Investigating the international prices of wheat and rice

Rezitis, Anthony N.; Ntinou, Anastasia G.; Pachis, Dimitris N. Article Investigating the international prices of wheat and rice Agricultural and Food Economics Provided in Cooperation with: Italian Society of Agricultural Economics (SIDEA) Suggested Citation: Rezitis, Anthony N.; Ntinou, Anastasia G.; Pachis, Dimitris N. (2015) : Investigating the international prices of wheat and rice, Agricultural and Food Economics, ISSN 2193-7532, Springer, Heidelberg, Vol. 3, pp. 1-17, https://doi.org/10.1186/s40100-015-0035-4 This Version is available at: https://hdl.handle.net/10419/179053 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ RESEARCH Open Access Investigating the international prices of wheat and rice Anthony N Rezitis 1,2* , Anastasia G Ntinou 2 and Dimitris N Pachis 2 * Correspondence: [email protected] 1 Economics and Management, University of Helsinki, Latokartanokaari 9, P.O.Box 27, Helsinki, Finland 2 Business Administration of Food and Agricultural Enterprises, University of Patras, G. Seferi 2, Agrinio 30100, Greece Abstract This paper investigates the international quarterly prices of wheat and rice from 1983(1) to 2012(4). The empirical analysis takes place with the structural time series methodology which decomposes the price series into their trend, cycle, seasonal and irregular components. The empirical results indicate that wheat prices present cyclical behavior while rice prices except for cyclicality are mainly governed by the irregular component. The results strain the importance of treating wheat and rice like two distinct commodities that require country specific and commodity specific policy measures. Finally, the impact of the shrinking Chinese grain stocks after 2002 is proposed as an important factor that resulted in the 2008 price spike. Keywords: Structural time series; International prices; Wheat; Rice Background From late 2006 through to mid-2008, the world witnessed unusual price increases in major agricultural commodities like wheat and rice (Dethier and Effenberger, 2011). However, until then, the world was used to generally low price variability in agricultural commodities especially after the volatile decades of 1970s and 1980s (Gilbert, 2006). This episode renewed the interest in agricultural commodities since they are tightly connected to food security, particularly in developing and underdeveloped countries. Moreover, developed countries were also concerned because price spikes adversely affect the welfare of producers and consumers. Therefore, the analysis of international agricultural commodity prices has once again been under the spotlight in agricultural economics research. This research has mainly focused on investigating the reasons for the price spike between 2006 and 2008. However, it has also enriched the literature with new insights into the drivers of agricultural commodities price changes. The price formation of agricultural commodities is mainly attributed to the variation of the market fundamentals, which are the supply and demand. Following Gilbert and Morgan (2010), natural shocks caused by weather conditions or diseases and the area planted are the main factors that affect production. Moreover, technological infrastructure is a key driver of the quantity produced. On the other hand, consumption varies because of changes in income levels or, in the prices of substitutes, or because of shifts in tastes. The extent to which given production and consumption shocks translate into the formation of prices depends on supply and demand elasticities, which reflect the responsiveness of producers and consumers to price changes. The supply and demand © 2015 Rezitis et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http:// creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. Rezitis et al. Agricultural and Food Economics (2015) 3:16 DOI 10.1186/s40100-015-0035-4 price elasticities are low in the short run (during the cropping period), especially if the stocks of agricultural products are low (Wright and Williams, 1991; Deaton and Laroque, 1992). Apart from market fundamentals, sudden shifts in policy (Christiaensen, 2009), input prices, exchange rates, and trading patterns as well as speculation, affect price formation (Interagency Report, 2011; Gilbert and Morgan, 2010). Furthermore, the ever strengthening relationship between agricultural commodities and energy prices (oil and biofuels) will continue to bind them together, while the effect of macroeconomic fundamentals should not be overlooked (OECD, 2008). The purpose of this