News impact for Turkish food prices

Chadwick, Meltem; Bastan, Meltem Article News impact for Turkish food prices Central Bank Review (CBR) Provided in Cooperation with: Central Bank of The Republic of Turkey, Ankara Suggested Citation: Chadwick, Meltem; Bastan, Meltem (2017) : News impact for Turkish food prices, Central Bank Review (CBR), ISSN 1303-0701, Elsevier, Amsterdam, Vol. 17, Iss. 2, pp. 55-76, https://doi.org/10.1016/j.cbrev.2017.05.001 This Version is available at: https://hdl.handle.net/10419/217306 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-nc-nd/4.0/ News impact for Turkish food prices * Meltem Chadwick * , Meltem Bastan Central Bank of the Republic of Turkey, Ankara, Turkey article info Article history: Received 27 March 2017 Received in revised form 4 May 2017 Accepted 4 May 2017 Available online 19 May 2017 JEL code: C32 C58 E31 Keywords: Food prices Asymmetric volatility EGARCH model News impact abstract Asymmetric volatility is a widely encountered concept particularly in financial series. It refers to the case that “bad news”generates more volatility than “good news”of equal magnitude. In an inflationary environment “bad news”is disclosed as increasing inflation that is expected to generate higher volatility. The present article examines whether unexpected price changes affect the volatility of prices asymmetrically for 90 retail food items of the Turkish consumer price index. These 90 food items have a weight of approximately 20 percent in headline consumer price index (CPI). We employ exponential generalized autoregressive conditional heteroscedastic (EGARCH) model to extract asymmetric volatility, using monthly data between January 2003 and January 2017. Our results reveal that volatility of food prices respond asymmetrically to unexpected price shocks for 62 percent of the retail food items. ©2017 Central Bank of The Republic of Turkey. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Food price volatility has become one of the hot topics for researchers and policy makers within the last decade due to its detrimental effect on macroeconomic stability, productivity of food prices and general well-being of consumers. It is well documented that increase in price volatility, which has distortionary effects on the welfare of both consumers and producers, affects the ability of market participants to forecast prices. In an inflationary framework, modelling price volatility inevitably translates into having proper information about inflation uncertainty, where the concept of “uncertainty”is proxied by volatility. There is a vast literature on the relation between inflation and inflation uncertainty, which has gained momentum with the increasing number of central banks implementing inflation targeting regime. The pioneering study in this field is that of Friedman (1977) designating a positive causality between the level of inflation and inflation uncertainty, with higher inflation leading to greater uncertainty. Ball (1992) formalizes Friedman's argument in the context of an asymmetric information game between the public and the policy maker. 1 Majority of the literature on the relation between inflation and inflation uncertainty employs both symmetric and asymmetric GARCH models. The impact of the news on volatility is captured by asymmetric GARCH models. While most of the studies on asymmetric news impacts in economic literature are on financial markets and other areas of macroeconomics, i.e. foreign exchange markets, there are some major studies that apply symmetric and asymmetric volatility models to inflation and one of those studies is by Kontonikas (2004) who analyses the inflation and inflation uncertainty in UK. His results support the Friedman-Ball hypothesis and he shows that inflation decreases with inflation * The views expressed herein are those of the authors and do not necessarily represent the official views of the Central Bank of the Republic of Turkey. *Corresponding author. E-mail address: [email protected].tr (M. Chadwick). Peer review under responsibility of the Central Bank of the Republic of Turkey. 1 In the light of these two seminal papers, many country and cross-country studies have been conducted on the direction and the sign of the relationship between inflation and inflation uncertainty. For example,  Zivkov et al. (2014) analyze 11 Eastern European countries' inflation and find that Friedman's hypothesis is confirmed for countries with flexible exchange rate regimes while refuted for countries with fixed exchange rate regimes. In his study on Turkey's inflation, Karahan (2012) studies the relationship for Turkey between 2002 and 2011 using GARCH type of models whose findings support Friedman's hypothesis. Contents lists available at ScienceDirect Central Bank Review journal homepage: http://www.journals.elsevier.com/central-bank-review/ http://dx.doi.org/10.1016/j.cbrev.2017.05.001 1303-0701/©2017 Central Bank of The Republic of Turkey. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/). Central Bank Review 17 (2017) 55e76 uncertainty which is asymmetric after the implementation of the inflation-targeting regime in UK. Another study is by Fountas et al. (2004), who analyze inflation and inflation uncertainty for the six European Union countries for the period 1960e1999, taking asymmetry in inflation uncertainty into consideration. While Friedman's hypothesis holds for countries except Germany, the asymmetric terms in their volatility equations are found to be significantly positive and authors discuss that such a result stems from the tough commitment of German monetary authority to price stability. 