Exchange Rate Change, Factor Market Distortion and Company Performance

3.1 Data Description

This paper mainly adopts the matched data of China’s national tax survey and customs from 2007 to 2011. Tax survey data is collected by China’s State Taxation Administration and Ministry of Finance to better monitor the tax base and assess the impact of tax policies. Therefore, the data are of high quality and accuracy (Liu and Mao, 2019). The database provides detailed financial and tax information at the company level, including 100-plus financial and corporate variables from balance sheets, income statements and cash flow statements. We process the data in the following manner. First, since we mainly assess how exchange rate changes affect manufacturing enterprises through imported intermediate goods, we retain only manufacturing companies in our sample; second, we remove observations with missing key information such as total assets, total liabilities, sales and fixed assets as well as a sample of noisy companies that clearly do not comply with accounting standards; third, since China’s national economy industry codes were changed in 2011, we adjusted the industry information in 2011 according to the corresponding change to make it consistent with the industry classification from 2007 to 2010. ^[1]

The purpose of this paper is to investigate the effect of exchange rate changes on the performance of companies with different proportions of imported intermediate goods. To depict the exchange rate shock more accurately to each enterprise, we also use China’s customs database to construct the REER of imported intermediate goods at the company level based on the annual imports of companies. China’s customs database records information on every import or export transaction by company, including transaction amount, number of transactions, the HS code of a product, trade mode, destination country, etc. By matching the HS codes with the BEC codes, ^[2] we can distinguish imports into capital goods, intermediate goods and consumer goods. In this paper, we focus on the cost channel of imported intermediate goods where the exchange rate has an impact on the performance of companies. Therefore, in constructing the REER of companies, we calculate it based on their import of intermediate goods from countries.

Since the national tax survey data and the customs database use different enterprise codes, we also need to match enterprise codes of the two databases. Referring to Yu (2015), we matched the tax survey data with the customs database. Finally, we obtained 143490 company-year observations for 62865 companies. The sales of enterprises in the matched data account for 45% of that of all enterprises in the tax survey database. The imports of intermediate goods by enterprises in the matched sample make up approximately 25% of the total imports of intermediate goods in the customs database. The matched data are highly representative.

3.2 Real Effective Exchange Rate

The purpose of this paper is to examine how exchange rate changes affect the revenue performance of enterprises through the cost channel of imported intermediate goods. To accurately describe the exchange rate changes faced by each enterprise, we draw on the approach of Hongbin Li et al. (2011) to construct an enterprise-level REER (reer_it). reer_it faced by the company i in period t is defined as follows:

where, E_jt indicates the exchange rate of the currency of j to RMB in the period t. We use the indirect quotation method, i.e., the foreign currency that one unit of RMB can be exchanged for. E_j0 denotes the exchange rate between the currency of j and RMB in the base period that is set in 2007 in this paper. CPI_t refers to China’s consumer price index in the period t (set at 100 for the base period of 2007) and CPI_jt the consumer price index for j in the period t (set at 100 for the base period of 2007). w _ijt represents the proportion of the amount of intermediate goods imported from the country j to the total amount of intermediate goods imported by the enterprise i in the t period. ^[1] When companies make production decisions, it is possible that they may involve changes in import countries. This affects the import weights of each country and thus the REER at the enterprise level, leading to the endogeneity problem in the regression (Dai et al., 2013; Dai and Xu, 2017; Wang and Lu, 2018). To avoid the possible endogeneity problem, we refer to Wang and Lu (2018). In the benchmark regression, we set w_ijt as the mean value of each enterprise in the sample interval. To put it differently, the import share does not change over time. By definition, an increase in reer_it indicates that an enterprise is facing an appreciation of the national currency, while a decrease in reer_it suggests that an enterprise is facing a depreciation of the REER. The bilateral exchange rate data and the consumer price index used in this paper are obtained from the International Monetary Fund’s International Financial Statistics Database. From 2007 to 2011, the mean values of the logarithm of enterprises’ REER in the sample were 4.605, 4.692, 4.732, 4.718 and 4.739, respectively. During the period, the REERs of companies in the sample were appreciated by roughly 13.4%. In the regression equation, the REER is taken in logarithmic form.

