2.1. Deregulation and increasing market competition

We motivate our study by first examining changes in the degree of market competition in 28 industrial sectors in China, for the period from 1998 to 2006. In Figure 1, each line represents the sectoral Herfindahl–Hirschman Index (HHI) of a specific two-digit sector, with the HHI of 1998 being normalized to 1 for all sectors.Footnote ⁵ Out of 28 sectors, only 4 sectors have a HHI that is greater than 1 in 2005, which means that other 24 sectors (86.7% of total) in China’s economy have experienced higher degrees of market competition in 2006 compared to 1998.Footnote ⁶ In addition, a breakdown

The increasing market competition in China can also be confirmed by examining the changes of sectoral average markups for the period between 1998 and 2006.Footnote ⁷ Figure 2 plots the average markup of each sector for the period between 1998 and 2006, where each line represents a specific two-digit sector. As the figure clearly shows, all 28 sectors have experienced significant declines in average markups during the aforementioned period, which is also evidence of higher market competition in China’s Economy.

The findings reported in Figures 1 and 2 suggest that the degree of market competition in China’s industrial sectors has substantially increased between 1998 and 2006. In the rest of the section, we examine the potential causes of this phenomenon.

2.2. The declines of SOE’s share in the upstream and downstream sectors

Coincided with the increasing market competition is the substantial decline in the share of State-Owned Enterprises (SOE). As Table 1 shows, while both upstream and downstream sectors have witnessed similar large declines in SOE’s share, the upstream sectors in China have noticeably higher HHI than the downstream sectors.Footnote ⁸ This indicates that the degree of market competition is lower in the upstream sectors.

Figure 1. Deregulation in China: changes in sectoral Herfindahl–Hirschman Index.

Figure 2. Deregulation in China: changes in sectoral average markups.

Another interesting and important observation is that SOEs in the upstream sector on average have substantially lower productivity than non-SOEs, whereas such differences in productivity do not exist in the downstream sector. To see this, we first compute TFP at the firm level and then normalize each firm’s TFP to its respective sector’s median TFP. Doing so allows us to consistently compare the productivity differences between SOEs and non-SOEs across different sectors. The results are summarized in Table 2. For instance, in 1998 the SOE’s TFP was 31% lower than the Non-SOE’s TFP in the upstream sector while the TFPs of the SOEs and Non-SOEs are much closer to each other in the downstream sector. Similar patterns of TFPs are also observed in other years during 1998–2006.

The findings from Table 2 highlight the important differences between the upstream and downstream sectors in China.Footnote ⁹

2.3. Notable events leading to deregulation

Given the facts outlined in the previous sections, one might ask whether we can point to specific events or policies that were directly responsible for the deregulation in China. Although we do not believe that a single policy or event had led to all the deregulation we have outlined previously, we argue that a number of policies can be viewed as sources of deregulation. In this section, we give a brief discussion of these events or policies.

First, between 1995 and 2002, the Chinese government had attempted to reform the massive SOE sector by instituting a policy known as “Guan Ting Bing Zhuan.” Quite literally, it means that the government asked a large number of SOEs to shut down, pause production, merge with other SOEs, or change production. As a result of this policy, large number of SOEs had disappeared and large number of workers had been laid off. Many believe that this reform represented a pivotal moment for SOEs’ development in China.

Table 1. Share of SOE and degree of competition (weighted by revenue)

Table 2. % Difference of SOE’s TFP and non-SOE’s TFP

Note: these numbers are referring to the TFP difference between SOE and non-SOE as a percentage of non-SOE’s TFP.

Closely related to the “Guan Ting Bing Zhuan” reform, the Chinese government in 1997 proposed a strategy called “Zhua Da Fang Xiao,” which is widely known as “grabbing the large ones and letting go the small ones.” Essentially, the goal of this strategy is to correct the fact that there were too many SOEs in too many industries, which created two main problems. First, a large number of SOEs means most SOEs were small with inefficient scale of production. Second, the presence of SOEs in almost all industries reduced the level of specialization. While SOEs could play a vital role in certain industries (e.g., industries deemed “strategic”), there was no doubt that non-SOEs were more competitive in many industries (e.g., manufacturing sectors such as textile or electronics). As a response to these two problems, the Chinese government had tried to make small SOEs to either merge with other small SOEs to take advantage of scale economies or exit the industries that SOEs were not competitive.

Perhaps more on the idealogy side and less on the policy side, the National Congress of China had made an important amendment to the constitution, clearly stating that “the non-public sector is an important component of the socialist market economy” (to be added: when). Although there were no immediate policies following the amendment, this event was significant to the development of China’s market economy as it provided the necessary foundation for the proliferation of the non-SOEs in China’s many industries.

