World trade interdependencies: a New Zealand perspective David Gillmore and Phil Briggs A key determinant of New Zealand s growth is its trade with the rest of the world. We have developed a world inputoutput table to investigate the way in which New Zealand and Australia depend on the rest of the world, and to look at the role of demand in China, Japan, the US, and the euro area. We show how higher demand in one economy affects production in another economy both directly, through its higher demand for final products, and indirectly, via its higher demand for intermediate products and its impact on production and demand in other economies. We highlight the impact of US demand on other economies and demonstrate the importance of Chinese and Australian demand on New Zealand. 1 Introduction International trade matters for long-run growth. This is especially true for small open economies such as Australia and New Zealand. The openness of one country (the amount of international trade a country engages in as a proportion of its GDP) is a measure of a country s inter-relatedness with the rest of the world. Given this ratio, a country s export shares (the percentage of its exports going to particular countries) can indicate the dependence on, and vulnerability to, specific countries. These measures therefore indicate the magnitude of the benefit each country may gain from the other s growth as well as revealing the extent of vulnerability to negative shocks from the other. For example, figure 1 illustrates how New Zealand is dependent on Australia and emerging Asia, based on merchandise export shares. Australia is highly dependent on emerging Asia, which in turn is very dependent on both the US and euro area. Thus, New Zealand is dependent on emerging Asia both directly and via its dependence on Australia. And, given developing-asia s dependence on the United States and euro area, both New Zealand and Australia have indirect linkages to these regions via emerging Asia. Figures on the arrows are the share of an economy s exports to trading partner countries/regions, while those in square brackets are an economy s total exports as a percentage of its GDP. Why use input-output analysis? While the export shares shown in figure 1 indicate how concentrated each country s exports are on particular trading partners, they give an incomplete picture of indirect trade linkages and vulnerabilities. A shock to one particular country affects output in other countries, not only through direct bilateral trade, but also indirectly via output fluctuations in other countries (Ohyama 2004). To gain a better understanding, what is also needed is (i) an understanding of intra-country output, (ii) how this Figure 1 Merchandise export shares illustrate inter-linkages 1! 1 In Figure 1, emerging Asia refers to China & Hong Kong, three NIE countries (Korea, Taiwan and Singapore) and ASEAN-5 (Philippines, Vietnam, Thailand, Malaysia and Indonesia). Trade figures for emerging Asia are based on extra-region trade only (i.e., intra-regional trade is not included in total trade). Figures are based on 2006 data. Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010 35
relates to trade, and (iii) third-country effects. One way of analysing these inter-linkages between countries is to use an international input-output table. Input-output analysis was developed by Wassily Leontief in the 1930s to study the inter-relationships between different sectors of an economy. The analysis centres on a matrix where suppliers of goods are represented by rows and users of goods by columns. One application of this method is to show the impact of a change to final demand on all supply chains within the economy. It is also used by government statistical agencies in constructing national accounts. 2 An international input-output table is a representation of total world output, showing the interaction between countries of interest. It is based on a matrix of trade flows between countries as well as internal demand within countries, both of which are divided into final and intermediate demand. Usually, output is broken down by industries in many countries, thus revealing the inter-industry linkages within and between countries. 3 A well-known example of an international input-output table is the Asian International Input-Output table developed by the Institute of Developing Economies at the Japan External Trade Organization (IDE- JETRO) in Japan. A major drawback of international input-output tables is that they are both time-consuming and expensive to construct. So, by the time they are published, the data that they contain are some years old. For example, the latest IDE- JETRO table is for the year 2000. To overcome this problem, economists have combined these tables with more recent data to estimate updated inputoutput tables. However, as the estimation procedure cannot accurately update the industry-level relationships, these updated tables are based on macro-level data (aggregate output and trade data) rather than industry-level data). 4 The world input-output table used in this paper is also a 2 For more detailed descriptions of input-output analysis, see Wixted et al. (2006, Section 2) and United Nations (1999). 