3.3 China’s Factor Market Distortion

Referring to the research by Gao et al. (2018), we use the difference between the regional factor market development and the benchmark factor market development as an indicator to measure factor market distortions. This indicator can reflect both the spatial change in factor market distortions and the dynamically temporal characteristics of factor market distortions by region. The formula is as follows:

where p denotes the province and t the year; FMD_pt is the factor market distortion at the province level; factormarket_pt is the factor marketization index at the province level; Max{factormarket_pt} denotes the maximum value of the factor marketization index in each province during the sample period.The marketization index of factors is derived from the Marketization Index of China’s Provinces by Wang et al. (2018). In 2007, the regions with the lowest factor market distortion were Shanghai, Tianjin and Beijing, while those with the highest were Xizang, Xinjiang and Guizhou.

3.4 Proportion of Imported Intermediate Goods

The proportion of imported intermediate goods is defined as the share of the annual amount of imported intermediate goods in that of intermediate goods input by companies. The amount of imported intermediate goods can be calculated from the customs database. The annual amount of intermediate inputs of enterprises can be obtained from the tax survey database. It is worth noting that the amount of imported intermediate goods is denominated in US dollars, while that of intermediate goods inputs is denominated in RMB. We convert the amount of imported intermediate goods into RMB-denominated prices based on the average annual exchange rates of RMB and US dollars.

3.5 Other Variables

The dependent variable in this paper refers to companies’ performance. In the regression, we use sales and total factor productivity as measures of their performance. Companies’ sales are defined as the logarithm of their revenue, and total factor productivity is estimated using the method of Levinhson and Petrin (2003). The main idea is to use intermediate goods as a proxy variable for productivity and performing a two-stage regression, thus addressing the simultaneity bias that may result directly from the estimation of the OLS method.

Moreover, we control a range of company-level variables in the regression, including export intensity, capital intensity, wages and number of employees. Export intensity is defined as the ratio of a company’s exports to revenue for that year. Capital intensity is defined as the division of a company’s total fixed assets for that year by the number of employees. Wage is defined as the division of total wages by the number of employees. Capital intensity, wages and number of employees are taken in logarithmic form in the regression equation.

3.6 Descriptive Statistics

Table 1 demonstrates the descriptive statistics of key variables in this paper.

3.7 Regression Equation Setting

To study the impact of exchange rate changes on the performance of import companies and distinguish heterogeneous responses of companies located in regions with different factor market distortions, we set the regression equation as follows:

Wherein i denotes an enterprise, t the year, and Perform_it the enterprise performance index. In the regression equation, we use sales and total factor productivity to represent a company’s performance. reer_it denotes the company-level REER, and FMD_p denotes the provincial factor market distortion indicator. To mitigate the endogeneity problem, we take the factor market distortion for each region at the beginning of the sample (2007). X_i,t Enterprise-level control variables include export intensity, capital intensity, enterprises’ wages and number of employees. μ_i denotes company fixed effects. θ_jt denotes the industry (CIC 2-digit code) × year fixed effect. δ_t represents year fixed effects. Standard error clustering is at the import enterprise level.

4.1 Baseline Results

The table reports this paper’s baseline results. Columns (1) and (2) report the regression results when the dependent variable is sales of an enterprise. In column (1), we analyze the relationship between the REER and an enterprise’s sales. The results show that the coefficient of REER is significantly positive at the 10% level. For every 10% appreciation of the REER, corporate sales increase by an average of 0.66%. In considering the median corporate sales of RMB 69 million, a 0.66% increase in sales corresponds to approximately RMB 460000. From the results of control variables, the coefficients of export intensity, capital intensity, enterprise wages and number of employees are significantly positive, indicating that enhancement of such variables significantly contributes to the sales expansion of enterprises.

In column (2), we further examine the effect of heterogeneity in the REER. Specifically, we study the differential impact of the REER on the performance of companies located in regions with different factor market distortions. We include in column (2) a cross term between the REER and the factor market distortion in each province. To mitigate the endogeneity problem, we select factor market distortions for all provinces at the beginning of the sample (2007). The results show that the cross-term coefficient of the REER and the factor market distortion are positive and significant at the 1% level. In other words, the enhancing effect of RMB appreciation on enterprise performance is more pronounced in regions with high factor market distortions. The result is equally significant in an economic sense. Take the coefficients in column (2) as an example. When RMB appreciates by 10%, the sales of companies in a province rise by an average of 1.52% (-0.105 × 0.1 + 0.431 × 0.596 × 0.1 = 1.52%) for the province with factor market distortions in the 90th percentile (factor market distortion = 0.596). For provinces with factor market distortions in the 50th percentile (factor market distortions = 0.317), the sales of enterprises in that province rise by an average of 0.32%. For provinces with factor market distortions in the 10th percentile (factor market distortions = 0.042), the sales of companies in that province fall by an average of 0.87%. This result suggests that the positive impact of RMB appreciation on the sales of enterprises in regions with higher factor market distortions.

