OIL PRICE VOLATILITY AND INVESTMENT RESPONSES: THE THAI EXPERIENCE OF PERMANENT AND TRANSITORY OIL PRICE UNCERTAINTIES

Transcription

1 OIL PRICE VOLATILITY AND INVESTMENT RESPONSES: THE THAI EXPERIENCE OF PERMANENT AND TRANSITORY OIL PRICE UNCERTAINTIES IKM Mokhtarul Wadud* School of Accounting, Economics and Finance, Deakin University Huson Joher Ali Ahmed School of Accounting, Economics and Finance, Deakin University ABSTRACT This paper investigates the nexus between investment and oil price volatility in the context of a developing industrialised economy, Thailand. Using a threshold based components generalized autoregressive conditional heteroskedasticity (CGARCH) model, oil price volatility is decomposed into permanent and transitory volatilities. The oil price volatility components are then analysed in a structural vector autoregression (SVAR) framework along with investment and other key macroeconomic variables. Dynamic impulse response functions obtained from the SVAR model reveal significant dampening effects of conditional and transitory oil price volatility shock on Thailand s aggregate as well as sectoral level investments. These findings appear reasonably robust to variable specification and measurements in the SVAR system, and have important policy implications for Thailand s growth sustainability. Keywords: Investment, Oil Volatility, Structural Vector Autoregression, CGARCH, Thailand JEL Classification: Q4, C51 * Corresponding author. Tel: mokhtarul.wadud@deakin.edu.au 1

2 OIL PRICE VOLATILITY AND INVESTMENT RESPONSES: THE THAI EXPERIENCE OF PERMANENT AND TRANSITORY OIL PRICE UNCERTAINTIES 1. INTRODUCTION Impact of exogenous shocks on any economy have been one of the most debated issues drawing significant research interests in the contemporary world. In particular, oil price changes have been at the forefront of such research agendas due to the widespread effects of such shocks on macro and microeconomic activities, both in the short and the long run, encompassing a range of sectors. It is important to perceive that oil price changes as well as the uncertainties inherent in the patterns of oil price changes tend to exert significant impact on economic activities. Bernanke (1983) maintains that oil price changes mainly affect the short run economic performance through disrupting purchases of expensive consumption and investment goods. Hamilton (23) shows that major oil price changes raises uncertainties, lowering consumers spending on cars, appliances, housing as well as firms spending on investment goods. While the effect of oil prices and oil price uncertainties are well endorsed in the literature, the pattern of such relationship remains largely inconclusive due to inter country differences in oil trade status, development stages and structures of oil dependent sectors. Hamilton (1983) made a seminal contribution to literature by modeling oil price shocks and by comprehensively examining their impact on the US economy. Hamilton (1983) showed that almost all the recessions in the US in the post-world War II period followed sharp rises in oil prices. Another intriguing aspect has been the asymmetric nature of the macroeconomic responses due to oil price shocks, which suggests that while output declines with increased oil prices, output and investment do not tend to rise as oil prices fall (Ferderer, 1996; Elder and Serletis, 29; Cunado and Gracia, 23; Cunado and Gracia, 25; Lee et al., 1995; Davis, 1987; Davis and Haltiwanger, 21; Mork and Olson, 1994). 2

3 Last couple of decades has seen a major surge in oil price uncertainty effects on economic activities (Ferderer, 1996; Elder and Serletis, 29; Pindyck, 1991). However, a sheer majority of these studies are conducted in the developed countries. There seems to be limited knowledge and information available with regard to oil price uncertainty effects in developing economies, although many of the developing economies record remarkable industrialization led by energy intensive sub-sectors. Thailand is one such economy that has seen economic growth through rapid industrialization along with other South East Asian nations. Despite being one of the worst affected economies during the Asian Crisis, Thailand recorded consistent recovery in the post crisis period poised on a sustained export led growth with contributions from the manufacturing industries. Following drastic decline in 1998 and 1999, Thailand s real gross fixed capital formation (total investment) began to increase by about 5.5% annually from 2 reaching somewhat close to its pre-crisis average investment level in 212. Despite the recent adversities of the subprime crisis and the political turmoil, Thailand seems to have been showing signs of its impending revival back on to the normal growth track in the new millennium. Much of the recent investment growth of the Thai economy is attributable to investment in metal products, machineries and other transportable goods. In view of growth of these industries with reasonably high energy dependence, Thai energy consumption and in particular, consumption of oil has increased significantly in the last decades. It is the second largest net oil importing economy in South East Asia after Singapore. Obviously, the pattern of investment sensitivities of the energy dependent industries in Thailand, due to uncertainties stemming exogenously such as the ones from oil price volatility calls for further investigation. The debate on oil price uncertainty and its impact on macroeconomic activities has pioneered by Bernanke (1983) and Henry (1974). They maintain that higher uncertainty could depress current investment and, hence, could exert a prolonged effect on aggregate output. 3