paper is to investigate the formation of the international prices of wheat and rice. Wheat is a staple that is produced and consumed mainly in temperate regions, while it is also used as an input into the production of meat products. The main producers and exporters of wheat are the United States, the European Union, Canada, Australia, Argentina, and more recently the Russian Federation, Ukraine, and Kazakhstan (FAO, 2009). Wheat is a commodity that is freely traded internationally. On the other hand, rice is a staple in Southeast Asian countries, in central and West Africa, in the Caribbean as well as in South America. Southeast Asian countries account for approximately 25 percent of global production (Baldwin et al., 2012). Furthermore, a small amount of rice is traded internationally; its price is not freely determined by the market but it is sold or bought in contracted prices (Timmer, 2010). The aforementioned characteristics of wheat and rice indicate that they are not correlated in terms of production, consumption, or trade patterns. Wheat is a staple in richer countries where meat consumption is also high, whereas rice is a staple in poorer countries. Thus, their price shocks are not interdependent, especially since rice is traded in the futures markets of Chicago and Bangkok at low volumes. The increased prices of these two staples have affected several countries, which in their efforts to protect their domestic markets have applied export restrictions and tariffs. However, these actions have reduced the amount of traded grains internationally, which is something that has exacerbated the price increases (Abbott et al., 2011). After the 2006–2008 crisis, the price levels decreased, but without going back to their previous levels. A researcher has the option of using different classes of models for the modeling of the observed as well as the unobserved price changes of a time series. A popular choice for the modelling of prices is the AutoRegressive Integrated Moving Average (ARIMA) model. This class of models was advocated by Box and Jenkins (1976) and its key attribute is the pursuit of stationarity with as much differencing as necessary. The problem with differencing is to know when it has eliminated enough of the trending and the seasonality of a time series, whilst the identification of the model by the sample autocorrelation function is imprecise due to its high sampling variability. However, ARIMA models are good for forecasting since the elimination of the trend and the seasonal component is not a problem (Durbin and Koopman, 2012). Another popular choice for the modeling of the relationships between prices is the Vector AutoRegressive model of Sims (1982). Nowadays, this class of models is based on the search for unit roots that indicate the presence of trending or seasonal behavior, and eliminate them by differencing as well as by the addition of a co-integrated relationship among the variables representing their common trend. The study of unobserved price changes in a series takes place by the use of (Generalized) AutoRegressive Conditional Heteroskedastic ([G]ARCH) models proposed by Bollerslev (1986) and Engle (1982), respectively. This Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 2 of 17 class of models was frequently used by researchers for the study of the price spike in agricultural commodities from late 2006 to mid-2008. These three classes of models attempt to model the price series formation and interactions by abandoning the structure of a fully specified model. A fully specified model attempts to use as many explanatory variables as possible for the modeling of a price series. However, no matter how many explanatory variables one could use, the researcher will be partly able to account for the variability of the time series. Thus, the fourth class of model, the Structural Time Series (STS) model, maintains the structure of the time series, which is composed of trend and, seasonal, cyclical, and irregular components, without however attempting to build a fully specified model. The trend determines the direction shifts occurring with permanent changes representing the long-term nature of the series. Seasonal effects are related to natural conditions. The cycle represents the magnitude and the length of fluctuations that occur in the short run, while the irregular component identifies unexpected events. Kalman (1960) introduced the STS models with the aim of addressing a wide variety of problems by means of their flexibility. A key aspect of the STS models is that the pursuit of stationarity is not necessary