Registration address change and corporate cost stickiness: from the perspective of policy preference and policy burden

ABSTRACT

A change of corporate registration address will directly affect the development of listed firms, and also the districts in and out of which the firms move, thereby ultimately leading to the reallocation of capital resources. Therefore, corporate registration address change attracts great attention from the theoretical and practical circles. From the perspective of policy preference and policy burden, using the sample of Chinese A-share listed firms that have experienced trans-provincial change in their registration address from 2000 to 2016, this study investigates the impact of registration address change on corporate cost stickiness and the mechanism through which registration address change affects corporate cost stickiness. The results indicate that: (1) with many factors controlled, registration address change greatly enhances corporate cost stickiness; (2) in the discussion of the mechanism, it is found that the increased policy preference and the aggravated policy burden are two ways via which registration address change affects cost stickiness. Therefore, for listed firms that experience both policy preference and policy burden simultaneously because of their registration address change, their cost stickiness increases the most. The study not only enriches the current literature on the economic consequences of registration address change, links corporate registration with cost management, and enlarges the scope of corporate cost management; it also has practical significance for corporate registration management and political resource allocation in China as well.

KEYWORDS:

• registration address change
• cost stickiness
• policy preference
• policy burden

1. Introduction

Corporate registration address is the address registered on the business licences of firms. According to the Company Law of the People’s Republic of China, the registration addresses of firms should be in accordance with their domiciles, and firms need to change their registration addresses on their licences as soon as there is any change in their office locations. Generally, firms need only to update their domicile change within the prefecture of the Administration for Industry and Commerce where they are located, which incurs low costs for firms. According to whether the registration address change goes beyond the jurisdiction of the Registration Authority or not, it can be divided into within-jurisdiction change and cross-jurisdiction change. As cross-jurisdiction change will directly lead to the reallocation of capital resources in different jurisdictions, it becomes the main concern of the theoretical and practical circles. According to the statistics (see Table 1 Panel A), a total of 111 listed companies experienced trans-provincial change in registration address in the Chinese A-share market from 2000 to 2016, among which 14 firms changed registration address several times. According to the statistics based on different economic regions in Panel B, the net inflow of firms in eastern and central China is positive, while it is declining continuously in western and northeastern China. According to the ‘4+3’ regional layout formulated in the 13th Five-Year Plan, eastern and central China have significantly greater economic advantages than western and northeastern China. As a result, the trend of registration address change shows that Chinese listed companies are inclined to move to more developed regions (Pan, Xia, & Liu, 2013). As is shown by the statistics based on provinces in Panel C, the preferential immigration places for listed firms are mainly concentrated in the developed eastern and central China, such as Zhejiang, Beijing, Shanxi and Shanghai, etc.

Table 1. The statistics of trans-provincial change of registration addresses in Chinese A-share listed firms from 2000 to 2016.

Panel A: Statistics based on years.
Years	No. of firms with registration address change	Years	No. of firms with registration address change	Years	No. of firms with registration address change
2000	0	2006	6	2012	5
2001	6	2007	10	2013	6
2002	4	2008	5	2014	12
2003	9	2009	9	2015	3
2004	5	2010	9	2016	9
2005	3	2011	10	Total	111
Panel B: Statistics based on economic regions.
Regions	No. of firms before the change	No. of firms after the change	Net inflow of firms	Regions	No. of firms before the change	No. of firms after the change	Net inflow of firms
East	62	79	17	West	22	11	–11
Central	11	16	5	Northeast	16	5	–11
Panel C: Statistics based on provinces.
Regions	No. of firms before the change	No. of firms after the change	Net inflow of firms	Regions	No. of firms before the change	No. of firms after the change	Net inflow of firms
Zhejiang	1	15	14	Guizhou	1	1	0
Beijing	10	21	11	Yunnan	1	0	–1
Shanxi	0	6	6	Shandong	5	4	–1
Shanghai	10	14	4	Chongqing	3	2	–1
Anhui	0	2	2	Inner Mongolia	3	1	–2
Guangdong	10	12	2	Jiangxi	2	0	–2
Tianjin	1	2	1	Hunan	5	3	–2
Xinjiang	0	1	1	Jilin	5	2	–3
Henan	1	2	1	Hainan	4	1	–3
Shaanxi	1	2	1	Heilongjiang	3	0	–3
Qinghai	0	1	1	Hebei	6	1	–5

Note: the eastern regions include: Beijing, Tianjin, Shanghai, Hebei, Shandong, Jiangsu, Zhejiang, Fujian, Taiwan, Guangdong and Hainan; the central regions include: Shanxi, Henan, Hubei, Anhui, Hunan, Jiangxi; the western regions include: Inner Mongolia, Xinjiang, Ningxia, Shaanxi, Gansu, Qinghai, Chongqing, Sichuan, Tibet, Guangxi, Guizhou and Yunnan; the northeast regions include: Heilongjiang, Jilin and Liaoning.

Referring to the definition of the nature of firms by Coase (1937), corporate operation is the mechanism of resource allocation. The resource endowment of a firm will be greatly different after a registration address change. Hence, the registration address change which leads to the change of the resources allocation schemes will bring about different economic consequence in the end.Cost stickiness reflects the asymmetrical change of corporate cost when the business volume of firms changes. When a firm changes its registration address, it adjusts its resource allocation plan, which incurs adjustment costs, thereby affecting the cost stickiness of the firm (Banker, Byzalov, & Chen, 2013). The effective reduction of cost stickiness can maximise corporate value creation ability and optimise resource allocation (Jiang & Hu, 2011). Therefore, it is particularly important to examine the economic consequences of the registration address change behaviour of firms from the perspective of cost stickiness. Further analysis of the ways through which registration address change affects cost stickiness shows that a registration address change inevitably leads to capital resource reallocation. The promotion tournamentof local governments forces government officials to find all possible ways to accomplish political achievements, compete for and protect the superior regional resources, and intervene in capital resource allocation (Sun & Luo, 2011). In order to compete for such scarce resources as listed firms, local governments are bound to provide various preferential policies and, at the same time, impose policy burden on firms as well. It is highly likely that a registration address change will affect corporate cost stickiness through adjustment costs and the expectation change of the management brought by policy preference and policy burden. In order to discuss the above issues, from the perspective of policy preference and policy burden, this study examines the impact of registration address change on corporate cost stickiness using the sample of Chinese A-share listed firms that have changed their registered addresses across province from 2000 to 2016. It is found that registration address change will significantly increase corporate cost stickiness with other factors controlled. And registration address change can increase corporate cost stickiness through the increase of policy preference and the aggravation of policy burden brought by the change. Therefore, for the listed firms with policy preference increased, and policy burden aggravated simultaneously after a registration address change, their cost stickiness increases the most.