4 Bernanke (1983) suggests that uncertainty about energy prices may induce optimizing firms to postpone investment decisions, thereby leading to a decline in aggregate output. More recently, Rafiq et al. (212) empirically examines the impact of oil price volatility on key macroeconomic indicators of Thailand. They show that oil price volatility has significant impact on macroeconomic indicators, such as unemployment and investment, over the period from 1993 to 26. Although, uncertainty in the oil price could have a dampening effect on output and investment, the literature on such a relationship has been limited both in developed and developing economies. Ferderer (1996) shows that oil price volatility, measured by monthly standard deviations of daily oil prices, helps to forecast aggregate output movements in the U.S. Moreover, part of the asymmetric relationship between oil price changes and output growth found in previous studies can be explained by the economy's response to oil price volatility. Huang et al. (25) examine the impact of both oil price shocks and oil price volatility shocks on macroeconomic performance. They document that oil price changes better explain macroeconomic activities than the volatility of the oil price. Elder and Serletis (21) examine the impact of oil price uncertainty on investment, consumer durable and aggregate output using U.S. data and suggest that uncertainty about oil prices tends to depress investment in the US. It could be worth mentioning, at this point, that despite a number of studies available on many countries, there has been an absolute dearth of research on how oil price affect output and investment from the perspective of emerging developing economies, including Thailand. There are two distinct aspects of our study in the context of the existing literature. First, although the role of uncertainty in macroeconomic performance has been widely investigated (Episcopos, 1995; Federer, 1993, 1996; Guo and Kliesen, 25; Ahmed et al., 212; and Ng, 212), the macroeconomic effects of oil price uncertainty, separated as permanent and 4

5 transitory, have been largely unexplored for a developing country such as Thailand. Given the limited studies conducted on the effect of exogenous shocks on the newly industrialised developing economies, this paper adds to literature with knowledge on how a representative developing economy such as Thailand responds to both permanent and transitory oil price uncertainties. Secondly, compared to the existing studies that considers macroeconomic responses, this paper makes a leap forward by examining investment dynamics at both aggregate and sectoral levels with regard to changes in permanent and transitory volatilities of oil prices. From policy viewpoint, having suffered through the global and region specific recessions, an inescapable option for a developing economy such as Thailand will be to finetune its course of development with appropriate immunity and precautionary measures adopted subsiding the detrimental effects from exogenous shocks or uncertainties. This study is expected to help address some of these concerns and provide fresher policy insights for Thai economy amid challenges in the new millennium. We consider component based GARCH measures developed by Nelson (1991) for oil price shocks beyond the scope of the study by Ferderer (1996). This measure allows to examine both transient and permanent volatility impact on aggregate investment and sectoral level investment. The stylized facts of conditional volatility are then used in the SVAR model to understand the impulse responses of macroeconomic factors to shocks emanating from oil price volatility. Our results mainly show that oil price shocks measured by conditional volatility have a negative and prolonged effect on both aggregate and sectoral level investments. The findings also show that negative impact on aggregate and sectoral investments are mainly caused by the transitory volatility of oil prices, especially for the relatively more energy dependent sectors. 5

6 The rest of the paper is structured as follows. Section 2 reviews the literature. Section 3 gives an overview of Thailand s energy consumption and sectoral development. Section 4 provides a description of the methodology of asymmetric-based CGARCH, the SVAR framework and data used to model oil price volatility and its effects. Section 5 provides detailed findings of the study and Section 6 concludes the paper. 2. RELATED LITERATURE The debate on oil price uncertainty and its impact on macroeconomic activities has stemmed from the theory of investment under uncertainty laid down by Bernanke (1983) and Henry (1974). They contend that higher uncertainty could depress current investment and, hence, could exert a prolonged effect on aggregate output. Bernanke (1983) suggests that uncertainty about energy prices may induce optimizing firms to postpone investment decisions, thereby leading to a decline in aggregate output. The neoclassical investment theory, the accelerator model of investment and q theory of investment tend to provide some explanations as to why increased oil price volatility depress economic activities. The negative impact emanating from oil price uncertainty on investment is the result of irreversibility of investment decision and firm s ability to postpone investments. Bernanke (1983), Pindyck (1991), Pindyck and Solimano (1993) and Dixit and Pindyck (1994) demonstrate theoretically a negative impact of uncertainty on firms investment. By contrast, assuming perfect competition, constant returns to scale and symmetric adjustment costs, Hartman (1972) and Abel (1983) show that an increase in uncertainty may instead boost firms real investment by raising the marginal value of capital. These contradicting theoretical predictions have attracted a great deal of empirical studies and the recurrence of various forms of uncertainties in recent years has further heightened interest in the subject. 6