3 See Claus & Li (2003) for a New Zealand example. 4 Recent papers using estimated macro-level AIIO tables include Sasaki & Ueyama (2009), Pula & Peltonen (2009), Mori & Sasaki (2007) and Takagawa & Okada (2004). macro-level table. However, there is an important way that it differs from the Asian International Input-Output tables. While the Asian International Input-Output tables include exogenous categories; four countries and an item for the rest of the world; our model contains no exogenous countries and the rest of the world item has been made endogenous. The input-output framework can show the effects of a shock to final demand in one country on production, trade, and income in other countries. It can distinguish between the direct effects of such a shock on output in other countries and the indirect effects that result from the initial shock. However, it is a static approach, in that it does not take into account the changes in employment, prices and spending patterns that would occur within a general equilibrium model. Box 1, overleaf, covers the input-output table in more detail. 2 Constructing the world inputoutput table Constructing our world input-output table involves four steps: 1. We derive a matrix of goods and services trade flows, using data from the IMF, OECD and UN. 2. We use country-level input-output tables (from the OECD and countries statistical agencies) to derive a set of parameters regarding tax rates, the ratio of value added to GDP, the ratio of intermediate goods to value added, and the ratio of imported intermediate goods to total imports: 3. We use these parameters in conjunction with IMF data on GDP and trade to produce a 2006 aggregated input-output table for each country, expressed in US dollars. 5 4. We combine these aggregated input-output tables with our matrix of goods and services trade flows to form our world input-output table. 5 The aggregated input-output table includes totals for: intermediate supply, final demand of locally produced products, imports of intermediate products, imports of final-demand products, tax on intermediate products, tax on final-demand products, and value added. 36 Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010
Box 1 Input-output methodology Table 1 shows a simplified three-country representation of the world input-output table used in this paper. It represents the production and demand of countries A, B, and C, divided between intermediate demand and final demand. These two categories relate to the destination country, i.e. they cover the intermediate demand products and final demand products that are required by the destination country. In both internal demand and final demand matrices, the items on the diagonals represent production that is sold domestically. Off-diagonal items of the rows represent exports, while those of the columns represent imports. The table decomposes output into intermediate demand and final demand. Total output of country A is represented by x a. This is the sum of those items shaded in grey: z aa is the output of country A that is used domestically as input into other production; and z ab and z ac represent the outputs of country A that are exported to countries B and C where they are used as inputs in production. Similarly, f aa is the output of country A that is consumed domestically as final product, while f ab is final product exported to country B. Finally, ya is the sum of the final-demand elements in that row (f aa and f ab and f ac ). The columns within the intermediate demand matrix represent inputs into production. For example, the column circled in blue includes inputs for country A s production z aa are inputs produced in country A, and z ba and z ca are imports of intermediate goods from countries B and C. Table 1 A typical world input-output table In this paper, when we talk about a shock to domestic final demand in one country, we change the items circled in red that is, the demand of residents of country A, which is satisfied by final goods produced both domestically (f aa ) and imported from other countries (f ba and f ca ). The lower three rows of the model include tax on intermediate demand and final demand, value added and total inputs. The table is symmetric. So total inputs (xa, the sum of the first column in the table above) equals the respective total output (the total of final demand and intermediate demand at the right of the table). In the input-output framework, total demand for output can be expressed as: AX + Y = X (1) where A is a matrix of input coefficients a ij = z ij /x j and X is the vector of total outputs for each country (the x a and x b and x c in our simplified input-output table above), and Y is the vector of total final demand for each country (the y a, y b, and y c above). AX equals the matrix of intermediate demand. Equation (1) can be rearranged as: X = [I A] -1 Y (2) where [I A] -1 is known as the Leontief matrix. With a shock to domestic final demand in one country, the values in vector Y increase. We use equation (2) to obtain a new value of total output (X). This step in effect involves solving a series of simultaneous equations that satisfy the new final demand values (Y) and the input coefficients of matrix (A). We are then able to calculate a new intermediate demand matrix, in which each element is z ij = a ij x j. This gives us an amended input-output table incorporating the impact of the shock to domestic final demand. Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010 37