This study makes the following contributions. First, a corporate registration address change will directly affect the development of listed firms, and the districts which the firms move from and to, thereby ultimately leading to the reallocation of capital resources. However, how efficient is the resource allocation? Only Pan, et al., (2013) and He (2016) have investigated the issue from the perspective of firm performance in current literature, but they reached different conclusions. Firm performance is a comprehensive index to measure corporate operational performance, which is affected by many factors, such as cost management, corporate governance and business environment. A direct study of the relationship between registration address change and firm performance cannot open the ‘black box’ of how registration address change affects firm performance. From the perspective of policy preference and policy burden, this study discusses in detail the influence of registration address change on corporate cost stickiness and the paths through which registration address change affects cost stickiness. This study not only enriches related current literature, but also enlightens studies on the economic consequences of registration address change. Second, the study on cost stickiness helps uncover the ‘black box’ of corporate cost management (Jiang, Di, & Yao, 2017). Through effectively connecting corporate registration with cost stickiness, this study expands the scope of corporate cost management. Third, a registration address change involves the redistribution of political and economic resources. Hence, this paper provides a reference for the government in terms of allocating policy preference and policy burden to different regions. Only when the government reasonably allocates governmental resources to different regions can it prevent the imbalanced distribution of firms in regions and promote the balanced development of regional economies. Finally, the study has realistic guiding significance for registration address management in China. Less developed regions can attract listed firms via offering such preferential policies as more government subsidies; however, such behaviours increase corporate cost stickiness as well, thereby leading to the inefficiencies of firms, which ultimately has a negative impact on overall economic development. Therefore, this study calls on the underdeveloped regions to fundamentally improve their soft environment quality, so as to create a fair competitive atmosphere for firms.

The rest of the paper is organised as follows. Section 2 reviews the relevant literature. Section 3 puts forward the hypotheses of this study on the basis of theoretical analysis. Section 4 describes the research design. Section 5 presents the empirical analysis. Section 6 further discusses the paths through which registration address change affects corporate cost stickiness. Section 7 gives conclusion and suggestion.

2. Institutional background and literature review

2.1. Registration address change

It is not common for listed companies to change their registered addresses in China. Current literature mainly focuses on the following three aspects: the geographical distribution characteristics, the motivation, and the economic consequences of a registration address change. In terms of distribution characteristics, Pan et al. (2013) found that the migration of listed firms in China was mainly upward (that is, firms moved to cities with higher administrative rank), and the regions along the east coast of the Yangtze River Delta and Beijing were the main regions where listed firms moved to. The move of headquarters and the change of registration addresses of listed firms were mainly due to mergers and corporate reorganisation. In terms of the motivation for registration address change, based on his empirical study of Chinese A-share listed firms from 2004 to 2013, He (2016) found that government subsidies obtained by firms after a registration address change were significantly increased, he suggested that obtaining government subsidies was one of the motivations for a registration address change in China. Based on the case study of the migration of Sany Heavy Industry Corporation from Changsha, Hunan province to Beijing on 9 January 2013, Bu and Liu (2017) believed that large private firms would change registration addresses on their own initiative to avoid local government intervention and seek a fair market environment. As for the economic consequences, Pan et al., (2013) suggested that the relocation of corporate headquarters did not lead to the significant growth in firm performance based on the univariate test. Even if it did lead to the growth in performance, the effect was very limited and usually lasted for a very short time. At the same time, the upward migration of corporate headquarters was not more helpful to improving firm performance than the parallel and downward migration. He (2016) believed that registration address change played a significant positive role in improving operational conditions and firm performance.

2.2. Cost stickiness

Anderson, Banker, and Janakiraman (2003) verified the existence of cost stickiness, taking American firms as the sample, and they defined cost stickiness for the first time. That is, costs increase more when activity rises than they decrease when activity falls by an equivalent amount. Wan and Xu (2012) verified the existence of cost stickiness in Chinese listed firms as well taking the Chinese A-share main board listed firms from 2004 to 2010 as the sample. Wang, Cao and Liu (2014) studied 1,044 listed firms and found that Chinese listed firms had cost anti-stickiness as well. Dalla and Perego (2015) studied small and medium-sized firms in Italy and found that cost stickiness existed only in listed firms. The current literature has intensively studied the factors influencing cost stickiness, which can be classified into the following three aspects: adjustment costs, the optimistic expectation of management, and agency problems (Banker, Byzalov, & Plehn-Dujowich, 2010; Jiang & Hu 2011).

In view of adjustment costs, Banker, Byzalov and Chen (2013) believed that costs are generated when a firm makes various decisions on resource allocation. When management changes the resource allocation plans, the adjustment costs will be generated accordingly. Adjustment costs will be generated from all aspects of business activities. Balakrishnan and Gruca (2010) studied the sample from Ontario hospitals and found that relatively high adjustment costs made the management more reluctant to cut the resources of core departments, so the cost stickiness of the core departments was greater than that of the auxiliary departments. According to Kim and Wang (2014), the higher the unemployment insurance the firms paid, the lower the cost stickiness they hold. Jiang, Di, and Zeng (2015) verified the negative correlation between financial constraints and cost stickiness from the perspective of financing cost. Liu and Liu (2014) and Jiang, Yao and Hu (2016) verified the existence of cost stickiness based on the implementation of the Labor Contract Law of the People’s Republic of China and the Provision on Minimum Wage respectively.

