Creation and diversion of income and jobs.

Part I: GDP and value added

Background

This visualisation provides access to a large array of results from my working paper "Creation and diversion of income and jobs: The case of the Korea - U.S. Free Trade Agreement". I described a new approach to quantify trade creation and trade diversion in an article published in The World Economy in 2021. The next step was to show how this new technique, that relies on the structural decomposition analysis in an inter-country input-output framework, works in a study of a bilateral trade relationship. The case of Korea – United States trade and their bilateral free trade agreement (KORUS) seemed appealing because of its significance and controversial negotiations. Hence this paper that calculates the economic benefits and losses of Korea and the United States that may be explicitly attributed to the creation, diversion and contraction of trade flows between them before and after the enforcement of KORUS.

What is trade creation (diversion, contraction)? A reminder and simplified example

The notions of trade creation and trade diversion emerged because of the need to describe the effects of preferential trade agreements. In his highly influential book (1950), Jacob Viner articulates that the trade-creating effect is the “shift in the locus of production” from the high-cost home country to a lower-cost partner country, and the trade-diverting effect is the “shift in the locus of production” between a low-cost third country and a high-cost partner country.

Let us consider a simplified example. Suppose you are in country A (home country) and there is a product that you produce and export. If the sales of the like product in country B (partner country) by your competitor established in that country drop becase your exports increase – what occurs is trade creation. You most likely benefit while your competitor loses, all other things being equal. The reverse effect – where your competitor's product replaces your product in country B's market – is not addressed by Viner as it is not expected to happen in a trade agreement. It can be easily recognised as import substitution, but I call it "trade contraction" for consistency with other terms. In this case, your competitor in country B is set to benefit, and you lose.

The above concerns home country and partner country. What about third countries? You may expand your exports to country B at the expense of reduced sales of the like product from country C (third country). This is trade diversion as defined by Viner. You benefit and your competitor in country C loses. We must not rule out the case where, vice versa, your sales in country B shrink because the demand in country B switches to the like product from country C. This is also trade diversion, harming you and benefiting your competitor in country C. To distinguish between these two cases, I call the former "trade diversion towards partner" or just "trade diversion in" and the latter "trade diversion away from partner" or "trade diversion out".

Note that in the simplified example above, I describe the effects that occur in partner country (B) market and affect you in home country (A). However, the same four effects may originate in your home market: your competitors' products may replace your product in country A's market or vice versa. In sum, you are affected by trade creation/diversion/contraction at home (country A) and at the competitor's market (country B). We may add up an effect, say, trade creation, that affects you at home and at the partner country market, to obtain net trade creation. Then net trade diversion in is typically positive, net trade diversion out is negative while net trade creation and contraction may be positive or negative depending on the balance between home and partner countries.

Data

The data input for this visualisation is from the 2018 edition of the OECD Inter-Country Input-Output (ICIO) tables – the latest edition at the time of writing. The tables cover 64 countries plus the rest of the world as a another country and 36 industries. Annual time series of the ICIO tables are at current prices for the period 2005-2015.

Understanding the visualisation

With the technique I describe in my papers, it is now possible to estimate the change of a macroeconomic variable at industry level induced by each of the trade substitution effects: creation, diversion and contraction. In the visualisation below, this variable is value added. Value added may be understood as gross incomes of industries generated from production activities prior to distribution. The sum of value added across all industries plus net taxes on products equals the gross domestic product (GDP). You may also explore another visualisation where the dependent variable is employment.

The bars in the charts below correspond to the changes in value added of selected industry or GDP of the selected economy that are explicitly attrubited to the underlying trade effects (see also a technical note below). "The country for which the effect is measured" is the country that experiences the said changes. You may choose "All effects, net" to see the interplay between net trade creation, net trade diversion towards the partner (in) and away from partner (out), net trade contraction. "Net" means that the results are summed from the home country and partner country perspectives (see the explanation above). If your choose one of the effects, you will see the underlying split between the effect that occurred in the home country market and that occurred in the partner country market.

The two charts are for the convenience of comparison: you may compare the results for the same industry in Korea and the U.S., or the results for two industries in Korea (or the U.S.), or the results for two different effects choosing one industry of either country.

Technical notes

Take note that each bar corresponds to the change in value added of selected industry induced by one of the four effects that involves the products of all industries. Suppose you have selected "Motor vehicles, trailers and semi-trailers". The chart output you see depicts the changes in value added of the industry producing motor vehicles induced by trade creation/diversion/contraction across the products of all industries. So it is not only shifts in the sales of motor vehicles that affect the performance of this industry, but also, indirectly, the shifts in the sales of the products of other industries. Clearly, this is another immplcation of domestic and global value chains. The estimation technique I used in this paper allows for a disaggregation of the products that were replaced by either trade effect, and I am likely to include it in one of my next investigations.
The above charts, unfortunately, do not provide a sense of significance of bilateral trade creation, diversion and contraction relative to the GDP growth of the two countries. These combined trade effects averaged at only +/-3% of Korea’s annual GDP change, reaching its maximum in 2014-2015 (12%). The share of the U.S. GDP change that is attributable to all trade effects with Korea did not exceed +/-1%.
The manufacturing industries in China and Mexico are additionally disaggregated in the OECD ICIO tables into those operating in standard mode and those operating in processing exports mode. The industry breakdown for China and Mexico was aligned here with the uniform 36-industry classification of other countries because details on processing exports are not relevant for this investigation.
An important note is that the OECD ICIO tables are not available at previous year prices which would allow to isolate the effect of price changes, and the structural decomposition analysis in my paper applies to data at current prices. However, this may not be a serious reason for concern as long as we compare trade creation with trade diversion and trade contraction, or trade creation (diversion, contraction) for one country as the home country, partner country or third country. It is reasonable to expect that price changes affect the relevant variables in the same way. That said, future research will surely benefit from time series of ICIO tables at previous year prices.

Acknowledgment

This visualisation was made in Observable wuth the D3 JavaScript library.