since the time series components such as trend and, seasonal and cyclical factors can accommodate evolving distributions over time. Moreover, an STS model does not aim to represent the underlying data generating process, but it represents the stylized facts of the price series in terms of the decomposition. Therefore, the structural nature of the model allows for direct interpretation (Harvey, 1989). Structural time series analysis has been used for the investigation of many price series that are quite different in their nature. This paper next points out a number of relevant studies that have analyzed the price series relevant to agricultural markets. In particular, Ubide (1997) investigated inflation in Mozambique and showed that the seasonal behavior of inflation is affected by agricultural products’seasonality. Faliva (1994) analyzed Italian unemployment in the agricultural sector. Shepherd (2006) used a structural time series model for the world cotton supply. Furthermore, Fatiga and Misra (2007) used a multivariate unobserved components model for cotton, wool, rayon, and polyester world prices. In their study, Crispin and Dale (1998) studied the US broiler industry and found that a significant role is played by the feed costs as well as technology advances. A similar study carried out by Chidmi and Fatiga (2007) examined the formation of US beef, pork, and poultry prices using a multivariate unobserved components approach. Farley and Murphy (1997) studied sockeye salmon stocks in Alaska and northern British Columbia with the intention of explaining the trending behavior of the sockeye salmon catch. Bhar and Hamori (2007) investigated corn, soybean and sugar futures’prices in order to extract information about the short and long-term dynamics of each series. Moreover, Heymans (2008) employed the unobserved component model in white and yellow maize futures prices. In a recent paper, Rezitis and Sassi (2013) analyzed the price movements of a commodity price index and provided future price predictions. Mirzabaev and Tsegai (2012) examined the impact of adverse weather conditions in Central Asia on wheat and potato price series and found that unfavorable weather conditions lead to higher wheat and potato prices, while the international price spikes of primary commodities negatively affect domestic prices. In this study, the international prices of wheat and rice are investigated with a univariate Structural Time Series model. The analysis decomposes the price series to their unobserved components of trend and, seasonal, cyclical, and irregular without searching Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 3 of 17 for any commonalities, since the market characteristics of the two products set them apart. Furthermore, the information acquired for the past behavior of the series are then used for forecasting. This study enriches the literature on the international price formation of agricultural commodities by utilizing a method that is not based on the properties of the time series but on the stylized facts that govern the series. Therefore, better forecasts can be achieved, whilst the interpretation of the results is straightforward due to the structural form of the model used. Moreover, the structural breaks and the temporal effects of the series can be identified and modeled leading to a model with a better fit to the data. The empirical analysis shows that wheat prices present cyclical behavior while rice prices, except for cyclicality, are governed by the irregular component. The results identify the role of the decreased Chinese stocks in the path of events that led to the 2008 price spike while underlining the importance of commodity and country-specific policy measures for wheat and rice. The paper proceeds as follows: in section 2 the theoretical framework of the model is presented; section 3 focuses on data presentation; the empirical results are provided in section 4; and, discussion and conclusions are presented in sections 5 and 6, respectively. Methods The structural time series methodology by Koopman et al. (2009) is used to decompose the wheat and rice price series into trend, seasonal, cycle and irregular components. Letting the logarithm of wheat or rice prices presented by y t then the stochastic linear model is given by: yt¼μtþγtþψtþX h j¼1 λjdj;tþεt;εt∼Ν0;σ2 ε  ð1Þ where μ t is the trend, γ t is the seasonal, Ψ t is the cycle, d j,t is an intervention (dummy) variable and ε t is the irregular component. In the present analysis the modelling procedure showed that the trend and seasonal components are