Causal Nexus Between Crude,Gold,Dollar and Stock Markets in India: Empirical Explorations Before,During and After the Global Recession

Abstract

This research study empirically examines the price linkages among oil, dollar, gold and stock markets in India over period from 1999:1 to 2019:12. We employ cointegrated vector error correction model (VECM) and Granger causality test to study the long-run and short-run relationships between commodity and financial markets before, during and after the global financial crisis. Our analysis finds the dependency on price movements in asset markets is time-varying and countercyclical in India. Findings suggest the asymmetric structure of price correlations among asset markets across three temporal periods on either side of the crisis. Our study offers useful insights into the strategic asset allocations to investors in response to economic cycles, to help optimise potential portfolio returns and provide protection towards some downside risks.

JEL Codes: C58, D53, F51

Keywords

Oil gold dollars stock returns asymmetric correlations VECM

I. Introduction

Research on macroeconomic effects in stock price behaviour is a long tradition in literature. For many years, academics have produced a vast number of studies, showing such relations in the context of traditional macroeconomic framework (Mukherjee & Naka, 1995). The economic impacts of oil, gold and currency prices on financial activities, however, have attracted considerable interests due to their importance in asset choices and portfolio diversifications (Raza et al., 2016). In fact, investors integrate alternative assets, including commodities with financial assets for implementing strategic allocations (Eychenne et al., 2011).

In investment finance, dynamic asset allocations enhance portfolio performance. As commodities have little correlations with stocks and bonds, they have the best risk-diversifying properties. The precious metals like gold can qualify as safe havens for portfolio optimisations (Baur & McDermott, 2016). Similarly, the prices of commodities like oil and gold affect the financial activities like stock investments and currency trading (Gokmenoglua & Fazlollahia, 2015). The price correlations between financial and commodity markets make stock and foreign exchange markets more sensitive to commodity price volatilities (Tang & Xiong, 2012). The inverted dynamics of the equity market with the price volatilities in commodity markets suggest strong evidence of diversification benefits to investors (Abid et al., 2019). The empirical shreds of evidence for the existence of time-varying commodity risk premiums related to business cycles are substantial in the literature (Deaton & Laroque, 1992). Hence, an analysis of the price correlations among commodity and financial markets at different economic episodes gains much significance today.

This article calls into question the causal nexus among oil, dollar, stock and gold markets in the Indian context. These assets possess significant diversification characteristics and share similar statistical properties, particularly in their price behaviour (Parimi, 2018). Their integrative price structure in India, one of the leading emerging markets of the world, is highly relevant. India has been among the highest recipients of foreign capital and most significant importers of oil and gold for the past many decades. Although financial markets play a decisive role in the foundation of a stable and efficient financial system of the economy, the prices of commodities like gold and oil significantly influence the prices of financial assets trading there. Commodity markets are not merely a haven to hedge against financial risk, but they are an alternative investment with a higher amount of certainty during periods of financial market instability (Le & Chang, 2012). Oil and gold are the most widely traded commodities in every emerging economy. People initially traded oil for its fundamental purposes but, over time, gained a permanent place in the investment portfolio (Kilian & Park, 2009). When the financial markets become volatile, investors tend to move out the risky assets like stock and invest into assets like gold (Gaur & Bansal, 2010). The development of future markets encourage investors to use oil and gold to diversify this portfolio or to attempt to achieve abnormal return (Zhang, 2013). However, as India needs to import both assets from the global market, the dollar fluctuations have their effects on the value of their transactions. These problems are faced in all countries, developed and developing, but they are more severe for developing nations. Oil shocks may have an impact on exchange rates, thereby the equilibrium in balance of payments. Thus, this article investigates whether oil, dollar, stock and gold markets in India are or are not cointegrated.

The debate over price linkages between asset markets raises a valid argument that the market participants with interests in investments are facing challenges in estimating the risk and trading, as the information from those markets are complex and uncertain. The increased interactions of financial and commodity markets have served as one fast transmission channel of the global financial crisis to the developing world (Nissanke, 2010). Hence, this article derives its motivation from the following considerations. First, most of the empirical studies in the area aim at bilateral or trilateral linkages among markets. Apart from the bidirectional causality analysis, using cointegration and VECM techniques, we carry out our investigation on price changes in four asset markets simultaneously. Second, empirically testing the expectation hypothesis as an indicator of developed and mature asset market seems necessary for Indian context, where the oil price shocks and gold price spikes expect to have more influence on stock prices than in other markets. One of the key contributions of this article is to test the market efficiency of select asset market characteristics with different functional structures as the process will help to identify and control for direct and indirect effects in portfolio choices. Third, the cointegration approach makes it possible to answer any questions connected with bilateral interactions. Indeed, the approach allows for relying on the potential linkage between markets through other indirect effects. Finally, compared with others, we implement analysis using a more extended period from January 1999 to December 2019 and divide it into several sub-sample frames. This is the first academic research study to evaluate the asset price correlations across three temporal periods: pre-crisis, during crisis and post-crisis. Accordingly, we can see the impacts from the economic cycle on price convergence or divergence among alternative asset markets through separately applying the casualty tests across expansion and recession periods. Such an application would facilitate direct comparisons on the directions and magnitudes of causal relations. Conversely, we attempt to answer the following questions: are the oil, dollar, stock and gold markets interdependent in emerging economies? If yes, what nature and direction of causality describes their interdependencies? Are the price convergences across asset markets uniform over economic cycles? In this research, we partly replicate the previous research studies, but with newer data covering various economic cycles, including boom, recession and resilience years.

The rest of this article proceeds as follows: Section II looks at the theoretical considerations and reviews the relevant empirical literature. Section III deals with descriptions of data and outlines the empirical methodology applied. After reporting the estimation outputs, Section IV discusses the empirical application of the model forecast with a framework of the methodology suggested before we conclude the study with the summary of our findings, policy implications and potentials for further research in Section V.

II. Theory and Empirical Literature

There are growing appeals for the price linkages between the commodity markets and financial markets. Most researchers consider assets such as crude (hereafter oil), gold, USD (hereafter dollar) and stocks as the sample for their study. The debate in most studies is much about nature and degree of causality among markets in terms of price variations. Moreover, there exists a lack of consensus among economists and researchers regarding the nexus between various markets (Kilian & Park, 2009). Empirical or applied research draws more often on the theoretical conceptualisations, hypotheses and framework posited by previous researchers. Since our research design is mainly empirical, the discussions in the literature review section foreground the existing theory and evidence on the price linkages among asset markets, and expect to provide a sound understanding of the price integration process as well as pass-through effect.