Our world input-output table includes New Zealand and 17 of its trading partners, plus a category for the rest of the world. Countries included are Canada, the US, Mexico, the UK, India, Japan, China, Korea, Taiwan, Singapore, Philippines, Vietnam, Thailand, Malaysia, Indonesia, Australia and New Zealand, while the euro area is included as one economy. Hong Kong is included within China. Most of these economies are in the Reserve Bank s GDP-16 basket of countries. We include Mexico in order to have a complete picture of the North American Free Trade Agreement countries and India because of its growing importance in the world economy. We used data for 2006 as that was the latest available year of services trade figures. Merchandise trade flows data Using export data from the IMF Direction of Trade Statistics, we constructed a matrix of merchandise trade flows, which has exports represented by values in the rows, while imports are represented by the columns. 6 However, there are two important data issues that must be addressed in constructing this matrix. The first issue relates to the lack of data for Taiwan, while the second concerns re-exports from Hong Kong and Singapore. Because of the political status of Taiwan, the IMF includes no section for this country in its Direction of Trade Statistics. However, we were able to extract the Taiwan figures from the Direction of Trade Statistics CD-ROM. Taiwan is included in the aggregate figures for the Advanced Countries category, and we simply calculated values for Taiwan exports as residuals (the difference between the aggregate Advanced Countries figures and the sum of the other countries that make up that category). 7 Because Singapore acts as a trading hub for the East Asian region, a high proportion (47 percent) of its merchandise trade consists of imports of commodities or manufactured products that are re-exported to other destinations with 6 Oct 2009 CD-ROM from the Cross Country Matrix View section. All figures are on an f.o.b. basis. 7 As a crosscheck of this approach, we found that our export figures matched exactly Taiwan official export figures. Table 122 of the Statistical Yearbook of the Republic of China 2007 (Edition 2008) includes exports in USD to 10 of the 17 countries in our sample. little or no added value. Thus, Singapore s exports data (as shown by the Direction of Trade Statistics figures) do not accurately reflect its own economic activity, and the data for other countries imports from Singapore do not reflect the true country of origin. We therefore re-allocated Singapore s re-exports to its trading partner countries. A similar issue exists for Hong Kong s trade data, where 95 percent of its imports from mainland China are re-exported. We overcame this issue by amalgamating Hong Kong and mainland China as one economy, China. Services trade flows data There is a severe lack of data on services trade flows between countries. To construct a matrix of services trade flows, we use data on bilateral flows where they exist, and estimate all other items. The existing services trade data falls into two categories. First, total exports and imports of services by country are published in the IMF Balance of Payments Statistics. Second, the UN and OECD publish bilateral services export and import flows, but for only a limited number of countries. There are many inconsistencies between these UN and OECD figures and between export and equivalent import figures. 8 We therefore averaged the export and equivalent import figures (where they exist) for each of the OECD and UN data sets; we then averaged these average OECD and average UN figures. However, given that the OECD and UN data do not cover all countries, it was necessary to estimate the missing data. We did this by estimating a second matrix for services flows, and then replacing relevant figures with the OECD/UN averages described above. 9 This ensures that we capture some 8 For example, the OECD figures for 2006: UK imports from the US is USD 28,535 million compared with US to UK exports of USD 52,604 million. 9 Technically, we derived the second matrix by applying an algorithm to scale our merchandise trade matrix so that each country s total services exports and service imports equal the figures from the IMF Balance of Payments Statistics. By basing this estimation on our merchandise trade matrix, we assume that services trade flows display similar patterns to merchandise trade. This assumption is consistent with findings that geographical distance is as important (or possibly more important) for services trade as for goods trade (Kimura & Lee 2004). 38 Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010
