Monetary policy effectiveness and stock market cycles in ASEAN-5

Abstract

This article examines the asymmetric effects of monetary policy on real output in bull and bear phases of stock market in five ASEAN economies (Malaysia, Singapore, Indonesia, the Philippines and Thailand) using the recently developed pooled mean group (PMG) technique. Stock market cycles are identified by employing Markov switching models and the rule-based nonparametric approach. Estimating the models using monthly data from 1991:1 to 2011:12, the results show that monetary policy (measured by short-term interest rate) has a negative and statistically significant long-run effect on real output in bull and bear market periods while the effects are stronger in bear periods than bulls. In the short run, there is no statistically significant relationship between monetary policy and real output. These results are consistent with finance constraints (capital market imperfection) models that predict that monetary policy is more effective during bear periods than bulls.

Keywords:

• monetary policy effectiveness
• stock bull and bear
• pooled mean group
• Markov switching

JEL Classification:

• E50
• E60

Notes

¹ Effects such as follows: (1) asymmetry associated with the direction of the monetary policy action, (2) asymmetry related to the size of the monetary policy, (3) asymmetry over the business cycle and (4) asymmetry during different inflation regimes.

² Several such asymmetries have been suggested in the literature, including the possibility that the response of stock returns depends on the direction of monetary policy shocks, or on the phase of the business cycle, or on the stock market being in a bullish or bearish period.

³ In the analysis, manufacturing production is used to proxy real output since monthly GDP was not available during the time dimension considered in the study.

⁴ See, for instance, Raghavan et al. (2012) who confirmed the role of asset prices in intensifying the effects of both interest rate and money shocks on output in Malaysia. Sriphayak and Vongsinsirikul (2007) found that asset prices play a central role in transmitting the effects of monetary policy to real output in Thailand.

⁵ See, for instance, Tan and Habibullah (2007) who studied the asymmetric effects of monetary policy over business cycle in four ASEAN countries, namely Malaysia, Indonesia, the Philippines and Thailand.

⁶ See Chen (2007) and Kurov (2010) for application.

⁷ See also Jansen and Tsai (2010) for application.

⁸ One possible explanation for this effect is that traders react faster to bad news when comparing to good news.

⁹ This is called the time-varying transition probabilities (TVTP) model.

¹⁰ The maximization algorithm for the Equations 1–5 is described in detail in Hamilton (1989).

¹¹ Cover (1992), Karras (1996) and Tan and Habibullah (2007) among others used monetary aggregates as measure of money supply shock. In these studies, residuals of money supply equations used to identify the stance of monetary policy.

¹² The identification problem with monetary aggregates, mentioned earlier, may remain present with the federal funds rate. The funds rates are not purely exogenous (Kakes, 1998; Chen, 2007). To solve this identification problem, many researchers have used orthogonalized innovations from the VAR models. See for instance: Bernanke and Blinder (1992), Kakes (1998), Garcia and Schaller (2002), Lo and Piger (2005), Chen (2007) and among others. In this research we also used Choleski orthogonalized innovations to interest rate from VAR models to measure monetary policy shocks, but we couldn’t find any significant relationship between this measure of monetary policy and real output. The results are available upon request.

¹³ Except for the Philippines, in which the high return state corresponds to volatile state.

¹⁴ The results of the mean group (MG) and dynamic-fixed effect (DFE) estimators are available upon request.

¹⁵ The results for estimating the models using MG and DFE estimators are available upon request.

¹⁶ All variables except inflation rate are in natural logarithms.

¹⁷ In this growth model, other control variables such as the level of initial development (as proxied by GDP per capita), physical capital, human capital and institutions are included. In this research, we utilize monthly data of ASEAN-5 countries and these variables are not available in a monthly frequency.