7 In the last two to three decades, there has been a sizeable body of empirical and theoretical research on the effects of oil prices on the macroeconomy. The majority of this research focuses on the effects of price level changes. However more recently, there has been dynamic shift in empirical literate that attract a number of researcher to examine the effect of oil price volatility on macroeconomic activities. In general, several authors have found that increased oil price uncertainty has a depressing effect on real GDP and investment spending (Federer, 1996; Guo and Kliesen, 25; Elder and Serletis, 21; Ahmed and Wadud, 211). More specifically, the literature on uncertainty related to various financial and macroeconomic variables have been considered. These include, among others, output uncertainty, inflation uncertainty, exchange rate uncertainty and stock market uncertainty. Federer (1993), Episcopos (1995) and Goel and Ram (1999) are among early studies evaluating the investment uncertainty relations for the US and other OECD countries. Federer (1993) utilizes risk premium embedded in the term structure of interest rate as a measure of uncertainty, and examines its relation to investment in durable equipment as well as contracts and orders of new plant and equipment for the US case. He finds their relations to be negative and significant. In a related study, Episcopos (1995) considers various sources of uncertainty emanating from leading indicator index, stock price, consumption, interest rate and the price level and find the inverse relationship between investment and source of uncertainty. Goel and Ram (1999) widen the analysis to a panel of 12 OECD countries but focus specifically on inflation uncertainty measured by 5-year moving standard deviation. They find evidence that the adverse effect of uncertainty does depend on the degree of investment irreversibility which is consistent with findings of Bernanke (1983) and Pindyck (1991) who have shown that in a partial equilibrium framework, the oil price uncertainty is important for investment decisions when firms consider an irreversible investment. 7

8 More recently, the analysis for developed economies is further undertaken by Byrme and Davis (24, 25a, 25b), Guo and Kliesan (25), Bulan (25), Elder and Serletis (21), Kellogg (21), Stockhammer and Grafl (21). Guo and Kliesen (25) find that increased uncertainty forecasts lower non-residential business investment. In another study, Elder and Serletis (21) show that uncertainty dampens the mining component of private fixed investment, but has no statistically significant impact on private fixed investment net of mining. Kellogg (21) finds that uncertainty has an important impact on drilling in the oil industry, leading credence to the result in Elder and Serletis (21). More importantly, Byrne and Davis (24) examine the impacts of inflation uncertainty on real investment but only for the US case. They consider distinction between temporary and permanent components of uncertainty using a switching regime model and find both components of uncertainty exert adverse effects on the US real investment but with stronger effect from the temporary component. To broaden the search on the impact of temporary and permanent uncertainty, Byrne and Davis (25a) extend the analysis to G-7 countries by adopting GARCH-generated measures of uncertainty and highlight the importance of sources of uncertainty. Built upon this finding, they further investigate whether temporary and permanent exchange rate volatilities have different implications on investment using the same data set (Byrne and Davis, 25b). Their results indicate adverse investment effects of only temporary exchange rate uncertainty. These studies provide further evidence showing the adverse effects of uncertainty on firms investment behaviour thereby adding a new dimension on the subject. As Byrne and Davis (25b) point out, firms should be able to cope with permanent volatility since insurance should be readily available. However, Ahmed et al. (212) emphasize that the sources of permanent volatility are shifts in fundamentals while those of temporary volatility stem from random events such as sudden disruptions in oil markets and political unrests. Based on these arguments, firms should be affected more by the transitory uncertainty than permanent uncertainty. 8

9 From the developing economies perspective, several studies investigate the uncertainty investment link (Serven, 23; Pradhan et al., 24; Ang, 21; Fatima and Waheed, 211 and Ibrahim, 211; Ibrahim and Ahmed, 214). However, the focus of these studies has been on the linkage between macroeconomic uncertainty and aggregate investment. In line with Byrne and Davis (25a), Serven (23) documents evidence supporting the adverse effect of exchange rate uncertainty on investment in developing countries. He further notes that a negative exchange rate uncertainty investment link is stronger for a country with higher openness and weaker financial system. In a more recent study, Fatima and Waheed (211) look at a variety of macroeconomic uncertainties and document significant and negative investment effects of uncertainties stemming from budget deficit, government consumption, total trade, exports and FDI. Like many other studies discussed above, they apply the GARCH methodology for volatility or uncertainty measurement. Pradhan et al. (24) assess the implications of exchange rate volatility on aggregate private investment in Indonesia, Malaysia, the Philippines and Thailand using annual data from 1972 to 21. They note a positive influence of exchange rate uncertainty on investment in Malaysia, the evidence for Thailand suggests otherwise. Adding to these findings for the developing markets, Ibrahim (211) finds the negative impacts of stock market uncertainty on Thai real investment in the long run as well as short run. While these studies differ in terms of countries covered (developed versus developing countries), data structure (time series versus panel data), time periods, and measurement and sources of uncertainty, they seem to provide a fairly consensus view that uncertainty depresses real investment. Still, looking back at the aforementioned studies, we may realize the absence of oil markets as a source of uncertainty and its implication on investment. However, more recently, Elder and Serletis (29) show that real investment s responses to oil price uncertainty is one of the reasons underlying the dampening output effect of oil volatility. In 9