As a milestone of China’s “reform and open up” policy, in 2001 China formally became a member of the World Trade Organization (WTO). In order to join the WTO, China had accepted a large number of conditions and promised to open its markets and reduce tariffs. It is difficult to overstate the importance of China’s entry to WTO. While our paper does not specifically model this event, our results do partly capture the profound impact of China’s entry into WTO on the Chinese economy.

3.1. Household

We consider a model inhabited with infinitely lived individuals of measure one. Time is discrete and denoted by $t\in \{0,1,\ldots,\infty \}$ . We assume that individuals are endowed with one unit of labor in each period and supply it inelastically. They make decisions of consumption and saving to maximize their lifetime discounted utility which are specified as,

\begin{equation*} \sum _{t=0}^{\infty } \beta ^{t} \frac {c_{t}^{1-\sigma }}{1-\sigma } \end{equation*}

where $\beta$ is the time discount factor.

The individual solves the following problem:

\begin{equation*} \max _{\{c_{t},k_{t+1}\}^{\infty }_{t=0}} \sum _{t=0}^{\infty } \beta ^{t} \frac {c_{t}^{1-\sigma }}{1-\sigma } \end{equation*}

subject to

\begin{equation*} c_{t} + a_{t+1} = w_{t} + (1+r_{t})a_{t} + N\pi _{mt}, \end{equation*}

where $c_{t}$ denotes consumption, $a_{t}$ is asset holdings, and $N\pi _{mt}$ is the upstream firms’ profits. $w_t$ and $r_t$ are representing wage rate and interest rate, respectively.

3.2. Production

The production side of the model features a vertical structure. That is, there are two sectors, the upstream and downstream sectors, with the upstream sector producing intermediate goods which in turn are used in the production of final goods in the downstream sector. In the following, we describe each of the two sectors respectively.

3.2.1. Downstream sector

The downstream sector is of perfect competition and features a representative firm. The downstream firm produces according to a Cobb-Douglas technology with capital, labor, and intermediate goods as its inputs, and the output can be used as either final consumption good or capital. Specifically, the representative firm solves the following profit-maximization problem:

\begin{equation*} \max _{K_{dt},L_{dt}, M_{t}} \pi _{dt} = A_{dt}\!\left(K_{dt}^{\alpha }L_{dt}^{1-\alpha }\right)^{\theta }M_{t}^{1-\theta } - (r_{t}+\delta )K_{dt} - w_{t}L_{dt}-P_{mt}M_{t}, \end{equation*}

where $K_d$ and $L_d$ are the capital and labor used in the downstream sector. Here, $M$ is intermediate goods with $P_m$ representing its price, and $A_d$ is the TFP in the downstream sector.

The profit-maximizing behaviors of the firm imply that,

\begin{align} w_t &=(1-\alpha )\theta A_{dt}\!\left(K_{dt}^{\alpha }L_{dt}^{1-\alpha }\right)^{\theta -1}M_{t}^{1-\theta }\!\left(\frac {K_{dt}}{L_{dt}}\right)^{\alpha } \end{align}

\begin{align} r_t & =\alpha \theta A_{dt}\!\left(K_{dt}^{\alpha }L_{dt}^{1-\alpha }\right)^{\theta -1}M_{t}^{1-\theta }\left(\frac {K_{dt}}{L_{dt}}\right)^{\alpha -1}-\delta \end{align}

\begin{align}P_{mt} &=(1-\theta ) A_{dt}\!\left(K_{dt}^{\alpha }L_{dt}^{1-\alpha }\right)^{\theta }M_{t}^{-\theta } \end{align}

3.2.2. Upstream sector

To capture the highly regulated feature of China’s upstream sector, we assume that the upstream sector in the model features Cournot competition.Footnote ¹⁰ Specifically, we assume that there exist $N$ symmetric firms in the upstream sector. Each of these firms produces a homogenous intermediate good, which can be used in the production of final goods in the downstream sector.

Denote $q_{t}$ as an upstream firm’s output at $t$ which is produced according to $q_{t}= A_{mt}k_{mt}^{\alpha }l_{mt}^{1-\alpha }$ and $P_{mt}$ as the price of intermediate good (or the output of the upstream sector) at $t$ . An upstream firm solves the following problem:

\begin{equation*} \max _{q_{t}} \pi _{mt} = P_{mt}q_{t} - (r_t+\delta )k_{imt}-w_tl_{mt}, \end{equation*}

with $q_{t}= A_{mt}k_{mt}^{\alpha }l_{mt}^{1-\alpha }$ .

It is important to note that the upstream firms engage in Cournot competition when making their production decisions. That is, at each period $t$ , upstream firms internalize the fact that its choice of $q_{t}$ will impact the price of intermediate good $P_{mt}$ and will choose $q_{t}$ that maximizes its profit $\pi _{mt}$ . Upstream firms’ optimal choices are determined in the equilibrium, which we will discuss further at the end of this section.