2 Univariate and multivariate GARCH type of models are used in economic literature very frequently to model the agricultural price volatility. An et al. (2016) use a multivariate GARCH model to analyze the volatility, asymmetry and spillovers among wheat and flour prices. Minot (2014) uses GARCH(1,1) model to examine the volatility of 167 food price series from 15 African countries. Rezitis and Stavropoulos (2010) employ several different symmetric, asymmetric and non-linear GARCH models to estimate volatility for the Greek beef market. Gardebroek et al. (2016) use multivariate GARCH approach to evaluate the time evolution and volatility transmission across corn, wheat and soybean price returns on a daily, weekly and monthly basis. Ait Sidhoum and Serra (2016) employ a multivariate GARCH model to study price transmission between consumer, producer and wholesale prices in the Spanish tomato market. 3 The effect of news on volatility is motivated by the pioneering works of Pagan and Schwert (1990) and Engle and Ng (1993).In their study, Engle and Ng (1993) define the news impact curve which measures how new information is incorporated into volatility estimates. The news impact curve depicts the impact of an unexpected shock on next period's volatility such that the impact of good and bad news are reflected on either sides of the curve with different slopes. The closest study to our paper is by Zheng et al. (2008), who analyze the volatility and the news impact with a particular focus on asymmetric news effects for the US food market. Across 45 retail food items, they find that price news destabilizes about a third of the markets such that unexpected price increases contribute more to the price volatility compared to unexpected price decreases. Global food price and volatility became more remarkable especially after the effects of serious food crisis of 2008 and 2011. 4 However, in Turkey food prices started to fall after 2011 and in February 2016 food prices dropped to a lowest value experienced after 2010. Shortly after, the food prices in Turkey reached its peak value in January 2017, deviating from historical trends and international food prices significantly. We observe a surge of 7.67 percent between January 2016 and January 2017 in food prices, while annual CPI inflation is found to be 9.22 percent. Since food items have the highest weight in the CPI basket (20.17 percent for year 2017), the path they follow has particular importance for policy makers. 5 The divergence of domestic food prices in Turkey from international levels is documented by Akcelik et al. (2016). They show that the level of divergence from European Union price levels and their volatility has been increasing since the global food crisis. Ogunc (2010) also shows that the volatility of the food prices in Turkey, particularly that of unprocessed food items, has been above that of CPI within the last decade. 6 For Turkey, it is well documented that the path of the unprocessed food items is the main factor behind the quick surge of the food prices. 7 In this vein, Atuk and Sevinc (2010) state that fresh fruits and vegetables in the CPI basket distinguish from others with their strong seasonality and the accompanying level of high volatility. They suggest that using constant weights would help diminish the volatility of CPI. Orman et al. (2010) also state that unprocessed food items exhibit a more fluctuating pattern compared to other sub-groups in the CPI basket. They attribute this observation to some structural factors such as high level of the dependency of production on climate conditions, high number of intermediaries, uncertainties around public support to agriculture, insufficient level of monitoring by the government, concentration of production in certain regions and fluctuations in external demand. 8 They conclude that a stable path could be attained by the implementation of medium to long-run policies. In Turkey, an evaluation committee for food prices was indeed established in 2014 to implement medium to long-run policies, while the official secretariat of the committee was transferred to the Central Bank of Republic of Turkey in December 2016. The “Food Committee”specifically examines every item with increasing price and volatility while presenting policy suggestions for different horizons. Due to their unpredictable and volatile pattern, the Committee closely tracks the prices of unprocessed food items, the prices of fresh fruits and vegetables in particular. The Committee has also designed and employed an early warning mechanism for this purpose. In this respect, similar to Zheng et al. (2008), this paper examines whether asymmetric news effects exist for 90 retail food items in Turkish food market. We choose 90 food items, which have a share of 19.87 percent in total consumer price index (CPI), out of total number of 129 items under headline CPI of Turkey. These 90 items have a share of 91.3 percent within the food prices of CPI and the choice is made considering data availability, i.e. we eliminate 39 items as they have missing data. 9 This study is essential as the existence of possible asymmetry in the behaviour of price volatility in the retail food prices of Turkey is so far unknown and such asymmetry in the retail price volatility can help policy makers to take some measures to meet the targets and also give useful information about retail market power. To the best of our knowledge, our paper is the first to observe news impact for the 2 Other papers that have results for asymmetric effect on prices include Zheng et al. (2008) and Rezitis and Stavropoulos (2010). 3 Some other studies on price volatility include Apergis and Rezitis (2003a,b,c), Fousekis and Grigoriadis (2016), Gouel (2013), Jha and Nagarajan (2002), Yang et al. (2003) and Rude and An (2015). 4 See FAO (2016) for details. 