10 other words, with escalation and sharp swings in oil prices in recent years, oil market volatility deserves empirical attention. We believe that the oil market could be a better source of uncertainty during recent period as compared to exchange rate uncertainty considered by Pradhan et al. (24) and Bhandari and Upadhyaya (21). To pursue on this path, Ibrahim and Ahmed (214) investigate the relationship between aggregate investment and oil volatility and its permanent and transitory components for a developing country, Malaysia. In the paper, the components generalized autoregressive conditional heteroskedasticity (CGARCH) model is utilized to decompose conditional oil volatility into permanent oil volatility and transitory oil volatility. Adopting a vector autoregression (VAR) framework to allow feedback effects between aggregate investment and its determinants, the paper documents evidence supporting the adverse effects of conditional oil volatility, permanent oil volatility and transitory oil volatility on aggregate investment and real output. One interesting aspect of their findings is that, contrary to the findings for the developed markets (US and OECD), the real effects of permanent oil volatility tend to be stronger for Malaysia. Hence, there is an indication that heightened oil volatility accounts for the slumps in Malaysia's aggregate investment after the Asian financial crisis. 3. THAI ENERGY CONSUMPTION AND SECTORAL DEVELOPMENT Thailand is a net importer of oil and natural gas, although the country is a growing producer of natural gas. Based on international energy statistics (IEA), Thailand has limited domestic oil production and reserves, thus imports make up a significant portion of the country's oil consumption. However, Thailand holds large proven reserves of natural gas, and recently natural gas production has increased substantially over the last few years. However, the country still remains dependent on imports of natural gas to meet growing domestic energy demand. Thailand s pace of expected economic growth poses a significant challenge to meet growing 1

11 energy need. Based on 211 report, Thailand was forecasted to grow at a rate of 5.5 percent in 212, and in turn, oil and gas production and consumption are expected to increase slightly in 212 and 213. As of 211, Thailand's primary energy consumption is mostly from fossil fuels, accounting for over 8 percent of the country's total energy consumption. The Table 1 reports energy consumption by energy mix. The petroleum remains second important source of energy consumption in 211 accounting for 36.2% of the country s energy need. INSERT TABLE 1 ABOUT HERE Table 2 provides a breakdown of Thailand s energy consumption by sectors. As the economy expanded and industrialized, Thailand continued to consume more oil for transportation and industrial uses which constitutes about 72% of total energy demand. The Table clearly suggests that oil and gas play important role as energy sources in the industry and transportation sectors. Hence, it would be worth exploring how oil price volatility could have demeaning effect on this two broadly categorized sectors. INSERT TABLE 2 ABOUT HERE Alternative Energy and Energy Mix With proven gas reserves, natural gas has replaced some oil demand and is the next largest fuel, growing to nearly a third of total consumption mix. Solid biomass and waste have played a strong role as an energy source in Thailand and comprise roughly 16 percent of energy consumption. Most biomass feedstock is from sugarcane, rice husk, bagasse, wood waste, and oil palm residue and is used in residential and manufacturing sectors. Thailand has promoted biomass for heat and electricity, though growth has been very gradual due to industry inefficiencies and environmental concerns. Thailand's new Alternative Energy Development 11

12 Plan calls for renewable energy to increase its share of total energy consumption to 25 percent by 222, in an effort to reduce dependence on fossil fuels. As Thailand continues to expand economically, it will place greater emphasis on energy supply security by diversifying its fuel slate and promoting upstream development of hydrocarbons including alternatives to conventional fuels. Oil production and consumption gap According to Oil & Gas Journal, Thailand held proven oil reserves of 453 million barrels in January 213, an increase of 11 million barrels from the prior year. In 211, Thailand produced an estimated 393, barrels per day (bbl/d) of total oil liquids, of which 14, bbl/d was crude oil, 84, bbl/d was lease condensate, 154, bbl/d was natural gas liquids, and the remainder was refinery gains. Thailand consumed an estimated 1 million bbl/d of oil in 211, leaving total net imports of 627, bbl/d, and making the country the second largest net oil importer in Southeast Asia. Figure 1 provides Thailand s dynamic consumptionproduction and net import patterns of oil over the period Thailand is a net importer of crude oil and a net exporter of petroleum products. Thus any changes in oil price or uncertainty in oil price may have a significant impact on its economy. The country imports over 6 percent of its total petroleum needs and almost 85 percent of its crude oil consumption, leaving Thailand highly dependent on global oil markets and volatile prices. About 78 percent of its crude imports originate from the Middle East, while another 8 percent are from other Asian suppliers. The country's oil import dependency has spurred the government to promote the use of other fuels such as natural gas, renewable sources, and biofuels as well as to boost crude oil and product stocks and to encourage investment in marginal field production. 12