fixed. Specifically, the trend component is defined as: μt¼μt−1þβt−1þηt;where ηt∼ΝID 0;σ2 η  ;σ2 η¼0ð2Þ where β t is the slope of the trend which is defined as : βt¼βt−1þζtwhere ζt∼NID 0;σ2 ζ  ;σ2 ζ¼0ð3Þ The irregular ε t , the level disturbance (η t ) and the slope disturbance (ζ t )aremutually interrelated, where η t and ζ t are normally and independently distributed white noise processes with zero means and variance σ2 ηand σ2 ζrespectively. Moreover, seasonality is the systematic calendar related influence that is captured by the seasonal component γ t . The seasonal component has a trigonometric deterministic seasonal structure, which is given by: Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 4 of 17 γt¼X s=2½ i¼1 γj;tð4Þ where each γ j,t is generated by : γj;t γ jt  ¼cosλjsinλj −sinλjcosλj  γj;t−1 γ j;t−1  þωj;t ω j;t  ;j¼1; ::::; s=2½;t¼1; ::::; Tð5Þ Note that the λ j =2π/s is the frequency, in radians, while the seasonal disturbances ω t and ω tare two mutually uncorrelated normally and independently distributed disturbances with zero mean and common variance σ2 ω. Since, the seasonal component has a deterministic form then σ2 ω¼0. For seven, the component at j= s/2 collapses to: γj;t¼γj;t−1cosλjþωj;tð6Þ The statistical specification of a cycle ψ t is given by: ψt ψ t  ¼ρψ cosλcsinλc −sinλccosλc  ψt−1 ψ t−1  þκt κ t  ;t¼1; ::::; Tð7Þ where ρ ψ is the damping factor (0 < ρ ψ ≤1) which reflects the speed with which various price fluctuations are dampened, λ c is the frequency (0 < λ c ≤π) in radians that defines the magnitude of the fluctuations of the price series and K t ,κ tare two mutually uncorrelated normally and independently distributed disturbances with zero mean and common variance σ2 κ. The duration of the cycle is 2π/λ c , showing the time length to complete the fluctuations. Higher order cycles are used for smoothing the extracted cycles. An nth-order univariate cycle is defined by ψ1;t ψ 1;t  ¼ρcosλcsinλc −sinλccosλc  ψ1;t−1 ψ 1;t−1  þκt κ t  ð8Þ ψi;t−1 ψ i;t−1  ¼ρcosλcsinλc −sinλccosλc  ψi;t−1 ψ i;t−1  þψi−1;t−1 ψ i−1;t−1  ;i¼2; ::::; nð9Þ κt κ t  ∼NID 0 0  ;σ2 κ 0 0 σ2 κ  ð10Þ where κt∼NID 0;σ2 κ  . The parameter ρis called the damping factor (0 < ρ≤1) while 0 <λ c ≤π. Finally, intervention variables d j,t are dummy (indicator) variables that are used to capture structural breaks or outlying (irregular) observations. The structural break is related to the unusual value of the level disturbance of the series and is modeled by a step intervention variable which is zero before the event and one after. An outlier can be captured by an impulse intervention that takes the value of one at the time of the outlier and zero elsewhere. Structural breaks represent unusual permanent changes while outliers represent unusual temporary changes. Moreover, a large value of the slope disturbance can be thought as a structural break in the slope and is modeled as a staircase intervention where the trend variable takes the values 1,2,3,…starting in the period after the break. The introduction of structural breaks and outliers in unobserved component models enhance the estimation of parameters and provide effective forecasts. Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 5 of 17 The data-set 1,2 used consists of quarterly data of the logarithm of the prices of wheat (Lhard-Wh) referring to Hard Red Winter and rice (LRice) referring to Thailand rice from 1983(1) to 2012(4). The aforementioned data are obtained from the World Bank, are measured in US Dollars per Metric Ton and are nominal. Figs. 1 and 2 show the evolution of the logarithm of wheat (Lhard-Wh) and rice prices (LRice). The graphs indicate that up to 2002 both rice and wheat prices exhibit cyclical behavior with a slight increasing trend while a seasonal pattern is observed. However, after 2002 the cyclical behavior is disrupted and the uprising trend becomes much steeper. In 2008 this upward movements reaches its peak. After 2008 prices decreased but they persisted in higher levels than before. The descriptive statistics of the price series in levels and logarithms are reported in Table 1. Results The structural time series models of wheat and rice are estimated by maximum likelihood. The price series are decomposed in their components by the smoothing algorithm proposed in Koopman et al. (2009). The most appropriate model for wheat consists of a trend with level and slope, a second order seasonal, two cycles of order one and interventions. The wheat decomposition is presented in Fig. 3. For rice, the model is comprised by a trend with level, seasonal, two cycles of order