Based on the optimistic expectations of the management, Banker, Byzalov, Ciftci, and Mashruwala (2014) held that optimistic expectations of management for the future will promote the resource input of firms, thus increasing cost stickiness. However, compared with male executives, female executives are more likely to pursue accounting conservatism and do not easily form optimistic expectation for future performance, which will help curb corporate cost stickiness (Quan & Tao, 2018). Liang (2015) studied the A-share non-state-owned listed manufacturing firms in China from 2001 to 2011, and found that the confidence level of the management was positively correlated with corporate cost stickiness, but debt constraints restrained the positive relationship. Zhou, Zhong and Xu (2016) divided the corporate competitive strategies into the Differentiation Strategy and the Low-Cost Strategy, and found that the cost stickiness of the former strategy was higher than that of the latter, and the expectation of the management regulated the relationship between corporate strategies and cost stickiness.

In view of agency problems, Kama and Weiss (2013) argued that when the management aimed not at maximising corporate benefit, their resource allocation will deviate from the optimum. However, the deviation depends on the different choices of the management, therefore, it’s possible that agency problems can increase or reduce cost stickiness. Based on the impact of a financial crisis on firms, Ma and Zhang (2013) found that the motivation of the management to maintain profitability weakened cost stickiness. The intensity of equity incentive inhibits cost stickiness by reducing agency costs (Liang, 2016). The Big Four Accounting Firms can restrain opportunistic behaviours of the management better, which helps restrain corporate cost stickiness (Liang, Chen, & Hu, 2015). And media coverage, as an important external governance mechanism, can effectively restrain corporate cost stickiness as well (Liang, 2017).

2.3. Policy preference and policy burden

Firms can get such policy preference as government subsidies, tax privilege, and preferential credit from government. From 2007 to 2010, the government subsidy expenditure was as high as 212.2 billion RMB (Bu & Yu, 2012). The allocation of government subsidies is affected by several factors, one of which is regional. Yu, Hui, and Pan (2010) found that firms in regions with poor institutional environment were likely to receive more government subsidies. Based on the study of all A-share listed firms in the Shanghai and Shenzhen Stock Markets from 2007 to 2010, Bu and Yu (2012) described the following geographical distribution features of government subsidies: the listed firms in western regions were increasingly getting policy support from the government, and those which locate at Beijing got the highest government subsidies. As a result, firms will take local government subsidies into consideration when they make decisions on corporate site selection (Ouyang, Zhu, & Zhang 2016). The above research shows that regional difference is an important factor influencing policy preference, and one of the motivations for firms to select or change their registration addresses is to maximise policy preference.

Lin and Li (2004) divided policy burden into strategic and social policy burden. The former referred to the direct intervention by government in investment decision, while the latter referred to the social welfare burden undertaken by firms such as providing employment. Many factors will affect the policy burden of firms, for example property rights (Lin & Li, 2004), political connection (Guo, 2011), hometown tie (Li & Sun, 2017), institutional environment (Guo, 2011), geographical location (Zhang, Huang, & Yan, 2016), etc. Bu and Liu (2017) found that one of the motivations for listed firms to change their registration addresses was to actively avoid local government intervention and seek a fair market environment, which helped firms reduce their policy burden. However, only large private firms have such a motivation, for example Sany Heavy Industry Corporation, while state-owned enterprises and small private firms in the growth stage would usually passively accept the policy burden imposed by the government.

Local governments can intervene in a market through both policy preference and policy burden. The registration address change, which is an economic activity related to capital resource reallocation, is bound to involve the change of policy preference and policy burden. Therefore, the study of the economic consequences of a registration address change definitely needs to analyse the above intervention behaviours of the government. To start with cost management, this study deeply explores the impact of registration address change on corporate cost stickiness, and discusses the ways through which registration address change affects corporate cost stickiness from the perspective of policy preference and policy burden.

3. Theoretical analysis and hypothesis development

Marketisation reform has led to decentralisation in the economic field. Under the established government management system, decentralisation has led to competition among local governments for economic resources (Zhou, 2003). Competition among local governments for official promotion enables them to take all measures to achieve political performance, compete for and protect superior regional resources (Sun & Luo, 2011). Attracting business and investment is the main strategy for local governments to compete with each other. Whether it is vertical or horizontal competition, obtaining capital resources is always the priority of local governments (Zhou, Feng, & Zhao, 2004). Listed firms are the driving force that promotes local economic development, and can bring higher quasi-rent for local governments; in other words, they are the powerful bargaining chips of government competition. The performance and influence of listed firms are among the factors that measure the performance of local officials. Therefore, local governments attach great importance to listed firms within their jurisdiction, and compete fiercely for the resources of listed firms (Sun & Luo, 2011). The direct listing (IPO) of a firm needs to go through such steps as filing, approval, sponsorship and road shows, which together usually takes a long time and is costly. In the context of government regulation on listing qualification, listing resources have become obviously scarce (Chen, Fan, & Liang, 2016). And, under this special institutional background, attracting listed firms in other jurisdictions to locate locally can not only avoid the listing risk, but also greatly promote the growth of local GDP and tax revenue in a short time, so listed firms have inevitably become the object for the competition of local governments.

However, although different starting points may determine whether local governments can relatively easily obtain competitive advantages in marketisation, what determines whether local governments obtain competitive advantages may ultimately lie in the local soft environment; namely, whether the institutional arrangement and administrative management are attractive enough for resources (Zhou & Zhao, 2002). For underdeveloped areas without any regional advantages, it is difficult to attract resources. In a comparative survey of some cities in Jiangsu province, Zhou and Zhao (2002) found that although adopting similar preferential policies, the areas in southern Jiangsu had greater advantages in attracting foreign capital due to their relatively high-quality soft environment. However, due to the poor soft environment in the underdeveloped areas of northern Jiangsu, domestic and foreign capital are rarely attracted even if preferential policies were introduced. The differences in the attractiveness of domestic and foreign capital in different regions are also confirmed by Bu and Liu (2017). Based on the case study of Sany Heavy Industry Corporation, which changed its registered place from Changsha, Hunan Province, to Beijing, Bu and Liu (2017) found that large private firms would proactively get rid of a policy burden through a registration address change, so as to avoid local government intervention and seek a fair market environment. On the contrary, in order to win fiscal competition, some local governments can only attract listed firms through tax exemptions, credit preference and more government subsidies due to the poor soft environment.