important bilateral services flows (such as the exceptionally high level of financial services trade in and out of the United Kingdom) that would not necessarily be reflected in our second matrix. The final step was to adjust our estimates to both directly, through its higher demand for final products, and indirectly, via its higher demand for intermediate products and its impact on production and demand in other economies. ensure consistency with the IMF country totals for services exports and imports. Completing the input-output table To complete our input-output table we use data from individual economies input-output tables. In particular, we use (i) percentage of total imports that apply to intermediate demand, (ii) domestic intermediate demand (the diagonal values of the Intermediate Demand (ID) matrix), (iii) taxes that apply to intermediate and final demand, (iv) value added, and (v) total output. From our goods and services trade matrix, we divide each flow into intermediate products and final products based on the percentage of total imports that apply to intermediate demand for each country. We then calculate each economy s domestic final demand and taxes. The complete table is shown in appendix 2. Two other points are worth noting: Because the country-level input-output tables for eight of the countries show zero tax figures, we also set the tax figures in our table to zero. 10 In order to include the rest of the world as an endogenous category, we assume the ratio of intermediate demand to total output is similar to the average of emerging/ developing countries. 3 Results of the input-output analysis We use the world input-output table to investigate the way in which New Zealand and Australia depend on the rest of the world, and to look at the role of demand in China, Japan, the US, and the euro area. We show how higher demand The effects of an increase in domestic final demand on output Figure 2 shows the impact of an increase in domestic final demand in each economy on the production (or total output) in its own economy and in other economies. Each bar of the chart represents the impact of a one-unit shock (of, say, one billion dollars) in that particular economy. The impact is measured in the same units. Figure 2 can be calculated in either of two ways. Firstly, it can be done by recalculating the input-output table adjusting the domestic final demand for the economy concerned, recalculating the table and calculating the change to the total output column. Alternatively, it can be done using a procedure outlined by Mori and Sasaki (2007). This involves constructing a production inducement coefficient matrix which demonstrates the impacts of a one-unit increase in domestic final demand in each economy in turn. Such a matrix is shown in table A3.1 in appendix 3. From figure 2, we see that for most economies the largest impact of an increase in their final demand is on local production. This is particularly so for large and advanced economies. On the other hand, most of the NIE and ASEAN-5 economies have a relatively high foreign impact which is Figure 2 Total production inducement from a one-unit increase in domestic final demand Units 3.0 Home economy Foreign economies 2.5 2.0 1.5 1.0 0.5 0.0 in one economy affects production in another economy 10 These countries are the US, Japan, China, Philippines, Vietnam, Thailand, Malaysia and Indonesia. Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010 39
consistent with the interdependent chains of production across economies in the region. 11 But one country stands out China. The impact of China s domestic final demand on domestic production as a ratio of its impact on foreign production is significantly higher than that of other countries. China s high domestic impact reflects both the size and dynamism of the domestic economy an indication that China is not completely reliant on final demand in foreign markets. The size of China s domestic economy appears more important than its exportled growth suggests. However, we need to be careful when interpreting the data presented in figure 2 and table A3.1. We can compare the impacts from a shock in one country on its own and other economies (comparing the values within the columns of the production coefficient inducement matrix), but comparing the effects of shocks in different countries (comparing values across rows in the production coefficient inducement matrix) can give a misleading picture. This is because the size effects of the economies from which the shocks originate are not We can break down the impacts on foreign countries. Taking China as an example, its one percent domestic finaldemand shock, which is equal to USD25.8 billion, results in its own domestic output increasing by USD56.9 billion and total foreign output by USD13.3 billion. A breakdown of this foreign impact (the red portion of the China bar in figure 3) is shown in figure 4. Note the high impact, in dollar terms, on Japan, the euro area, and the US, as well as the rest of the world. Figure 4 Impact on foreign output from a one percent shock to domestic final demand in China USD billions 3.00 2.50 2.00 1.50 1.00 0.50 taken into account. Obviously, the large demand of the US and the euro area will have scale effects on world trade. Figure 3 shows the production (domestic and foreign) induced by an increase of one percent of domestic final demand in each economy. Here, we see the importance of the US and the euro area because of the size of their economies. Figure 3 Impact on output from a one percent