two, interventions and irregular while is exhibited in Fig. 4. The smoothing algorithm showed very strong convergence for both price series. Table 2 presents the diagnostics and goodness-of-fit statistics of the log-likelihood (LogL), the normality test following a χ 2 distribution with 2 degrees of freedom (Nχ2 2  ), a heteroskedasticity test following a F distribution with (33,33) degrees of freedom for wheat (H 35 (F 35,35 )) and (35,35) degrees of freedom (H 35 (F 35,35 )) for rice. Moreover, the classical Durbin-Watson test statistic (DW), a Box-Ljung statistic based on the first 16 Fig. 1 Evolution of Wheat price index. (Lhard-Wh) from 1983(1) to 2012(4) Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 6 of 17 autocorrelations tested against a χ 2 distribution with 9 degrees of freedom (Q( 16,9 )) and the coefficient of determination (R 2 ) are presented for both price series. The residuals are assumed to be normally and independently distributed in a correctly specified model. The standardized residuals as well as their correlogram, spectral density and density are shown in Figs. 5 and 6 for wheat and rice prices, respectively. The statistics exhibited in Table 2 and the graphs presented in Figs. 3 and 4 show that the estimated models are robust. More specifically, the correlogram and spectral density suggest that the residuals are not autocorrelated since the theoretical spectrum for white-noise residuals is a horizontal straight line. Furthermore, the aforementioned statistics do not give evidence of misspecifications in the estimated models. Table 3 presents the q-ratios showing the variances of the disturbances of the components affecting the structure of the price series. For wheat, cycles 1 and 2 account for the majority of the fluctuations since the trend, the seasonal and the irregular components are fixed. Regarding rice prices, most of the fluctuations are attributed, apart from cycles 1 and 2, to the irregular component since trend and seasonal are fixed. Furthermore, the q-ratios indicate that the fluctuations of cycle 2 are mainly responsible for variations of wheat prices while the fluctuations of irregular are mainly responsible for variations of rice prices. The parameters of cycle 1 and 2 for wheat and rice prices are depicted in Table 4. More specifically, the shorter cycle for wheat (cycle 1) has a variance of 0.002, a period of 1.75 years and a damping factor of 0.986 while the longer cycle (cycle 2) has a variance of 0.019, a period of 7.7 years and a damping factor of 0.906. The shorter cycle for rice (cycle 1) has a variance of 0.001, a period of 1.1 years and a damping factor of 0.515 while the longer cycle (cycle 2) has a variance of 0.002, a period of 15.5 years, Fig. 2 Evolution of Rice price index. (LRice) from 1983(1) to 2012(4) Table 1 Descriptive statistics Variables Means Standard Deviations Variables (logarithms) Means Standard Deviations Wheat 173.0 65.83 Lhard-Wh 5.096 0.32 Rice 314.9 134.9 LRice 5.679 0.36 Notes: Rice stands for the rice price index (2005 = 100), Hard Wheat stands for the Hard Wheat price index (2005 = 100) Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 7 of 17 and a damping factor of 0.814. The cycles of both wheat and rice prices show high degree of persistence as is clearly depicted in Table 4. The maximum likelihood estimates of the final state vector and the regression effects (intervention dummies) of wheat and rice are presented in Table 5 while the state vector anti-log analysis is presented in Table 6. With respect to wheat, the level μ T of the price index at the end of the period (2012(4)) is 269.9 (see Table 6) which is well above the mean value of 173.0 (see Table 1). Furthermore, the slope component is significant indicating that the growth rate of wheat prices is 4.5 % per year (see Table 5). The Fig. 3 Decomposition of Wheat price index (Lhard-Wh) into trend, seasonal, cycle 1 and cycle 2 Fig. 4 Decomposition of Rice price index (LRice) into trend, seasonal, cycle 1 and 2 plus the irregular Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 8 of 17 the same rationale should be developed by less developed countries so that they will be able to have access to the hedging markets of futures and options. On the other hand, richer countries are not so worried about their food security as much as the effects of food prices on inflation. Despite the fact that developed countries are both more open to world markets and more capable of isolating their farmers from world market volatility, they would also benefit from cooperating with the less developed countries on