5 We observe that, due to their weight in the basket, the changes in the price of certain items such as fresh fruit and vegetables and veal became more prominent in the last couple of years. The price of veal for example rose by 146 percent between January 2009 and January 2017. 6 Ogunc (2010) examines the structural problems behind price volatility in Turkey, suggesting that long chains of logistics, vastness of informal economy, structural problems in irrigation, storage and packaging capacities contribute to the high level of volatility in prices. In addition to them, absence of big producers who would not have financial problems in mitigating sudden shocks, insufficient capacity of insurance for farmers and big numbers of sellers in the retail sector are also listed as structural problems leading to highly volatile food prices. 7 Close examination of the Central Bank of the Republic of Turkey's (CBRT) inflation reports, summary of monetary policy meetings and open letters written to the government illustrate that the main focus has been the rapid increase in food prices and especially their volatility that are outside the control of CBRT. 8 Since 2016 the vulnerability of the unprocessed food items to external demand has been explicitly experienced with the restrictions imposed on Turkish exports of food products to Russia. Even though decreasing exports is considered to have favorable effects on domestic inflation, accompanying fluctuations and increase in volatility have become inevitable. 9 Some items like peach do not have price data available for winter as they are only produced and consumed in the summer, therefore we excluded those type of seasonal food items. There are some other items that have been included in the CPI basket just recently and those items are also excluded from the analysis. M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e7656 subcomponents of CPI inflation. Therefore, present article contributes to the price volatility literature by testing the asymmetry of 90 retail food items employing EGARCH model of Nelson (1991).In line with Friedman-Ball hypothesis, we will suggest that news of high prices to be more stabilizing compared with the news of low prices for the food items that we study. Our results show that among 90 items, there are 5 items which do not exhibit any time-varying variance all of which belong to mature markets. We also find that out of 56 items for which news effect is detected, 32 of them exhibit asymmetric volatility such that the total weight of these items in the CPI basket amounts to 7 percent. Between these two extremes, there are food items which exhibit time varying volatility with no news effect and with symmetric news effect, respectively. The striking point of the results is the downward sloping news impact curve that is detected for 10 items such as tomato, eggplant and zucchini. We consider that even though the primary focus should be on the items with asymmetric volatility, the case of these items should also not be ignored. The remainder of the article is organized as follows. Section 2 describes the EGARCH(1,1) model and the news impact employed in this study to test the asymmetry in the volatility of 90 food items. Section 3gives the detailed summary of the results and their implications. Section 4concludes. 2. Methodology Inflation volatility is often claimed to be one of the most important costs of inflation, since it distorts the allocation decision of market players by redistributing wealth between debtors and creditors and by reducing the effectiveness of relative prices in co-ordinating economic actions. In this respect, the importance of a correctly specified volatility model is essential for the valuation of future inflation. Early studies use unconditional volatility measures, i.e. standard deviation of inflation, and such measures have a drawback in the sense that higher variability need not necessarily imply higher uncertainty. 10 Ever since the seminal papers of Engle (1982) and Bollerslev (1986), inflation volatility is proxied by ARCH and subsequent GARCH models. 11 There are certain advantages of GARCH type of models to proxy volatility. 12 Contrary to other measures of volatility, GARCH type of models allow the researcher to formally test for constant volatility and asymmetric volatility over the sample period. In this part of the article we will concern ourselves with EGARCH model to measure volatility and employ news impact analysis. Following Engle and Ng (1993), let y t be the log difference of food price. 13 Let F t1 be the past information set containing the realized values of all relevant variables up to time t1. The relevant expected price change and volatility are the conditional expected value of y t , given F t1 . These are denoted by m t ≡Eðy t jF t1 Þand h t ≡Varðy t jF t1 Þ. Therefore, the unexpected price change at time tis ε t ≡y t m t . In this paper, ε t is treated as a collective measure of news at time tand opposite to Engle and Ng (1993) a positive ε t (an unexpected increase in price) suggests the arrival of bad news, while a negative ε t (an unexpected decrease in price) suggests the arrival of good news. 14 Given that the predictable volatility is dependent on past news an ARCH(1) model ala Engle (1982) will be: ht¼ u þ a ε2 t1 where a and u are constant parameters. Bollerslev (1986) generalizes the ARCH model to the GARCH model, such that a GARCH(1,1) model will be: ht¼ u þ a ε2 t1þ b ht1 where a , b and u are constant parameters. Indeed, the GARCH(1,1) model is like an ARMA(1,1) model for the variance and therefore a GARCH model is an infinite order ARCH model. Unfortunately, ARCH and GARCH models cannot capture some important aspects related to the data. The most important aspect not measured with these models is the leverage or asymmetric effect of news. 15 Nelson's 1991 exponential GARCH (EGARCH) model captures such asymmetric effects and the EGARCH(1,1) model can be represented as: logðhtÞ¼ u þ b logðht1Þþ g εt1 ffiffiffiffiffiffiffiffiffiffi ht1 pþ a "jεt1j ffiffiffiffiffiffiffiffiffiffi ht1 pffiffiffiffiffiffiffiffiffi 2= p p# (1) where u , b , g and a are constant parameters. 