Theory

This study needs to discuss theoretical literature on the possible nexus among four asset markets at first. The impact of oil price changes on stock prices is generally through the inflation route. The increase in oil prices causes an increase in the cost of production that leads to an increase in consumer price levels in the economy (Brown & Yucel, 2002). Rising inflation adversely affects the corporate income due to rising costs and slowly adjusting output prices, reducing profits and, therefore, the share price (DeFina, 1991). The linkage between exchange rates and stock price is explained by two approaches. The first one is ‘good market approach’, which suggests that the exchange rate changes expect to have a direct bearing on stock prices as it affects both firms’ profit earnings and cost of capital on external borrowings (Dornbusch & Fischer, 1980). The second one is ‘portfolio balance approach’ that emphasises the role of capital account transactions in an economy. The augmented capital inflows improve the supply of dollars leading to exchange rate appreciation, while increased capital outflows affect the demand for dollars leading to exchange rate depreciation.

During the crises with financial uncertainties and risks, investors seek strategies to diversify their assets. They consider gold as a substitute asset and a safe place for their risk aversion (Gokmenoglua & Fazlollahi, 2015). Investors are motivated to find gold as a safe investment to hedge their risks (Arfaoui & Ben, 2017). There are three direct pass-through channels of oil prices to exchange rates—terms of trade, wealth effect and the portfolio reallocation (Buetzer et al., 2016). The terms of trade channel links the price of oil to the inflation, affecting real exchange rate. When oil prices rise, wealth is transferred from oil-importing nations to oil-exporting nations (in dollar terms), the reflection of which will be in the foreign trade and current account balances. Then, the currencies of oil-exporting countries appreciate, and currencies of oil-importing countries depreciate. Oil exporters’ relative preferences for dollar assets can decide their exchange rates according to portfolio effects (Coudert et al., 2008). The causality from exchange rates to oil prices is the fact that the supplier of oil price denominates its value in dollar terms. Accordingly, an appreciation of the dollar increases the price of oil, measured in terms of the domestic currency, and this lowers the demand for oil outside the USA, resulting in a drop in the oil price, all else equal. Studies explain the causality between gold and oil prices through the inflation channel. When oil prices rise, inflation rises, and the price of gold (as a good) goes up as well. When the value of the dollar appreciates against the Indian rupee, the price of gold tends to fall. It is because gold becomes more expensive in Indian rupee; hence, its demand recedes.

Empirical Studies

Price changes in oil markets act as an essential determinant of stock market returns. As pioneering research in this area, Kling (1985) concludes that the increases in the price of oil could lead to stock market declines. Supporting this finding, about 1 years later, Jones and Kaul (1996) observed consistent negative relations between oil price changes and aggregate stock returns in US and Canadian markets. However, studies carried out by Wei (2003) and Huang et al. (1996) fail to produce significant causality between oil and stock returns in various market contexts of the USA, the UK, Japan and Canada. Kilian and Park (2009) observe that the reaction of US stock returns to oil price shocks depends significantly on whether the changes in oil prices are due to demand-side or supply-side factors. Basher et al. (2012), based on a structural vector autoregressive (SVAR) modelling on monthly data for the period from 1988 to 2008, provide time series evidence on the negative short-run impact of oil price increases on emerging market stock returns and exchange rates. In a global perspective, Arfaoui and Ben (2017) found an inverse correlation between oil and stock prices.

The studies carried out by Rangasamy (2017) produced evidence of pass-through effects from energy prices to inflation. Empirical studies by Chen et al. (2005) and Benakovic and Posedel (2010) reveal that inflation impacts stock markets negatively. However, crude oil prices influence Indian stock markets negatively (Sharma et al., 2018). Zarei et al. (2019) provide empirical evidences on the inverse relationship between the exchange rate and stock returns. Tian and Ma (2010) employ the cointegration autoregressive distributed lag (ARDL) approach to study the relationship between the exchange rate and the Chinese share market. The result showed that the exchange rate and money supply influence the performance of stock markets positively. In the Indian context, while Saji and Harikumar (2013) and Mahapatra and Bhaduri (2019) found stock returns react significantly to foreign exchange rate fluctuations in India.

There is empirical evidence on the rise in gold rates when equity prices fall, particularly during turbulent times (Gaur & Bansal, 2010). Investors are motivated to find gold as a safe investment to hedge their risks (Arfaoui & Ben, 2017). Stock prices have a significant effect on gold prices in selected Asian markets during the global financial crisis in 2008 (Ziaei, 2012). International gold prices positively affect the stock prices, while the oil price affects them negatively (Singhal et al., 2019). Singh and Kishor (2014) for the period from 2002 to 2013 and Tripathy and Tripathy (2016) for the period from 1990 to 2016 have produced some contradicting results on the linkage between gold and equity prices in India. Sahu et al. (2014) reveal that the Indian stock markets and oil prices are strongly exogenous.

Oil shocks affect all countries, but their effects on the exchange rate closely relate to the asymmetries between the economies (Krugman, 1983). Exchange rate shock has a significant negative impact on oil prices in the short run, while oil price shocks have a significant impact on exchange rate changes in the long run (Brahmasrene et al., 2014). Oil prices negatively influence the exchange rate in the long run in emerging markets (Singhal et al., 2019). However, Yiew et al. (2019) suggest that the adjustment process between the exchange rate and the oil price is constant for a certain period that facilitates the prediction of the direction of exchange rate movement. Zhang and Wei (2010) suggest long-term equilibrium between the oil and gold markets where the change in the prices of former linear Granger causes the volatility of the gold prices. Simakova (2011) established the existence of a long-term relationship between gold and oil prices. Fattouh (2010) examined the asymmetry in the spread adjustment process for oil and metals. Arfaoui and Ben (2017) reveal direct linkage between gold rate and oil prices, and the ARDL cointegration methodology adopted by Gokmenoglua and Fazlollahia (2015) find bilateral causality between oil and gold prices. Jain and Biswal (2016) produce evidence in favour of fall in gold prices and crude oil prices, which cause a fall in the value of the Indian Rupee and the benchmark stock index, that is, Sensex.

