Modelling the symmetrical and asymmetrical effects of global oil prices on local food prices: A MENA region application

This paper explores the complex nexus between the global oil prices and the food prices of the Middle East and North Africa (MENA) region during the period 2000–2020. Both linear and nonlinear models of the autoregressive distributed lag (ARDL) approach are adapted into panel data form to investigate the symmetrical and asymmetrical influence of oil prices on food prices. The key results are summarized: (i) the long-term effect of oil prices on food prices is significantly positive including both oil-exporting and oil-importing nations. The positive impact on oil exporters—due to higher oil revenues—is greater than importing nations, leading to an increased demand for food. Additionally, the effect on oil exporters is negative and significant in the short term but not significant for importers. (ii) The nonlinear ARDL panel analysis for the whole MENA sample confirms the presence of negative short-term asymmetric behaviour due to the heterogeneous response within the oil-importing and oil-exporting samples, while in the long term, the asymmetric effect is positive, indicating that food prices increase regardless of fluctuations in oil prices. (iii) The nonlinear ARDL results using time series affirm the absence of an asymmetric nexus among oil and food prices for some oil-exporting nations (including Kuwait, Saudi Arabia, United Arab Emirates) and Tunisia within the oil-importing group. However, the food prices of other countries are asymmetric to the oil price. This study provides recommendations that are useful to MENA countries to establish a stable mechanism for oil and food prices to ensure food security in the region.


Introduction
As important strategic resources, food and oil have been widely studied by scholars (Dalheimer et al. 2021;Mokni and Ben-Salha 2020;Sarwar et al. 2020). In the world's first half of 2008, food and oil prices increased dramatically. The international actual prices including all main food crops have exceeded their peak values for almost three decades, as indicated by the Food and Agriculture Organization (FAO, 2008) report. In the first 3 months of 2008, the food price index for the FAO was 53% higher than the same 3 months of the year before. The increasingly rapid food higher prices have led to greater deprivation for poor people in both advanced and emerging economies. Many reasons contribute to the rising prices of food; however, high oil prices are a major factor. The expansion of biomass energy-such as biodiesel and bioethanol-in response to climate change has caused a crowding-out effect on food production for people, while high oil prices have increased agricultural costs incurred in the production and transportation processes (Gardebroek and Hernandez, 2013;Sarwar et al. 2020;Dalheimer et al. 2021). Rising food prices have posed a substantial threat to food security especially for countries that rely heavily on importing food. This severe situation and the double crisis caused by rising energy prices threatened the international community's plan to achieve their Millennium Development Goals in 2015. Consequently, recognizing the internal Responsible Editor: Roula Inglesi-Lotz relationship between food and oil prices is an important prerequisite to stabilizing food prices and establishing an early warning mechanism. However, due to the huge differences in terms of resource endowment, agricultural productivity and economic status among regions of the world, the relationship between the two prices can have a distinguished mechanism. For example, countries rich in agricultural products will be much less affected by high oil prices than countries with poor supply capacity of agricultural products. Therefore, the internal relationship between food and oil prices should be analysed according to the research objective.
As for the region of the Middle East and North Africa (MENA), researching how oil prices affect food prices is even more important. The main reasons are (1) most countries are net importers of food commodities as food crops are difficult to cultivate due to insufficient water supply and limited access to arable land. They are particularly vulnerable to fluctuations in international commodity markets and are most seriously affected by rising prices. Therefore, stable food prices and reliable sourcing of imports are the basis for food security in these countries.
(2) Some countries that are rich in oil resources but lacking in food have implemented the strategy of the 'Oil for Food programme'. This policy allowed Iraq to export a specific portion of its oil in order to benefit from its revenues in purchasing the food needs of its people, under the supervision of the United Nations. The rising oil price will attract great economic benefits, but more will have to be spent on buying food. Thus, the social welfare of these countries is uncertain given the volatility of oil and food prices. (3) Although MENA countries are geographically concentrated, the resources and economic conditions of these countries are vastly different. Only certain countries are rich in oil products. Therefore, for oil-importing countries (oil importers) and oilexporting countries (oil exporters), international oil prices can exhibit differing impact mechanisms on local food prices. Researching the influence of oil prices on the MENA region's food prices-where food is primarily dependent on importsis of great significance in establishing a stable food price mechanism and ensuring regional security.
Consequently, this study will focus on exploring the distinctive mechanisms impacting oil and food prices in the MENA region by comparing oil exporters to oil importers. This research is among the first to focus on examining this from a MENA perspective. Previous studies often used time series which cannot capture inter-group information and eliminate the individual fixed effect; this study explores such relationship using a panel data sample. Considering food prices may appear similar or dissimilar as oil prices fluctuate, the research builds linear and nonlinear autoregressive distributed lag (ARDL) modelling methods to address symmetrical and asymmetrical interlinkage between the prices of food and oil. The contributions of this study are (i) focusing on a specific group of countries in the MENA region to determine the impacts of the oil price on their food prices and (ii) focusing on a comparative exploration on the subject by comparing oil exporters with oil importers. (iii) Simultaneously, the symmetrical and asymmetrical interactions in the food-oil price nexus are captured by applying panel data from oil exporters and importers of the MENA region. The remainder of this paper has the following structures: in 'Literature review', the research literature relating to food-oil prices nexus is summarized; in 'Indicators and data', the data and indicators used for the study are presented; 'Methodology' highlights the panel symmetric and asymmetric ARDL models; 'Results and discussions' discusses the empirical application for MENA countries from 2000 to 2020; and the paper concludes with 'Conclusions and policy implications', which discusses policy implications.