16 Given F t1 , we can examine the implied relation between ε t1 and h t with the help of the news impact curve and can see the asymmetric effects of increasing and decreasing food prices on the next periods' volatility. 17 Given the EGARCH(1,1) model in Equation (1) we can embed a parametric test for the asymmetry hypothesis, i.e. when g ¼0 there are symmetric effects and when g is positive (negative) high (low) price news generates more volatility. We finish describing our methodology by illustrating the mean and error equations given by: yt¼ m þX k i¼1 r iytkþεi(2) εt¼ n tffiffiffiffiffi ht p(3) where y t is the first difference of log price, kis the lag length and n t is a white noise process with unit variance. The lagged terms in the autoregressive AR(k) process defined by Equation (2) capture the predictable components of the price changes. 3. Empirical results In this study, we use monthly retail price data that is publicly 10 Kontonikas (2004) states that this will be the case only if agents don't possess the relevant information to predict part of the increased variability. 11 There are also survey-based proxies for inflation volatility where dispersion in inflation forecasts of individual respondents are aggregated. 12 Although not related to inflation, Day and Lewis (1992) show that implied volatility from the Black-Scholes model cannot capture the entire predictable part of future stock price volatility relative to some GARCH and EGARCH models. 13 See the unit root tests for price series at the Appendix. 14 GARCH type of models are mostly used to evaluate financial returns and a positive shock for an asset return is good news, yet it will not be considered good when the prices considered are related to food. Therefore the definition of good news is the opposite of general literature on news impact. 15 The leverage effect suggests that a symmetry constraint on the conditional variance function in past ε’s is inappropriate. 16 EGARCH model is asymmetric as the level of ε t1 =ffiffiffiffiffiffiffiffiffiffi h t1 pis included with a coefficient g . 17 See Engle and Ng (1993) for the details and shape of news impact curve. It is important to note that, Engle and Ng (1993) show that the shape of the curve depends on very strict parameter assumptions. M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e76 57 available at Turkish statistics agency (Turkstat). We only use food items that do not have missing data problem and that restricts us to 90 food items under CPI. These 90 food items have a 20 percent of share in CPI. Data for all the food items is available after January 2003, therefore all the EGARCH models are estimated between January 2003 and January 2017 (last available data). 18 Some of the food items under CPI exhibit seasonality and we use TRAMO-SEATS for seasonal adjustment. GARCH type of models require stationary data for estimation, therefore our data is transformed using logarithmic difference, i.e. y t ¼lnðp t =p t1 Þ. We choose klags in the AR(k)-EGARCH(1,1) of each food item so that there does not remain any serial correlation in the residuals of the estimation and we come up with the most parsimonious specification for the mean and variance model illustrated with Equations (2) and (1). In all the cases kup to 4 lags provide good approximation to the data generating process. Additionally, EGARCH(1,1) provides a proper approximation for the time-varying variance as there does not exist any remaining heteroscedasticity in Table 1 Maximum likelihood estimates of EGARCH(1,1). 18 See the descriptive statistics of the series at the Appendix. M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e7658 the residuals of the estimated models. 19 Table 1 and Table 2 lists the maximum likelihood estimates of a , b and g . We use 5 percent level of significance for all the estimated parameters. When we check the significance of the parameters alpha and beta, we observe that most of the time one or the other is statistically significant pointing to existence of time-varying variance in 94 percent of the food items of the CPI. For the food items listed in the gray shaded rows of Tables 1 and 2, we cannot observe time-varying variance and this happens for 5 items including biscuit, milk, cucumber, baking powder and tea 20 With a closer inspection, it can be seen that these 5 food items with constant variance belong mostly to mature markets where competition and entry is difficult with no incentive for extra profit or margin. Therefore, the growth prospect and volatility for these items are more stable in the short-run. The weight of these 5 constant variance food items in the headline CPI amounts to 1.55 percent, whereas it amounts to 7.12 in the food and non-alcoholic beverages’sub-component of Table 2 Maximum likelihood estimates of EGARCH(1,1) continued. 19 Residual diagnostics of EGARCH(1,1) models are illustrated with Table C.1 and Table C.2 at the Appendix. 20 We declare an item to have constant variance if neither the individual significance nor the joint significance of coefficients are verified. For 5 items listed as constant variance items, a , b and g are not significant jointly and individually. M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e76 59 CPI. It is remarkable to see that 3 of these food items, i.e. biscuit, milk and tea, are produced in markets with high levels of concentration. Such items could lead to further consideration about the impact of market structure on price determination dynamics. 