The studies of Kilian and Vega (2010) claimed that different kinds of news might produce a variety of stochastic arrival processes and, thus, communicate varying effects on gold price behaviour. The studies performed by Singhal et al. (2019) indicate that in an emerging market, gold price does not have any significant impact on the exchange rate. Samanta and Zadeh (2012) found strong cointegration relationship between gold and dollar prices. Shiva and Sethi (2015) investigate the relationship among the gold, dollar and stock prices in India, and their findings convey unidirectional causality that runs from gold prices to stock prices to dollar prices. Mensah et al. (2017) showed that there is an inverse relationship between the dollar and oil prices. Sathyanarayana et al. (2018), in their research, have tried to investigate three major macroeconomic indicators such as oil, gold and forex (Dollar) on the performance of Indian stock and concluded that the dollar, gold and oil prices are significant in the transmission of the volatility of the chosen indices and have the power to transmit shock on stock prices in India.

III. Methodology

Data

We use monthly data for the sampling period, spanning from January 1999 to December 2019. The data are Brent oil price, gold price, the monthly average of the foreign exchange values of the dollar against the Indian rupee and the National Stock Exchange (NSE) Nifty stock market index. We have accessed the data on oil prices, and stock index from the Federal Reserve Bank of St. Louis, and National Stock Exchange, respectively. The gold and dollar prices have been sourced from the Reserve Bank of India. All the price data series used in the empirical analysis are in rupee terms. Gold, dollar and stock price data are readily available in rupee terms, but crude prices are in dollar terms. Hence, we multiply crude price data with rupee–dollar exchange rates so as to get whole price data series in rupee terms. We compute the monthly percentage change of asset prices, using the conventional method of log-transformation of price series.

To assess the price convergence or divergence among markets across business cycles, we divide the whole sample period into three sub-sample frames or panels. The first period or panel A covers the period from 1999 to 2008 to capture the effects of the capital market boom in India. The second period or panel B spans the period from 2009 to 2014 to reflect the distressed financial conditions due to the recessionary pressures, and the third sample period panel C ranges from 2015 to 2019 to allow for the effects of regained momentum in price trends after fading off the effects of recession. Along with the full sample period of panel D, we re-estimate all the models designed in this research study using each sub-sample data to measure the degree of both long-run and short-run effects of the covariations in prices among all markets observed in this study. Such an analysis with changing dynamics of economic cycles expects to make out that the level of integration promising diversification benefits through rational asset choices over different markets.

Model Estimation

This research study measures the co-movement of prices among oil, dollar, stock and gold markets in emerging markets like India. The relationship between asset prices is either long term or short term, and our research aims to assess the market integration behaviour in terms of prices. Precisely, we are looking into the long-run equilibrium relationship and the short-term causality among the four asset markets. Accordingly, our empirical design is based on two estimation procedures: the cointegration test suggested by Johansen (1991) to investigate the asymmetry of price integration among markets and the causality test under vector autoregression (VAR) modelling proposed by Granger (1969) to measure the correlations in their prices. The procedure for executing cointegration and causality has already been well documented in the literature; hence, we do not give much elaboration on our empirical procedure. Instead, we explain only the main equations to be estimated for our market integration analysis.

Cointegration Analysis

We adopt the maximum likelihood approach of Johansen to test the long-run price relations of selected asset markets. If the five stock price indices share a common stochastic trend, then they are said to be cointegrated (Christensen & Nielsen, 2006). The presence of cointegration relation forms the basis of the vector error correction (VEC) specification. Fundamentally, the test begins its cointegration estimation with a VAR(k) model for a set of ‘g’ variables as follows:

Y_{t} = A_{1} Y_{t - 1} + A_{2} Y_{t - 2} + \dots + A_{p} Y_{t - k} + {BX}_{t} + ε

(1)

In order to use the Johansen test, the VAR (Equation [1]) needs to be turned into a vector error correction model (VECM) of the form:

Δ Y_{t} = Π Y_{t - k} + \sum_{j = 1}^{k - 1} r_{j} Δ Y_{t - j} + ε

(2)

where

Π = (\sum_{i = 1}^{k} β_{j}) - I_{g} r_{i} = (\sum_{j = 1}^{i} β_{j}) - I_{g}

The VAR contains g variables in a first-differenced form on the left-hand side (LHS) and k − 1 lags of the dependent variables (differences) on the right-hand side (RHS), each with a ${'r coefficient matrix}^{'}$ attached to it. The lag length employed in VECM can affect the Johansen test; hence, we should select the lag length optimally, using the suitable lag selection criterion.

The cointegration test procedures centre on examination of the ′Π matrix’. Π $Π$ can be interpreted as a long-run coefficient matrix, since, in equilibrium, all the $Δ Y_{t - i}$ will be 0, and setting the error terms ′ε $' ε$ ’ to their expected value of 0 will leave $Π Y_{t - k} = 0$ .

There are two statistics for cointegration under the Johansen approach, which are formulated as follows:

\hat{λ_{trace} (r) = - T \sum_{i = r + 1}^{g} In (1 - {\hat{λ}}_{i})}

(3)

and

\hat{λ_{t max} (r, r + 1) = - T In (1 - {\hat{λ}}_{r + i})}

(4)

$λ_{trace}$ is a joint test where the null hypothesis is that the number of cointegrating vectors is less than or equal to r against an unspecified alternative hypothesis that there are more than r. It starts with ρ eigenvalues, and then, successively, the largest is removed. $λ_{trace} = 0$ when all the $λ_{i} = 0$ , for i = 1, …, g.

Error Correction Model

If the asset price series are cointegrated, these equations would need an additional error correction term, and the desired model would be:

Δ Y_{t} = α_{0} + δ_{1} (Y_{t - i} - γ X_{t - i}) + \sum_{i = 1}^{k} α_{1} Δ Y_{t - i} + \sum_{i = 1}^{k} α_{2} Δ X_{t - i} + ε_{1 t}

(5)

Δ X_{t} = β_{0} + δ_{2} (Y_{t - i} - γ X_{t - i}) + \sum_{i = 1}^{k} β_{1} Δ Y_{t - i} + \sum_{i = 1}^{k} β_{2} Δ X_{t - i} + ε_{2 t}

(6)

where

$(Y_{t - i} - γ X_{t - i})$ is an error correction term computed from the level-form estimate of the long-run relationship between Y and X. Causality now can be asserted by the significance of the parameters α₂, $β_{1,}$ δ₁ and δ₂. If δ₁ is significantly different from 0, and δ₂ is not, then if Y and X drift apart, the Y variable will correct to restore equilibrium. If δ1 is not significantly different from zero, but δ2 is, then X variable corrects. If both δ1 and δ2 are significant, then both Y and X will have a correction to restore equilibrium (Hamilton, 1994). Error Correction Model (ECM) confirms the dynamic relationship between the prices. The short-run adjustment parameter for this model measures the speed of price transmission, while the long-run multiplier indicates the degree of price transmission from one price to the other.