3.3. Competitive equilibrium

A competitive equilibrium consists of prices $\{w_{t},r_{t},P_{mt}\}^{\infty }_{t=0}$ and allocations for the representative household $\{c_{t},a_{t+1}\}^{\infty }_{t=0}$ , for upstream firms $\{k_{mt},l_{mt}\}^{\infty }_{t=0}$ , and for the downstream firm $\{K_{dt},L_{dt},M_{t}\}^{\infty }_{t=0}$ such that:

1. (a) Given prices $\{w_{t},r_{t}\}^{\infty }_{t=0}$ and dividends from the upstream firms $\{\pi _{mt}\}^{\infty }_{t=0}$ , allocations $\{c_{t},a_{t+1}\}^{\infty }_{t=0}$ solve the household’s problem.

2. (b) Given prices $\{w_{t},r_{t},P_{mt}\}^{\infty }_{t=0}$ , allocations $\{K_{dt}, L_{dt}, M_{t}\}^{\infty }_{t=0}$ solve the downstream firm’s problem.

3. (c) Given prices $\{w_{t},r_{t}\}^{\infty }_{t=0}$ and the number of upstream firms $N$ , allocations $\{k_{m,t}, l_{m,t}\}^{\infty }_{t=0}$ solve individual upstream firms’ problem.

4. (d) Markets clear:

   Intermediate Goods Market:

   \begin{equation*} Nq_{t} = M_{t}; \end{equation*}
   Labor Market:
   \begin{equation*} 1 = L_{dt} + Nl_{mt}; \end{equation*}

   Capital Market:

   \begin{equation*} a_{t} = K_{dt} + Nk_{mt}. \end{equation*}

3.4. Equilibrium analysis

The detailed analysis of the upstream and downstream firms’ optimization problems in the equilibrium can be found in the appendix. Their optimizing behaviors imply that in the equilibrium, the aggregate amount of labor employed in the upstream sector is given by

\begin{eqnarray*} L_{mt} & = & \left (1 - \frac{1}{N}\right )(1-\theta ) \end{eqnarray*}

where $L_{mt}=Nl_{mt}$ , and $l_{mt} = \left (\frac{N-1}{N^{2}}\right )(1-\theta )$ . Let $\Upsilon$ denote the share of aggregate labor allocated to the downstream sector. It is easy to see that $\Upsilon =1-L_{mt}$ .

Note that a key component of the production side is the Cournot competition among $N$ symmetric firms in the upstream sector. Here, we consider a few special cases of the Cournot competition to gain further understanding of the problem.

3.4.3. The degree of competition in the upstream market

One common measure of the degree of competition in a given market is the HHI. It is defined as

\begin{equation*} \textit{HHI} = \sum _{i=1}^{n}s_{i}^{2}, \end{equation*}

where $s_{i}$ is firm $i$ ’s output share of the industry. In the case of symmetric firms as in this model, $s_{i} = 1/N$ for all $i$ . Therefore, the HHI implied in the model is simply $HHI=\frac{1}{N}$ .

4.1. Calibration

We calibrate the model to the Chinese economy. Specifically, our calibration strategy consists of two steps. In the first step, we predetermine the values of some standard parameters based on the existing literature. In the second step, we calibrate the rest of the parameters to match key moments of the Chinese economy from 1998 to 2006. We discuss the details of our calibration strategy in the following.

One period is assumed to be one year. The utility function is assumed to take the following form: $u(c)=\frac{c^{1-\sigma }}{1-\sigma }$ where $\sigma$ is set to 2.0. The subjective time discount factor $\beta$ is set to 0.95. The capital depreciation rate $\delta$ is set to 10% and the capital share $\alpha$ is set to 0.5 based on the estimates in Bai et al. (Reference Bai, Hsieh and Qian2006), and Song et al. (Reference Song, Storesletten and Zilibotti2011). The TFPs $A_{mt}$ and $A_{dt}$ in 1998 are normalized to one.

Our main goal is to study the impact of deregulation on the Chinese economy from 1998 to 2006. Thus, we assume that the initial steady state of the benchmark model mimics the Chinese economy in 1998 with the value of $\theta$ , which is the value-added share of the downstream sector, chosen to match its counterpart in 1998.

The key step in our calibration procedure is to obtain reasonable estimates of the level of competition in the upstream sector in 1998 as well as that in 2006. To do this, we set the value of $N_{1998}$ , which is the number of firms in 1998 as well as the inverse of the HHI in 1998, so that the labor share of the upstream sector in the benchmark matches that in the data (0.9353). The rationale behind this calibration strategy is that this labor share is directly impacted by the level of competition in the upstream sector in the model. To be specific, as a higher level of competition will lead to higher output in the sector, the employment share of the upstream sector will also be higher. Lastly, we calculate the percentage change of HHI in China’s upstream sector in the data. We then obtain the value of $N_{2006}$ , which is the number of firms in 2006, by applying the same percentage change.