Literature review
Substantial research on food and oil markets was stimulated by the 2008 global food crisis. The volatility of the crude oil price affects food commodities through increasing fertilizer costs, transportation costs and fuels for agricultural machinery and so on (Chen et al. (2019). The existing studies contend that food prices are significantly and positively related to the oil price. For instance, Chen et al. (2010), using the ARDL model, detected that crude oil fluctuations and other grain market fluctuations have a significant impact on the price of grain. Esmaeili and Shokoohi (2011) used a principal component analysis to investigate the nexus between the global oil price, world prices of food and other macroeconomic variables. Their findings revealed that the price of oil influences food prices in an indirect way, as confirmed by Pal and Mitra (2018), Gohin and Chantret (2010) and Ciaian and Kancs (2011). However, some scholars have found that food prices are not influenced by oil price fluctuations. For example, Reboredo (2012) examined co-movements among the global oil price and those of wheat, corn and soybeans by applying copulas. His results indicated that the increase in food price was not caused by volatility in oil prices. Therefore, it supports neutrality against oil price fluctuations in agricultural markets. Baumeister and Kilian (2014) confirmed that shocks in oil prices are not linked to food prices in the USA due to the small contribution of agricultural commodities to total food prices. Another analysis containing the volatility transmission of corn, oil and ethanol prices was performed by Gardebroek and Hernandez (2013). They found that there is no major fluctuations in the energy market that boost price variability on the corn market.
Due to the differences in the selected samples and the complexity of the relationships between them, it is difficult to reach a consensus from the literature. Increasingly, scholars believe this is a complex nonlinear relationship. Using Malaysia as a case, Ibrahim (2015) reported the presence of short-and long-run asymmetries in food prices behaviour when oil price increases which led to increasing food prices-but there was no association between the decline in oil prices and food prices. The potential asymmetry for transit from the price of oil to food is also analysed in an Indian context (Pal and Mitra 2016). For an oil-dependent emerging economy as Nigeria, Nwoko et al. (2016) investigated the causal link between food and oil prices volatility and concluded that the causality from oil prices to aggregate food price fluctuations is unidirectional. Olayungbo and Hassan (2016) examined the symmetric interactions between food and oil prices of oil exporters by applying a panel ARDL approach. The long-run finding indicated that oil prices affected food prices positively and significantly, while the shortterm impact was similarly positive and significant. These results were verified by Meyer et al. (2018).
As the hub connecting Asia, Europe and Africa and the most important oil-producing region in the world, the MENA region has garnered much attention. Ek Fälth et al. (2021) explored the impact of nuclear energy, land availability and the expansion of transmission on the cost of electricity from renewable sources by comparing the MENA region to Europe. Kassouri and Altıntaş (2020) examined the trade-off regarding environmental and human well-being issues. Bellakhal et al. (2019) focused on the interaction impact between governance and trade openness on investments in renewable energy in the MENA region over the period 1996-2013. Apergis et al. (2014) investigated the Dutch disease impact of oil rents on added value for the agricultural sector in oil producers of MENA region, establishing a long-run negative relationship that relatively slowly re-equilibrium with an added agricultural value following a boom in oil rents. Although previous literature has investigated many aspects of the MENA region, the mechanism influencing food and oil prices has been neglected.
In conclusion, although there are considerable researches on the effects of oil price on food price, there are few studies that examine and explore how oil price influences food prices using a sample of MENA nations. Previous studies on this issue have always used time series which cannot capture group information or mitigate individual fixed effects. Consequently, this study explores this relationship using panel linear and nonlinear ARDL models to assess both short-term and long-term dynamic interactions between food and oil prices for MENA countries.