Granger Causality

We employ causality test suggested by Granger (1969) to test cross-covariance in price distributions among four asset markets. The Granger causality test in our research estimates short-run causality between asset price series and decides whether one price series (X) is useful in the prediction of the other (Y). If the values of X carry statistically significant information content about the future values of Y, then we infer that the time series X Granger-cause Y. Granger causality test for the stock price series acts as the first-step estimation with the following equations:

Δ Y_{t} = α_{0} + \sum_{i = 1}^{k} α_{1} Δ Y_{t - i} + \sum_{i = 1}^{k} α_{2} Δ x_{t - i} + ε_{1 t}

(8)

Δ X_{t} = β_{0} + \sum_{i = 1}^{k} β_{1} Δ Y_{t - i} + \sum_{i = 1}^{k} β_{2} Δ x_{t - i} + ε_{2 t}

(9)

where Y_t and X_t are asset price series, Y_{t − i}> and X_{t − i} are the lagged variable vectors, matrices α and β are to be estimated and ε_t is uncorrelated white-noise error term.

IV. Results and Discussions

Descriptive Statistics

Table 1 explains the statistical moments of monthly price changes or returns of nifty, dollar, oil and gold in percentage terms. The return volatility is higher in oil and nifty, while the same is lower in dollar and gold investments. Skewness is negative for nifty and oil, while the same is positive for dollar (except in panel A) and gold assets. However, both the skewness and the kurtosis measures of the nifty are significantly different from those of other assets. The lower degree of negative skewness disregards the probability of significant decreases in nifty prices. The return distributions of nifty are almost symmetrical, as, in most cases, we find the value of its skewness hovering around zero. Interestingly, the frequency distribution of price changes in all the assets during the past 5 years of the study is normal, which is further evidenced by their coefficients of kurtosis. We could make a similar observation, concerning the price changes, particularly Nifty and oil in panel A. However, for the next few years, their distributions have become more leptokurtic. Except in panel C, the descriptive analysis indicates the most riskiness associated with oil in terms of the leptokurtic structure of price distributions.

Table 1.

Summary Statistics.

Panel	Variables	Mean	S.D.	Skewness	Kurtosis
	Nifty	1.671	6.345	–0.548	2.950
Panel A (1999:2008)	Dollar	–0.056	1.030	–0.804	6.328
	Oil	2.005	7.808	–0.550	3.289
	Gold	1.186	4.117	0.633	4.625
	Nifty	0.841	8.102	–0.047	5.075
Panel B (2009:2014)	Dollar	0.395	2.776	0.303	3.483
	Oil	1.346	8.097	–0.996	5.272
	Gold	1.658	4.209	0.100	3.778
	Nifty	0.989	3.860	–0.078	2.77
Panel C (2015:2019)	Dollar	0.228	1.585	0.017	2.89
	Oil	–0.237	9.291	–0.418	2.85
	Gold	0.344	2.993	0.388	4.007
	Nifty	1.220	6.209	–0.283	5.354
Panel D (1999:2019)	Dollar	0.217	1.988	0.628	5.901
	Oil	1.103	8.251	–0.645	3.747
	Gold	0.929	3.928	0.525	4.502

Source: The author.

Cointegration Analysis

Most asset prices trend over time, that provides non-stationary data, producing spurious regressions (Brooks, 2008). The causality and cointegration tests assume stationary variables or the presence of random behaviour of the series. Therefore, our analysis checks the order of integration with augmented Dickey–Fuller (ADF) test estimates at first and with a less restrictive test of Phillips–Perron (PP) after that. The results of the unit root tests reported in Table 2 indicate that each of series in their level form is non-stationary in two alternative models, with the presence of an intercept, and with a deterministic trend and an intercept. However, first differencing the series induces the stationarity in all variables, suggesting that all our price series are integrated of order one. The results of PP tests are not fundamentally different from the individual ADF results. Analytically, the results from tests at the level point to the presence of a unit root in all cases. The results, after first differencing the series, reject the null hypothesis of non-stationary or the presence of a unit root at the 1% significance level, suggesting that all the price series of assets included in the sample are first-order integrated. Hence, we treat them as I(1) process and proceed with the causality and cointegration analysis.

Table 2.

Unit Root Test Results.

Variables	Augmented Dickey–Fuller Test Results				Phillips–Perron Test Results
	Level		First Difference		Level		First Difference
	With Intercept T	With Trend and Intercept	With Intercept	With Trend and Intercept	With Intercept	With Trend and Intercept	With Intercept	With Trend and Intercept
Nifty	0.6365 (0.9908)	–2.5053 (0.3256)	–15.2891 (0.0000*)	–15.3529 (0.0000*)	0.6102 (0.9848)	–2.4460 (0.3549)	–15.2891 (0.0000*)	–15.3639 (0.0000*)
Dollar	0.3094 (0.9602)	–1.4115 (0.858)	–13.9110 (0.0000*)	–13.9114 (0.0000*)	–0.3138 (0.9195)	–1.7182 (0.7474)	–13.8751 (0.0000*)	–13.8751 (0.0000*)
Oil	–2.2079 (0.2035)	–2.6852 (0.2427)	–10.4779 (0.0000*)	–10.4734 (0.0000*)	–1.9245 (0.3207)	–2.3263 (0.4277)	–10.2701 (0.0000*)	–10.2632 (0.0000*)
Gold	–0.1764 (0.9381)	–1.8763 (0.6637)	–14.1432 (0.0000*)	–14.1222 (0.0000*)	–0.1772 (0.9380)	–1.9044 (0.6491)	–14.0831 (0.0000*)	–14.0606 (0.0000*)

Source: The author.

Note: *Significant at the 1% level.

