Foreign and Domestic Shocks: Macroeconomic Responses of ASEAN-3 Countries

Abstract

The study provides new empirical evidence on the relative importance of foreign and domestic shocks on selected ASEAN-3 (Malaysia, Indonesia, and Thailand) macroeconomic variables. Three structural vector auto regression models are estimated for each country. The focal point is given on the formulation of the sources of foreign factors. The first model uses trade-weighted foreign variables of both US and Japan to represent the foreign factors. The other two models use US and Japan by themselves, respectively, to represent the foreign factors. Two important results are emerged. First, foreign sectors play an important role in influencing macroeconomic variables of each of the ASEAN-3 country, especially in the medium and the long-run horizon. Second, most of the time, the Japanese factors are more dominant than the US factors in influencing domestic output and inflation for each of the ASEAN-3 countries.

Key Words:

• Structural vector autoregressive (SVAR) model
• foreign shocks
• monetary policy

Jel Classification:

• C51
• C54
• F41
• F43
• F62

Acknowledgement

The authors thankfully acknowledge financial support from the UKM research grant: UKM-GGPM-PLW-016-2011.

Notes

1. The monetary authority is BNM in Malaysia, BOT in Thailand, and BI in Indonesia.

2. These statistics are based on author calculation. These data are collected from the Balance of Payments Statistics and Direction of Trade Statistics, International Monetary Fund database. On average since 1980 until 2010, the share of Malaysia's export to Japan and US as a percentage of total export is 15% and 18%, respectively. In total, it contributes 33% of Malaysia total export. In Thailand, the average share of export to Japan and US is 15% and 18%, respectively. In Indonesia, their export is heavily dependent on Japan, in which the average share of export is 30.85% (since 1980–2010), and Japan continues to be the Indonesia's main export destinations. The average share of Indonesia's export to US and is 16.30%.

3. Monetary policy strategy of the BNM was based on targeting the monetary aggregates before changing it to interest rate targeting in mid-1990s. In Indonesia, BI changed the monetary policy strategy from base money targeting to inflation targeting in 1999 and recently to interest rate targeting. For Thailand, monetary policy strategy of BOT was monetary targeting after the 1997 crisis and later on has changed to inflation targeting where the 14-day repurchase rate has been used as the key policy rate.

4. Some examples of related studies for Malaysia are Azali and Matthews (1999), Ibrahim (2005), and Tang (2006). For Thailand, studies by Disyatat and Vongsinsirikul (2003), Hesse (2007), Charoenseang and Manakit (2007), and Kubo (2008) are of related. As in Indonesia, there is no empirical research that has been published in modelling monetary policy using VAR/SVAR methodology.

5. Ramaswamy and Sloek (1997) have discussed extensively whether the VAR model should be estimated in level, difference or in vector error correction model.

6. For example, there are studies on the US economy by Sims (1986), Blanchard and Quah (1989), Gali (1992), Gordon and Leeper (1994), Christiano et al. (1996), Bernanke and Blinder (1992), Bernanke and Mihov (1998), and Sims and Zha (2006).

7. The liquidity puzzle emerges when a positive innovation in monetary aggregate (money supply) appear to be associated with an increase instead of a decrease in nominal interest rates. The price puzzle happens when an increase in interest rates (monetary tightening) is associated with an increase in the price level rather than a decrease. The exchange rate puzzle occurs when a monetary contraction (e.g. a positive innovation in interest rates) is associated with an impact depreciation of the domestic currency rather than an appreciation.

8. The interest rates puzzle or output puzzle occurs when a monetary contraction (an increase in interest rates) is associated with an increase in output level rather than a decrease.

9. Recently the Malaysian government has reduced the oil price subsidy, thus making the domestic oil price more closely related to the world oil price.

10. Since during recessions, a policy change is deemed to be significant and can affect the behaviour of an agent as postulated by the Lucas critique, the use of dummy variables to account for the economic recessions is important as this will allow only the modest policy change by the Bank Negara to be investigated (see Leeper and Zha, 2003). Nevertheless, using the dummy variables may affect the model's results as the economic recession episodes are dummied out from the model analysis. Secondly, the inclusion of the dummy variables accounts for the mean shift (downward) that occurred in GDP growth over the two recessions. It is a simple way to account for a mean shift in GDP growth over the two recessions. Inclusion of the two dummy variables improves the results from the diagnostic tests on the reduced form residuals.

11. Walsh (2003) argues that whether an interest rate shock has a contemporaneous effect on output will depend on the frequency of the data observations. For example, it would be more plausible to assume a contemporaneous effect when using monthly data as opposed to using annual data.

12. According to Kuang (2008), the greater speed and magnitude of interest rate pass-through is a result of the significant improvements in the level of efficiency in the banking system, led by the changes in the financial infrastructure and policy framework during the past several years.

13. Sims (1992) orders the interest rate variable first, followed by a monetary aggregate, a consumer price index, an industrial production index, an exchange rate index, and a commodity price index for each model of the five developed countries under study. The innovations to the interest rate variable (as a measure of monetary policy) are assumed to affect all other variables contemporaneously but not vice versa.

14. In RATS, B is estimated using the Broyden, Fletcher, Goldfarb, and Shanno algorithm. The initial starting values for B are found using the genetic method.

15. A total of 2500 random samples (with replacement) are drawn from the original sample data.

16. Using recursive identification structure often yields price and the exchange rate puzzles. The impulse responses functions for this identification scheme are not shown but are available upon request.