The differences in the soft environment and resource endowment make local governments attract listed resources through differentiated preferential policies, which will inevitably lead to the change of political resources enjoyed by listed firms after their registration address change. On one hand, firms may be supported directly by preferential policies from local governments after registration address change. According to the optimistic expectation of the management, massive government subsidies can directly increase the free cash flow of firms. At the same time, government subsidies mitigate the financial constraints of firms (Ren & Lu, 2014), and the alleviated financial constraints will directly lead to the increase of corporate cost stickiness (Jiang, Hu, & Zeng, 2015). On the other hand, a corporate registration address change may aggravate or alleviate the burden of firms (Xiao, Li, & Yuan, 2018). Based on the perspective of adjustment costs, firms may need to undertake diversified goals such as providing employment and donation under the constraints of a policy burden. These goals will increase corporate adjustment costs such as human resources, etc., thus increasing corporate cost stickiness. At the same time, under the constraints of a policy burden, the management needs to change the resource allocation plans with the consent of both firms and local governments, or even firms need to bear excess costs at the expense of corporate interests, which will increase the adjustment costs and finally lead to the corresponding increase of corporate cost stickiness (Banker et al., 2013). Of course, it’s possible as well that firms change their registration addresses to avoid local government intervention and seek a more fair competition environment (Bu & Liu, 2017; Xiao, Li, & Yuan, 2018). When firms do not need to bear too much policy burden, their cost stickiness will also be reduced. Based on the above analysis, a corporate registration address change will affect the cost stickiness of firms in many ways, which depend on such comprehensive factors as policy preference and policy burden from local governments. Therefore, the study proposes the following competitive hypotheses:

H1a: Ceteris paribus, a registration address change will increase corporate cost stickiness.

H1b: Ceteris paribus, a registration address change will reduce corporate cost stickiness.

4. Research design

4.1. Sample and data

The initial sample contains 111 Chinese A-share listed firms that changed their registration addresses across provinces from 2000 to 2016.¹ For the 14 listed firms that changed registration addresses many times, this study keeps only the sample firms before and after their first change. In the remaining sample, 59 listed firms (including one listed financial firm) were either backdoor listed or merged and restructured in the year of the change, and their number of employees fluctuated greatly from 43,726 at maximum to only three at minimum in different years. Therefore, it’s possible for firms to change completely after registration address change. In order to avoid the incomparability resulting from the change, this study eliminates all the sample firms with fewer than 200 employees in any year, referring to Zeng and Chen (2006). Moreover, the study eliminates the sample firms in financial and insurance industries and with missing data as well, and finally gets 693 firm-year observations of 49 listed firms for regressions. The data of registration address change and policy preference used in this study are manually collected from the annual report. Other data are from the China Stock Market and Accounting Research (CSMAR) database. To avoid the effects of extreme values, all continuous variables are winsorised at both the top and bottom 1% level. And STATA12 software is used to process the data.

4.2. Model setting and variable definitions

Based on the relevant studies of Anderson, Banker and Janakiraman (2003), Liang, Chen and Hu (2015) and Jiang, Yao and  Hu (2016), this study constructs model (1) to test the research hypothesis:

(1) $\begin{array}{rl}& \mathrm{\Delta }Lnsga=a_0+a_1\mathrm{\Delta }Lnsale+a_2\mathrm{\Delta }Lnsale\times D+a_3\mathrm{\Delta }Lnsale\times D\times Post\\ & +a_4\mathrm{\Delta }Lnsale\times D\times Market+a_5\mathrm{\Delta }Lnsale\times D\times D\mathrm{\_}Twoyear\\ & +a_6\mathrm{\Delta }Lnsale\times D\times Gdpgrowth+a_7\mathrm{\Delta }Lnsale\times D\times Einten\\ & +a_8\mathrm{\Delta }Lnsale\times D\times Ainten+a_9\mathrm{\Delta }Lnsale\times Post+a_{10}\mathrm{\Delta }Lnsale\times Market\\ & +a_{11}\mathrm{\Delta }Lnsale\times Gdpgrowth+a_{12}\mathrm{\Delta }Lnsale\times Einten+a_{13}\mathrm{\Delta }Lnsale\times Ainten\\ & +\sum Firm+\sum Year+\epsilon \end{array}$(1)

The main variables in the model are defined as follows.

1. The dependent variable: ΔLnsga indicates the cost change, which is measured as the differences between the natural logarithm of corporate cost (the sum of management cost and sales cost) in the current year and the natural logarithm of corporate cost in the previous year.

2. The explanatory variables. (i) ΔLnsale indicates the change of income, which is measured as the differences between the natural logarithm of corporate operational income in the current year and the natural logarithm of corporate operational income in the previous year. (ii) D is a dummy variable, which indicates the change of operational income in the current year compared with that in the previous year, and it equals 1 if corporate operational income in the current year is increased compared with that in the previous year, and 0 otherwise. (iii) Post indexes the registration address change, it equals 1 in the year when a firm changes its registration address and in the years after the change, and 0 otherwise.