21 The last column of Tables 1 and 2 illustrates the Lagrange Multiplier (LM) statistics for a joint significance test of parameters a , b and g and only 7 food items out of 90 fail the LM test statistics. The last two columns of Tables 1 and 2 that are shaded in colour red are for food items with both significant g and LM test statistics. Table 3 lists these items separately with both significant statics of g and LM test, i.e. that are shaded red in Tables 1 and 2. In the empirical literature, a significant g parameter of Equation (1) is usually considered a sign of asymmetric volatility, however we behave conservative and regard items to have asymmetric effect if both the g and the LM statistic is significant together. Columns one and three of Table 3 shows coefficients for high and low price news respectively. This table is extremely important to observe the asymmetric effect of a decrease and Table 3 Effect of news on food price volatility. Food Item High price Low price news a þ g t-stat. ( a þ g ) news a  g t-stat. ( a  g ) Baby food 0.77 5.62 0.74 5.37 Boiled and pounded wheat 0.34 4.26 0.02 0.13 Cracker 0.83 4.76 2.46 16.86 Wafer 1.34 11.96 0.41 1.94 Thin dough 0.62 3.78 0.18 0.92 Macaroni 0.01 0.24 0.19 2.39 Cereal 0.63 2.83 0.23 1.31 Veal 0.21 2.84 0.24 2.65 Offal 0.36 2.26 0.23 2.23 Sausage 0.71 3.64 0.17 0.88 Fresh fish 0.50 2.77 0.02 0.14 White cheese 0.31 3.49 1.04 3.84 Kasar cheese 0.24 3.60 0.04 0.83 Tulum cheese 0.03 0.22 0.46 3.81 Egg 0.09 0.58 0.42 2.27 Olive oil 0.45 3.74 5.63 15.35 Sun-flower oil 0.32 7.14 0.10 1.08 Corn oil 0.37 6.77 0.01 0.09 Apple 0.78 2.98 0.16 0.65 Lemon 0.65 4.76 1.92 9.50 Banana 0.73 3.03 0.05 0.20 Hazelnut (without shells) 0.20 5.80 0.42 12.09 Pistachio 0.59 4.68 0.07 0.67 Peanuts 0.79 6.01 0.30 1.70 Roasted chick-pea 0.19 3.59 1.20 6.05 Sun flower seed 0.05 0.77 1.12 4.07 Pumpkin seed 1.31 6.77 0.39 1.81 Tomato 0.05 0.42 1.18 7.78 Zucchini 0.12 0.94 1.30 6.15 Onion 0.56 4.98 0.38 3.28 Lettuce 0.44 4.53 0.99 4.82 Parsley 0.90 6.36 0.85 3.73 Eggplant 0.07 0.46 0.73 3.70 Garlic 0.14 0.81 0.70 3.93 Green onion 0.23 2.71 1.03 5.90 Potato 1.15 7.26 0.20 1.17 Other pulse 0.53 2.50 0.55 3.47 Tomato sauce 0.86 3.19 0.48 2.09 Olive 0.09 4.28 0.35 22.81 Chips and appetizers 0.38 4.18 1.06 4.85 Granulated sugar 0.67 3.66 0.09 0.52 Cube sugar 0.29 1.75 0.30 1.96 Jam 0.18 1.80 0.42 3.74 Honey 0.52 2.23 0.55 2.49 Halvah 0.53 3.50 0.21 1.11 Turkish delight 0.08 0.44 0.58 3.62 Holiday candy 0.41 4.66 1.50 12.65 Ice-cream 0.49 4.99 0.19 2.89 Condiment-spices 0.49 6.94 0.52 4.86 Packaged soup 0.19 33.92 2.25 138.80 Turkish Coffee 0.83 3.96 0.11 0.61 Ready-made coffee 0.75 8.93 1.77 11.23 Water 3.50 45.15 2.60 14.26 Carbonated fruity beverages 1.82 18.71 0.18 0.94 Coke 2.29 11.68 0.22 0.87 Fruit Juice 0.23 28.14 0.03 3.63 21 See € Onder (2016) and Kaynak (2016). M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e7660 increase in prices on the next period volatility. For EGARCH(1,1) model the effect of high price news on conditional variance can be quantified by a þ g and the effect of low price news can be quantified by a  g . When we examine Table 3 closely, estimated high price and low price effects are 0.21 and 0.24 respectively for veal. If we quantify the effects for veal, we can claim that an unexpected price increase measured by a unit increase in the standardized residual with ε t1 >0 increases volatility by 21 percent and an unexpected price decrease with ε t1 <0 decreases volatility by 24 percent. This asymmetric effect for food item veal can be observed clearly from the news impact curve at Appendix D.Appendix D shows that, for veal the news impact curve is upward sloping for high price news and downward sloping for low price news. The asymmetry is sharper for eggs, i.e. high price effects increase the volatility only by 9 percent, yet low price news decrease price by 42 percent. Unexpectedly, news impact curve is downward sloping for some food items and this is more common for negative shocks. One example is baby food, which has similar high and low price effects but the news impact curve is sloped downwards for the low price news. News impact curve is downward sloping both for the high and low price news for 10 food items, which is quite extraordinary. For these 10 items, i.e. sausages, white cheese, kasar cheese, tomato, zucchini, eggplant, olives, holiday candy, packaged soup and instant coffee, any change in price will decrease volatility. There does not exist any common features of these 10 food items that will cause their volatility to respond to negative and positive shocks in such an unexpected behaviour. 22 Out of 56 food items listed in Table 3,fornearlyhalfofthem high price news effects are larger than the low price news effects in absolute values, which means for half of the food items consumers respond disproportionately given the high price news. Indeed, Table 4 sums up the news impact analysis of all EGARCH(1,1) models estimated for the 90 retail food items. For the food items that takes place in the first column of the table, low price and high price news have the same effect on the price. The 10 shaded items on the first column of the table are food items having negative sloped news impact curve for both high price and low price news. Considering the price of veal, which is one of the food items in the first column of Table 4, it should be stated that the news effect detected for veal should be examined closely in light of the price increases observed in recent years and the subsequent import decision, to prevent the repercussions of increased uncertainty. As can be seen in the news effect structure of the veal in Appendix D,an unexpected decrease in the price of veal decreases volatility whereas an unexpected increase in price causes a jump of volatility in an almost exponential manner. This vulnerable structure validates the special focus of policy makers to prevent unexpected increases in the price of veal, and similar sensitive items in this category. The second column of Table 4 have food items with significant news impact and the effect of low price and high price news are asymmetric. A similar pattern is observed for food item egg, such that unexpected price increase in the previous period leads to higher uncertainty compared to unexpected price decrease in Table 4 Classification of food price variance response. 