Estimating Elasticities of Exports and Imports for Transpacific Countries

3.1 Data and Methodology

This section presents estimates of aggregate trade elasticities estimated using a consistent methodology. Estimates are presented for the US and for PRC; Japan; Korea; Malaysia; Philippines; Thailand; and Taipei,China. For India and Indonesia, data were not available over a long enough time period to estimate the model using the same methodology.

Equation (1) presented export functions based on the imperfect substitutes model. Import functions based on the imperfect substitutes model can be written as:

where im_t represents the log of real imports, rer_t represents the log of the real exchange rate, and rgdp_t represents the log of domestic real income.

Quarterly data on aggregate exports and imports are available from the International Monetary Fund's International Financial Statistics (IFS). For most countries these data are measured in dollars and were deflated using the US producer price index from IFS. For Japan they are measured in yen and are deflated using Japanese export and import prices from IFS. For the US they are measured in dollars and are deflated using US export and import prices from IFS.

To calculate rest of the world income for each exporting country we constructed a weighted index of income changes in the top 10 export destinations. The index was calculated using the following formula:

where rgdp_t^* is rest of the world income for an exporting country, the subscript i indexes the 10 largest export markets, rgdp_i is income in country i, and wi is the share of exports going to country i relative to exports going to the 10 largest export markets. The weights were calculated using annual data from the CEPII-Chelem database and converted to quarterly data using linear interpolation. The index was set equal to 100 in 1981Q1.

Data on real income and the real effective exchange rate were obtained from various sources. These are discussed in the Appendix 2 [ PDF 44.9KB | 1 page ].

Export and import functions are estimated using DOLS:

where the variables are defined above.

For most countries we used data from 1981Q1 to 2008Q3. We estimated Equation (6) using a DOLS (2, 2) model. Because this involves using two leads and lags of the first differences of the right-side variables, the actual sample period for the estimation is 1981Q4–2008Q1. In the case of PRC, Philippines, and Thailand we could only obtain consistent data over the 1990Q1–2008Q3 period and the actual sample period for the estimation is thus 1990Q4–2008Q1.

3.2 Results and Discussion

Table 4 [ PDF 48.2KB | 1 page ] presents the results. Income coefficients for every country are positive and statistically significant. These results indicate that for each country an increase in domestic income would increase imports and an increase in rest of the world income would increase exports.

There is evidence of an asymmetry between the income elasticities for imports and exports. For the US, the income elasticity of imports is greater than the income elasticity of exports. On the other hand, for all the East Asian countries except Japan, the income elasticity of imports is less than the income elasticity of exports.

The exchange rate elasticities take on the unexpected signs in many cases for Asian countries but take on the expected signs and are statistically significant for the US. For Asian countries, the exchange rate estimates may be distorted because of the nature of East Asian production networks (see, e.g., Kamada and Takagawa 2005). For the US, the results indicate that a 10% appreciation of the dollar would reduce exports by about 4% and increase imports by about 3%.

There are a few problems with the aggregate estimates reported in Table 4. One, already mentioned, is that these estimates can be distorted by the nature of processing trade within Asia. For example, many of the imports into Asian countries are parts and components that are used to assemble goods for re-export to the rest of the world. An exchange rate appreciation in the assembly country that reduces exports will also reduce the demand for imported goods that are used to produce the exports. This can cause estimated exchange rate coefficients in import equations to be biased toward zero. A second problem, noted long ago by Orcutt (1950), is that estimates of aggregate trade elasticities may be biased if elasticities differ by sector. A third problem, emphasized by Kenen (2007), is that it may be necessary to disaggregate trade flows by country or by group of countries to avoid biased results. A fourth problem is that employing the same specification for all countries does not take into account country-specific characteristics.

It is thus useful to review other evidence concerning how exchange rate changes affect exports and imports. Below we discuss additional evidence for both the US and for Asian countries.

Chinn (2004, 2005a, 2005b) and others, in a series of valuable studies using cointegration techniques, uncovered several stylized facts concerning US trade elasticities:

(i) Exchange rate elasticities for US exports in real terms are precisely estimated and range from 0.68 to 0.84.

(ii) Exchange rate elasticities for US imports in real terms are not statistically significant unless computers and oil are excluded. These amount to 15% of total imports. If they are excluded, price elasticities for the remaining 85% of imports are statistically significant but low. They range from 0.29 to 0.49.

(iii) The sum of the export and import elasticities just barely exceeds one (1.15 if we use the midpoints), implying that the Marshall-Lerner condition for a depreciation to improve the trade balance in nominal terms is just barely met.