mitigating the undesirable effects of high food prices. The above analysis indicates that policy makers should use the right mix of policies, depending on the staple food products of their country as well as their country’s economic strength. In this way, structural time series analysis can be a valuable tool for policy makers. Fig. 9 Prediction testing for the Wheat price index (Lhard-Wh) Fig. 10 Prediction testing for the Rice price index (LRice) Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 15 of 17 Conclusions The present study has analyzed the international prices of wheat and rice from 1983(1) to 2012(4) using the structural time series analysis. The price series were decomposed into their trend, seasonal, cyclical, and irregular components. The decomposition process provided knowledge on the stylized facts of the international wheat and rice quarterly prices. The importance of this study is that it has given size and shape to the cyclicality and irregular component of the price series of wheat and rice, which are shown to be the drivers of the price formation mechanism. Moreover, the structural time series analysis is not based on achieving stationarity, something that could be difficult in the presence of an intense price spike such as that of 2008. Furthermore, identifying the interrelationship of the decrease of the Chinese grain stocks with the 2008 wheat price spike would not be possible without the information acquired by the structural analysis. The paper does not state that if the Chinese stocks had not decreased, the price surge would not have taken place, but it has identified that a major factor that held back the growth rate of prices ceased to exist. Finally, the outcome of this study emphasizes the importance of considering wheat and rice as separate commodities requiring country-specific and commodity-specific policy measures. Endnotes 1 Both monthly and quarterly prices were used for the estimation of the STS models. However, the models estimated with the use of monthly price data series did not pro399 vide well specified statistical results. Therefore, quarterly price data series were used 400 because they generated well specified empirical results. 2 The Hard Red Winter wheat as well as the Thailand rice prices are used as benchmarks of international prices for wheat and rice since they have the larger weight in 403 world production of wheat and rice, respectively. Competing Interests The authors declare that they have no competing interests. Authors' Contributions AR was the contributor of the conception and the design of the paper while he made critical revisions in its final draft. Also, AR has given the final approval of the version to be published and agreed to be accountable for all aspects of the work related to its accuracy or integrity. Furthermore, AR has contributed to the estimation of the model and the interpretation and analysis of the empirical results. AN and DP acquired the data, estimated the models, analyzed and interpreted the results while they involved in the drafting of the paper. Acknowledgements This research was supported by a grant from a Regional program of Western Greece. This research has been co-financed by the European Union (European Social Fund - ESF) and Greek national funds through the Operational Program "Education and Lifelong Learning" of the National Strategic Reference Framework (NSRF) - Research Funding Program: Heracleitus II. Investing in knowledge society through the European Social Fund. Table 8 Post-sample prediction tests on hard wheat (Lhard-Wh) and rice (LRice) price index Predictive test Wheat Failure χ2 8test 11.626 [0.168] Cusum t(8) test −0.596 [1.432] Rice Failure χ2 8test 4.751 [0.783] Cusum t(8)test 0.526 [0.613] Notes: Values in brackets are p-values Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 16 of 17 Received: 19 June 2014 Accepted: 8 May 2015 References Abbott PC, Hurt C, Tyner WE (2011) What’s driving food prices in 2011? Farm Foundation Issue Report (FFIR). Available at: http://www.farmfoundation.org/news/articlefiles/1742-FoodPrices_web.pdf Baffes J, Haniotis T (2010) Placing the 2006/08 commodity price boom into perspective. The World Bank: Policy research working paper 5371. Available at SSRN: http://ssrn.com/abstract=1646794 Baldwin K, Childs N, Dyck J, Hansen J (2012) Southeast Asia’s Rice Surplus. USDA (United States Department of Agriculture) Outlook: A Report from the Economic Research Service. Available at: http://www.clientadvisoryservices.com/ Downloads/RCS-12-31-2012_Special_Report.pdf Bhar R, Hamori S (2007) Component structures