Our objective is to test the cointegration, thereby long-run causality, among four asset markets included in our sample. The Johansen method is sensitive to the lag length; hence, we estimate the VAR system that comprises prices of nifty, dollar, oil and gold for various lag lengths. The values that compute for the relevant Akaike information criterion (AIC), Schwarz Bayesian criterion (SBC) and Hannan–Quinn criterion (HQC) decide the appropriate lag length for the cointegration analysis. We estimate six alternative VAR (p), p = 1, 2, 3, …, 6 models over the four sample periods, including the full period analysis and report the corresponding results in Table 3. As expected, the maximised values of the log-likelihood increase with p and all the three information criteria indicate that the optimal lag length is one in every panel we have formed. Therefore, we test cointegration using one lag in the VAR system.

Table 3.

Selection of Optimal Lag in the VAR System.

Period	Lag	LL	P(LR)	AIC	SBC	HQC
Panel A(1999:2008)	1	785.1466	–	–15.1514*	–14.6335*	–14.9418*
	2	790.1001	0.8714	–14.9327	–14.0005	–14.5553
	3	796.5148	0.6852	–14.7429	–14.3985	–14.1978
Panel B(2009:2014)	1	429.4679	–	–12.5990*	–11.9299*	–12.3350*
	2	439.5178	0.2158	–12.4159	–11.2116	–11.9407
	3	448.6274	0.3112	–12.2039	–10.4644	–11.5175
Panel C(2015:2019)	1	451.2439	–	–15.4016*	–14.6782*	–15.1211*
	2	460.4187	0.3038	–15.1578	–13.8558	–14.6530
	3	474.1059	0.0375	–15.0752	–13.1945	–14.3461
Panel D(1999:2019)	1	1663.0241	–	–13.9237*	–13.6304*	–13.8056*
	2	1671.0117	0.4547	–13.8560	–13.3277	–13.6430
	3	1680.2585	0.2958	13.7988	–13.0356	–13.4911

Source: The author.

Note: *Optimal lag length.

We also need to determine the appropriate restrictions on the intercept and trends in our estimation models. The study follows the Pantula principle of model selection, that is, we estimate all three alternative models that are moving from the most restrictive to the least restrictive model. Comparing the trace or maximal eigenvalue test statistics against the corresponding critical values, we choose that estimation model, which cannot reject the null hypothesis for the first time. Table 4 presents the results from three estimating models for each panel from which we have to choose the model, which is more robust in prediction. The cointegration model to estimate causality among asset prices is consistent in the analysis. Accordingly, in all the four panels, for the first time, we cannot reject the null hypothesis is for the first model, ‘restricted intercept and no trends’ in the levels of data. Further, irrespective of the panels used, we can see that both trace and maximal eigenvalue test statistics suggest the existence of one cointegrating relationship among price movements of assets included in the sample.

Table 4.

The Pantula Principle Test Results, K = 1.

	R	n-r	Model 1 Restricted Constant		Model 2 Unrestricted Constant		Mode3 Restricted Trend
	R	n-r	λ _trace	λ _max	λ _trace	λ _max	λ _trace	λ _max
Panel A(1999:2008)	0	4	47.933	25.626	53.784	26.475	42.255	21.118
	1	3	22.317*	12.724*	27.39*	14.854	21.137	14.133
	2	2	9.533	7.051	12.435	8.714	7.003	4.9632
Panel B(2009:2014)	0	4	58.949	29.716	50.397	27.478	45.735	24.026
	1	3	31.240*	16.604*	22.919	11.231	21.709	10.739
	2	2	14.635	10.652	11.688	8.785	10.970	9.598
Panel C(2015:2019)	0	4	52.055	28.767	41.971	20.876	55.077	22.074
	1	3	26.288*	18.137*	21.094	14.733	33.002	15.753
	2	2	13.228	8.0410	6.361	6.264	17.249	11.605
Panel D(1999:2019)	0	4	66.029	32.799	45.526	29.444	55.212	30.642
	1	3	33.229*	21.000*	16.081	10.232	24.569	12.505
	2	2	12.229	9.185	5.849	5.139	12.065	8.292

Source: The author.

Note: * Indicates not rejecting the null hypothesis of no cointegration for the first time at least at the 5% level.

We have one cointegrating vector of four asset prices since we cannot reject the null hypothesis of one cointegrating vector against the alternative hypothesis of two or more cointegrating vectors in panels. We present the results of the cointegration models in Table 5. We have one cointegrating vector each in panels, which is presented in Table 6. From these results, we observe a positive relationship of Indian stock prices with dollar prices and negative relationship with oil and gold prices (except in panel C). The magnitude of the relationship is higher during the post-crisis period once the recovery becomes evident. The results again provide evidence in favour of the hypothesis that there is a link between the stock market and other asset markets in emerging economies. The models seem to capture most of the implications of the underlying economic theory and are consistent with the findings of previous studies, testing the cointegration for each of the variables, reflecting the market integration. The presence of a cointegrating relationship between markets suggests that there are some factors common to stock, dollar, oil and gold markets that drive their prices in a developing country like India although the oil price effects weakly persist in the analysis.

Table 5.

Johansen Test for Cointegration.

H₀	Eigenvalue	λ _trace	p-Value	λ _max	p-Value
r = 0	0.2129	47.933	0.0061	32.799	0.0080
r ≤ 1	0.1254	22.317	0.0865*	21.000	0.0934*
r ≤ 2	0.1036	9.533	0.1374*	9.185	0.1435*
r = 0	0.3213	58.949	0.0159	29.716	0.0423
r ≤ 1	0.2085	31.240	0.1258*	16.604	0.2674*
r ≤ 2	0.1393	14.635	0.2537*	10.652	0.2886*
r = 0	0.3400	52.055	0.0176	28.767	0.0425
r ≤ 1	0.2536	26.288*	0.1235*	18.137*	0.1280*
r ≤ 2	0.1216	13.228	0.4567*	8.0410	0.3830*
r = 0	0.127739	66.029	0.0030	32.799	0.0136
r ≤ 1	0.083783	33.229	0.0802*	21.000	0.0751*
r ≤ 2	0.037548	12.229	0.4285*	9.185	0.414*

Source: The author.

Note: * Indicates the existence of cointegration at least at the 5% level.

Table 6.

Cointegrating Equations of Nifty with Dollar, Oil and Gold Prices.