(iv) The income elasticity of demand for US exports is between 1.7 and 2.0.

(v) The income elasticity of demand for US imports is 2.4.

(vi) The Houthakker-Magee effect (i.e., the finding that income elasticities for US imports substantially exceed income elasticities for US exports) is still present in the estimates. The difference in the income elasticities, however, appears to have fallen since Houthakker and Magee's original work in 1969.⁵

Chinn concluded based on these estimates that a depreciation of the dollar, if not accompanied by a decrease in expenditures in the US or an increase in expenditures in the rest of the world, would be unlikely to substantially reduce the US trade deficit.

For Asian countries it has been useful to disaggregate imports and exports by category. Some of the key import categories are parts and components, capital goods, and consumption goods. Some of the key export categories are labor-intensive manufacturing goods and technologically-sophisticated goods.

Parts and components trade within East Asia has exploded along with the growth of regional production and distribution networks. IMF (2005) noted that the flow of imports for processing is driven by the demand for final exports in the rest of the world. It estimated that the exchange rate elasticity for imports for processing in Asian processor economies is small (about 0.1).

While the level of the exchange rate may not matter much for parts and components trade within East Asian production networks, both theoretical and empirical evidence indicates that exchange rate volatility matters a lot. Theoretically this effect arises because the service link cost for production blocks separated by national borders is an increasing function of risk and uncertainty, and exchange rate volatility increases risk and uncertainty. Empirically, using panel DOLS techniques over the 1985–2005 period, Thorbecke (2008b) reported that a one standard deviation increase in the coefficient of variation of the exchange rate between two Asian countries would reduce electronic components imports on average by US$300 million per year. Similarly, estimating a gravity model over the 1992–2005 period, Hayakawa and Kimura (2009) found that exchange rate volatility as measured by the standard deviation of the rate of change of the exchange rate decreased parts and components trade in Asia more than tariff barriers did. Ito et al. (2008), surveying Japanese multinational corporations, reported that exchange rate stability between Asian countries is essential for the uninterrupted flow of parts and components within regional production networks.

In capital goods trade there is essentially a complimentary relationship between firms in Japan; Korea; and Taipei,China and firms in developing Asia. Sophisticated capital goods are produced in developed Asia and exported to developing Asia, and many of these goods are difficult to procure elsewhere. Thorbecke (2008a, 2009a), using a gravity model and annual data over the 1982–2003 period, found that an appreciation in developed Asia relative to developing Asia would cause a large drop in capital goods trade. This would harm firms in developed Asia by reducing their exports and also harm firms in developing Asia by making it harder to purchase vital inputs that are difficult to procure elsewhere.

Consumption imports also have the potential to play an important role in Asia. In the PRC, for instance, the lion's share of imports represent inputs into the production process and consumption imports accounted for only 8% of the total in 2007.⁶ An increase in consumption goods exports would benefit PRC consumers. Thorbecke (2009b), using a panel DOLS model and annual data over the 1985–2006 period, reported that a 10% appreciation of the yuan would increase the PRC's consumption imports from the rest of the world by 13%.

Two key export categories within Asia are final electronics goods such as consumer electronics goods and computer equipment and labor-intensive manufactures such as clothing, furniture, and footwear. Final electronics goods are produced largely within East Asian production and distribution networks. Japan; Korea; and Taipei,China produce sophisticated technology-intensive intermediate goods and ship them to the PRC and ASEAN for assembly and re-export. Value-added in the assembly countries is typically small.⁷ Labor-intensive manufactures, on the other hand, are typically produced in developing Asia largely using domestic inputs.⁸

For final electronics goods, the evidence indicates that exchange rates in the countries producing parts and components are an important determinant of exports from developing Asia. For instance Ahmed (2009) and Thorbecke and Smith (2010) found that an appreciation in East Asian supply chain countries would cause a large drop in processed exports. Thorbecke (2009c) reported similar results for the East Asian computer industry.⁹

For labor-intensive manufacturing goods, Thorbecke (2009d) and Thorbecke and Zhang (2009) found that an appreciation in the developing Asian country exporting the goods would cause a large drop in exports. The results indicate that there is also substantial competition between different countries in Asia that export labor-intensive goods to third markets. Similar results, reported in Section 2, indicate that a depreciation in ASEAN countries would cause a large drop in the PRC's exports to the US.

Download this Paper [ PDF 247.9KB| 28 pages ].

Comment(s)

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