3. The control variables. (i) Marketisation (Market). Following Fan, Wang, and Zhu (2011) and Wang, Fan, and Yu (2017), the ranking of the Marketisation Index overall rating of different regions is used to measure the differences in the institutional environment before and after the registration address change of the listed firms.² After a registration location change, if the ranking of the Marketisation index overall rating of a  province in which a firm is registered is lower than it is before, it equals 1, and 0 otherwise. (ii) D_Twoyear is a dummy variable. It equals 1 if the operational income declines for two consecutive years, and 0 otherwise. (iii) Economic growth (Gdpgrowth), which is measured as the current year’s GDP subtracted from the previous year’s GDP divided by previous year’s GDP. (iv) Einten indexes the employee density, which is equal to the ratio of the number of employees of a firm to its operating income in the current year (per million, in RMB). (v) Ainten indexes the asset density, which is equivalent to the ratio of the final total assets of a firm to its operating income in the current year. In addition, the study adds the firm and year fixed effects in all models as well.

5. Analysis of empirical results

5.1. Descriptive statistics

Table 2 reports the descriptive statistics of the main variables. It is shown that the mean and the median of ΔLnsga are 0.147 and 0.119, respectively, and the mean and the median of ΔLnsale are 0.124 and 0.096, respectively, which are basically identical with the statistical results of Gong, Liu, & Shen (2010), and Liang (2015, 2017). The sample proportion of the income decline D was 35.6%, slightly higher than that of Gong et al. (2010), Liang, Chen, and Hu (2015) and Liang (2017). The mean value of Post is 0.456, indicating that the sample is roughly equal before and after registration address change. The mean value of Market is 0.371, indicating that about 37.1% of listed firms change their registered addresses from regions with higher marketisation to regions with lower marketisation, reflecting that listed firms in China are inclined to move to developed regions (Pan et al., 2013). In addition, firms with operating income that declined for two consecutive years account for 15.2% of the sample.

Table 2. Descriptive statistics of the main variables.

Variable	Obs#	mean	Standard deviation	Min.	P25	Median	P75	Max.
ΔLnsga	693	0.147	0.478	–1.376	–0.053	0.119	0.290	2.113
ΔLnsale	693	0.124	0.440	–1.146	–0.096	0.096	0.274	1.787
D	693	0.356	0.479	0.000	0.000	0.000	1.000	1.000
Post	693	0.456	0.498	0.000	0.000	0.000	1.000	1.000
Market	693	0.371	0.483	0.000	0.000	0.000	1.000	1.000
D_Twoyear	693	0.152	0.359	0.000	0.000	0.000	0.000	1.000
Gdpgrowth	693	0.094	0.019	0.067	0.079	0.094	0.101	0.142
Einten	693	3.358	6.629	0.098	0.708	1.498	2.964	48.66
Ainten	693	2.384	1.989	0.373	1.168	1.947	2.823	12.63

5.2. Multiple regression analysis

Table 3 reports the regression results of model (1). The dependent variable for Columns (1) to (3) is ΔLnsga. Column (1) contains only the income change ΔLnsale and the cost stickiness ΔLnsale*D, which aims at testing whether there is cost stickiness in the sample firms. As shown, the coefficient of ΔLnsale is significantly positive at the 1% level, indicating that the management costs and sales costs of the sample firms generally change synchronous with the change of income to the same direction; the coefficient of cost stickiness ΔLnsale*D is significant and negative at the 5% level, indicating that the decline in the costs when income declines is less than the increase in the costs when income rises, which means that stickiness is generally existing in the sample firms. To examine the effect of registration address change on corporate cost stickiness without controlling other factors, the interaction term ΔLnsale*D*Post is added in Column (2) based on column (1). The results show that the coefficient of ΔLnsale*D*Post is significant and negative at the 1% level, indicating that corporate cost stickiness has increased significantly after registration address change. With other control variables added based on column (2), the regression results of column (3) show that the coefficient of the interaction term ΔLnsale*D*Post is significantly negative at the 1% level. According to the above results, with the trans-provincial change of registration addresses of listed firms, corporate cost stickiness has generally increased significantly, thus supporting H1a.

Table 3. Registration address change and cost stickiness.

Variable	(1) ΔLnsga	(2) ΔLnsga	(3) ΔLnsga
ΔLnsale	0.672***	0.673***	1.457***
	(13.564)	(13.664)	(4.641)
ΔLnsale*D	–0.261**	–0.012	–2.268***
	(–2.310)	(–0.083)	(–3.780)
ΔLnsale*D*Post		–0.465***	–0.797***
		(–2.892)	(–3.860)
ΔLnsale*D*Market			0.000
			(0.001)
ΔLnsale*D*D_Twoyear			0.242
			(1.482)
ΔLnsale*D* Gdpgrowth			25.082***
			(4.585)
ΔLnsale*D*Einten			0.019
			(0.727)
ΔLnsale*D*Ainten			–0.008
			(–0.213)
ΔLnsale*Post			0.262***
			(2.952)
ΔLnsale*Market			0.163
			(1.479)
ΔLnsale* Gdpgrowth			–9.931***
			(–3.368)
ΔLnsale*Einten			–0.006
			(–0.240)
ΔLnsale*Ainten			–0.030
			(–1.177)
Constant	0.787***	0.687***	0.688***
	(3.620)	(3.137)	(3.046)
Firm Fixed Effect	Yes	Yes	Yes
Year Fixed Effect	Yes	Yes	Yes
Obs#	693	693	693
Adj_R2	0.2943	0.3025	0.3378
F Value	5.44	5.55	5.64

Note: *, ** and *** represent the significance levels of 10%, 5% and 1% respectively (two-tails). The t value is in the parenthesis, here in after the same.

5.3. Robustness test³

In order to improve the reliability of the study conclusions, the following robustness tests are conducted. First, the two-stage Heckman regression is conducted to control the effect of possible self-selection behaviours of firms that experience registration address change on the conclusion of the study. Second, the study retests the hypothesis by removing the 15 listed firms that have their actual controllers or main business changed before and after registration address change. Third, when testing the differences in cost stickiness before and after the registration address change of listed firms, the study includes all the years within the sample period in the regression analysis. However, there are extreme cases. If the listed firms changed their registration address in year 2001 (2016), there is only one-year sample before the registration address change, but a 16-year sample after the registration address change, which would probably cause cost stickiness to be affected by factors other than registration address change. Therefore, this study retests the hypothesis with only the sample of 3 years before and after registration address change, and 5 years before and after registration address change, respectively. The regression results of the robustness tests are generally consistent with the main regression, indicating that the conclusions of this study are relatively stable.