shock to domestic final demand USD billions 300 The impact on total output in each economy can be further broken down into direct and indirect effects. Figure 5 shows the initial direct effects on production, and the subsequent, indirect, effects on domestic intermediate demand and foreign intermediate demand for both Australia and New Zealand. For New Zealand, the initial direct impact on output (USD3.6m) is the amount of exports of final products from New Zealand in order to satisfy the rise in domestic final demand in China. This rise in domestic final demand 250 200 150 100 50 0 Home economy Foreign economies in China also generates additional economic activity both in China and in its trading partners. To supply inputs to this additional activity, New Zealand exports additional intermediate products to both China (USD11.2m) and its trading partners (USD4.7m). Finally, the increase in activity in New Zealand as a result of the rise of both final demand and foreign intermediate demand generates additional domestic 11 NIE (newly industrialising economies) are Korea, Taiwan and Singapore. ASEAN-5 includes Philippines, Vietnam, Thailand, Malaysia and Indonesia. demand for intermediate products (USD17.0m). 40 Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010
Figure 5 Direct and indirect impacts on Australia and New Zealand Australia New Zealand AU USD millions USD millions 400 40 NZ 350 35 300 30 250 25 200 20 150 15 100 10 50 5 0 China direct Other indirect China indirect AU domestic 0 China direct Other indirect China indirect NZ domestic What is noteworthy about figure 5 is the difference in size between the direct impact on final demand and the total output generated. In the case of the Chinese shock on New Zealand, the effect on total output is ten times the size of the initial direct effect. Additionally, the total amount of exports generated (USD19.5m) is five times the size of those generated by the initial increase in final demand. It is possible that the total indirect effects are being underestimated, because increases in final demand in each country from a higher level of demand for intermediate products are not being accounted for in the input-output model. When we shock the model, final demand remains the same, except in the country being shocked. However, final demand could be expected to rise in each country, given the additional income that would be generated by higher exports. It could be argued, though, that the indirect effects, even as shown here, are higher than they would be in reality. This is because the input-output approach assumes that there are no constraints to supply in meeting the demand induced by the initial shock. Overall, the results from the input-output approach should be seen as being indicative rather than definitive. impact on GDP. 12 Output figures include the value of inputs as well as value added and so are higher than GDP figures, which are a measure of the value added alone and therefore reflect better the level of economic activity in an economy. Figure 6, overleaf, shows the impact on GDP of changes to domestic final demand in the US, the euro area, Japan and China. 13 The first chart shows the prominent influence of US final demand on GDP in other economies, largely reflecting the size of the US economy. While the euro area is also a large economy, a one-percent shock has a lower impact on Asian economies than the US. However, it has a larger impact than the US on the rest of the world, which includes eastern Europe and Russia, and is a sizable proportion of the world economy. We now focus on the impact of a separate shock in each economy on Australia and New Zealand. Figure 7, overleaf, highlights the strength of the influence of the US and Australian economies on New Zealand. 14 It also shows the dominant influence of Japan and China on Australia. We have also included the impacts on Japan and China, to demonstrate US and euro area impact on these economies. Elasticities: the effects of an increase in domestic final demand on GDP In the previous section, we have shown the impacts of shocks to domestic final demand on the output in other economies. Input-output analysis can also give us an indication of the 12 In appendix 3, table A3.2 is the matrix of elasticities, where each cell is the percentage change to GDP in one economy as the result of a one-percent change to domestic final demand in another economy. 13 The charts in figure 6 are constructed from the us, ea, jp, and cn columns in table A3.2. 14 The charts in figure 7 make use of the data in the nz, au, jp, and cn rows of table A3.2. Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010 41
Figure 6 Impact on GDP in other economies from a one percent shock to domestic final demand in the US, Euro area, Japan and China United States 0.25 euro area 0.25 0.20 0.20 0.15 0.15 0.05 0.05 Japan 0.25 China 0.25 0.20 0.20 0.15 0.15 0.05 0.05 Figure 7 The impact of one percent shock to domestic final demand in other economies on GDP in Australia, New Zealand, Japan and China Australia New Zealand 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 Japan China 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 42 Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010
