Dissecting the Effects of Terms of Trade Shocks on the Korean Economy

ABSTRACT

Micro- as well as macro-level analyses on the terms of trade (TOT) for Korea are conducted. We demonstrate that the deteriorated TOT since the mid-1990s are largely attributable to declines in export prices of manufacturing goods and surges in energy import prices. A further investigation using a vector autoregressive (VAR) model identified by sign restrictions on impulse responses suggests that the structural innovation that reduces export prices and increases import prices is the most significant driver of the TOT fluctuations in Korea. Although the shock deteriorates the TOT, it is clearly associated with an expansionary effect on output, which is more pronounced at longer horizons.

KEY WORDS:

• Decomposition of terms of trade
• sign-restricted VAR
• terms of trade

JEL CLASSIFICATIONS:

• E32
• E52
• F40

Acknowledgments

We would like to thank two anonymous referees and seminar participants at the Bank of Korea for helpful comments. This joint work was funded by a financial support from the Bank of Korea while Joonyoung Hur was working at California State University Northridge as an assistant professor. The findings, interpretations, and conclusions expressed in this material represent the views of the author(s) and are not necessarily those of the ASEAN+3 Macroeconomic Research Office (AMRO) or its member authorities. Neither AMRO nor its member authorities shall be held responsible for any consequence of the use of the information contained therein. And the views expressed in this article are solely the responsibility of the authors and should not be interpreted as reflecting the views of the Bank of Korea.

Notes

^1. In 2014, however, the net contribution of exports to GDP shrunk to 1.0%, exceeded by that of domestic demand, 2.3%, in terms of net of import inducement effects, mainly due to weak global demand (The Bank of Korea 2015).

^2. The decomposition results remain almost unaltered if a ratio-based term of trade measure, $P_t^x/P_t^m$, is considered instead.

^3. For the decomposition of the terms of trade by industrial types, all the commodity categories are further broken down into five subgroups: (1) Mining, (2) Transportation equipment, (3) Electrical & Electronic equipment, (4) Fiber & Leather, and (5) Others (Agricultural, Forestry & Marine, Food & Beverages, Wood & Paper, Coal & Petroleum, Chemical, Non-metallic Mineral, Basic Metal, Metal, General Machinery, Precision Equipment, and Other Manufacturing Products).

^4. A pitfall of using sign restrictions for VAR analyses is often associated with a wide interval for impulse responses, which makes VAR-based inferences less plausible. For example, Kilian and Murphy (2012) illustrate that merely imposing sign restrictions may not sufficiently narrow down the set of impulse responses to support a specific economic theory of oil price determination.

^5. Following Uhlig (2005), we conduct inferences based on the median responses. However, using median responses for inferences faces a formidable challenge by Fry and Pagan (2011) and Inoue and Kilian (2013). These studies reveal the problem that there is no unique structural model generating the identical impulse responses to the median responses. An alternative approach by Fry and Pagan (2011)proposes a method to pin down a single structural model that renders the closest impulse responses to the median responses.

^6. Notice that the response pattern of output and consumption contrasts to the so-called Harberger-Laursen-Metzler (HLM) effect. Harberger (1950) and Laursen and Metzler (1950)establish an argument that a fall in the terms of trade would decrease real income, which in turn causes a drop in private savings and a deterioration of the current account. The positive responses of output and consumption, at least in terms of the median estimates, to globalization shocks suggest that the HLM effect may not hold for the Korean economy over the sample period.

^7. The share of exports in GDP over the sample period is on average 0.35 for Korea, with a systematic upward trend over time. This value is substantially higher than that of the advanced small open economies such as Australia, Canada, and New Zealand. The average shares for these countries from the early 1980s to the late 2000s are about 0.18, 0.28, and 0.29, respectively.

^8. In the context of news shocks, Barsky and Sims (2011) demonstrate that the contribution of the shocks on US business cycles varies widely across the short- and long-run identification restrictions.