6. Further discussion: the ways through which registration address change affects cost stickiness

6.1. Registration address change and policy preference

Local governments in China play a very important role in regional economic growth, and their enthusiasm for using all possible sources of investment to promote local economic development is unusual in the world (Zhou, 2007). The main strategy of local government competition is to invite business and attract investment. Whether it is vertical or horizontal competition, obtaining capital resources is the first priority of local governments (Zhou et al., 2004). As the registration address determines the jurisdiction of corporate tax, the most direct impact of registration address change is to greatly promote the GDP and the tax revenue growth of the area into which the firms move. As an important measure to interfere in local economic development, policy preference such as a government subsidy is one of the most important means for local governments to compete for capital. In order to compete for the scarce resources of listed firms, local governments will certainly provide them with various preferential policies such as government concessional loans, government subsidies, low transfer fees for land-use, etc., which are more obvious in firms that migrate from relatively developed regions such as eastern China to relatively underdeveloped regions such as western China (Bu & Yu, 2012). The registration address change can significantly improve policy preference (government subsidies), which is also verified by the related research of He (2016). Policy preference includes various government subsidies such as R&D subsidies, special funds, interest subsidy loans, tax relief and rapid low-interest loans, and they bring about cash inflow or cost saving for firms. With preferential policies, firms can increase the resources available, reduce operating costs and improve development potential in the future, which promotes the optimistic expectation of the management for the future. Based on the optimistic expectation, the management may spontaneously increase resource input, resulting in the increase of cost stickiness (Banker, Byzalov,  Ciftci, & Mashruwala, 2014). In other words, the increase of policy preference brought about by registration address change is likely to be one of the ways that increases corporate cost stickiness.

In order to explore the ways through which registration address change affects corporate cost stickiness from the perspective of policy preference, the study first conducts a univariate test on the policy preference (government subsidies) before and after registration address change. There are three ways to measure government subsidies in current research. Subsidy, the government subsidies standardised by the operational revenue in the current year (Tang & Luo, 2007); Subsidy1, the government subsidies standardised by the total assets at the end of the year (Wang Li, & Xing, 2014);Subsidy2, the natural logarithm of (government subsidies plus 1) (Zeng, Bu, & Rao 2017). Table 4 reports the results of the univariate test on government subsidies before and after registration address change with the three measurements. It is shown that the Wilcoxon test is significantly negative at the 1% level for all the measurements. And the mean T-test is negative for the three, but significantly negative only for Subsidy2. According to the univariate test, the policy preference of firms has increased greatly after the registration address change of firms, compared with that before the change⁴

Table 4. Univariate test of policy preference before and after registration address change.

	Post=0 (Before the change)			Post=1 (After the change)
Variable	Obs#	Mean	Median	Obs#	Mean	Median	Mean T test	Wilcoxon test
Subsidy	377	0.006	0.001	316	0.006	0.001	–0.919	–4.478***
Subsidy1	377	0.003	0.000	316	0.003	0.001	–0.798	–3.833***
Subsidy2	377	9.221	12.94	316	13.58	15.51	–8.709***	–8.957***

To test whether policy preference is one of the reasons leading to the increase of cost stickiness after the change, the study divides the whole sample into two groups according to the change directions of the policy preference before and after registration address change. They are the group with policy preference increased (the mean value of policy preference in all the years after the change is greater than before) and the group with policy preference decreased (the mean value of policy preference in all the years after the change is less than before). If the interaction term of registration address change and cost stickiness ΔLnsale*D*Post is negatively related with cost change ΔLnsga, and the negative relationship of the two is more significant in the group with policy preference increased, it is confirmed that policy preference is indeed one of the reasons that leads to the increase of cost stickiness. Table 5 reports the grouping regression results. It can be seen from the sample distribution that there are 472 firm-year observations in the group with policy preference increased, while there are 221 firm-year observations in the group with policy preference decreased, indicating that most of the listed firms have obtained more government subsidies compared with that before the change, which is consistent with the conclusion of He (2016). The regression results of the two groups show that the negative relationship between the interaction term ΔLnsale*D*Post and ΔLnsga exists only in the group with policy preference increased, whereas the coefficient differences of the two groups are not significant, which to some degree verifies the above speculation. That is to say, policy preference is one of the ways through which registration address change affects cost stickiness.

Table 5. Registration address change and cost stickiness: group by policy preference.

Variable	(1) Group with policy preference increased	(2) Group with policy preference decreased
ΔLnsale	1.451***	1.215**
	(3.422)	(2.271)
ΔLnsale*D	–2.046***	–2.829
	(–2.771)	(–1.581)
ΔLnsale*D*Post	–0.518**	–0.631
	(–2.013)	(–1.151)
ΔLnsale*D*Market	–0.215	–0.241
	(–0.661)	(–0.358)
ΔLnsale*D*D_Twoyear	0.316	0.249
	(1.620)	(0.563)
ΔLnsale*D* Gdpgrowth	20.023***	39.140**
	(2.963)	(2.285)
ΔLnsale*D*Einten	0.044	0.080*
	(1.215)	(1.778)
ΔLnsale*D*Ainten	0.043	–0.336***
	(1.000)	(–2.972)
ΔLnsale*Post	0.049	0.481***
	(0.419)	(2.843)
ΔLnsale*Market	0.304**	0.230
	(2.039)	(1.087)
ΔLnsale* Gdpgrowth	–8.023**	–10.961**
	(–1.974)	(–2.324)
ΔLnsale*Einten	–0.043	–0.025
	(–1.309)	(–0.629)
ΔLnsale*Ainten	–0.042	0.043
	(–1.447)	(0.739)
Constant	0.045	0.213
	(0.242)	(0.656)
Firm Fixed Effect	Yes	Yes
Year Fixed Effect	Yes	Yes
Obs#	472	221
Adj_R2	0.3331	0.3890
F Value	4.92	4.18

