The impact of China's pension insurance contributions on consumption and savings

Residents' savings rate remains high, becoming a "Chinese mystery" that is of great concern to both domestic and foreign. In the context of the financial tsunami, the excessive savings rate of Chinese families has once again become the focus. Promoting economic growth by stimulating consumption is generally seen as the key to promoting healthy and sustainable economic development. The academic community has put forward various explanations for the reasons for the high savings of Chinese families. One of the most common explanations is related to China's inadequate social security system: families need to make preventive savings for a variety of possible income shocks, including savings for pensions, education, and large medical expenses that may occur. Although academic research has not yet determined how important this saving motivation is, I would like to thank the colleagues at the Tsinghua University Academic Luncheon for their comments. Thanks to the anonymous reviewer for their valuable comments.

Bai Zhongen thanked the National Social Science Fund Major Project (Project No. 10zd 007) for funding. Wu Binzhen thanked the Natural Science Youth Fund (Project No. 70903032) for funding. Responsibility.

Scholars have suggested that consumption should be stimulated by raising the level of social security. This proposal has also received policy support. In the “package plan” of the government's response to the financial crisis, the level of social security has been greatly improved and it has been listed as one of the four major measures to deal with the crisis. 1 However, some scholars have pointed out that the current fundamental of expanding the total domestic demand is to increase the disposable income of ordinary people and redistribute national income. 2 Increasing the level of social security will ultimately increase the tax burden of enterprises or residents, and further squeeze the current period of residents. Disposable income. In fact, the social security burden of employees and enterprises is already very heavy. The comprehensive proportion of employee social insurance as a percentage of total wages is close to or even more than 40%. Specifically, pension insurance accounts for about 28% of wages. Medical insurance About 8% of wages, unemployment insurance accounts for about 3% of wages, work injury insurance accounts for about 1% of wages, and maternity insurance does not exceed 1% of wages. This ratio even exceeds some high-welfare countries in Europe. Although in this 40% ratio, employees only pay about 11%, and the rest are paid by the company, but economic theory tells us that under some reasonable assumptions, the part of the enterprise’s payment will be converted into the employee’s pre-tax salary. The decline, that is, the tax burden of the company may be passed on to the employees.

It can be seen that if we want to measure the impact of increasing the level of social security on consumption, we need to study the impact of increasing the burden of social security contributions on consumption. Since pension insurance is the main body of social security, we will focus on the payment of pension insurance. A simple theoretical model predicts that the impact of pension insurance contributions on consumption depends on the benefits of pension insurance, the extent of borrowing constraints, the main motivation for savings, and other characteristics of the pension system.

Most of the current empirical studies at home and abroad assume that the impact of pension insurance is reflected by the income constraints of the contributor's life. Therefore, it is necessary to analyze the net income level of pension insurance (the discounted value of all income minus the discounted value of all contributions). The impact of consumption and savings. However, when there is a credit constraint in the market and the family cannot smooth the consumption by borrowing the future pension insurance income, the impact of the current pension insurance contribution and the discounted value of the future income cannot be simply added or subtracted. Therefore, the current consideration is directly considered. The impact of pension contributions is necessary. Given that China's capital market is still underdeveloped, and China's pension insurance contribution rate is already high, it is of great practical significance to directly examine the impact of current pension contributions on consumption.

We will use the data of nine provinces and cities in China from 2002 to 2009 to estimate the impact of current pension contributions on household consumption savings. The difficulty with empirical estimates is that the level of contributions is often related to some unobservable work characteristics. We use the instrumental variable method to solve this endogeneity problem. Gong Baizhong, Qian Zhenjie: Who is occupying the income of the residents? Analysis of the income distribution pattern of the Chinese nationals. The Chinese Social Sciences section refers to the State Council’s notice on the printing and improvement of the pilot program for improving the urban social security system. The State Council has established the construction of urban vehicle variables and utilized various cities. The difference in pension contribution policies over time.

We found that before the pension insurance reform in 2006, given the income level before the pension payment, and given that the individual was covered by the pension insurance system, the current pension contribution burden significantly inhibited the consumption of the contributor family. effect. However, the estimated magnitude of the impact varies greatly depending on the method of estimation.

The estimated value given by the regression of the instrumental variables shows that the employee pension contribution rate is increased by 1 percentage point (ie, the disposable income after payment is reduced by about 284 yuan), and the household consumption is reduced by about 1.75% (about 424 yuan); and the OLS estimate It is much smaller. By correlating with industrial enterprise data, we found that the employee contribution rate increased by 1 percentage point, accompanied by an increase in corporate contribution rate of about 0.34 percentage points. Therefore, the decline in disposable income due to a 1 percentage point increase in employee contribution rate may reach 381. yuan. The estimated value of regression of instrumental variables means that the decline in disposable income due to the increase in pension contribution rate has basically translated into a decline in consumption, and the amount of savings has remained basically unchanged. In addition, we found that being covered by pension insurance itself would significantly stimulate consumption. The negative impact of pension contributions on consumption may have weakened after the 2006 reforms, but the significance of this change depends on the model setting, so it is not yet possible to give a definitive answer. Finally, we found that the increase in the employee pension contribution rate did not lead to a significant increase in the current pension income. The elderly’s consumption propensity was lower than that of the pension contributor. Therefore, it can be considered that the current pension contribution burden is rising. The overall level of consumption has a major inhibitory effect.

The structure of this paper is as follows. The second part introduces the related characteristics and the characteristics of China's pension insurance system and its changes. The third part gives a simple theoretical model, pointing out that the low income of pension insurance, the constraints of borrowing and the existence of target saving motives may lead to the current pension payment restraining consumption. The fourth part introduces the measurement model. The fifth part introduces the data and gives a statistical description. The sixth part discusses the empirical results and conducts a robustness test. The seventh section estimates the impact of pension contributions on total consumption. Section 8 summarizes and discusses the policy implications.

II. Relevant and institutional background The theory of pension insurance tells us that the relationship between pension insurance and savings is very complicated. If there are no borrowing constraints and uncertainties, there is an alternative relationship between private savings and pension insurance. When the income of pension insurance and the income of private savings are equal, the current consumption is not affected by the payment of pension insurance, and the private savings and pension insurance contributions have a complete replacement relationship, that is, the reduction of private savings is exactly equal to the contribution of pension insurance. . If the income of pension insurance is greater than the income of private savings, the consumption will increase, and the reduction of private savings will be more than the payment of pension insurance. However, when the income of pension insurance is less than the income of private savings, the current consumption will decline, and the reduction of private savings will be less than the payment of pension insurance. In all three cases, pension insurance will reduce the private savings rate (savings/distributable income).