22 It would be interesting to study and find the reason behind this unexpected behaviour, which is beyond the scope of this study. M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e76 61 the same period. It is seen that as the amount of unexpected price decrease gets larger, the impact on volatility gets smaller. Detection of such a structure shows that developments such as the unexpected increase in the price of egg through the end of 2016 should be monitored closely by policy making authorities in order to limit its impact on volatility and further price increases. 23 Items such as apple, onion and potato are also found to exhibit news impact structure and this structure reveals that price declines do not cause any change in volatility whereas price increases definitely contribute to that. In Turkish food market, it is known that some food products are stored either by producers or middlemen to sell them at a higher price in subsequent periods. Since these three items are among those products, it could be suggested that any unexpected decline in the price leads to increase in the stored amount with no impact on price level volatility. In other words, since the actors in those markets already know that their products would be stored in the face of a price decline, any unexpected decline might not end up with increased volatility as all producers (or middlemen) would (surely) wait until the prices recover to a desired level. 24 The third column of the table has only one food item that have an insignificant LM test statistics, yet g for this item is individually significant. For the food items listed on the last two columns of Table 4, either there are no price news effects (fourth column with both insignificant LM test statistics and insignificant asymmetry parameters for the EGARCH model), or there is constant variance (5 food items that are shaded gray in Tables 1 and 2). The fourth column,i.e. with no news impact but time-varying variance, includes items such as bread, rice, wheat flour, mutton and poultry. It is evident that the first three of these food items are among basic necessity foods, which need further consideration by policy makers. Although the weight of these items amount to 7.2 percent in the headline CPI, it amounts to 33 percent within the food and non-alcoholic beverages’sub-component of CPI. 25 Monitoring the volatility of the price of poultry is also a matter of significance since poultry sector is closely connected with the exchange rate and export dynamics along with the high level of imported input materials. 26 Because inference in GARCH type of models depends on the correct specification set and the validity of the functions used to represent the conditional mean and the conditional variance we conduct an additional test on asymmetry by employing the regression given as: s2 t¼ d 0þ d 1st1þ d 2st2þ d 3st3þ h t(4) where s t ¼ðε t =h 1=2 t Þ 0 is the predicted standardized residual from Equation (2). A test for any additional symmetry will be a joint significance test for the null hypothesis of d 1 ¼ d 2 ¼ d 3 ¼0. 27 Table 5 Diagnostic regression. 23 It is publicly known that the sudden increase in the price of egg was triggered by the sudden increase in export demand accompanied by the decline in the demand for eggs from the countries with avian influenza disease. It is also known that the demand for Iranian egg was substantially directed to Turkish egg market. 24 Though this scenario is a valid and vastly observed one in Turkey, how long producers would wait for the prices to recover and at what price they would be satisfied is a question that needs to be examined on a market level. Admittedly, this is out of the scope of this paper. 25 The weight of bread has particular importance with 2.22 percent in the CPI basket and 10 percent in the food and non-alcoholic beverages' basket. 26 It would be plausible to decompose the volatility of the price of poultry as it could be affected by many factors simultaneously. 27 See Enders (2008) for details. M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e7662 .00000 .00005 .00010 .00015 .00020 .00025 .00030 .00035 .00040 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Sausage .00004 .00008 .00012 .00016 .00020 .00024 .00028 .00032 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Salami .000 .001 .002 .003 .004 .005 .006 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Fresh fish .00000 .00004 .00008 .00012 .00016 .00020 .00024 .00028 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Milk .000100 .000125 .000150 .000175 .000200 .000225 .000250 .000275 .000300 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Yoghurt .0000 .0004 .0008 .0012 .0016 .0020 -25 -20 -15 -10 -5 0 5 10 15 20 25 News/epsilon(t-1) Response/Sigma^2(t) White cheese .00000 .00004 .00008 .00012 .00016 .00020 .00024 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Kasar cheese .0000 .0005 .0010 .0015 .0020 .0025 .0030 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Tulum cheese .000 .004 .008 .012 .016 .020 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Egg .0001 .0002 .0003 .0004 .0005 .0006 .0007 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Butter