Data and Methodology

3.1 Specifying Export and Import Functions

According to the imperfect substitutes model of Goldstein and Khan (1985), export and import functions can be represented as:

where ext represents real exports, rert represents the real exchange rate, yt* represents foreign real income, imt represents real imports, yt represents domestic real income, and all variables are measured in natural logs.

In the case of the PRC's processing trade, it is necessary to modify these equations. Below I consider some of the other factors that should affect imports for processing and processed exports.

For imports for processing, the International Monetary Fund (IMF) (2005) noted that the price elasticity should be small because the intermediate goods are not produced domestically, resulting in little potential for import substitution. However, the huge surpluses in processing trade that have emerged since 2005 suggest that firms have been able to source more intermediate goods from within the PRC. Thus the demand for imports for processing may have become more price elastic in recent years.

The IMF (2005) also argued that imports for processing should vary one-for-one with processed exports. Imports for processing should thus flow elastically into the PRC in response to an increase in the demand for processed exports in the rest of the world. Processed exports are therefore included as a right hand side variable to explain imports for processing. Since imports for processing are not intended for the domestic market but only for the assembly of processed exports, the preferred specification below includes processed exports but not PRC income.

Foreign direct investment (FDI) flows and multinational corporations (MNCs) also play important roles in processing trade (see Gaulier, Lemoine, and Unal-Kesenci 2005). As discussed above, 84% of the PRC's processed exports in 2006 were produced by foreign-invested enterprises (Feenstra and Wei 2009). FDI is thus included as a right hand side variable. As Marquez and Schindler (2007) noted, the effect of FDI on imports could be positive or negative depending on whether the investment generates substitution effects or complementary effects.

Following previous authors (e.g., Garcia-Herrero and Koivu 2007), a World Trade Organization (WTO) dummy variable is included as a right hand side variable. The PRC's WTO accession may have given foreign firms more confidence to enter into longer-term relationships with PRC firms. Garcia-Herrero and Koivu (2007) posited that the PRC's WTO accession began affecting the PRC's trade after it became certain that the PRC would join the WTO in the beginning of 2000. The WTO dummy variable is thus set equal to one beginning in 2000.

For processed exports much of the value-added comes from imported inputs, especially inputs produced in other East Asian countries. A generalized appreciation in East Asia would thus have a larger effect on the costs of the PRC's processed exports measured in the importing country's currency than a unilateral appreciation of the CNY. A unilateral appreciation would only change the relative foreign currency cost of the PRC's value-added in processed exports.5 An integrated exchange rate is thus included that weights exchange rate changes in supply chain countries by the countries' value-added in processing trade.

3.2 Constructing an Integrated Exchange Rate

Following Tong and Zheng (2008), the PRC's value-added in processing trade can be measured as the difference between the value of the PRC's processed exports (VPEt) and the value of imports for processing from all supply chain countries (ΣiVIPi,t):

where VAChin,t equals the PRC's value-added in processing trade. Annual data on the total value of processed exports and the total value of imports for processing is used to calculate the PRC's value-added.

To calculate the value-added in supply chain countries this paper focuses on the nine leading providers of imports for processing to the PRC. These are Germany, Japan, the Republic of Korea (hereafter Korea), Malaysia, the Philippines, Singapore, Taipei,China, Thailand, and the US. For these suppliers weights (w_i,t) are calculated by dividing their contribution to PRC imports for processing by the amount of imports for processing coming from the nine major suppliers together. These weights are used to calculate a weighted exchange rate (wrer_j,t) between the PRC and each country j that purchases processed exports from the PRC by calculating the inner product of the weights and the bilateral real exchange rates between the countries supplying imports for processing and country j:

where rer_i,j,t is the bilateral real exchange rate between supply chain country i and country j purchasing the final processed exports.