If there are borrowing constraints and uncertainties, the alternative relationship between private savings and pension insurance will become more complicated. For example, Hubbard and udd pointed out that borrowing constraints reduce the negative impact of pension insurance on household savings, and may even increase the savings rate. 2 They also pointed out that pension insurance can reduce preventive pension savings. Samrnck further pointed out that the existence of risks such as large medical expenses may lead to the motivation of target savings or the buffering of buffers. This kind of motivation makes residents have a target value of savings, so the contribution of pension insurance increases. At the same time, despite the reduction in disposable income, residents are still reluctant to reduce their savings, which in turn may lead to a decline in consumption and an increase in the savings rate. 3 In addition, haler also suggested that self-binding savings and short-sighted reasons may also lead to an increase in the savings rate of the pension insurance. 4 Because of the relationship between pension insurance and savings rate, there is no theoretical conclusion, and empirical research is particularly important.

At present, most of the empirical research on the relationship between pension insurance and savings at home and abroad does not consider the impact of borrowing constraints and uncertainty, but only considers the pension insurance to affect the current savings by affecting the budgetary constraints of the client's life, so pay attention to the net pension. The impact of wealth on consumer savings. There are three main methods used in empirical research. The first one is represented by the early research of Feldstem, using time series data to analyze the impact of the net wealth of pension insurance on current consumption. 5 This method is difficult to rule out the effects of other changes in the same period. The second is to use the difference in the cross-section of net wealth of pension insurance to directly examine the crowding-out effect of net wealth of pension insurance on private savings (stock). 6 The main problem with this method is that the level of pension insurance contributions is often related to other job characteristics of the parties. Since it is impossible to control all these job characteristics, it is difficult for empirical research to completely separate the impact of pension insurance itself. The differences in research conclusions based on the first two methods are also relatively large. For example, Kotlkoff found that pension net wealth has little effect on private savings, while Feldstein and Pellechio and Gale found that 1, 1990, pp. 193-205. Pension contributions may also be stimulated by affecting employment (such as encouraging employees to retire early) Savings. Moreover, the old-age security system is mostly accompanied by the redistribution of income between generations and generations. Different income groups and intergenerational generations have different marginal propensity to consume. The older generation has the motive of bequests. Therefore, pension security is for society. The direction of the overall savings rate is difficult to determine.

The replacement rate of pension net wealth and private savings is between 39% and 100%. 1 A recent study has used a third approach, using natural experiments to obtain comparatively exogenous pension insurance differences, such as an examination of changes in the pension system in Italy or the United Kingdom. 2 These studies are close to Gale's conclusions. As a rare empirical study on the impact of China's pension insurance on savings, He Lixin et al. and Feng, He and Sato use this method. In response to the reform of China's pension insurance system from 1995 to 1997, they studied the changes in the net wealth of pension insurance caused by policy changes and the impact on household savings rate. 3 It was found that the decline in the net wealth of pension insurance significantly increased the household savings rate, ranging from 30% to 40%, and the difference between the households varies with the parameters. Chamon and Pasad indirectly consider the impact of pension insurance on savings by studying the relationship between family savings and family demographics, and suggesting that imperfections in social security are important reasons for the high savings rate of Chinese residents. 4 In order to fully examine the relationship between pension insurance and savings, we need to consider the impact of credit constraints and uncertainty. Only analyzing the impact of the net wealth of pension insurance on savings can not achieve this goal. It is necessary to directly examine the impact of current pension insurance contributions on current savings, which needs to be supplemented in this regard. 5 He Lixin, Feng Jin, and Sato Hir: The impact of pension insurance reform on household savings rate: China's empirical evidence, economic research Kotlikoff distinguishes the net income of pension insurance and the impact of paid pension contributions on family wealth, and found that it has been paid The pension contribution does reduce the accumulation of wealth, but there is no significant difference between this negative impact and the general negative impact of savings on savings. He found that the net income of pension insurance had no significant effect on wealth. L.Kotlikoff, “Testing (II) Institutional Background Foreign studies rarely discuss the impact of the pension contribution rate itself. One of the important reasons is that foreign pension insurance taxes are mostly nationally unified, and it is difficult to find lateral differences in exogenous However, in China, pension insurance is basically in a state of co-ordination at the city and county levels before 2007. 1 And in order to encourage institutional innovation tailored to local conditions, the central government has allowed local governments to choose their own implementation plans under the unified guiding principles, thus creating The situation of the endowment insurance system in all parts of the country.

Although institutional differences between regions may be related to many unobservable regional characteristics, regional differences in China's pension insurance policies over time allow us to effectively control regional differences that do not change over time.

More specifically, before 1997, there was no uniform regulation for individual employee contribution rates. Individual contributions began to be gradually implemented after 1991, not exceeding 3% of wages. On July 16, 1997, the State Council established a unified basic pension insurance system for enterprise employees. The decision requires that “the proportion of individual contributions should not be lower than 4% of the salary of the individual in 1997. From 1998, it will increase by 1 percentage point every two years, and finally reach 8% of the salary of the person. The conditional areas and wages will grow faster. In the year, the rate of increase in the proportion of individual contributions should be appropriately accelerated. The proportion of corporate contributions should generally not exceed 20% of the total wages of employees (including those allocated to individual accounts), and the specific proportion should be determined by the people's governments of provinces, autonomous regions, and municipalities directly under the Central Government. 2 Such a policy has led to different growth rates of personal pension contributions in various localities. This also prompted the Ministry of Labor and Social Security to issue a document in 2003, requiring individuals with a rate of 8% to reach the rate, and take measures as soon as possible to achieve this ratio. 3. In this way, after controlling the fixed effects of various regions, the regional differences in such policy changes are given.

Based on the data of the 2002-2009 urban household survey, the changes in pension contribution rates and coverage rates in different provinces are given. It can be seen that there is a big difference in pension coverage and contribution rates, and the trends are not the same. During this period, the average contribution rate of the city is about 61%, the variance is 2%, the average annual contribution rate of the city is 0.86%, and the variance of the growth rate is 1.7%. This is why we study the current contribution rate to consumption. In 1998, the State Council required all provinces, municipalities and autonomous regions to gradually implement provincial-level coordination of endowment insurance. However, by 2007, except for a few provinces and cities such as Beijing and Shanghai, which achieved real-level provincial coordination, most of the cities in 2004 had achieved an individual contribution rate of 8%, but there were also many The city still implements a 7% contribution rate (such as Jinan, Wuhan, Yantai, etc.), and even a city with a 5% contribution rate (Xiamen), see 50 major cities wage and social insurance payment information inquiry, "http :// =Xiamen. With regard to the decision to improve the basic endowment insurance system for enterprise employees, since January 2006, the proportion of individuals paying basic old-age insurance premiums has been unified to 8%. , providing better natural experiments (a) inter-provincial differences in coverage over time (b) inter-provincial differences in contribution rates over time, and the time trends of pension coverage and contribution rates in inter-provincial differences (2002- 2009) Source: The authors calculated according to the 2002-2009 data of the Urban Household Survey.