M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e76 69 .000 .001 .002 .003 .004 .005 .006 .007 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Margarine 0E+00 1E+46 2E+46 3E+46 4E+46 5E+46 6E+46 7E+46 8E+46 9E+46 -25 -20 -15 -10 -5 0 5 10 15 20 25 News/epsilon(t-1) Response/Sigma^2(t) Olive oil .000 .001 .002 .003 .004 .005 .006 .007 .008 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Sun-flower oil .000 .002 .004 .006 .008 .010 .012 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Corn oil .00 .04 .08 .12 .16 .20 .24 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Apple 0.0E+00 2.0E+09 4.0E+09 6.0E+09 8.0E+09 1.0E+10 1.2E+10 -15.0 -10.0 -5.0 0.0 2.5 5.0 7.5 12.5 News/epsilon(t-1) Response/Sigma^2(t) Lemon .00 .01 .02 .03 .04 .05 .06 .07 .08 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Banana .000 .001 .002 .003 .004 .005 .006 .007 .008 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Wal nut (wit hout shells) .000 .001 .002 .003 .004 .005 .006 .007 .008 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Hazelnut (without shells) .0 .1 .2 .3 .4 .5 .6 -14 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 News/epsilon(t-1) Response/Sigma^2(t) Pistachio M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e7670 .00 .05 .10 .15 .20 .25 .30 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Peanuts .0000 .0001 .0002 .0003 .0004 .0005 .0006 .0007 .0008 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Roasted chick-pea .00000 .00004 .00008 .00012 .00016 .00020 .00024 -14 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 News/epsilon(t-1) Response/Sigma^2(t) Sun flower seed 0 4 8 12 16 20 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Pumpkin seed .00 .01 .02 .03 .04 .05 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Raisin 0 2 4 6 8 10 12 14 16 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Sweet green pepper 0 100 200 300 400 500 600 700 800 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Green pepper .00 .01 .02 .03 .04 .05 .06 .07 .08 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Tomato .00 .01 .02 .03 .04 .05 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Zucchini 0.0 0.2 0.4 0.6 0.8 1.0 1.2 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Onion M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e76 71 .00 .02 .04 .06 .08 .10 .12 .14 .16 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Lettuce .0 .1 .2 .3 .4 .5 .6 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Parsley .00 .01 .02 .03 .04 .05 .06 .07 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Eggplant .02 .04 .06 .08 .10 .12 .14 .16 .18 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Cucumber 0 1 2 3 4 5 6 7 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Garlic .00 .01 .02 .03 .04 .05 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Green onion 0 10 20 30 40 50 60 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Potato 0 1 2 3 4 5 -15.0 -10.0 -5.0 0.0 2.5 5.0 7.5 12.5 News/epsilon(t-1) Response/Sigma^2(t) Dry bean .0000 .0002 .0004 .0006 .0008 .0010 .0012 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Chickpea .0004 .0008 .0012 .0016 .0020 .0024 .0028 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Lentils M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e7672 .00 .01 .02 .03 .04 .05 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Other pulse 0.0 0.2 0.4 0.6 0.8 1.0 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Canned vegetables .00 .04 .08 .12 .16 .20 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Tomato sauce .00000 .00001 .00002 .00003 .00004 .00005 .00006 .00007 .00008 .00009 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Olive 0.0 0.2 0.4 0.6 0.8 1.0 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Chips and appetizers .00 .01 .02 .03 .04 .05 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Granulated sugar .0000 .0001 .0002 .0003 .0004 .0005 .0006 .0007 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Cube sugar .0000 .0004 .0008 .0012 .0016 .0020 .0024 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Jam .00 .01 .02 .03 .04 .05 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Honey 0 1 2 3 4 5 6 7 8 9 -15.0 -10.0 -5.0 0.0 2.5 5.0 7.5 12.5 News/epsilon(t-1) Response/Sigma^2(t) Grape molasses M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e76 73 .000 .004 .008 .012 .016 .020 .024 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Halvah .0000 .0004 .0008 .0012 .0016 .0020 .0024 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Chocolate cream .0000 .0002 .0004 .0006 .0008 .0010 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Turkish delight .0000 .0001 .0002 .0003 .0004 .0005 -20 -16 -12 -8 -4 0 4 8 12 16 20 News/epsilon(t-1) Response/Sigma^2(t) Holiday candy .0000 .0005 .0010 .0015 .0020 .0025 .0030 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/ epsilon(t-1) Response/Sigma^2(t) Ice cream .00 .02 .04 .06 .08 .10 -14 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 News/epsilon(t-1) Response/Sigma^2 (t) Condiment spices .000 .004 .008 .012 .016 .020 .024 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/epsilon(t-1) Response/Sigma^2(t) Salt .00 .01 .02 .03 .04 .05 -30 -20 -10 010 20 30 News/epsilon(t-1) Response/Sigma^2 (t) Baking powder .00 .04 .08 .12 .16 .20 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Ketchup .000 .004 .008 .012 .016 .020 .024 .028 .032 .036 -40 -30 -20 -10 010 20 30 40 News/epsilon(t-1) Response/Sigma^2(t) Packaged soup M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e7674 .0 .1 .2 .3 .4 .5 .6 .7 .8 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Turkish coffe e .0000 .0004 .0008 .0012 .0016 .0020 .0024 .0028 .0032 -30 -20 -10 0 10 20 30 News/epsilon(t-1) Response/Sigma^2(t) Instant coffee .0000 .0002 .0004 .0006 .0008 .0010 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/ epsilon(t-1) Re sp onse /Sigm a^2(t) Tea .00000 .00001 .00002 .00003 .00004 .00005 .00006 .00007 .00008 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 News/epsilon(t-1) Response/Sigma^2(t) Cocoa .0000 .0002 .0004 .0006 .0008 .0010 .0012 -10 -8 -6 -4 -2 0 2 4 6 8 10 News/ epsilon(t-1) Re sp onse /Sigma^2(t) Cocoa beverages 0.0E+00 2.0E+22 4.0E+22 6.0E+22 8.0E+22 1E+23 1.2E+23 1.4E+23 -25 -20 -15 -10 -5 0 5 10 15 20 25 News/epsilon(t-1) Response/Sigma^2(t) Water .0000 .0004 .0008 .0012 .0016 .0020 .0024 .0028 .0032 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Mineral water 0 20,000 40,000 60,000 80,000 100,000 120,000 -14 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 News/epsilon(t-1) Response/Sigma^2(t) Carbonatedfruitybeverages 0.0E+00 2.0E+13 4.0E+13 6.0E+13 8.0E+13 1.0E+14 1.2E+14 -20 -16 -12 -8 -4 0 4 8 12 16 20 News/epsilon(t-1) Response/Sigma^2(t) Coke .00002 .00004 .00006 .00008 .00010 .00012 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 News/epsilon(t-1) Response/Sigma^2(t) Fruit juice M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e76 75 References Ait Sidhoum, A., Serra, T., 2016. Volatility spillovers in the Spanish food marketing chain: the case of tomato. Agribusiness 32, 45e63. Akcelik, F., Yucel, C.Y., et al., 2016. Turkiye’de Gida Fiyatlari: Uluslararasi Bir Karsilastirma. Tech. rep., Research and Monetary Policy Department, Central Bank of the Republic of Turkey. An, H., Qiu, F., Zheng, Y., 2016. How do export controls affect price transmission and volatility spillovers in the Ukrainian wheat and flour markets? Food Policy 62, 142e150. Apergis, N., Rezitis, A., 2003a. Agricultural price volatility spillover effects: the case of Greece. Eur. Rev. Agric. Econ. 30, 389e406. Apergis, N., Rezitis, A., 2003b. Food price volatility and macroeconomic factor volatility:‘heat waves’or ‘meteor showers’? Appl. Econ. Lett. 10, 155e160. Apergis, N., Rezitis, A., 2003c. Mean spillover effects in agricultural prices: the case of Greece. Agribusiness 19, 425e437. Atuk, O., Sevinc, O., 2010. TUFE’de Sabit ve Degisken Agirlik Sistemi Yaklasimlari: Turkiye Taze Meyve-Sebze Fiyatlari Uzerine Bir Uygulama. TCMB Ekonomi Notlari 10, 15. Ball, L., 1992. Why does high inflation raise inflation uncertainty? J. Monetary Econ. 29, 371e388. Bollerslev, T., 1986. Generalized autoregressive conditional heteroskedasticity. J. Econ. 31, 307e327. Day, T.E., Lewis, C.M., 1992. Stock market volatility and the information content of stock index options. J. Econ. 52, 267e287. Enders, W., 2008. Applied Econometric Time Series, second ed. John Wiley &Sons. Engle, R.F., 1982. Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica 50, 987e1007. Engle, R.F., Ng, V.K., 1993. Measuring and testing the impact of news on volatility. J. Finance 48, 1749e1778. FAO, 2016. World food situation. Food Price Indices. February, 04/02/2016. Fountas, S., Ioannidis, A., Karanasos, M., 2004. Inflation, inflation uncertainty and a common European monetary policy. Manch. Sch. 72, 221e242. Fousekis, P., Grigoriadis, V., 2016. Joint price dynamics of quality differentiated commodities: copula evidence from coffee varieties. Eur. Rev. Agric. Econ. Friedman, M., 1977. Nobel lecture: inflation and unemployment. J. Political Econ. 451e472. Gardebroek, C., Hernandez, M.A., Robles, M., 2016. Market interdependence and volatility transmission among major crops. Agric. Econ. 47, 141e155. Gouel, C., 2013. Food price volatility and domestic stabilization policies in developing countries. NBER Work. Pap. Ser. Jha, R., Nagarajan, H.K., 2002. Wholesale spreads and the dynamics of retail price volatility in Indian rice markets. Appl. Econ. Lett. 9, 387e390. Karahan, O., 2012. The relationship between inflation and inflation uncertainty: evidence from the Turkish economy. Procedia Econ. Finance 1, 219e228. Kaynak, S., 2016. Giyim esyasi imalati piyasa tapisi ve yogunlasma orani: Tukiye’nin ilk 500 sanayi kurulusu uzerine bir uygulama. Uluslararasi Yonetim Iktisat ve Isletme Dergisi 12, 21e37. Kontonikas, A., 2004. Inflation and inflation uncertainty in the United Kingdom, evidence from GARCH modelling. Econ. Model. 21, 525e543. Minot, N., 2014. Food price volatility in sub-Saharan Africa: Has it really increased? Food Policy 45, 45e56. Nelson, D.B., 1991. Conditional heteroskedasticity in asset returns: a new approach. Econometrica 59, 347e370. Ogunc, F., 2010. Turkiye’de islenmemis gida enflasyonunda oynaklik durum tespiti. TCMB Ekonomi Notlari 10. € Onder, K., 2016. Türkiye Bisküvi, Çikolatalıve S¸ ekerli Mamuller Sekt€ orü: Firma Yo gunlas¸ ma Analizi. Dokuz Eylül Üniversitesi _ Iktisadi ve _ Idari Bilimler Fakültesi Dergisi 31. Orman, C., € O günç, F., Saygılı,S¸., Yılmaz, G., 2010. _ Is¸ lenmemis¸Gıda Fiyatlarında Oynaklı  ga Yol Açan Yapısal Fakt€ orler. TCMB Ekonomi Notları16. Pagan, A.R., Schwert, G.W., 1990. Alternative models for conditional stock volatility. J. Econ. 45, 267e290. Rezitis, A.N., Stavropoulos, K.S., 2010. Modeling beef supply response and price volatility under CAP reforms: the case of Greece. Food Policy 35, 163e174. Rude, J., An, H., 2015. Explaining grain and oilseed price volatility: the role of export restrictions. Food Policy 57, 83e92. Yang, J., Zhang, J., Leatham, D.J., 2003. Price and volatility transmission in international wheat futures markets. Ann. Econ. Finance 4, 37e50. Zheng, Y., Kinnucan, H.W., Thompson, H., 2008. News and volatility of food prices. Appl. Econ. 40, 1629e1635.  Zivkov, D., Njegi c, J., Pe canac, M., 2014. Bidirectional linkage between inflation and inflation uncertainty-the case of Eastern European countries. Baltic J. Econ. 14, 124e139. M. Chadwick, M. Bastan / Central Bank Review 17 (2017) 55e7676