6.2. Registration address change and policy burden

Local governments intervene in the market directly through a policy burden, which includes a strategic policy burden as the intervention in the investment decision-making of firms and a social policy burden requiring firms to provide excessive employment and staff welfare (Lin & Li, 2004). Generally speaking, the better the institutional environment, the lower the degree of government intervention, and the lower the policy burden borne by firms; for example, the developed regions in eastern China (Zhang, Huang, & Yan 2016). Bu and Liu (2017) found that one of the motivations for the registration address change of listed firms was to avoid local government intervention and seek a fair market environment, thus reducing their policy burden. However, only large private firms had such a motivation, most of the state-owned enterprises or smaller private firms that were in their growth stage would usually passively accept a policy burden imposed by the government. In order to explore the ways through which a registration address change affects cost stickiness from the perspective of policy burden, this study first conducts a univariate test on the policy burden before and after registration address change. Following Li and Sun (2017), the study estimates the following model and measures policy burden as the absolute value of the residual μ

(2) $C{i}_t=b_0+b_1\mathrm{L}\mathrm{n}siz{e}_{t-1}+b_{2}Le{v}_{t-1}+b_{3}Growt{h}_{t-1}+b_{4}RO{A}_{t-1}+b_{5}Capita{l}_{t-1}+\sum Province+\sum Industry+\sum Year+\mu$(2)

Ci_t indexes capital intensity, and is the ratio of net fixed assets (per million, in RMB) to the number of employees; Lnsize_t−1, Lev_t−1, Growth_t−1, Roa_t−1 and Capital_t−1 represent firm size, financial leverage, growth rate of total assets, return on assets and asset structure of period t–1, respectively. Province, Industry and Year are dummy variables.

Table 6 reports the results of the univariate test of the policy burden before and after registration address change. It is shown that both the mean T-test and the Wilcoxon test are significantly negative at the 10% level. This indicates that political resources work like a double-edged sword (Choi & Thum, 2009). While the government gives more policy preference to firms that change their registration addresses, it accordingly imposes a policy burden on the firms as well. And when firms invest in the non-economic benefit maximisation projects or provide excessive employment due to the consideration of social benefit, their resources are not optimally allocated. From the perspective of adjustment costs, the management needs to change resource allocation plans with the consent of both firms and the government, or even bear the excess costs at the expense of corporate interests, which increases the adjustment costs and finally leads to the increase of corporate cost stickiness (Banker, Byzalov, & Chen, 2013). In other words, a policy burden increase brought about by a registration address change is likely to be another way that increases the cost stickiness of firms.

Table 6. Univariate test of the policy burden before and after registration address change.

	Post=0 (Before the change)			Post=1 (After the change)
Variable	Obs#	Mean	Median	Obs#	Mean	Median	Mean T test	Wilcoxon test
Burden	360	0.650	0.449	316	0.753	0.433	–1.627*	–1.813*

To test whether a policy burden is another reason for the increase of cost stickiness after the change, this study divides the whole sample into two groups according to the change directions of the policy burden before and after the registration address change. They are the group with policy burden increased (the mean value of the policy burden in all the years after the change is greater than before) and the group with the policy burden decreased (the mean value of policy burden in all the years after the change is less than before). If the interaction term of the registration address change and cost stickiness ΔLnsale*D*Post is negatively related with cost change ΔLnsga, and the negative relationship of the two is more significant in the group with policy burden increased, it is confirmed that policy burden is indeed one of the reasons that leads to the increase of cost stickiness. Table 7 reports the grouping regression results. It can be seen from the sample distribution that there are 376 firm-year observations in the group with the policy burden increased, while there are 317 firm-year observations in the group with the policy burden decreased, indicating that more than half of the listed firms bear more policy burden after registration address change. The regression results of the two groups show that the negative relationship between the interaction term ΔLnsale*D*Post and ΔLnsga exists only in the group with policy burden increased, and the coefficient differences of the two are significant at the 1% level (chi2(1)=5.4, prob>chi2=0.0202), which verifies the above speculation. That is to say, policy burden is one of the ways through which a registration address change affects cost stickiness⁵

Table 7. Registration address change and corporate cost stickiness: group by policy burden.

Variable	(1) Group with policy burden increased	(2) Group with policy burden decreased
ΔLnsale	1.265***	1.959***
	(2.841)	(4.034)
ΔLnsale*D	–1.482	–3.315***
	(–1.621)	(–3.294)
ΔLnsale*D*Post	–1.263***	–0.333
	(–4.135)	(–1.065)
ΔLnsale*D*Market	–0.097	0.379
	(–0.183)	(0.943)
ΔLnsale*D*D_Twoyear	0.182	0.134
	(0.818)	(0.507)
ΔLnsale*D* Gdpgrowth	18.382**	33.107***
	(1.984)	(4.087)
ΔLnsale*D*Einten	0.036	–0.038
	(0.896)	(–0.848)
ΔLnsale*D*Ainten	0.042	–0.028
	(0.807)	(–0.341)
ΔLnsale*Post	0.364***	0.097
	(2.874)	(0.730)
ΔLnsale*Market	0.111	–0.014
	(0.704)	(–0.076)
ΔLnsale* Gdpgrowth	–7.966*	–13.473***
	(–1.818)	(–3.166)
ΔLnsale*Einten	–0.036	0.058
	(–1.028)	(1.421)
ΔLnsale*Ainten	–0.037	–0.049
	(–1.218)	(–0.812)
Constant	0.275	0.536**
	(1.310)	(2.151)
Firm Fixed Effect	Yes	Yes
Year Fixed Effect	Yes	Yes
Obs#	376	317
Adj_R2	0.3044	0.3921
F Value	3.98	5.16