In a given area, pension contributions may vary between employees at different wage levels. More specifically, the individual payable social insurance premium is the product of the social insurance contribution base and the contribution ratio. In general, the social insurance contribution base for individual employees this year is the monthly average of the total annual salary income of individuals. However, if the individual's salary is lower than the lower limit of the contribution base (60% of the average wage of the local employees), 60% of the local average wage is used instead of the employee's actual salary as the payment base; and the individual salary exceeds the payment base limit (local average salary) The portion of 300%) is not included in the contribution base. Therefore, the contribution rate will change nonlinearly when the individual's salary reaches the upper or lower limit of the contribution base. When the wages of employees are lower than the lower limit of the payment base or above the upper limit, the actual contribution rate decreases with the increase of wages. When the wage is between the upper limit and the lower limit, the actual contribution rate is equal to the local payment ratio. Our data shows that 35% of non-retired, non-institutional employees are below the base of the payment base, and 0.6% of such personnel are above the upper limit of the payment base.

Therefore, the pension contribution rate of many families is affected by the upper and lower limits. In addition, it is worth mentioning that many enterprises and employees have evasive behaviors, especially under-reporting social insurance contributions. 1 This tax evasion behavior is often related to the nature of the enterprise and the regulatory enforcement of each city.

Considering that the payment of pension insurance is a kind of compulsory savings with future benefits, the changes in people's behavior depend on future earnings. In the 1997 reform, employees who worked after July 1, 1998, 11% of the employees' wages and salaries entered the personal accounts of pension insurance, in which individual contributions were all entered into personal accounts, and the remaining 1 enterprises and employees evaded paying pension insurance premiums. The phenomenon is quite serious. The most common way is to report and under-report the salary base. In January and June of 2005, a total of 12.684 billion yuan of under-reporting payment was found in all parts of the country. Five insurance premiums were paid out of the enterprise payment; the rest of the enterprise's payment was included in the overall account. 1 For those who have accumulated 15 years of payment, the pension income after retirement is composed of the basic pension of the unified account and the personal account pension. The former is calculated based on 20% of the average monthly salary of the local employee in the previous year. The personal account accumulates the principal and interest of 1/120. Therefore, if the number of years of income is large, the yield of personal accounts will be larger. However, since a total of 11% of the wages enter the personal account, the insufficient part of the employee's payment is supplemented by the enterprise, and the employee's payment part will squeeze out the payment part of the enterprise. This crowding effect reduces the rate of return of the employee's contribution. The marginal rate of return of the pension contribution to the pooled account is 0. After 15 years of payment, regardless of the amount paid by the company, the employees receive a 20% social average wage.

Another major institutional change occurred in the January 2006 policy change. 2 Among them, there are two most important changes. The proportion of personal contributions is 8%, and the personal account is adjusted from 11% of the paid wages to 8%, that is, all the contributions of the enterprises are included in the overall account. Second, the basic pension calculation method has been adjusted: the basic pension for the overall account is based on the average monthly salary of the local employees in the previous year and the average monthly average payment of the index, based on the individual accumulated contribution period. Calculate the proportion of the calculation, the payment is paid to 1% every 1 year, and the average monthly payment of the index is increased by the average individual's contribution wage. Personal account pension is the number of months of personal account storage divided by the average life expectancy of urban population, my retirement age, and interest. It can be seen that the new accounting method reflects the principle of more pensions and more contributions, which helps to increase the enthusiasm of residents. 3. A simple theoretical model This section uses a simple life cycle model to provide a theoretical basis for our empirical research. This simple model considers three important factors that are closer to reality. (1) An individual must pay a pension during work (younger), a pension benefit after retirement (in old age), (2) an individual is subject to borrowing constraints, and (3) an individual who wants to purchase large consumer goods or other in the future The reason is that there is a motive for the target savings. Among them, target savings are generally derived from preventive savings for future uncertainties. However, if the theoretical model specifically refers to the State Council's decision to improve the basic endowment insurance system for enterprise employees, http:// Policy changes in 2006 for pension benefits The impact is more complicated. The first policy change canceled the contribution of enterprises to individual accounts, so the income of employees on personal accounts declined. However, due to the elimination of the crowding effect of employee contributions on corporate contributions, the marginal rate of return of employees' contributions increased. The second policy change increases the marginal rate of return of pension contributions on the pooled account. Due to the adoption of the “step-by-step” approach, this reform basically guarantees that the income of the old-age insurance for the elderly and the middle-aged people has not decreased, and the uncertainty is introduced, it is difficult for us to obtain an explicit solution. Therefore, we have considered another Reasons that may lead to target savings: Some commodities, such as housing, are bulk consumer goods, and need to pay a relatively large amount in one lump sum. Because of credit constraints, if you want to consume such goods, consumers need to save a certain amount. Buy a house as an example to illustrate this problem.

Suppose an individual needs to make three decisions. In each period, they need to decide whether to spend, save, and whether to buy a house (0-1 choice). If you buy a house, you need to pay P yuan. In addition, people are faced with borrowing constraints, without losing the generality, we assume that people can not borrow. The problem of individual intertemporal choice can be expressed by the following equation: where U is the utility function, P is the discount rate, Ct is the consumption of t period, and Ht is the period of t. At the same time, each worker is required to pay wage income tax tw and social security tax tss.B for the total pension paid by the individual (considering interest income), and the constraint gives the calculation formula of B. b is the rate of return on pension insurance. For the sake of simplicity, we assume that 匕 is a constant. Let P = 1 / (1 + r). If b=1, the pension insurance is the same as the risk-free savings in the capital market; if b>1, then the pension insurance is higher than the capital market; on the contrary, it is lower.

In order to understand this optimization problem, we first need to find the optimal consumption savings option (C, A-,) after a given purchase, and then compare the utility values ​​under different purchase options to find the best purchase choice and corresponding consumption. Savings choices. The ultimate optimal choice depends on the distribution of income, house prices, and initial assets. 1 To simplify the analysis, we focus on the following situations: Suppose the housing is given a sufficiently high utility, people's most tss) W2, and P>bXB. We further assume that if we do not consider the changes in the pension system here. If the pension insurance policy changes, individuals will be re-optimized according to the new policy, so the impact of the policy depends on the number of periods that are re-optimized, in other words, depending on the age of the individual.

