Level and Volatility of Stock Prices and Aggregate Investment: The Case of Thailand

Abstract

The present paper analyzes the aggregate investment behaviour for Thailand and its relations to real stock prices and stock market volatility. In the analysis, we focus on their long run relations as well as their dynamic causal interactions by means of time series techniques of cointegration and vector autoregression (VAR). Our basic framework consists of real aggregate investment, real output, lending rate, real stock prices and stock market volatility. We obtain evidence for their long run relation and that, in the long run, real aggregate investment is positively related to real stock prices and negatively related to the stock market volatility.The generalized impulse-response functions (IRF) generated from the VAR also paint similar picture in that the real aggregate investment reacts positively to shocks in real stock prices and negatively to innovations in stock market volatility. These results tend to be robust when we extend the framework to include alternatively real credits, real effective exchange rate and real government spending.

Keywords:

• Real aggregate investment
• stock market variables
• cointegation
• generalized impulse-response functions
• Thailand

Keywords:

• E22
• O16

Acknowledgements

We would like to thank the referees of the journal for constructive comments and suggestions. We, however, remain responsible for the remaining errors.

Notes

1. It should be noted that the level VAR is valid in the context of cointegrated series. See the arguments given by Ramaswamy and Slok (1998).

2. The referee suggests extending the sample period to cover most recent years, i.e. 2008–2010. However, the data-set of the concerned variables has recently been updated to only 2008 with 2009–2010 data still being preliminary figures. We believe that extending the sample period by only a year is unlikely to materially affect the results we obtain.

3. The X12 procedure is an ARIMA-based method developed by the US Census Bureau to seasonally adjust a time series variable. More specifically, it applies moving averages to decompose the time series into its main components, trend and seasonality.

4. It may be argued that the observed relations between stock market variables and real investment may be due to the Asian crisis, which concurrently dragged both the stock prices and investment to nosedive. We re-estimate the basic model by adding the Asian crisis dummy variable taking the value 1 for 1997.Q3-1998.Q4 and 0 otherwise to account for independent influences of the crisis on the stock market. The results we obtain are largely similar with the coefficients of real stock prices and stock market volatility to be, respectively, 0.312 and −0.155, both are significant at 1% level.