The key parameter values are summarized in Table 3.

Table 3. Parameter values and calibration

4.2. Comparisons of steady state

A comparison of two steady states ( $N_{1998}$ and $N_{2006}$ ) is summarized in Table 4. As the table shows, the differences in the level of competition in the upstream market can lead to about 43% differences in steady states.

Table 4. Comparisons of steady state

Of course, comparing steady states is somewhat misleading as China is clearly not on steady states. To further examine the impacts of deregulation as well as the importance of the vertical structure, we conduct two counterfactual exercises in which we vary upstream sector’s level of competition and the structure of the model.

In order to conduct the two counterfactual exercises, we first need to derive the evolution of TFP growth between 1998 and 2006. For this purpose, we assume the following: the Chinese economy was at a steady state with $N = N_{1998}$ in 1998, and the number of firms in the upstream market increased linearly to $N = N_{2006}$ in 2006 and will remain at that level after 2006. We assume that the economy takes 50 years to transition to the new steady state. With the assumptions above, we can derive the average TFP growth rate from 1998 to 2006, which was 4.12%.

4.3. Counterfactual 1: the importance of increased competition in the upstream sector

To examine the importance of increased competition in the upstream sector, we simply run a counterfactual exercise where the only difference with the benchmark case is that the number of firms did not increase between 1998 and 2006 (and keep the evolution of TFP derived from calibration).

Comparing the evolution of real GDP between the benchmark case and the counterfactual exercise lets us examine the effect of deregulation in the upstream sector. The quantitative results are included in Table 5. We can see that, assuming no changes in the upstream sector has occurred, the percentage growth rate per year has decreased from 8.99% to 7.18%, which is about 20% drop in annual growth rate. This result suggests that the increased competition in the upstream sector can account for about one-fifth of China’s growth between 1998 and 2006.

Table 5. Comparisons of transition path

4.4. China’s high returns to capital

Song et al. (Reference Song, Storesletten and Zilibotti2011) have documented an increase of the returns in capital during the transition period of China’s economy. Their paper has provided an explanation, in which the authors argue that financial frictions were the main reason that not only that returns to capital did not decrease has China accumulated more capital, but increased instead. In this paper, we provide an alternative explanation. Namely, the increase in competition in the upstream sector can sustain a period of high return to capital despite the increase in aggregate capital stock.

To see this, we compare calculate the returns to capital assuming deregulation occurred in the upstream sector. Figure 3 shows this case. If this did not happen, returns to capital in China would simply remain at a constant level.

Figure 3. Returns to capital: deregulation in the upstream sector.

4.5. Further discussion

Our main focus in this paper is on the role of the deregulation in upstream sectors in understanding China’s growth experience, with special attention to its spillover effects on downstream sectors. To keep the model tractable, we have abstracted away from modeling some other potential drivers of economic growth during this period of time. For instance, our model cannot separately identify the effects of increased competition in downstream sectors on economic growth, as it features a perfectly competitive downstream sector. However, our model does not necessarily exclude these other drivers of growth that we did not specifically model. Note that our quantitative strategy assumes the residual (i.e. the part of growth cannot be accounted for by upstream deregulation) is all attributed to TFP growth. Thus, the other possible drivers we do not explicitly model, such as increased competition in downstream sectors, are implicitly captured by this residual (TFP growth), which accounts for a major share (i.e. 80%) of growth between 1998 and 2006 in our benchmark model.

One concern about our current modeling strategy is that we equally attribute the TFP change between downstream and upstream sectors, while other possible drivers of growth, such as increased competition in downstream sectors, are more likely to concentrate in the downstream sector. To address this concern, we conduct additional robustness check exercises in this section. Specifically, we consider an alternative model in which we assume that the TFP change consists of two components: (1) a component equally attributed across sectors, and (2) a component only occurring in downstream sectors to mimic the effect of increased competition in this sector. Quantitatively, we set the value of the second component of TFP change so that its effect on growth is the same as that of deregulation in the upstream sector. We then attribute the rest of growth to the first component of TFP change (the residual). Compared to our benchmark model, this alternative model features extra TFP changes in the downstream sector to capture the increased competition in it. In this alternative model, we redo the same experiments and re-calculate the same statistics as in our benchmark model. We find that the effect of deregulation in upstream sectors on growth between 1998 and 2006 remains similar. In addition, the alternative model delivers similar implications for returns to capital over time.Footnote ¹²