To discuss the results. 2 First consider the case of credit constraint relaxation (notbmdmg), when people do not borrow (A2>0).

Consumption depends on the budget constraints of a lifetime, so it depends on whether the income of the pension insurance is greater than the return of the capital market.

Define the savings rate as savings divided by the disposable income after the pension payment, and Yt indicates the household disposable income after the social security contributions. When the pension contribution rate rises, such as (the income of the pension insurance is greater than the return of the capital market), the consumption rises and the savings rate decreases; if b<1 but 匕>/(1―〖), the consumption and savings rates decline, private The decline in the savings rate is because the (compulsory) savings have been made for retirement by paying pensions; if 匕 (1 tw), consumption has fallen, but the savings rate has risen. The reason is that if P is high, people have incentives to save for housing and retirement, so the amount of savings declines is relatively less than the amount of disposable income, and the savings rate rises. 3 We noticed that the savings rate is unlikely to rise if no purchases are introduced (and thus the target savings motive), because b is unlikely to be less than the balance between consumption in the third period, but only to balance consumption between the first two periods. Therefore, the income of pensions is not covered by the insurance system. When the pension contribution rate rises, the income of the first two periods declines, and the consumption of the first two periods also decreases. 4 As consumption and disposable income decline, the final change in the savings rate is uncertain, depending on P and VIII. (1 + 2 relationship. When 卩 > 八. (1 + 2, the savings rate rises with the increase in the pension contribution rate (even if b>1) otherwise the savings rate declines. The intuitive explanation is that 卩> eight. (1 +2 means that the family needs to be the most likely to appear in the standard savings motivation.

The demand for second-phase purchases.

The specific proof is not reported here due to space limitations, and interested readers can request it from the author.

Another savings rate (s2) is defined as savings (YtC) divided by disposable income before pension contributions. Since ds2/dtss=(1twtss)dS/dtss-s, it is more likely that the impact of pension contributions on s2 will be negative. If the purchase is regarded as consumption and removed from the savings, the impact of the pension payment on the savings rate after deducting the purchase is always negative.

Here we assume that the pension insurance yield does not change with the amount of contributions. In addition, the optimal choice may change due to the coverage of pension insurance. When there is no pension insurance system and it becomes a positive pension contribution rate, consumption may not necessarily decline.

IV. Measurement Model Setting Based on the above discussion, we will directly examine the impact of the current pension contribution rate on the current consumption and savings of the contributor. Since there is only household-level consumption, the sample of empirical analysis is based on households. In order to avoid the main body of pension payment is not the decision-making body of consumption and savings, we only consider those households whose spouses and their spouses are the mainstay of family pension contributions (the proportion of households and spouses paying more than 80% of the total pension contribution of the family). The pension contribution rate is the ratio of the total amount paid by the head of household and their spouse to their wage income. In addition, if the head of household or spouse is covered by pension insurance, we define that the family is covered by pension insurance. 1 Since there is no information in our data on the payment of pension insurance, we only consider the payment of pensions for employees. In order to simplify the statement, unless otherwise specified, the subsequent pension insurance payment refers to the employee payment portion.

The basic model is set as follows: For the pension insurance contribution rate, we mainly examine the impact of the contribution rate, and also use the logarithm of the contribution amount for robustness analysis. Y is the disposable income of the family before the social security payment. Under the premise that the disposable income before the payment is fixed, ai represents the rate of change in consumption after increasing the pension contribution rate by 100%. Considering the work of pension insurance and the work without pension insurance, there is often a big difference. We control the “denying insurance coverage” dummy variable (Dpension) to control the linear impact of this job characteristics on consumption. Therefore, ai reflects the difference in consumption between households with a relatively heavy pension contribution burden and a lighter burden when they are covered by pension insurance, given that they have the same disposable income before payment. In addition, after controlling the logarithm of household disposable income before payment, the coefficient of each variable in the equation also represents the impact on the consumption rate log Dg (consumption/pre-distributable income).

The biggest difference between our model and the model we talk about is that we don't have direct control over the net benefit of pension insurance. This is because the policy of China's pension insurance system changes very frequently, people are hard to know about future earnings, and the lack of information on corporate contributions in the data, we can not accurately calculate pension wealth. 2 On the other hand, if we only pay attention to the net impact of pension contributions on consumption, we do not have to estimate the net wealth of pensions. Given that people pay a pension, there is some expectation of the income of the pension, and ai already contains the positive impact of the expected changes in pension income on consumption. At the same time, ai also includes the negative impact of payment and consumption on factors such as borrowing constraints and target savings. In other words, ai represents the combined impact of current pension contributions on consumption.

These definitions have no essential impact on the outcome. We also tried to use 60%, 50% as a sample selection condition, and to control the number of people covered by pension insurance in the family. The results are very similar.

Most of the foreign countries are targeting the pension insurance system in developed countries, and most of them do not have personal accounts. They estimate the pension insurance income of each household based on the current pension insurance income formula, assuming the parameter values ​​of wage growth rate, life expectancy, and discount rate. But in a fast-growing country like China, wage growth is harder to predict.

Pension contribution rates are also related to many other job characteristics, including wages, industry, occupation, and job stability. At the same time, these job characteristics often directly affect consumption. Therefore, we control these operating characteristics as much as possible in X to avoid estimation bias. More specifically, we control the wage income of the head of the household and the spouse and the dummy variables of the wage-quantity group in the city (5 groups), the working life of the head of the household, the square of the working years, the industry (16 groups), occupation (8 groups) and the nature of the enterprise (state-owned enterprises, urban collective enterprises, individual or private enterprises, other types of enterprises). The group variable that adds wages here is to control the non-linear effects of wages. 1 We also control the demographic characteristics of heads of households that may affect job characteristics, including age, age grouping (6 groups), education grouping (9 groups), gender, marital status, ethnicity, and household registration type. 2 In addition to these factors, pension contributions may also be positively related to other insurance project contributions, including medical insurance, unemployment insurance, and housing provident fund. If the contributions of these projects directly affect consumption, and the impact of each project on consumption is the same, then the impact of pension contributions will be overestimated. Therefore, we try to control the payment of other projects in the model to detect the existence of such overestimation.