	Dollar	Oil	Gold
Panel A	0.0927	0.0953	–1.224
Panel B	0.5972	–0.2964	–0.5842
Panel C	0.4281	–0.3265	0.0800
Panel D	0.3208	–0.0840	–0.2287

Source: The author.

Table 7 reports summary results from the VECM and the basic diagnostics about the residuals of each error correction equations. More specifically, we provide the coefficients and the corresponding t-statistics for the ECM components, which, in some cases, have the expected signs and are statistically significant. During the pre-crisis period, the dollar market is the only one market that holds long-run relationship with the Indian stock market. However, in resilience, while the decline in gold prices in global markets boosted the nifty prices by 15% , oil appreciated well with the gradual recovery of the global economy from the shackles of the crisis. For the past 5 years, the Indian stock prices have almost decoupled from the variations in dollar and gold prices. However, the oil price fluctuations to an extent influence nifty price movements. Again, in full sample period analysis, VECM results reveal the long-run equilibrium relationship between nifty and oil prices. The insignificance of the ECM component for the remaining asset variables indicates that they are weakly exogenous to the models.

Table 7.

Summary Results from the VECMs and Diagnostic Tests.

	∇Nifty	∇Dollar	∇Oil	∇Gold
ecm(–1)	0.007 (2.626)	–0.009 (–2.326)	0.0114 (0.8933)	0.0116 (3.854)
R ²	0.061	0.049	0.008	0.124
S.E. of Regression	0.063	0.010	0.041	0.075
ecm(–1)	0.0332 (0.9360)	–0.0048 (0.6912)	0.0509 (2.8082)	–0.1521 (–4.7724)
R ²	0.052	0.002	0.102	0.245
S.E. of Regression	0.081	0.028	0.042	0.074
ecm(–1)	–0.1121 (–5.0378)	0.007 (0.0734)	–0.007 (–0.745)	0.0132 (1.975)
R ²	0.185	0.031	0.026	0.034
S.E. of Regression	0.078	0.012	0.026	0.0384
ecm(–1)	–0.0798 (–2.1296)	0.0032 (0.5025)	–0.0126 (–1.0164)	0.0499 (3.5106)
R ²	0.1999	0.1262	0.0700	0.2406
S.E. of Regression	0.0895	0.0151	0.0297	0.0339

Source: The author.

Granger Causality

Since VECM fails to produce evidence on the long-run equilibrium relationship among many asset markets in panels, we want to capture the short-run causality. The integrative structure of asset markets often assumes the bidirectional causality between the price series. The fundamental principle of causality expects the interdependency of price series across markets continually. Therefore, the research, next, examines the evidence of short-run causality between the price series. Causality in asset markets refers more to the efficiency of one market to predict the other. We employ the reasonably simple test procedure of causality suggested by Granger (1969) to look into the direction of causality between the price series. Table 8 presents the causality test results. The causality between the price series is quite divergent across panels, and the analysis fails to produce a consistent relationship in this regard.

Table 8.

Granger Causality Test Results.

Direction of Causality	Panel A 1999:2008	Panel B 2009:2014	Panel C 2015:2019	Panel D 1999:2019
Nifty causes Dollar	4.1469** (0.0428)	6.9572* (0.0096)	0.4952 (0.4840)	4.6457** (0.0352)
Dollar causes Nifty	5.5178* (0.0196)	0.2250 (0.6362)	1.2971 (0.2587)	0.0346 (0.8530)
Nifty causes Oil	0.6076 (0.4365)	1.0884 (0.3001)	0.6901 (0.4090)	1.6294 (0.2068)
Oil causes Nifty	0.0139 (0.9061)	7.5705* (0.0072)	11.5752* (0.0001)	1.9352 (0.1694)
Nifty causes Gold	1.4869 (0.2239)	7.2269* (0.0084)	0.3241 (0.5709)	2.8771*** (0.0862)
Gold causes Nifty	2.6219*** (0.0921)	0.3579 (0.5543)	1.0572 (0.3075)	9.0327* (0.0039
Dollar causes Oil	5.5522** (0.0193)	3.0261*** (0.0849)	6.4124 ** (0.0136)	5.1183** (0.0162)
Oil causes Dollar	1.1060 (0.2940)	9.3901*** (0.0028)	0.6433 (0.4353)	0.0221 (0.8822)
Dollar causes Gold	0.9999 (0.3183)	1.7286 (0.1965)	1.2929 (0.2595)	0.1215 (0.7286)
Gold causes Dollar	4.4396 (0.0362)	9.2628* (0.0030)	4.1676** (0.0451)	5.9243** (0.0180)
Oil causes Gold	0.0433 (0.8354)	3.8807*** (0.0515)	0.2249 (0.6348)	1.0574 (0.3080)
Gold causes Oil	2.6674*** (0.0892)	3.0387*** (0.0843)	2.8836*** (0.0941)	5.6503 (0.0207)

Source: The author.

Note: *, ** and *** Significant at the 1%, 5% and 10% level, respectively.

We find bidirectional causality between nifty and dollar and between nifty and gold in panel A and panel D, respectively. Similarly, bidirectional causality exists between oil and dollar, and oil and gold in panel B, the period after the global financial crisis of 2008. Nifty continuously cause rupee–dollar rates, except during the past 5 years of panel C. While nifty index changes in India affect gold prices in resilience and full period analysis, the oil price changes in global markets do not expose to price shocks in Indian markets as we expected. However, the effects of oil price fluctuations are evident in Indian stock prices during the resilience and the subsequent period. The persisting impact of dollar fluctuations on gold and oil prices is evident in all panels; the research study hardly produces the evidence of reverse causality in both cases. Thus, the results of causality tests on price structures in asset markets produce mixed results. A close observation reveals that the dependency on price movements across many asset markets is time-varying and countercyclical.

V. Conclusion

This research study, using the causality and cointegration tests, assessed cyclical effects of the short-run and long-run causality among dollar, oil, gold and stock prices in India for the period from 1999 to 2019. Here, it is of our utmost interest to examine the interactions of Indian stock markets with the oil, forex and gold markets. The causality results confirm the theoretical postulation that it is the fluctuations in stock prices that lead to exchange rate variations. A booming stock market would attract capital flows from global economies that push up the demand for domestic currency. The reverse would occur, when the market goes bearish, where the investors would sell off their holdings to avoid further losses and repatriate their investments to home in dollar terms that create demand for foreign currency, leading to depreciation of the domestic currency. As a result, rising (declining) stock prices would lead to an appreciation (depreciation) in exchange rates. In fast-growing emerging markets like India, this is quite persisting, and this finding is consistent with the observations of Tabak (2006).

