FC is negative, indicating that HP can improve IME through decreasing FC. Therefore, hypothesis 4 has been verified.

From the perspective of REI, column (5) shows that HP has a positive impact on REI at a significance level of 1%, with an impact coefficient of 0.479. Column (6) shows that REI hinders IME at a significance level of 1%. Therefore, it can be concluded that REI plays a mediating role, whereby with the increase of REI, HP can inhibit IME, and thus hypothesis 5 has been verified.

However, traditional mediation models may have some shortcomings, such as limited applicability and potential endogeneity concerns. When dealing with mediation effects, the testing ability of the Sobel test is generally higher than that of stepwise regression method (Wen and Ye, 2014). Therefore, we followed the approach of Combs et al (2007) in conducting the Sobel test on the mediating effect. The results are shown in Table 7.

According to the results in Table 7, the values of the Soble test of the three mediating variables all passed the 1% significance level test, as did the values of Goodman-1 (Aroian) and Goodman-2 of the three mediating variables. In addition, the proportion represents the proportion of the indirect effect to the total effect, indicating the strength of the mediating effect. The proportion of these three mediating variables all exceeded 50% and passed the 1% significance level test. This once again indicates that these three variables do indeed play a mediating role and validates hypotheses 3 to 5.

To further understand the heterogeneity of the impact of HP on IME, we conducted a heterogeneity test and classified 275 cities in 27 provinces of China into eastern cities and central and western cities for regression analysis. Based on China’s administrative regional planning and academic conventions, we divided the 27 provinces of China into two regions: the eastern region and the central and western regions. The eastern region includes 11 provinces, namely, Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan. The central and western regions include 16 provinces: Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, and Ningxia. By matching with ME, there are 1132 valid samples in eastern cities and 632 valid samples in central and western cities, with the results shown in Table 8.

According to the regression results, whether in eastern cities or central and western cities, HP has a significant impact on IME, and the relationship between the two still shows an inverted U-shape. However, this impact also has regional differences. The results show that high HP in eastern cities have a greater inhibitory effect on IME than in central and western cities, thus verifying hypothesis 6. The possible reason for this is that high HP in the eastern cities exert a significant inhibitory effect on IME through mechanisms such as increasing enterprise costs, suppressing talent inflows, squeezing financing resources, and guiding short-term corporate behavior. In contrast, in the central and western cities, due to lower HP and less cost pressure on enterprises, as well as limited innovation resources, the inhibitory effect of HP on IME is relatively weak.

We conducted robustness tests on the empirical results of this study through four methods: lagging the independent variable by one period, replacing the dependent variable, changing the sample observation time, and using the instrumental variable (IV) method.

This study finds that: (1) There is an inverted U-shaped relationship between HP and IME. When the level of HP is low, the rise of HP plays a promoting role in innovation for manufacturing enterprises. When the overall level of HP is too high and exhibits a bubble-like trend, the rise of HP tends to inhibit IME. Therefore, the level of HP in China can only play a positive role in IME when it matches the level of the national economy. (2) HP affects IME through RA, FC, and REI. Among them, HP hinders IME through RA and REI, while HP promotes IME through alleviation of FC. (3) The impact of HP on IME varies regionally. When HP is at a high level, rapid rise of HP has a significant inhibitory effect on IME in eastern and central western cities of China. Among them, the inhibitory impact of HP on IME in eastern cities is more significant than in central and western cities.

Based on the above research conclusions, this study proposes the following policy recommendations.

There is an inverted U-shaped relationship between HP and IME, which means that HP only plays a positive role when the level of HP in China matches the level of the national economy. Therefore, the government should actively stabilize real HP, establish a sound real estate supervision and management system, regulate the financial investment attributes of real estate, and crack down on illegal speculation. Additionally, the government should establish a dynamic monitoring and management mechanism for the relationship between HP and IME and flexibly adjust policies according to actual situations. For example, the government regularly evaluates the relationship between HP and IME, identifies reasonable ranges for HP, and adjusts real estate supervision and management policies and manufacturing support policies in a timely manner based on monitoring results.

HP hinders IME through RA and REI but promotes IME through reduction of FC. Therefore, on the one hand, the government can continuously optimize the RA of the housing supply system, improve the housing system that combines renting and purchasing, increase public rental housing to meet the basic housing needs of residents, and alleviate the drawbacks of high HP on IME. The government should also continuously optimize the allocation of financial resources, optimize the financing scale allocation between various sub sectors of the manufacturing industry, improve the financial service capabilities for small and medium-sized ME, and encourage more financial resources to flow into the innovation field of the manufacturing industry. On the other hand, the government should allow for moderate rise of HP in areas where HP is too low. Such an approach can enhance the market value of real estate for enterprises, strengthen their asset collateral capabilities, and bring about positive chain reactions such as improving the credit rating of enterprises in the capital market, easing FC, promoting the acquisition of external funds, and enhancing the financial stability of enterprises.

Finally, compared to cities in the central and western regions, HP has a more significant inhibitory impact on IME in eastern cities. The government should implement differentiated HP adjustment policies according to local conditions. For eastern cities, especially large cities with many ME, the government should strictly control HP to prevent excessive HP from having adverse effects on IME. For example, the government can implement subsidy policies to mitigate the adverse effects of high HP on IME. In addition, we recommend establishing manufacturing bases in the central and western cities with lower HP, providing policy support and infrastructure support, and encouraging the establishment of R&D centers or branch offices in these areas.