We also control the characteristic variables of some families, including the number of family members, the number of children under the age of 18, and the number of people over the age of 60. Given that the low-wage family pension contribution rate is relatively high, and the marginal consumption propensity of low-income families tends to be relatively low, this negative correlation will cause a to show a negative sign. This bias may persist after controlling the linear impact of household income logarithms. Therefore, in addition to controlling the non-linear effects of wages, we also improve households that allow different per capita incomes, with different income consumption elasticity. More specifically, we introduce the numerator group (high school, low school group) in which the per capita income is located in the city, and the intersection of the logarithm of the disposable income before the family payment. Finally, we also control urban fixed effects and year fixed effects to control urban differences that do not change over time and the common time trends everywhere. We also tried to add cross-cutting terms for the province's dummy variables and years to allow for different potential trends in the province's consumption.

Although we have maximized control over the observable work characteristics associated with pension contributions, there are still some unobservable factors associated with pension contributions that affect consumption. For example, enterprises with relatively high pension contributions are often relatively mature enterprises, and their work is relatively stable. Households with higher risk aversion are more likely to choose social insurance. The existence of these factors leads to OLS's estimation of “missing variables”. The estimated deviation brought. We use the instrumental variable method to solve this potential problem.

The instrumental variables we consider are the average contribution rates for each city in each year. As mentioned above, after 1997, the proportion of contributions in various cities began in principle from 4%, an increase of 1 percentage point every two years. After controlling wages and other job characteristics, an important source of differences in personal pension contribution rates Is the non-linear change that occurs when the wage rate reaches the upper or lower limit of the payroll base. But the impact of wages themselves may be non-linear, so if you do not control the non-linear effects of wages, the coefficient before the pension contribution rate may have an estimated bias.

In the base model, we do not control the relevant characteristics of the spouses because their corresponding characteristics with the head of the household are often highly correlated. In the robust test, we tried to control these characteristics of the spouse. For families without a spouse, these values ​​were set to 0, and the results changed little.

Up to 8%. But the progress of each city can be adjusted according to the local economic situation. Therefore, the average annual contribution rate of the city reflects the relevant policies of each year, which obviously affects the contribution rate of each household.

Since the city and annual fixed effects have been controlled, we have actually used the difference in average contribution rates over time between cities. Although the difference in the rate of increase in the rate of contribution between cities means that the institutional environment of each city is different, if the difference does not change with time, or changes linearly with time, then the urban fixed effect and the linear trend of the province have been basically controlled. The impact of this. Similarly, for the variable “covered by pension insurance”, the annual average coverage rate of each city is an effective tool. V. Data and statistical description. The data from the China National Bureau of Statistics' China Urban Household Survey (2002- 2009). The urban household survey covers all provinces in China, and samples are obtained by sampling (stratified by prefecture-level cities, county-level cities, counties). All samples are rotated 1/3 per year, and all samples are rotated every three years. The survey collects data by means of a survey of households. The data includes basic information for each family member: income and expenses, including various social security contributions. There is also detailed household income and expenditure information in the data.我们使用的数据包含北京、辽宁、浙江、安徽、湖北、广东、四川、陕西和甘肃9个省市，分别来自于东中西三个地区，具有全国代表性。

我们首先使用个人层面的数据，描述养老保险缴费和参与情况，并重点关注除离退休人员、事业单位以外的城镇就业人员，包括城镇在岗职工及城镇私营和个体就业人员。①其中，个体工商户和灵活就业人员在2006年改革之前大都没被纳入社保体系。事业单位实施与企业单位完全不同的养老保险制度，因此我们剔除了在事业单位工作的样本。我们还剔除了一些异常值，包括没有工资收入的个体及养老金缴费率大于1的样本。最终样本为128329人，其中在岗职工为99347人。

2002―2009年间，除离退休人员、事业单位以外的城镇就业人员的平均工资、养老保险覆盖及缴纳情况。可以看到，养老保险在这类人员中的平均覆盖率约为80%，而所有城镇就业人员的覆盖率约为78%，这些数据和官方公布的覆盖率非常接近。如果剔除2008年，这覆盖率在2006年达到谷底，而后开始上升。除2008年之外，职工个人的缴费额度和缴费率逐年上升，2006年分别达到在岗职工包括在国有、城镇集体、联营、股份制、外商和港澳台投资、其他经济单位及其附属机构就业，并由其支付工资的各类人员，不包括再就业的离退休人员、私营和个体从业人员、乡镇企业、农村及其他从业人员（如现役军人、农民工等）。

劳动和社会保障部2007年发布的近年来我国社会保险基本情况指出，2006年城镇企业基本养老保险的覆盖率为76%. 1076元和6.3%，与政策规定的缴费率8%仍有一定差距，到2009年也才达到7.5%.①如果考虑其他社会保障支出，包括医疗保险、失业保险、工伤保险、生育险等，职工的个人缴费负担也基本呈现逐年上升的趋势。到2009年，总的社保个人缴费率达到9.6%，如果进步考虑住房公积金，总的个人缴费率达到13.表1除离退休人员、事业单位外的城镇就业人员养老金覆盖和缴费情况（样本：128329）年份样本量（个）平均工资（元）养老金覆盖率官方所有城镇就业人员平均工资（元）所有城镇就业人员覆盖率养老金缴费额（元）养老金缴费率社保总缴费率加入住房公积金的总缴费率总计注：（1）所有城镇就业人员包括离退休人员、事业单位及无工资收入的从业人员。资料来源：中华人民共和国国家统计局编：中国统计年鉴（2010），北京：中国统计出版社，2010年。

表2给出了养老金缴费率和其他社保项目缴费率及收入之间的相关性。可以清楚地看到，各种社保项目的缴费之间存在正相关关系，而且在5%的显著水平上显著。养老金缴费和所得税之间存在负相关关系，但相关系数比较小。养老保险缴费率和工资收入呈现负相关关系，说明养老金缴费率是累退的，原因是低工资职工更可能在缴费基数下限以下；而高工资职工更可能超过缴费基数上限。但养老保险覆盖率、缴费金额以及住房公积金缴费率，随工资的上升而上升。

我们粗略估算了养老保险的收益（由于篇幅限制，表格没有报告），发现以当前退休工人的养老保险收益和该城市在岗职工平均工资的比重，所定义的替代率平均达到了63.5%，在2006年之前有所下降，2009年有所上升。我们还发现，养老保险抚养系数，即领取养老金的人数和养老金缴费人数之比，大致呈现上升趋势（从2002年的45.7%上升至2009年的50%），说明养老保险体系的压力在加大。