The results show the evidence of causality running from oil price changes to stock market prices during specific years, despite the integration between two markets being time-varying and inconsistent. Moreover, no evidence of causality exists in either direction between two markets in full sample period analysis, and the findings are directly in line with the previous findings of Bhunia (2013). Similarly, during the same period, Granger causality test, confirms the presence of causality between gold prices and Indian stock prices, and this result ties well with previous studies performed by Shiva and Sethi (2015). Additionally, the results from the VECM provides evidence for the existence of a long-run relationship between the Indian stock prices and gold prices. Thus, the Indian stock market is relatively independent of dollar markets, and its prices are less sensitive to the price fluctuations. However, the research study produces conclusive evidence of long-term causality from gold to nifty since 2009, suggesting that stock prices in India share a common stochastic trend with gold prices. At this stage of understanding, we believe that the investors in emerging markets like India can consider gold as an alternative investment with inherent value and can hold that value when equities decline in markets.

The policy implications of the findings of this research study are explicit on several grounds. The asymmetry among asset performance revealed by our study signals weak integration, promising diversification benefits, through rational asset allocations for investors. From the strategic side, the loosely linked asset markets can contribute to the portfolio optimisations as they deviate significantly from the long-run equilibrium path. The investors can improve their portfolio performance by adjusting for the business cycles in their asset allocations. For example, portfolio allocations can be more towards stocks and crude oil during expansions and to gold during market downturns. Further, the oil price does cause the stock price to change, and gold has the potential of an effective hedge and diversifier against any uncertainty as the recovery from the current pandemic conditions prolongs. This would significantly help monetary authorities and policymakers to adjust the policy mechanism for better price regulations and financialisation of major commodity markets. The process ultimately helps economies to achieve their twin economic objectives—exchange rate stability and stock market growth.

The findings of the study are not free from limitations. Even though the current findings can be a good method to make out the short-run and long-run relations among four asset markets, we cannot assure the repetition of the trend in future. Unquestionably, new oil supplies, alternative energy resource discoveries, the continuation of current sagging economic conditions and the emerging geopolitical trends, all can be significant in shaping the factors encouraging dependence structure of asset prices beyond the scope of our findings. Moreover, another apparent limitation of this research study is that the degree of stock market interdependencies with the rest of the markets can often be sector-specific. Hence, an extension to this study is to explore the influence of the selected asset price variables on the stock prices of different sectors, such as manufacturing, transport, energy and power in several emerging markets, which can provide more useful and valuable policy recommendations.

Footnotes

Declaration of Conflicting Interests

Funding

The author received no financial support for the research, authorship and/or publication of this article.

References

Abid

, Dhaoui

, Goutte

, & Guesmi

(2019). Hedging and diversification across commodity assets. Applied Economics . https://doi.org/10.1080/00036846.2019.1693016

Arfaoui

, & Ben

R. A.

(2017). Oil, gold, US dollar and stock market interdependencies: A global analytical insight. European Journal of Management and Business Economics , 26, 3, 278–293.

Basher

S. A.

, Haug

A. A.

, & Sadorsky

(2012). Oil prices, exchange rates and emerging stock markets. Energy Economics , 34, 227–240.

Baur

D. K.

, & McDermott

T. K. J.

(2016). Why is gold a safe haven? Journal of Behavioral and Experimental Finance , 10, 63–71.

Benakovic

, & Posedel

(2010). Do macroeconomic factors matter for stock returns? Evidence from estimating a multifactor model on the Croatian market (Working Paper No. 1012). University of Zagreb.

Bhunia

(2013). Cointegration and causal relationship among crude price, domestic gold price and financial variables: An evidence of BSE and NSE. Journal of Contemporary Issues in Business Research , 2, 1–10.

Brooks

(2008). Introductory econometrics for finance . Cambridge University Press.

Brown

P. A.

, & Yucel

M. K.

(2001). Energy prices and aggregate economic activity: An interpretive survey. The Quarterly Review of Economics and Finance , 42(2), 193–208.

Brahmasrene

, Huang

J. -C.

, & Sissoko

(2014). Crude oil prices and exchange rates: Causality, variance decomposition and impulse. Energy Economics , 44, 407–412.

10.

Buetzer

, Habib

M. M.

, & Stracca

(2016). Global exchange rate configurations: Do oil shocks matter? IMF Economic Review, 64(3), 443–470.

11.

Chen

M. H.

, Kim

W. G.

, & Kim

H. J.

(2005). Macro and non-macro explanatory factors of Chinese hotel stock returns. International Journal of Hospitality Management , 24, 243–58.

12.

Christensen

B. J.

, & Nielsen

M. O.

(2006). Asymptotic normality of narrow-band least squares in the stationary fractional cointegration model and volatility forecasting. Journal of Econometrics , 133(1), 343–371.

13.

Coudert

, Mignon

, & Penot

(2008). Oil price and the dollar. Energy Studies Review, 15(2), 45–58.

14.

Deaton

, & Laroque

(1992). On the behaviour of commodity prices. Review of Economic Studies , 59(1), 1–23.

15.

DeFina

R. H.

(1991). Does inflation depress the stock market? Federal Reserve Bank Philadelphia Business Review , 11, 3–12.

16.

Dornbusch

, & Fisher

(1980). Exchange rates and the current account. American Economic Review , 70, 960–71.

17.

Eychenne

, Martinetti

, & Roncalli

(2011). Strategic asset allocation . White Paper. https://www.lyxor.com

18.

Fattouh

(2010). The dynamics of crude oil price differentials. Energy Economics , 32(2), 334–342.

19.

Gaur

, & Bansal

(2010). A comparative study of gold price movements in Indian and global markets. Indian Journal of Finance , 4(2), 32–37.

20.

Gokmenoglua

K. K.

, & Fazlollahia

(2015). The interactions among gold, oil, and stock market: Evidence from S&P500. Procedia Economics and Finance , 25, 478–488.

21.

Granger

C. W. J.

(1969). Investigating causal relations by econometric models and cross spectral methods. Econometrica , 37(3), 424–38.

22.