Columns (3) and (4) report the regression results when the dependent variable is total factor productivity. The results in column (3) show that the coefficient of REER is significantly positive at the 5% level, indicating that RMB appreciation also contributes to the increased total factor productivity of enterprises. The results in column (4) show that the cross-term coefficient of REER and factor market distortion are also significantly positive, indicating that the effect of RMB appreciation on companies’ total factor productivity is stronger in regions with high factor market distortions. Regarding the coefficient of control variables, the coefficients of export intensity, capital intensity and companies’ wages are significantly positive, while the coefficient of number of employees is significantly negative.

The results in Table 2 are consistent with expectations. Factor market distortions raise the cost of domestic intermediate goods by reducing the efficiency of resource allocation. Consequently, companies are less likely to purchase domestic intermediate goods and more likely to import them. As such, through the import channel, the effect of higher REER on higher enterprise performance become stronger in regions with high factor market distortions.

4.2 Channel Test

In previous regressions, we found that RMB appreciation results in the higher performance of companies and this phenomenon is concentrated in regions with high factor market distortions. According to the previous analysis, one of the channels at play is the effect of domestic factor market distortions on companies’ imported intermediate goods decisions: factor market distortions make companies more inclined to use imported intermediate goods by raising the cost of producing domestic intermediate goods. The higher the proportion of imported intermediate goods, the greater the impact of import exchange rate changes on companies’ performance. Therefore, through the import channel of intermediate goods, the effect of higher REER on higher enterprise performance is strong in regions with high factor market distortions. This subsection provides an empirical test of the channel.

4.3 Expansion Test: The Impact of Exchange Rate Changes on Upstream Enterprises

The previous analysis has shown that the national currency appreciation promotes enterprise performance by cutting the cost of imported intermediate goods. However, the national currency appreciation may also cause domestic enterprises to rely excessively on imported intermediate goods, thereby reducing the use of domestic intermediate goods. If this hypothesis holds, then the national currency appreciation may worsen the performance of domestic upstream enterprises that provide intermediate goods. To verify whether this phenomenon exists, we include domestic production networks in our empirical analysis and examine how exchange rate changes faced by downstream enterprises affect upstream enterprise performance. Referring to Liu and Wang (2020), we define the exchange rate change faced by an upstream enterprise as the weighted average of the exchange rate changes faced by each downstream industry of that enterprise, where the weight is the proportion of intermediate goods in each (upstream) industry that are allocated to each downstream industry. ^[1]

Specifically, we first measured the change in the REER by industry (CIC-4 digit). The REER at the industry level is defined as the average of the REERs (logarithm) of all companies in the industry. Based on the input-output table, we can measure the proportion of intermediate goods from each (upstream) industry that are allocated to each other industry. The higher the proportion, the more intermediate goods are allocated to the downstream industry, and the greater the impact of REER changes faced by that downstream industry on enterprises in that upstream industry. By using the allocation ratio as a weight, we calculate the (downstream) changes of REER faced by each upstream industry based on the REER of each downstream industry. The changes are defined as the weighted average of REER of each downstream industry that has a production network association with that upstream industry.

Table 5 demonstrates the impact of exchange rate changes on the performance of upstream companies. The core explanatory variable is the downstream exchange rate. In accordance with the previous analysis, the downstream exchange rate faced by each enterprise is the weighted average of REERs of each downstream industry of that enterprise’s sector, where the weight is the share of intermediate goods produced by the industry that are allocated to each downstream industry. If our previous hypothesis holds that the national currency appreciation causes downstream industries to be overly dependent on imported intermediate goods, which in turn reduces the demand for intermediate goods from upstream companies and depresses the performance of upstream companies, then we should observe a significantly negative downstream exchange rate coefficient.

We use sales, profitability, value added and total factor productivity to measure enterprise performance. The results in Table 5 show that the downstream exchange rate coefficients are all significantly positive. When the downstream exchange rate increases, the performance of upstream companies improves, as evidenced by sales expansion, increased profitability, and significant increases in value added and total factor productivity. The economic intuition is that the national currency appreciation first promotes improved enterprise performance by reducing the co st of imported intermediate goods. The improved performance of downstream companies spills over to upstream industries through production network effects (Barrot and Sauvagnat, 2016), which in turn improves the performance of enterprises in upstream industries.