除离退休人员、事业单位外的城镇就业人员社保缴费率与收入的相关关系（样本：128329）12养老金缴费率医疗保险缴费率其他社保缴费率住房公积金缴费率所得税率工资收入养老金缴费率医疗保险缴费率其他社保缴费率住房公积金缴费率所得税率工资收入注：代表5%显著水平上显著。

由于城镇调查数据中没有家庭成员个人的消费支出信息，我们的回归分析都以家庭为单位。在家庭成员过多时，家庭决策的情况变得复杂，因此我们去掉了样本中家庭成员个数大于9人的家庭。同时，我们剔除了户主年龄不处在劳动力年龄（男16―60周岁，女16―55周岁），或户主已经离退休的家庭。为了尽可能地避免养老保险体系的差异带来的影响，我们剔除了那些户主没有工作、为私营企业主、个体工商户或在事业单位工作的家庭。我们还删除了户主及配偶的养老金缴费额占家庭总缴费额的比例小于80%的样本，以避免养老金缴费主体并非消费和储蓄决策主体的情况。经过这些样本筛选后，我们的最终样本包含了9个省市的112个城市，共计51691户家庭。我们对所有的名义变量都进行了物价调整，以2006年的CPI为100.回归涉及变量的描述性统计，由于篇幅限制没有报告。

六、回归结果本节报告主要的回归结果。所有回归都考虑了稳健性方差，并控制了In（家庭缴费前收入），人均收入在城市内的分位组（高中低三组，基准组是中等收入家庭）及其与n（家庭缴费前收入）的交叉项，n（户主和配偶缴费前工资收入），户主和配偶工资在其居住城市内所处的分位组（5组），户主的年龄，年龄分组（6组），教育分组（9组），工作年限，工作年限的平方，行业（16组），职业（8组），以及企业的性质（国有企业、城镇集体企业、其他类型企业），户主性别，婚姻状态，民族及户口类型，家庭人口数，工作的人数，家庭人口数，18岁以下小孩数，60岁以上老人数。收入组的分类是基于家庭人均收入在该城市内的分位组（3组）。

（一）养老金缴费对消费的影响我们首先关注养老保险缴费率对消费的平均影响。考虑到养老保险制度在2006年有重大变化，人们对养老金缴费的回报率预期可能发生本质变化，因此，表3首先基于2002―2005年的数据来估计。前四列都采用了工具变量（IV）回归。第1列没有控制“被养老保险覆盖”这个哑变量，IV为该样本所在城市平均的养老金缴费率。第2―4列控制了“被养老保险覆盖''，并在IV中加入了城市平均的养老保险覆盖率。两个IV第一阶段回归的F值基本都大于10，说明它们是强的工具变量。这一结论也被其他检验所证实。①我们还把IV加入基础模型作为额外的控制变量，结果显示，这些IV对消费没有直接的显著影响。

第1列结果表明，平均而言，养老金缴费率对消费的影响为负，但不显著。第2列显示，如果在被养老保险覆盖的家庭间比较，养老金缴费率上升1个百分点，消费下降206%，在5%的显著水平上显著，95%的置信空间为（一3. 83%，一0.29%）。由于第2列已经控制了工资收入（对数）以及五个工资分位组组别的影响，因此养老金缴费率的负向系数，并非来源于工资和养老金缴费率的负相关关系。

而且我们在前2列中已经控制了家庭人均收入组别（分高中低3组，基准组是中等收入家庭）及其与缴费前收入对数的交叉项，这一结果也不太可能来源于家庭消费收入弹性的非线性影响。我们还发现，如果不控制家庭消费收入弹性的非线性影响，或者工资的非线性影响，结果变化很小。②第3列控制了其他保险项目的缴费率，用来避免养老金缴费的系数包含了其他项目的影响。为了避免其他缴费项目的内生性，我们在IV中加入了各个项目的城市平均缴费率；而第4列将其他项目合并成项，以减少共线性问题。我们看到，养老金缴费率对当期消费的影响依然显著为负（在10%的水平上显著），不过幅度下降至1.75%，90%的置信区间为（一3.43，一0.06）。

第4列结果的点估计值意味着，养老金缴费率上升1个百分点，缴费后可支配收入下降约284元，消费下降约424元。这一结果有些出人意料。然而这里我们没有控制企业的养老金缴费，而职工的养老金缴费和企业缴费可能存在正相关关系。城镇住户调查数据中无法观察到企业的养老保险缴费，因此我们首先利用2004年国家统计局关于规模以上工业企业调查的数据，计算了分县市、行业及企业性质（共12924个组别）的企业缴纳养老和医疗保险费占工资的比例，然后利用相同年份的城镇住户调查，计算了分县市、行业及企业性质（共904个组别）的个人养老及医疗保险缴费率。③简单回归分析表明，个人的养老医疗保险缴费率上升1个百分点时，企业的养老医疗缴费率上升0.34个百分点（方差为0.16），%的水平上拒绝了两个IV是弱IV的假设。我们检查了对工具变量强弱不敏感的LIML回归，结果和用两阶段回归的结果非常接近。这些都证实，这里基本可以排除工具变量不够强的顾虑。

如果不控制家庭消费收入弹性的非线性影响，缴费率的系数为一2. 08，如果不控制工资分位组别，缴费率的系数为一2. 05，都在5%的水平上显著。

企业数据无法区分养老保险和医疗保险。这一相关性在非国有、非集体的企业中最为明显。在这类企业中，个人缴费率上升1个百分点，企业缴费率上升0.60个百分点。

在5%的显著水平上显著。两个数据库2005年的信息也给出了类似的结果。因此，个人养老医疗缴费率上升1个百分点，实际上代表总的缴费率上升1.34个百分点。如果企业的缴费最终完全转嫁给职工，这意味着养老金缴费后，可支配收入下降达到381元，与消费下降1.75%的量相差不大。这样，表3的估计结果意味着，养老金缴费率上升导致的可支配收入下降，基本全都转化成了消费的下降，储蓄基本没有变化。

表3的后三列给出了最小二乘（OLS）回归结果，以作比较。OLS的结果同样显示，缴费率对消费的影响显著为负，但在幅度上与IV回归估计值的差别比较大。比如第6列显示，养老金缴费率上升1个百分点，消费下降0.25%.而在控制其他缴费项目后（第7列），这一影响进一步下降至0.13%.然而，第7列的估计值仍处在IV回归估计值的90%置信区间内。由于OLS面临内生性问题，我们更加偏好IV回归结果。我们进一步采用基于Hausman检验原理的C统计，检验零假设OLS估计结果是一致的“问题。①结果显示，除了第1列外，其他模型设定都可以在10%的水平上拒绝这一零假设。因此可以认为，IV回归结果更为可靠。