13 INSERT FIGURE 1 ABOUT HERE 4. MODELLING OIL PRICE VOLATILITY The literature on the choice of the measure for uncertainty has been diverse and to-date there seems to be an apparent lack of consensus with regard to choice of a single measure. Ferderer (1996) used conventional standard deviation to develop oil price volatility in an effort to examine the impact of oil price uncertainty on macroeconomic activities, while Kuper (22) presents the GARCH (1,1) model to measure oil price volatility at daily and monthly frequency. However, Kuper only focuses on measuring volatility, not the volatility effect of the oil price on economic activities. As opposed to GARCH, we use CGARCH as introduced by Engle and Lee (1999). The rationale for the preference of CGARCH over GARCH (1,1) is twofold. First, the standard GARCH models assume that positive and negative error terms have a symmetric effect on volatility. In other word, good news and bad news have the same effect on the volatility in the model. However, given the oil price behavior, oil price volatility may not react to both bad and good news at a similar fashion due to asymmetric nature of the effects of oil price volatility. Second, as its name suggests, the CGARCH decomposes the volatility into its transitory and permanent components. Hence, given its relevance to the present analysis, we adopt the CGARCH model. Since GARCH modelling and estimation requires the relevant variable to be stationary, we tested oil price for stationarity. Both the ADF and Phillips-Perron tests revealed that the oil price series possesses unit root at levels and it is stationary at first difference. Hence the first difference of oil price is used for CGARCH modelling. The CGARCH specification for the oil price (OPt) is given below: OP 2 t p t = + µ i OP t i + i= 1 µ ε (1) 2 2 ( ε w) + β ( σ w) = w + α t 1 t t σ (2) 2 t q t = α 2 2 ( ε q ) + β ( σ q ) t 1 t 1 1 σ (3) q t ( q w) + ϕ( ε σ ) t = w + t 1 t 1 t 1 t 1 ρ (4) 13

14 Where is the first difference operator and εt is the error term with a time-varying variance 2 (i.e. a measure of conditional volatility) σ t. Equation (1) is the conditional mean equation specified to follow an autoregressive process, of the order (p) which is selected based on uncorrelated errors. Equations (2) represents the conditional volatility of oil prices to a constant mean (w). Equations (3) and (4) capture respectively the transitory and permanent components of conditional volatility. As proposed by Engle and Lee (1993, 1999), the CGARCH model allows for the reversion of the conditional volatility to a time-varying trend component q. This is represented by equations (3) and (4), which capture the transitory and permanent components 2 of conditional volatility, respectively. Note that the conditional volatility σ t is mean-reverting around the permanent volatility (qt), and accordingly, σ q measures the transitory 2 t t component of volatility with the speed of mean reversion represented by ρ. Since the permanent volatility is more persistent than the transitory volatility, it is assumed that < (α + β) < ρ < 1. The standard CGARCH model is based on the assumption of symmetry between positive and negative shocks. For example, there could be empirical evidence of a negative shock of oil causing increased uncertainties due to heightened expectations of a speculative attack. To accommodate this possibility in our model, we allow for temporary asymmetry in our explanatory variables, in order to test whether a shock due to the bad news effect (price reduction) can explain conditional volatility of oil price. Ideally, this asymmetric effect is incorporated in the transitory volatility equation. The threshold based CGARCH model is given as follows, σ q = αε ( q ) + γε ( q ) d + βσ ( q ) t t t 1 t 1 t 1 t 1 t 1 t 1 t 1 q = w + ρ( q w) + ϕε ( σ ) 2 2 t t 1 t 1 t 1 (5) 14

15 Where d is a dummy variable indicating a negative shock and takes a value of 1 ifε < t 1, and otherwise. Note that α and ( α + γ) measure the effect of a positive and a negative news on volatility, respectively. 4.1 Econometric model Incorporating the key macroeconomic variables and oil price volatility, we define the vector of endogenous variables as follows: YY tt = [VVVVVV, IIIIII, CCCCCC, IIII, EEEE] VOP represents volatility of oil price in general, which could be specified in terms of conditional, transitory and permanent volatility. INV is the aggregate and sectoral level investment, CPI is the price level, IR is the interest rate and EX measures exchange rate of Thai local currency against US dollars. In this paper we develop and examine a 5-variable SVAR model based on a modified Kim and Rubini (2) model that accounts for external shock emanating from oil price shocks. Such a model is expected to more appropriately portray how a small open economy such as Thailand, with vulnerabilities to exogenous shocks, responds to oil price uncertainties. To capture the sensitivities of the Thai economy to oil price uncertainties we include Thailand s aggregate and sectoral investments as well as the core consumers price indexes. Interest rate is used to help evaluate monetary policy responses while exchange rate is used to represent Thailand s external transactions. A SVAR has the following general form: AX ( ) t = A1 L Xt + Bε t (6) Where Xt represents n-vector of relevant variables as follows: XX tt = [VVVVVV tt, IIIIII tt, CCCCCC tt, IIII tt, EEEE tt ] 15

16 The A and B are 5 x5 matrices of coefficients; and A q i 1(L) A L 1i i= 1 = represents a matrix polynomial in the lag operator with AA 1ii being a 5 x 5 matrix of coefficients. Here matrix A is used to define the impulse responses of endogenous variables to structural shocks, denoted by εε tt = [εε tt VVVVVV, εε tt IIIIII, εε tt CCCCCC, εε tt IIII, εε tt EEEE ]. Matrix B contains the structural form parameter of the model. ε t is an n-vector of serially uncorrelated, zero mean structural shocks with an identity covariance matrix, = E ε t ε ] ε [ t = I Reduced form VAR The reduced form of the VAR model can be represented as: X = CLX ( ) + u (7) t t t Where C(L) = A -1 A1(L) with: AA uu tt = BBεε tt (8) The residuals ut in the reduced model are also presumed to be white noise, but they may be correlated with each other due to the contemporaneous effect of the variables across equations. We estimate the AB model proposed by Amisano and Giannini (1997). 4.2 Identification and contemporaneous restriction In order to recover the structural parameters, the reduced form equation system (7) has to be estimated. However, restrictions need to be imposed for equation (8) to be identified. We 16