of agricultural commodity futures traded on the Tokyo Grain Exchange. Asia-Pacific Finan Markets 13:1–9 Bollerslev T (1986) Generalized autoregressive conditional heteroskedasticity. J Econ 31:307–327 Box GEP, Jenkins GM (1976) Time series analysis: forecasting and control. Revised edn. Holden–Day, San Francisco Chidmi B, Fatiga ML (2007) Dynamics of Price–cost Margins in the US Meat Industry. Paper prepared for presentation at the American Agricultural Economics Association Annual Meeting Portland OR July 29-August 1. Available at: http://core.kmi.open.ac.uk/download/pdf/7025801.pdf Christiaensen L (2009) Revisiting the global food architecture: lessons from the 2008 crisis. WIDER discussion paper 2009/04. UNU-WIDER, Helsinki Crispin MK, Dale C (1998) Modeling US broiler supply response: a structural time series approach. Agri Resource Econ Rev 27(2):241–251 Deaton A, Laroque G (1992) On the behavior of commodity prices. Rev Econ Stud 59:1–23 Dethier JJ, Effenberger A (2011) Agriculture and development - a brief review of the literature. Policy Res Working Paper 5553:1–58 Durbin J, Koopman SJ (2012) Time series analysis by state space methods. 2nd eds. Oxford, UK: Oxford University Press Engle RF (1982) Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica 1982(50):987–1006 Faliva M (1994) Trend-cycle detection as a filtering problem. J ltalian Statistical Soc 3:315–338 FAO (2009) Wheat flour. Agribusiness handbook. Available at: https://www.responsibleagroinvestment.org/sites/ responsibleagroinvestment.org/files/FAO_Agbiz%20handbook_Wheat%20Flour.pdf Farley EV Jr, Murphy JM (1997) Time series outlier analysis: evidence for management and environmental influences on sockeyes salmon catches in Alaska and northern British Columbia. Alaska Fishery Res Bull 4(1):36–53 Fatiga ML, Misra SK (2007) Common trends, common cycles and price relationships in the international fiber market. J Agric Resour Econ 32(1):154–168 Gilbert CL, Morgan CW (2010) Food price volatility. Philos Trans R Soc B 365:3023–3034 Gilbert CL (2006) Trends and volatility in agricultural commodity prices. In: Sarris A, Hallam D (eds) Agricultural commodity markets and trad. Edward Elgar, Cheltenham UK, pp 31–60 Harvey CA (1989) Forecasting, structural time series models and the Kalman filter. Cambridge, UK: Cambridge University Press Heymans A (2008) Managing an Agricultural Commodities Portfolio in South Africa with pairs trading. Thesis submitted to the centre for Business Mathematics and informations of the North-West University (Potchefstroom Campus) in fulfillment of the requirements for the degree of Philosophiae Doctor (Risk Management). Available at reserch gate: http://www.researchgate.net/publication/ 38115421_Managing_an_agricultural_commodities_portfolio_in_South_Africa_with_pairs_trading__Andr_Heyman Inter-agency Report (2011) Price Volatility in Food and Agricultural Markets: Policy Responses. Available at OECD: http:// www.oecd.org/tad/agricultural-trade/48152638.pdf Kalman RE (1960) A new approach to linear filtering and prediction problems. J Basic Eng Transac ASME Series D 82:35–45 Koopman SJ, Harvey AC, Doornik JA, Shephard N (2009) STAMP 8.2: structural time series analyser, modeller and predictor. Timberlake Consultants, London Mirzabaev A, Tsegai D (2012) Effects of weather shocks on agricultural commodity prices in Central Asia. Zentrum für Entwicklungsforschung (ZEF) Center for Development Research Discussion Papers on Development Policy No. 171. Available at SSRN: http://ssrn.com/abstract=2188353 OECD (2008) Rising Agricultural prices: causes, consequences and responses. Policy Brief Rezitis AN, Sassi M (2013) Commodity Food Prices: Review and Empirics. Economics Research International 2013(694507):15 Hindawi Publishing Corporation. Available at: http://dx.doi.org/10.1155/2013/694507 Shepherd B (2006) Estimating Price Elasticities of Supply for Cotton: A Structural Time-Series Approach. Munich Personal RePEc Archive (MPRA) Paper No. 1252. Available at: http://mpra.ub.uni-muenchen.de/1252/ Sims C (1982) Macroeconomics and reality. Econometrica 48(1):1–48 Timmer CP (2010) Management of rice reserve stocks in Asia: analytical issues and country experience. Commodity Market Review 2009–10. FAO, Rome Italy, pp 87–120 Ubide A (1997) Determinants of Inflation in Mozambique. IMF (International Monetary Fund) Working Paper/97/145. Available at: http://www.imf.org/external/pubs/ft/wp/wp97145.pdf Wright BD, Williams JC (1991) Storage and commodity markets. Cambridge University, Cambridge, UK Rezitis et al. Agricultural and Food Economics (2015) 3:16 Page 17 of 17