表3养老金缴费率对消费的影响被解释变量：log（家庭消费）工具变量（IV）回归最小二乘法（OLS）回归养老金缴费率被养老保险覆盖医疗保险缴费率其他社保缴费率住房公积金缴费率所有其他缴费率续表3被解释变量：log（家庭消费）工具变量（IV）回归最小二乘法（OLS）回归Log（缴费前收入）样本数R平方第一阶段IV的F值城市缴费率城市覆盖率内生性检验P值注：括号内是稳健型方差，分别表示1% 5%1%显著水平显著。

我们发现，“被养老保险覆盖”本身显著剌激了消费。结合给定被养老保险覆盖后，养老金缴费率对消费的抑制作用，一个合理的解释是，人们认可参加养老保险的收益，因为一旦被覆盖，就可以享受一些基本保障（比如2006年之前，基本养老保险收益几乎和缴费多少无关），但是给定有了基本保障，养老金缴费率的上升反而会抑制消费，因为养老金缴费的边际回报率很低，同时（或者）有信贷约束及目标储蓄动机的存在。另外，结果显示，收入消费弹性的估计值在66%左右，与相一致。①我们还发现，在IV估计中，医疗保险缴费也对消费有抑制效果，不过在OLS估计中，这一影响不再显著。其他社保项目和住房公积金的影响都不显著。

表4第一部分基于2002―2009年的数据，估计了养老金缴费率的影响，各列的模型设定和表3―样。结果同样显示，养老金缴费负担对消费有显著的抑制作用。更具体的，第1列表明，平均而言，养老金缴费率显著抑制了消费。

第2列则显示，在控制养老保险覆盖状态后，这种抑制作用更为明显。即给定缴费前的可支配收入，养老金缴费率上升1个百分点（约380元），消费下降1.02%（约315元），估计值的95%置信空间为（一1.76%，一0.28%）。在控制其他保险缴费项目后（第4列），这一影响下降至0.89%（约276元），95%置信空间为（一1.68%，一表4养老金缴费率对消费的影响：2006年之前及2006年之后的效果差异被解释变量：log（家庭消费）工具变量（IV）回归最小二乘（OLS）回归养老金缴费率被养老保险覆盖所有其他缴费率Log（缴费前收入）控制其他各项项目缴费否是否样本数R平方第一阶段IV的F值城市缴费率城市覆盖率内生性检验p值养老金缴费率2006年后被养老保险覆盖2006年后养老金缴费率被养老保险覆盖所有其他缴费率所有其他缴费率2006年后控制其他各项项目缴费否是否样本数R平方注：括号内是稳健型方差，分别表示10% 5%1%显著水平显著。表4的后三列给出了OLS回归结果。与表3类似，OLS的结果也证实，缴费率对消费的影响显著为负，但在幅度上与IV回归估计值的差别比较大。第6列显示，养老金缴费率上升1个百分点，消费下降0.26%.而在控制其他缴费项目后（第7列），这影响进步下降至。

16%.但第7列的估计值仍处在IV回归估计值的95%置信区间内。这里的C统计检验显示，我们无法在10%的水平上拒绝前2列模型设定的OLS结果可信的假设，虽然可以在15%的水平上拒绝第3列的OLS估计。然而在考虑其他保险缴费项目后，我们可以在1%的水平上拒绝OLS回归结果。

比较表3和表4第部分的IV回归结果，可以看到，养老金缴费对消费的抑制效果在2006年之后明显减弱，虽然OLS的估计值差别并不大。①基于2006―2009年数据的回归结果也验证了这一点。更具体的，2006―2009年养老金缴费率对消费影响的IV估计值都不显著，而OLS估计都显著为负，且在幅度上与2002―2005年的估计值差别不大。2006―2009年“被养老保险覆盖”的影响也不再显著。我们在表4的第二部分，通过加入养老金缴费与2006年之后“哑变量的交叉项，以检验2006年前后的差别是否显著。IV中也加入了城市平均缴费率及覆盖率与2006年后”哑变量的交叉项。结果有些意外：无论是IV还是OLS回归，养老金缴费率对消费的影响，在2006年前后的差别都不显著。因此，养老金缴费率的影响在2006年前后的变化，对模型设定比较敏感。

总之，养老金缴费对消费有显著的抑制效果这一结论，在2006年之前基本没有疑问，虽然幅度的大小在不同的估计方法下有较大的差异。更为可信的IV估计值意味着，养老金缴费率上升导致的可支配收入下降，基本全都转化成了消费的下降，储蓄变化很小。这一结果并不意外。2006年之前，养老保险统筹账户部分，多缴费并不能多得，只要被养老保险覆盖，即得到同样的收益。而个人缴费部分，由于在制度设计上对企业缴费有挤出效应，因此收益率也很低。

这既解释了“被养老保险覆盖”对消费的正向剌激作用，也解释了给定被覆盖后，养老金缴费率的上升对消费有显著的抑制效果。2006年的改革强调了“多缴费多收益”的原则，同时个人缴费对企业缴费的挤出效应也不再存在，这或许改变了人们对养老金缴费收益率的认识。但是由于估计结果对样本选择和估计方法的敏感性，我们还不能肯定养老金缴费对消费的负面影响，在2006年之后有所减弱甚至消失了。

我们后面的分析将集中考虑2002―2005年的样本，并主要报告IV的估计结果。

（二）养老金缴费对储蓄率的影响前面的结果表明，2006年之前，给定养老金缴费前的收入，当我们提高养老金缴7费，减少当期的可支配收入时，消费会减少。理论模型指出，其背后的机制有两种可能性。一是养老保险的收益率小于市场利率；二是有信贷约束同时有目标储蓄动机，因此当期的可支配收入减少时，人们可能会减少消费并增加储蓄率。理论模型预测，如果没有信贷约束和购房需求，只要b>0，储蓄率就会下降。如果第二种机制存在，+r）2，储蓄率也会随养老金缴费率的上升而上升。

表5养老金缴费对储蓄率的影响：工具变量（IV）回归被解释变量储蓄率1储蓄/缴费后可支配收入储蓄率2：储蓄/缴费前可支配收入养老金缴费率被养老保险覆盖所有其他项目缴费率Log（缴费前收入）样本数R平方IV第一阶段F值城市缴费率城市覆盖率注：括号内是稳健型方差,分别表示10% 5%1%显著水平显著。