17 impose the restrictions on contemporaneous relations among the variables in the SVAR model based on Kim and Roubini (2). To examine the impact of oil price volatility on investment at the aggregate and sectoral levels, we consider the following identification scheme and contemporaneous restriction uu VVVVVV uu IIIIII uu CCCCCC uu IIII uu EEEE = εε VVVVVV εε IIIIII εε CCCCCC εε IIII εε EEEE (9) Where the asterisks represent the coefficients to be estimated in the above systems (9). The residuals obtained from the reduced form VAR are represented by the vector of u s presented on the left hand side of both the above equation systems. The structural shocks, the ε s are shown on the right hand side of the system representing shocks to oil price volatility, investment, CPI, interest rate and exchange rates. Row (1) of the SVAR system represents (exogenous) shocks from oil price fluctuations. Row (2) of the SVAR system represents the aggregate and sectoral investment levels to be contemporaneously affected by the oil price shock. The theory of investment under uncertainty, developed by Henry (1974) and Bernanke (1983), suggests that higher uncertainty could depress current investment and, hence, is likely to exert a prolonged effect on aggregate output. We expect a possible negative demand shock generated from oil price uncertainty as a result of individuals and households postponement of consumption for big ticket items such as cars, housing, appliances, and investment goods. Friedman (1977) proposes the idea that when mean inflation grows, it creates greater uncertainty about future inflation, and hence, subsequent negative impacts on output due to the sub-optimum allocation of resources. Bernanke (1983) suggests that it is optimum for a firm to postpone irreversible investment expenditures when they experience increased uncertainty about the future oil price, and that firms investment 17

18 uncertainty may create cyclical fluctuations in aggregate investment. Row (3) of the system shows that consumer prices are contemporaneously affected by both oil price uncertainty, investments and interest rates. Row (4) of the SVAR system (9) shows that Thai interest rates and hence Thai monetary policy is contemporaneously affected by exogenous shocks of oil price fluctuations and investment. The last row of the system (row 5) represents the external sector and shows that Thai exchange rates are contemporaneously affected by oil price volatility, CPI and interest rates. Hence, overall, we allow the oil price volatility shock to contemporaneously affect Thailand s domestic investment, price level, monetary policy and the exchange rate. 4.3 Impulse response functions The impulse response functions are derived and used to examine the dynamic responses of the variables (VVVVVV, IIIIII, CCCCCC, IIII, EEEE) to various shocks within the SVAR system. Having identified the structural shocks, we can then find the impulse response of a variable to a onetime shock to any variable included in the model, which can be obtained from the following: X = C*( L) ε t C*( L) = CLA ( ) t 1 Where CC (LL) 1 = CC(LL)AA generates the impulse response function of Xt to structural shocks to εt. It would be worth mentioning at this stage that since the primary objective of using the SVAR system is not to estimate the VAR coefficients, but rather to examine the impact of dynamic shocks generated by oil price change and its volatility, the estimates of VAR coefficients are deemed unbiased without transforming the variables into stationary. 18

19 4.4 Data and Variable descriptions Following Hamilton and Herrera (24) and Bernanke et al. (1997), in this paper we use quarterly data from 1994 to 212. The choice of data starting from 1994 is partly due to the fact that Thailand s macroeconomic and sectoral level data with a monthly or quarterly frequency are mainly available from , depending on the type of the data sought. All variables except for the interest rate are expressed in natural logarithms. The data on aggregate and sectoral investments are obtained from Office of the National Economic and Social Development databases. Data on exchange rate and interest rates are obtained from the Bank of Thailand. The core CPI series is used to measure Consumer prices and is taken from Bureau of Trade and Economic Indices, Ministry of Commerce, Thailand. Oil price volatility is estimated using crude oil price index derived from the simple average of three spot prices; viz., Dated Brent, West Texas Intermediate and the Dubai Fateh. Total investment and sectoral investments are measured as seasonally adjusted real gross fixed capital formation expressed in 1988 constant prices. The sectoral investments are categorised as the type of capital. Exchange rates and interest rates are real effective exchange rate measures in Thai Baht per US dollar and Bank of Thailand s policy rates, respectively ESTIMATIONS AND RESULTS In this section, we first briefly discuss estimation results of CGARCH model. This is followed by the baseline results based on SVAR framework. 1 Since 25, policy rates published by the Bank of Thailand is used for constructing the interest rate variable. For period prior to 25 interbank lending rates available for the maximum time horizon upto 6 months are used due to non-availability of policy rates published by the Bank of Thailand. However, these figures closely correspond to the Bank of Thailand s policy rates. 19

