ارتباط متقابل بین بازارهای نفت خام و کالاهای کشاورزی

In this paper, we investigate cross-correlations between crude oil and agricultural commodity markets. Based on a popular statistical test proposed by Podobnik et al. (2009), we find that the linear return cross-correlations are significant at larger lag lengths and the volatility cross-correlations are highly significant at all of the lag lengths under consideration. Using a detrended cross-correlation analysis (DCCA), we find that the return cross-correlations are persistent for corn and soybean and anti-persistent for oat and soybean. The volatility cross-correlations are strongly persistent. Using a nonlinear cross-correlation measure, our results show that cross-correlations are relatively weak but they are significant for smaller time scales. For larger time scales, the cross-correlations are not significant. The reason may be that information transmission from crude oil market to agriculture markets can complete within a certain period of time. Finally, based on multifractal extension of DCCA, we find that the cross-correlations are multifractal and high oil prices partly contribute to food crisis during the period of 2006–mid-2008

Oil price surges in recent years inspire people to develop the alternative energy. The bioethanol and biodiesel extracted from corn and soybean, respectively, are considered as the appropriate substitutes of crude oil. Thus, increases in oil prices can result in the increases in corn and soybean prices and finally lead to the increases in prices of other agricultural commodities as the planting acreage is limited in a certain period of time. Additionally, higher crude oil prices will lead to higher production costs. In this sense, people suspect that large increases in agricultural commodity prices in 2006–2008 may be caused by rising crude oil prices [1], [2] and [3]. As a response of higher oil prices, the central banks will adjust the interest rate [4] and [5] which may also lead to changes in commodity prices according to the standard theory of cost carry. In this paper, we will investigate the cross-correlations between crude oil and agricultural commodity markets. The investigation of related issue can be seen in a plenty of recent studies and their conclusions are mixed. Some researchers show the significant oil–agricultural commodity price relationships. For example, Mitchell [3] finds that energy price contribute 15%–20% of total cost of the US agriculture. Hence, the food crisis in 2006–2008 is partly due to persistent increases in oil prices. Harri and Hudson [6] find that crude oil market volatility spills over corn futures market. Harri et al. [7] investigate the cointegration between oil and primary agricultural commodity prices. They show that corn, cotton and soybean prices are significantly correlated with oil prices. Nazlioglu et al. [8] consider that volatility spillover from oil market to agriculture market is significant only after the food crisis. This finding is further supported by Du et al. [9] in the framework of a bivariate stochastic volatility model and by Ji and Fan [10] based on a bivariate EGARCH model. Some economists document that oil–agricultural commodity price linkages are not significant. For example, Nazlioglu and Soytas [11] find that the response of Turkish agriculture price to oil price changes is not significant. Reboredo [12] finds that the tail dependence between oil and agriculture returns is very weak. Gilbert [13] shows that oil price changes do not Granger cause agriculture price changes. A major limitation in existing studies is that they investigate the return or volatility relationships in a linear framework.1 As the nonlinearity in asset prices and volatilities has been generally accepted [14], it is more appropriate to investigate the relationships between crude oil and agricultural commodity markets in a nonlinear framework. In this paper, we will fill this gap and do it from a completely fresh perspective. We borrow the methods from statistical physics and therefore in the area of econophysics our work is the first that investigates oil–agriculture price linkages. In this paper, we analyze nonlinear cross-correlations between crude oil and agricultural commodity markets. Although cross-correlations in financial markets have been detected in many studies [15], [16], [17], [18], [19] and [20], to the best of our knowledge the cross-correlations between crude oil and agricultural commodity markets have not been investigated in existing studies. Our empirical procedure contains four main steps. First, using a statistical test proposed by Podobnik et al. [21], we qualitatively examine whether the linear cross-correlations between crude oil and agricultural commodity markets are significant. Our results show that at smaller lag lengths, the return cross-correlations are not significant and at larger lag lengths, the return cross-correlations are significant at 10% level. The volatility cross-correlations are highly significant at each of the lag lengths under consideration. Second, we use a detrended cross-correlation analysis (DCCA) proposed by Podobnik and Stanley [22] to analyze nonlinear cross-correlations. We find that return cross-correlations are persistent for corn and soybean and anti-persistent for oat and wheat. The volatility cross-correlations are strongly anti-persistent. Third, we use a cross-correlation coefficient proposed by Zebende [23] and find that the cross-correlations are weak but significant for smaller time scales. For larger time scales, the cross-correlations are not significant. The reason may be that the information transmission from crude oil market to agriculture markets completes within a certain period of time. Finally, we use the multifractal form of DCCA [24] and find that both return and volatility cross-correlations are multifractal. By analyzing the cross-correlations during the period of recent food crisis, we find that high oil prices adequately contribute to increases in agriculture prices during the period from 2006 to mid-2008. We also give some economic and modeling implications based on the empirical findings. The remainder of this paper is organized as follows. Section 2 gives a description of methodology. Section 3 shows data and some preliminary analysis. Section 4 shows empirical results. Section 5 is some relevant discussions. Section 6 concludes.