While figure 7 gives us the elasticities of a change to domestic final demand in other economies on GDP in New Zealand and Australia, they do not take account of the actual growth seen in individual economies in recent times. Figure 8 shows real growth rates by region from 2006 to 2007. Figure 8 Real GDP growth rates in each economy (2006-07) 14 12 10 8 6 4 2 0 In figure 9 we adjust the shocks to domestic final demand to represent these growth rates. Note, though, that our procedure assumes that the shock to each country occurs exogenously independently from shocks in the other economies. Figure 9 is therefore only illustrative of the influence of the growth in each economy on Australia and New Zealand. We see the dominant impact of the growth in China s domestic final demand on Australia. For New Zealand, China and Australia are the dominant economies. 4 Conclusion In this article, we have described the development and use of a world input-output table to better understand the interrelationship of New Zealand and its main trading partners, and the importance of demand from China, Japan, the US, and the euro area. We have shown how higher demand in one economy affects production in another economy both directly, through its higher demand for final products, and indirectly, via its additional demand for intermediate products and its impact on production and demand in other economies. And we have found that the total trade-related impact may be many times greater than the initial direct effect would suggest. Our analysis has confirmed the importance of both the US and euro area domestic final demand as well as those of Japan and China within the world economy. It has also demonstrated the huge influence of the Chinese and Australian economies on New Zealand economy. References Claus I and K Li (2003), New Zealand s production structure: an international comparison, New Zealand Treasury Working Paper, 03/16. Kimura, F and H-H Lee (2004), The gravity equation in international trade in services, European Trade Study Group Conference, University of Nottingham. Figure 9 Impact of 2006-07 growth in other economies on GDP in Australia and NZ Australia New Zealand 0.30 0.25 0.20 0.15 0.05 0.30 0.25 0.20 0.15 0.05 Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010 43
Mori, T and Sasaki H (2007) Interdependence of production and income in Asia-Pacific economies: an international input-output approach, Bank of Japan Working Paper, Series No. 07-E-26. Ohyama S (2004) Evolution of output multipliers: An analysis with a particular emphasis on Asia, Bank of Japan Working Paper Series No. 04-E-12. Pula, G and T A Peltonen (2009) Has emerging Asia decoupled? An analysis of production and trade linkages using the Asian international input-output table, European Central Bank Working Paper, 993. Sasaki, H and Ueyama S (2009) China s industrial structure and its changes in recent years: An analysis of the 1997-2005 input-output tables, Bank of Japan Working Paper Series No. 09-E-2. Takagawa I and T Okada (2004), Estimation of input coefficients of the extended international IO table and analysis of interdependency in the Asia-Pacific economy, Bank of Japan Working Paper Series No. 04-J-6 (in Japanese). United Nations (1999) Handbook of input-output table compilation and analysis, Handbook of National Accounting, Series F, No.74. New York. Appendix 1 Abbreviations used in tables ca Canada us mx gb ea in jp cn kr tw sg ph vn th my id au nz row United States Mexico United Kingdom euro area (includes the 12 countries that were its members in 2006: Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal and Spain) India Japan China (includes mainland China and Hong Kong) Korea Taiwan Singapore Philippines Vietnam Thailand Malaysia Indonesia Australia New Zealand Rest of the world Wixted B, N Yamano and C Webb (2006) Input-output analysis in an increasingly globalised world: Applications of OECD s harmonised international tables, OECD DSTI/ DOC(2006)7. 44 Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010
X is the total output for each country (the sum of each row of ID and FD). Y is the output for each country that goes to final demand (the sum of each row of FD). This table is available in Excel format from the authors on request.! Appendix 2 World input-output table (2006, USD billions) Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010 45
Appendix 3 Production and GDP inducement coefficient matrices. Table A3.1 Production inducement from one-unit increase in final demand. Each cell is the additional production in country i that is induced by a one-unit increase in domestic final demand in country j (where i represents the rows and j the columns). For example, a one unit-increase in domestic final demand in Canada (ca) induces an increase of 0.341 units of output in the US. Table A3.2 GDP inducement coefficient matrix Each cell is the percent of GDP in country i that is induced by a one-percent increase in domestic final demand in country j (where i represents the rows, and j the columns). For example, a one-percent increase in domestic final demand in Canada (ca) induces an increase of 0.016 percent of GDP in the US. 46 Reserve Bank of New Zealand: Bulletin, Vol. 73, No. 2, June 2010