20 5.1 CGARCH Results The estimated results for the CGARCH model are presented in Table 3. The results show that the autoregressive parameter ρ is highly significant. This indicates a slow convergence of the permanent oil price volatility. The parameter ϕ in the permanent volatility equation is also highly significant, which suggests that arrival of new information about oil rice affects the trend component of oil price volatility. An interesting finding is that the parameter α is also highly significant, which indicates that both a positive and negative oil price shocks raise oil price volatility temporarily. Likewise, the threshold parameter γ is not statistically significant indicating no obvious asymmetric effects of a shock in oil price. The estimates in Table 3 also shows that the parameter β is statistically insignificant, which suggests that there is no GARCH effect in the transitory component of oil price volatility. INSERT TABLE 3 ABOUT HERE Figure 2 plots the dynamics of oil price volatility for the sample period. The permanent volatility looks relatively more stable compared to conditional volatility and this is an expected phenomenon, since the earlier is meant to portray longer term trend pattern of the oil price volatility. While the permanent volatility tends to follow a more smoothed out trail, there are periodic rise in the permanent volatility as well as frequent sparks in the transitory volatility. Also, both conditional and permanent volatility of oil price seem to have increased slightly in more recent years. An important observation, however, is that the upswings in transitory volatility and rise in permanent volatility tend to coincide with various major financial and political events such as the 1997/1998 Asian crisis, Dot Com bubble in early 2, 9/11 attack and subsequent war on terrorism in 21, Enron bankruptcy in 21, and the latest subprime mortgage crisis that led to the global financial crisis and subsequent recession since 28. 2

21 Given the circumstances, it would be interesting to see whether exogenous shocks such as the oil price volatilities act in detriment to investment growth in Thai industries and, if it does, whether the temporary or permanent components of oil price volatility exerts a more significant impact on Thai investment and other macroeconomic fundamentals. INSERT FIGURE 2 ABOUT HERE 5.2 Impulse responses Impulse Responses to Oil Price Volatility Shocks Figure 3 presents the impulse responses of total investment, CPI, interest rates and exchange rates due to shocks in the conditional oil price volatility. The first of the diagrams shows that a one-time increase in the conditional oil price volatility results in a fall in total investment. Precisely, investment declines significantly immediately following the rise in conditional volatility, until the 2 nd quarter following the shock. The effect on CPI is similar, as the consumer prices decline significantly from quarter 1 to quarter 4 following the shock. Interest rates tend to decline significantly immediately after the increase in conditional volatility spanning up to quarter 3. The responses of the exchange rate, however, does not look significant. The responses due to increase in permanent volatility, as shown in Figure 4, seem to be generally inconspicuous. As permanent volatility rises, fall in total investment is marginally significant for about six months. Expansionary monetary policy responses with marginal statistical significance is noticeable for a very brief period of a quarter following the shock in permanent volatility. There seems to be no significant dampening impact of a shock in permanent oil price volatility on the exchange rate in a shorter time horizon. However, exchange rate tends to depreciate between 6 to 8 quarters after the rise in permanent volatility. 21

22 Further, note that the response of CPI to a one period shock in permanent volatility is not significant (Figure 4). The evidence of the responses of investment and other macroeconomic variables as presented in Figure 5 appears to be more overt and intriguing. The impulse responses clearly indicate that as the temporary volatility in oil price rises, the total investment falls significantly up to about six months following the volatility shock. Hence, the postponement of investment by the Thai firms in view of heightened transitory volatility is quite apparent and the shortlived nature of the adverse impact is indicative of the firms fast paced adjustment accommodating the effects of increased temporary volatility. As expected, with increased transitory volatility exerting detrimental pressure on aggregate demand, CPI also falls significantly for about a year or so after the shock in temporary volatility. Similarly, there is a significant dampening effect of the interest rates reflecting monetary expansions as anticipated policy response in view of negative impact of increased temporary volatility on investment and aggregate demand (Figure 5). However, no significant effect of rise in temporary oil price volatility on the exchange rate is detected. Overall, there appears to be clear evidence that it is the transitory volatility of oil prices, and not the permanent volatility, that exerts significant effects on some of the key macroeconomic variables in Thailand such as investment, CPI and interest rates. In other words, the significant adverse effects on investment and other variables due to increase in conditional or total volatility are primarily caused by the adverse responses of these variables to increased transitory volatility. INSERT FIGURE 3 ABOUT HERE 22

23 INSERT FIGURE 4 ABOUT HERE INSERT FIGURE 5 ABOUT HERE Investment responses by sectors The effect of oil price uncertainty provides a general idea as to how the firms and other economic agents behave in response to changing uncertainty. However, this information could be supplemented by delving into the sectoral responses to oil price uncertainty changes, which could ideally lay foundations for more effective policy formation. For Thailand s economy, a steady recovery in the post Asian Crisis period has been central to further its move up along the development ladder. Besides, the advent of the new millennium marks a number of policy liberalisations leading to freer international trade with escalated competition. Hence, investment sensitivities of various sectors to external shocks such as oil price convey important information from policy perspective. Figure 6a and 6b report the impulse responses of various sectors due to shock in oil price volatility. The responses of investment in food and textiles products (sector 1) are reported along the first column of Figure 6a, i.e., along the first three vertically stacked graphs. It is evident from these graphs that there is a significant dampening effect on investment in food and textiles products within the first three quarters, due to shocks in both conditional and transitory volatilities of oil prices. In contrast, investment barely tends to decline due to shock in permanent volatility, for a short period of about a quarter. The graphs stacked vertically along the middle column of Figure 6a represent impulse responses of sector 2, which is other transportable goods primarily comprising wood products, furniture, cork, straw and plaiting materials. As per these graphs, investment in these products declines significantly for about a quarter due to a shock in the conditional volatility. However, investment tends to decline 23