表5给出了养老金缴费率对储蓄率的影响。①我们考虑两种储蓄率。一种是模型中考虑的储蓄除以缴费后的可支配收入（储蓄率1），一种是中常用的储蓄除以缴费前的可支配收入（储蓄率2）。结果表明，在控制是否“被养老保险覆盖”的状态后，缴费率对这两个储蓄率的影响都为正向，虽然都不显著。这和模型预测是一致的，因为在有预算约束时，缴费负担导致储蓄率增加的前提SP>A.（1这个条件可能只适用于一部分人群。另外，结果显示，养老金缴费负担对储蓄率1的正向影响，比对储蓄率2的影响在幅度上要强，这也与模型预测相致。

为了检验之前结果的可靠性，我们做了一些稳健性检验，结果都增强了基本结示，当缴费前的可支配收入给定，被养老保险覆盖的家庭在养老金缴费金额上升1个百分点后（约13元），会显著减少消费约0.06个百分点（列1，约15元），而在控制其他项目的缴费后，会显著减少消费约0.12个百分点（列2，约29元）。OLS结果也显示了显著的抑制效应。

其次，我们考虑了在消费中剔除医疗和教育支出。医疗和教育支出弹性比较小，家庭可能不会因为为将来的住房储蓄而降低这一类消费，同时在我们关注的时间段内，医疗卫生领域也经历了不少的政策变化，考虑医疗支出可能会引入不必要的噪音。估计结果显示，缴费率对消费的影响依然显著，幅度有所下降，但变化不大。

这与我们发现缴费负担对医疗和教育支出的影响不显著相一致。

接着，我们用被养老保险覆盖的在岗职工家庭来估计，即我们剔除了户主在私营企业从业或灵活就业的家庭及没有被养老保险覆盖的家庭。剩下的样本更为同质，可以最大限度地控制企业性质以及其他方面的差异。结果与表3依然类似，只是估计值和显著性略有下降。

最后，养老金缴费抑制消费的作用机制，对于其他减少当期可支配收入的税种也同样成立。因此，所得税应该也会对消费产生负面影响。而且，所得税没有直接的收益，因此它对消费的负面影响应该比养老金缴费更强。实证结果证实了这些设想。更具体的，我们用城市平均所得税税率和交税的就业人员比例为IV估计了所得税对消费的影响。结果显示（表格没有报告），所得税税率增加1个百分点，消费减少2.5%，在5%的水平显著；在控制是否交税后，消费减少3. 3%，在10%的水平显著。OLS的结果同样显示，所得税显著抑制消费，但幅度要小很多（消费减少大约0.4%）。

七、养老金缴费对总消费的影响前面的讨论只考虑了养老金缴费负担对缴费者的消费和储蓄的影响。另一个重要的问题是：养老金缴费对整体的消费有怎样的影响。如果现有的退休工人因为当前的缴费率上升而得到了更好的福利，进而导致他们增加消费，也许总的消费并没有下降很多。因此有必要分析城市平均的缴费率，如何影响养老金领取者的消费行为。

表6的第一部分，分析了同一时点上城市平均的个人缴费率与养老金收益（用9替代率来衡量）之间的关系。结果显示，在控制了城市和年度固定效应以后，城市平均个人缴费率与城市平均养老金收益没有显著的相关性。平均而言，当前的养老金领取者并没有从当前缴费率的增加中受益。①第二部分显示，在控制了城市和年份固定效应之后，城市平均缴费率的上升并没有提高养老金领取家庭的平均消费倾向。②第三部分则显示，养老金领取者的消费倾向比当前缴费者的消费倾向更低，这结论与研究中国居民储蓄行为的研究发现相致。③这些结果意味着，职工缴费率的上升并没有导致当前养老金领取者的收入有所增加。即使有所增加，由于老年人的消费倾向更低，总的消费也不太可能因此而上升。因此基本可以推断：养老金缴费率对宏观消费主要起着抑制的作用。10表6养老金缴费率对总消费的影响第一部分：被解释变量：城市平均替代率第二部分：被解释变量：平均消费倾向（有养老金收益的家庭）城市平均养老金缴费率城市抚养系数年份和城市固定效应否是否是样本数R平方第三部分：被解释变量：log（家庭消费）有养老金收益的家庭没有养老金收益且户主小于55岁的家庭ln（可支配收入）家庭特征否是否是户主特征否是否是样本数R平方注：括号内是稳健型方差，分别表示10% 5%1%显著水平显著。

八、结本文利用中国城镇住户调查2002―2009年9个省市的数据，考察中国目前职工的养老金缴费率对家庭消费储蓄的影响。我们利用各个城市养老金缴费政策随时间的变化，以及这种变化在城市间的差别来构建工具变量，进而解决缴费率与一些不可观察的工作特性相关导致的内生性问题。

我们发现，2006年之前，给定养老金缴费前的工资及收入水平，同时给定被养老保险系统所覆盖，养老金缴费率的上升对缴费家庭的消费有显著的抑制效果。影响的幅度受估计方法影响较大。其中IV回归的点估计值显示，养老金个人缴费率上升1个百分点，消费下降1.75%，意味着养老金缴费的上升基本都转化成了消费的下降，储蓄变化很小。在2002―2005年间，家庭的职工养老金缴费率上升了1.1个百分点（从4.8%上升到5.9%），而平均的家庭消费倾向（消费/缴费前收入）下降了4.9个

Ungrouped

Juheng has been providing technological innovations to the world for more than 30 years. Operating on a global scale, Juheng designs and manufactures products ranging from ships, railroad cars, aircraft and motorcycles to industrial plants, steel structures and hydraulic equipment.

The Juheng with over 30 years of experience in the hydraulics industry. Juheng manufactures and markets a wide variety of hydraulic products for mobile, industrial, marine and other applications. Dedicated to global customer support, Juheng has operations in Japan, England, Korea, India, China and the United States. In addition to total commitment toward global customer support, Juheng places great emphasis on quality. The hydraulic production facilities in Japan and England are ISO 9001 and 14001 certified.

Juheng's Ungrouped have been continuously developed to meet the needs of the era such as high efficiency, low noise and compact design.

Juheng offers variety of variable displacement Ungrouped that have high performance, reliability and ease of maintenance.
Juheng focuses on understanding and assessing customers` ever-changing needs. By taking full advantage of Juheng`s worldwide resources, our staff works closely with factory engineers to provide technical solutions to customers` hydraulic requirements. Through technological leadership and a customer-centered philosophy, Juheng Precision Machinery has built relationships with top companies around the world who value the high performance and reliability that Juheng Hydraulic Products deliver.

High Quality Excavator Parts,High Quality Ungrouped,Good Excavator Parts

Jining Juheng Hydraulic Machinery Co., Ltd. , https://www.jhexcavator.com