6.3. Policy preference, policy burden and corporate cost stickiness

According to the above analysis, an increase of policy preference or policy burden can lead to an increase of corporate cost stickiness. Therefore, it is possible that if the policy preference and the policy burden of firms increase simultaneously after a registration address change, their cost stickiness should increase substantially. In order to verify this inference, this study divides the whole sample into four groups according to the change directions of policy preference and policy burden before and after registration address change (whether the mean value of policy preference or policy burden in all the years after the change is greater than that of before): (1) the group with both preference and burden increased; (2) the group with preference increased but burden decreased; (3) the group with preference decreased but burden increased; (4) the group with both burden and preference decreased. Table 8 reports the grouping regression results. It can be seen from the sample distribution that the group with both policy preference and policy burden increased has the largest sample size (268 firm-year observations), which indicates that listed firms that get more policy preference are imposed with more policy burden as well after the change of registration addresses. This is consistent with the analysis above. According to the comparison of the regression results of the four groups, the negative relationship between the interaction term ΔLnsale*D*Post and ΔLnsga is only significant in the group with both policy preference and policy burden increased, and the coefficient differences compared with that of the other three groups is at least at the 10% level, which further verifies that policy preference and policy burden are indeed two important ways to increase corporate cost stickiness after registration address change. And for firms with both policy preference and policy burden increased simultaneously after the change, their cost stickiness increases most.

Table 8. Registration address change and cost stickiness: group by policy preference and policy burden.

Variable	(1) Group with preference and burden increased	(2) Group with preference increased but burden decreased	(3) Group with preference decreased but burden increased	(4) Group with preference and burden decreased
ΔLnsale	1.052	2.011***	1.470*	2.127**
	(1.540)	(3.433)	(1.866)	(2.281)
ΔLnsale*D	–1.034	–3.832***	–3.424	–8.234**
	(–0.868)	(–3.479)	(–1.356)	(–2.308)
ΔLnsale*D*Post	–1.033**	–0.039	–0.575	–0.121
	(–2.495)	(–0.098)	(–0.742)	(–0.122)
ΔLnsale*D*Market	–0.127	0.041	0.825	0.364
	(–0.168)	(0.089)	(0.737)	(0.282)
ΔLnsale*D*D_Twoyear	0.031	0.460	0.734	–0.897
	(0.118)	(1.298)	(1.174)	(–1.265)
ΔLnsale*D* Gdpgrowth	12.684	32.077***	36.419	88.643***
	(1.036)	(3.603)	(1.545)	(2.685)
ΔLnsale*D*Einten	0.023	0.072	0.197	0.020
	(0.499)	(0.641)	(1.151)	(0.342)
ΔLnsale*D*Ainten	0.069	0.060	–0.246	–0.220
	(1.193)	(0.431)	(–0.970)	(–1.103)
ΔLnsale*Post	0.272	–0.176	0.152	0.364
	(1.384)	(–0.986)	(0.445)	(1.559)
ΔLnsale*Market	0.171	0.186	–0.387	0.006
	(0.660)	(0.814)	(–0.881)	(0.016)
ΔLnsale* Gdpgrowth	–5.317	–12.424**	–9.262	–18.566**
	(–0.773)	(–2.391)	(–1.339)	(–2.274)
ΔLnsale*Einten	–0.026	–0.084	–0.189*	0.025
	(–0.615)	(–1.020)	(–1.993)	(0.443)
ΔLnsale*Ainten	–0.052	0.017	0.134	0.040
	(–1.475)	(0.208)	(1.411)	(0.274)
Constant	0.214	–0.078	–0.256	0.240
	(0.961)	(–0.367)	(–0.852)	(0.709)
Firm Fixed Effect	Yes	Yes	Yes	Yes
Year Fixed Effect	Yes	Yes	Yes	Yes
Obs#	268	204	108	113
Adj_R2	0.2293	0.4479	0.4439	0.4338
F Value	2.69	5.02	3.37	3.38

7. Conclusion and summary

The change of corporate registration address will directly affect the development of listed firms, and also the districts from and to which the firms move, which finally brings about the reallocation of capital resources. Therefore, it has important research value. Cost control is one of the main concerns of corporate management and an important factor for economic decision-making (Liu, 2006). Starting with cost stickiness, this study examines the economic consequences of corporate registration address change, and the ways through which the change affects cost stickiness from the perspective of policy preference and policy burden. First, with the sample of Chinese A-share listed firms that have experienced trans-provincial change in their registration addresses from 2000 to 2016, this study conducts an empirical analysis on the relationship between registration address change and corporate stickiness using the ABJ cost stickiness model. The results show that registration address change significantly increases the cost stickiness of listed firms. Second, the study divides the sample into two groups according to the change directions of policy preference and policy burden, and the results show that policy preference and police burden are two ways through which registration address change affects cost stickiness. Moreover, for listed firms that get both policy preference and policy burden because of their registration address change, their cost stickiness increases the most. This paper effectively links registration address change with corporate cost stickiness and expands the scope of corporate cost management. At the same time, it constitutes a reference for the government in making a decision on how to distribute policy preference and policy burden to different regions. Only through reasonably allocating governmental resources to different regions can the government prevent the imbalanced distribution of firms in regions and promote the balanced development of regional economies.

Acknowledgements and funding

We appreciate helpful comments from anonymous reviewers and acknowledge financial support from the National Natural Science Foundation of China (Project No. 71602191).

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Listed firms in the same provinces or under the jurisdiction of the same local governments before and after registration address change are not included.

² Market index data of Wang, Zhou, & Rao (2017) are only up to 2014, while the sample of this study is up to 2016. For 2015 and the year after, data in 2014 are used instead.

³ Limited by the length of the article, the results of this part are not listed. Please contact the authors for these results.

⁴ The analysis in the following part of this paper is based on Subsidy (government subsidies standardised by the operational revenue in current year), and the test results using the other two measurement methods remain unchanged.

⁵ Referring to Bu and Liu (2017), the study measures the policy burden of firms as the ratio of the sum of ‘the cash paid to and for employees’ and ‘the taxes and fees paid’ to operating income. The regression results are consistent with the above results, which proves that the above conclusions are relatively robust.