24 significantly in these products for an extended period of about 5 quarters following a one-time shock in transitory volatility. While the investment responses generally remain insignificant due to an increase in permanent volatility, there seems to be a small increase in investment between the 6 th and 1 th quarters following the shock in permanent volatility. Given the large export oriented furniture and wood products industries in Thailand and relatively low dependence of these industries on oil consumption, it is possible that investment responds positively over a longer horizon due to increased trend volatility with influx of investment from other sectors and consistent demand growth. The three graphs in the right most column of Figure 6a show the impulse responses of investment in metal, machineries and equipment. It is apparent that investment in these products decline significantly spanning up to about a year following an increase in conditional volatility. It is also evident that such response is overwhelmingly influenced by the declining investment due to a rise in transitory volatility. Compared to this, effect of permanent volatility on investment in the sector appears to be significantly negative for a shorter duration of a couple of quarters. Overall, there seems to be a more intense and prolonged adverse impact on investment in this sector due to increase in transitory volatility than to an increase in trend volatility of oil price (Figure 6a). This is an expected result, especially in view of the fact that Metal, machineries and equipment heavily embed use of oil exerting significant bearing on investments in Thailand s growing manufacturing and other industries. INSERT FIGURE 6a ABOUT HERE INSERT FIGURE 6b ABOUT HERE The impulse responses of investment in construction and land development sector (sector 4), as shown in the first column of Figure 6b, indicate that investment growth in this sector remains insignificant due to volatility shocks, for about a year or so, and then tends to 24

25 rise significantly for about a year, due to increase in trend volatility, and for about 6 months due to increase in transitory volatility. Thai construction and land development industries comprise the second largest sector of the country after the metal, machineries and equipment sector in terms of total investment. Also, investment in Thai construction and land developments have seen a consistent growth in the post Asian Crisis period, with little or negligible adversities exhibited in view of the more recent crises or global economic downturn such as the GFC. It is evident that while investment in this sector remains rather less responsive to short term external shocks, there could be a rise in investment in this sector due to its long term return prospects as investment in more vulnerable and oil dependent sectors falters with increased oil price uncertainty. Finally, investment in the business services sector (sector 5) that comprises investment in real estate services exhibits insignificant effect due to shocks in conditional, permanent and transitory volatilities of oil prices. This is an expected result since real estate services are unlikely to respond with reduced investment and growth as oil price uncertainty rises, due to low energy dependence nature of this sector. 6. CONCLUSIONS This paper empirically examines the effects of oil price volatility on the total and sectoral investments as well as a range of macroeconomic variables in Thailand based on a structural vector autoregressive (SVAR) model. Our analysis proceeds with developing and estimating a threshold based component GARCH to model oil prices thereby alienating the conditional, permanent and temporary oil price volatility. We model these volatility measures along with aggregate and sectoral real investment, consumer prices, real policy rate and exchange rate in a SVAR framework, generating impulse-response functions for predicting and forecasting effects of volatility shocks on investment and other macroeconomic variables. The results indicate that oil price volatility adversely affects aggregate real investment in Thailand, much 25

26 in conformation with the prediction by the irreversible theory of investment. We find that both the consumer prices and real interest rate tend to decline significantly due to positive shocks in oil market volatility, indicating reduced aggregate demand and required monetary expansions by the Bank of Thailand. A key contribution of this paper emanates from the analysis of volatility responses of various disaggregated sectors in Thailand. We find that investment significantly falls in textiles, wood and furniture, and machineries and equipment industries in response to an increase in oil price volatility. Another important finding of this study is that aggregate and sectoral investment as well as consumer prices and interest rates are found to have been overwhelmingly affected by positive shocks in transitory oil price volatility, and not by shocks in permanent volatility. Essentially, this evidence is in accord with the documented finding for the US and other developed markets. However these results contradict with those of Ibrahim and Ahmed (214) who find permanent oil price volatility shocks to be adversely affecting aggregate investments in Malaysia. The findings of this study have important implications for policy. Firstly, since Thai economy is not immune to oil price volatility, it is important to take recourse to prudent management of the economy in view of reduced investment and deflationary pressures. Secondly, while firms should be able to better cope with the increased temporary oil price volatility, the significant and stronger adverse effects of temporary oil volatility points to the absence of insurance markets to guard against any volatility risks or the lack of managerial expertise to identify and accommodate the adversities of heightened transitory or permanent oil price volatility. Thirdly, firms operating in the energy dependent industries such as textiles, furniture, wood products and machineries ought to remain vigilant and foresighted with regard to changes in oil price and hence the oil price volatility. These firms need to pursue well developed dynamic plans to weather and manage the effects of increased transitory volatilities. Fourthly, to create a more conducive environment so as to foster investment growth in Thailand 26