Hamilton

J. D.

(1994). A new approach to the economic analysis of nonstationary time series and the business cycle. Econometrica , 57(2), 357–384.

23.

Huang

, Masulis

, & Stoll

(1996). Energy shocks and financial markets. Journal of Futures Markets , 16, 1–17.

24.

Jain

, & Biswal

P. C.

(2016). Dynamic linkages among oil price, gold price, exchange rate, and stock market in India. Resources Policy , 49, 179–185.

25.

Johansen

(1991). Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models. Econometrica , 59(6), 551–580.

26.

Jones

C. M.

, & Kaul

(1996). Oil and the stock markets. The Journal of Finance , 51(2), 463–491.

27.

Kilian

, & Park

(2009). The impact of oil price shocks on the US stock market. International Economic Review , 50(4), 1267–1287.

28.

Kilian

, & Vega

(2010). Do energy prices respond to US macroeconomic news? A test of the hypothesis of predetermined energy prices. The Review of Economics and Statistics , 93(2), 660–671.

29.

Kling

J. L.

(1985). Oil price shocks and stock market behavior. The Journal of Portfolio Management , 12(1), 34–39.

30.

Krugman

P. R.

(1983). Oil shocks and exchange rate dynamics. In Frankel

J. A.

(Ed.), Exchange rates and international macroeconomics (pp. 259–284). University of Chicago Press.

31.

T. H.

, & Chang

(2012). Oil price shocks and gold returns. International Economics, 131(3), 71–104.

32.

Mensah

, Obi

, & Bokpin

(2017). Cointegration test of oil price and us dollar exchange rates for some oil dependent economies. Research in International Business and Finance , 42, 304–311.

33.

Mahapatra

, & Bhaduri

S. N.

(2019). Dynamics of the impact of currency fluctuations on stock markets in India: Assessing the pricing of exchange rate risks. Borsa Istanbul Review , 19(1), 15–23.

34.

Mukherjee

T. K.

, & Naka

(1995). Dynamic relations between macroeconomic variables and the Japanese stock market: An application of a vector error correction model. Journal of Financial Research , 18, 223–37.

35.

Nissanke

(2010). Commodity market structures, evolving governance and policy issues. In Nissanke

, Mavrotas

(Eds.), Commodities, governance and economic development under globalization. Palgrave Macmillan. https://doi.org/10.1057/9780230274020_4

36.

Parimi

(2018). Factors influencing the gold prices: An empirical investigation in the Indian context. Theoretical Economic Letters , 8, 3444–3456.

37.

Rangasamy

(2017). The impact of petrol price movements on South African inflation. Journal of Energy in Southern Africa , 28(1), 120–132.

38.

Raza

, Shahzad

S. J. H.

, Tiwari

A. N.

, & Shahbaz

(2016). Asymmetric impact of gold, oil prices and their volatilities on stock prices of emerging markets. Resource Policy , 49, 290–301.

39.

Sahu

, Bandopadhyay

, & Mondal

(2014). Crude oil price, exchange rate and emerging stock market: Evidence from India. Jurnal Pengurusan , 42, 75–88.

40.

Saji

T. G.

, & Harikumar

(2013). Can macroeconomic factors lead stock returns in India? Artha Vijnana , 55(3), 289–307.

41.

Samanta

, & Zadeh

(2012). Co-movements of oil, gold, the US dollar, and stocks. Modern Economy , 3(1), 111–117.

42.

Sathyanarayana

, Harish

S. N.

, & Gargesha

(2018). Volatility in crude oil prices and its impact on Indian stock market: Evidence from BSE Sensex. SDMIMD Journal of Management , 9(1), 65–76.

43.

Sharma

, Giri

, Vardhan

, Surange

, Shetty

, & Shetty

(2018). Relationship between crude oil prices and stock market: Evidence from India. International Journal of Energy Economics and Policy , 8(4), 331–337.

44.

Shiva

, & Sethi

(2015). Understanding dynamic relationship among gold price, exchange rate and stock markets: Evidence in Indian context. Global Business Review , 16(5), 93–111.

45.

Simakova

(2011). Analysis of the relationship between oil and gold prices. Journal of Finance , 51(1), 651–662.

46.

Singhal

, Choudhary

, & Chandra

B. P.

(2019). Return and volatility linkages among International crude oil price, gold price, exchange rate and stock markets: Evidence from Mexico. Resources Policy , 60, 255–261.

47.

Singh

R. P.

, & Kishor

(2014). Co-integration of gold price movement with nifty indices: A study in Indian context. Transnational Corporations Review , 6(1), 42–57.

48.

Tabak

(2006). The dynamic relationship between stock prices and exchange rates: Evidence for Brazil (Working Paper No. 124). Banco Central Do Brazil.

49.

Tang

, & Xiong

(2012). Index investment and the financialization of commodities. Financial Analysts Journal , 68(6), 54–74.

50.

Tian

G. G.

, & Ma

(2010). The relationship between stock returns and the foreign exchange rate: The ARDL approach. Journal of the Asia Pacific Economy , 15(4), 490–508.

51.

Tripathy

, & Tripathy

(2016). A study on dynamic relationship between gold price and stock market price in India. European Journal of Economics, Finance and Administrative Sciences , 88, 23–35.

52.

Wei

(2003). Energy, the stock market, and the putty-clay investment model. American Economic Review , 93, 311–323.

53.

Yiew

T. -H.

, Yip

C. -Y.

, Tan

Y. -L.

, Habibullah

M. S.

, & Khadijah

C. A.

(2019). Can oil prices predict the direction of exchange rate movements? An empirical and economic analysis for the case of India. Economic Research , 32(1), 812–823.

54.

Zarei

, Ariff

, & Bhatti

M. I.

(2019). The impact of exchange rates on stock market returns: New evidence from seven free-floating currencies. The European Journal of Finance , 25(14), 1277–1288.

55.

Ziaei

S. M.

(2012). Effects of gold price on equity, bond and domestic credit: Evidence from ASEAN +3. Procedia—Social and Behavioral Sciences , 40, 341–346.

56.

Zhang

Y. -J.

, & Wei

Y. -M

. (2010). The crude oil market and the gold market: Evidence for cointegration, causality and price discovery. Resources Policy , 35(3), 168–177.

57.

Zhang

Y. -J

. (2013). The links between the price of oil and the value of US dollar. International Journal of Energy Economics and Policy , 3(4), 341–351.