In this paper, we have investigated cross-correlations between crude oil and agricultural commodity markets. We focus on both return and volatility cross-correlations.Weuse a statistical test proposed by Podobnik et al. [21] to examine whether the linear cross-correlations are significant. Our findings indicate that the return cross-correlations are not significant for smaller lag lengths and are marginal significant for lag lengths larger than 100. In existing studies, vector autoregression models (VAR) or vector error correction (VEC) models are always used to capture linear relationships between crude oil and agricultural commodity prices. Therefore, the lag lengths of these models should be longer than 100 business days. For example, if daily data is employed, the lag lengths of VAR or VEC models should be longer than 100 to well capture crosscorrelations. If monthly data is employed, the lag lengths should be longer than 4. Using Podobnik’s Q test, we also find the highly significant cross-correlations between oil and agricultural commodity volatility series. This can be explained by the strong auto-correlations in individual volatility series. To quantify the cross-correlations, we use the detrended cross-correlation analysis proposed by Podobnik and Stanley [22]. Our evidence indicates that the cross-correlations between oil and corn and soybean price returns are persistent, whereas the cross-correlations between oil and oat and wheat returns are anti-persistent. That is, an increase in crude oil price is more likely to be followed by an increase in corn (soybean) price and a decrease in oat (wheat) price. The existence of long-range cross-correlations implies that considering the past changes in crude oil prices can improve the predictability of agricultural commodity prices. The predictability of commodity prices is in contrast with the efficient market hypothesis in Fama’s sense. The cross-correlations for volatility series are highly persistent. Our findings on long-range cross-correlations also have important modeling implications. For example, economic models incorporating long-range cross-correlations may better capturing interactions between crude oil and agricultural commodity markets. In this sense, the fractionally cointegrated model is more appreciated than the conventional VEC model. Then, we use a cross-correlation coefficient proposed by Zebende [23] to measure cross-correlations. Our results indicate that the cross-correlations for returns and volatilities are relatively weak but significant for smaller time scales. The insignificant cross-correlations for larger time scales may be due to the fact that the information transmissions from crude oil market to agriculture markets can complete within the period of a year. We investigate the cross-correlations during the period of recent food crisis. Our results show that during this period, cross-correlation coefficients between crude oil and agricultural commodity markets are stronger than those during common periods, indicating that high oil prices partly contribute to this food crisis. Finally, we use the extension of DCCA to investigate the multifractality in cross-correlations. We find that the generalized Hurst exponents depend on the fluctuation order, suggesting the existence of multifractality. We would like to conclude this paper by outlining two points which deserve our further investigations. First, there are several papers that show evidence of long memory and multifractality in interest rates [44–46]. Commodity prices should be linked to interest rates. This could be a potential source of persistent behavior-through interest rates. Quantitatively measuring the role of interest rate in cross-correlations between oil price and agricultural commodity price changes is interesting. Second, several important papers by Cajueiro and Tabak [30,31,48,49] have shown that long-range dependence in financial and commodity markets display time-varying behaviors. Therefore, whether oil–agriculture cross-correlations are also time-varying deserves our further investigation.