The weighted exchange rate wrer_j,t is then combined with the bilateral exchange rate between the PRC and country j (rer_Chin,j,t) to calculate a single integrated exchange rate (irer_j,t) measuring how exchange rate changes affect the entire cost of the PRC's exports of processed goods to country j:

To calculate irer in this way it is necessary to measure exchange rates using a common numeraire. This can be done by employing the real exchange rate variables constructed by the Centre D'Etudes Prospectives et D'Information Internationales (CEPII). The CEPII real exchange rate between countries i and j is calculated by first dividing gross domestic product (GDP) in US dollars for country i by GDP in purchasing power parity (PPP) for country i and doing the same for country j. The resulting ratio for country i is then divided by the ratio for country j. This variable measures the units of consumer goods in country i needed to buy a unit of consumer goods in country j. It can be compared across countries as well as across time. Because it is comparable across countries, it can be used in equation (2) to calculate irer. Higher values of wrer and irer represent stronger exchange rates in the PRC and in supplier countries.

The other independent variables are the PRC's capital stock in manufacturing, the stock of FDI, and a WTO dummy variable. Cheung, Chinn, and Fujii (2010) have found that the PRC capital stock helps to explain PRC exports. As discussed above, the FDI stock and the PRC's WTO accession may help to explain the increase in processing trade.

The dependent variables are PRC imports for processing and PRC processed exports. These are obtained from the PRC's Customs Statistics. Following Cheung, Chinn, and Fujii (2010), the Hong Kong, China to PRC re-export unit value index is used to deflate the PRC's imports and the Hong Kong, China to US re-export unit value index is used to deflate the PRC's exports. The data are discussed in more detail in the Data Appendix [ PDF 15.2KB | 1 page ].

3.3 The Econometric Model

Panel A of Table 2 [ PDF 34.8KB | 2 pages ] reports the results from a battery of panel unit root tests.⁶ Column (1) presents the Im, Peseran, and Shin W-statistic, column (2) the asymptotically distribution free (ADF) Fisher Chi-square statistic, column (3) the Phillips-Perron Fisher Chi-square statistic, column (4) the Levin, Lin, and Chu t-statistic, and column (5) the Hadiri heteroscedastic consistent Z-statistic. For the first four tests, the null hypothesis is that the variable has a unit root while for the fifth test the maintained hypothesis is that the variable is stationarity. In most cases the results indicate that the series have unit roots. Panel unit root tests are not conducted for the series with no cross-sectional variation (i.e., the PRC capital stock, PRC inward FDI, and PRC income).

Panel B of Table 2 reports the results of the Kao residual cointegration test.⁷ For both the export and the import equation the results indicate that the null hypothesis of no cointegration can be rejected. Panel dynamic ordinary least squares (DOLS) estimation, a technique for estimating cointegrating relations, is thus employed.

DOLS involves regressing the left hand side variable on a constant, the right hand side variables, and lags and leads of the right hand side variables. The individual import equations have the form:

Here im_i,t represents real imports for processing from country i to the PRC, irer_i,t represents the integrated real exchange rate, rgdp_C,t equals real income in the PRC, tex_t represents the PRC's total real processed exports to the world, FDI_t denotes the stock of foreign direct investment, WTO is the WTO dummy variable, μ_i is a country i fixed effect, and p represents the number of leads and lags. im_i,t, irer_i,t, rgdp_C,t, tex_t and FDI_t are measured in natural logs. im_i,t and irer_i,t, vary both over time and across countries and rgdp_C,t, tex_t and FDI_t only vary over time.

The individual export equations have the form:

Here ex_i,t represents real processed exports from the PRC to country i, irer_i,t represents the integrated real exchange rate, rgdp_i,t equals real income in the importing country, K_t represents the Chinese capital stock in manufacturing, FDI_t denotes the stock of foreign direct investment, WTO is the WTO dummy variable, μ_i is a country i fixed effect, and p represents the number of leads and lags. ex_i,t, irer_i,t, rgdp_i,t, K_t and FDI_t are measured in natural logs. ex_i,t, irer_i,t, and rgdp_i,t vary both over time and across countries and K_t and FDI_t only vary over time.

Annual data over the 1992 to 2008 period are used. One lead and lag is employed in the DOLS estimation.

Download this Paper [ PDF 208.8KB| 22 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.