Indicators and data
To explore the association of food and oil prices in MENA countries, this study employs panel data of food prices (FDP) from 2000 to 2020 for 14 countries of the MENA region and annual time series data of the global oil price (OLP). Referring to previous literature (e.g. Meyer et al. 2018;Olayungbo and Hassan 2016;Taghizadeh-hesary et al. 2019), additional panel data such as the inflation rate (INF), the degree of trade openness (TO) and urbanization levels (URB) are chosen as control variables. Given data availability, we only obtain time series annual data for the global oil price (the average of two major types of crude oil: Brent and West Texas Intermediate (WTI)). Food prices in the MENA region are transformed by consumer prices indices to actual values for the base period (2015 = 100). The degree of trade openness (TO) is expressed as a percentage of total imports and exports to gross domestic product (GDP), and the urbanization level (URB) is measured as the urban population ratio. The relevant data was derived from the Energy Information Administration (EIA), the BP Statistical Review of World Energy, the Food and Agriculture Organization (FAO) and the World Bank Database. Specific indicators for variables are shown in Table 1. The sample countries were divided into two groups to analyse the data more precisely: group 1, oil exporters group (Algeria, Bahrain, Iraq, Kuwait, Oman, Qatar, Saudi Arabia and UAE); and group 2, oil importers group (Jordan, Lebanon, Tunisia, Morocco, Egypt, West Bank and Gaza). Table 2 provides descriptive statistics. For the oil exporters sample, the mean of food prices (FDP) is greater than for the oil importers (81.702). The oil price (OLP) has volatility of 25.704 based on its standard deviation, which denotes some shocks between 2000 and 2020. The inflation rate (INF) among the oil-exporting group has a maximum value of 53.231% and a minimum of −10.067%, whereas those values for oil importers reach 29.602% and −3.749%. The degree of trade openness (TO) among the MENA region has a mean of 92.768%, with a maximum of 191.872% and a minimum of 30.247%. Finally, the mean of the urbanization level (URB) for oil exporters (83.294%) is higher than for oil importers (69.049%). Overall, different connected mechanisms for food and oil prices in the two types of nations are clearly indicated. Table 3 shows the correlation matrix among variables. For the MENA countries, there is a positive correlation between food prices (FDP) and oil prices (OLP). The inflation rate (INF) is positively correlated with food prices (FDP), except for oil exporters where there is a negative correlation. Furthermore, the degree of trade openness (TO) negatively correlates with food prices (FDP) for oil importers and the MENA region, and the relation between urbanization level (URB) and food prices (FDP) is positive for oil exporters. These results indicate that there is no potential multicollinearity problem.

Methodology
This article specified the recent method proposed by Pesaran et al. (2001) of the autoregressive distributed lag (ARDL) boundary testing to determine the influence of global oil Note.
(1) The CD test is based upon the null hypothesis that the cross-sectional independence tends to N (0,1). A p value near zero indicates the correlation between panel sets (2) Parentheses denote probability values, while ***, ** and * represent 1%, 5% and 10%, respectively, of significance. If no special instructions, the following symbols are the same prices on MENA countries' local food prices. This approach has many advantages compared to classical co-integration models (e.g. Engle and Granger 1987;Johansen and Juselius 1992), including (i) estimations of both short-and long-term coefficients can be captured simultaneously; (ii) it is practicable even if I(0) or I(1) or combination of any of the regressors are used; and (iii) prevent endogeneity issues by taking into consideration a small sample and producing better outcomes over other co-integration methods. All variables are logarithmically transformed to address the potential heteroskedasticity problem. We construct panel linear ARDL and nonlinear ARDL models according to Shin et al. (2014) and Salisu and Isah (2017) to detect the existence of symmetrical and asymmetrical relationships among global oil prices and local food prices.

The panel linear ARDL model
Given the assumption that oil prices have a symmetric influence on food prices-the effect is similar if oil prices increase or decrease-and referring to Salisu and Isah (2017), the following formula is expressed by a symmetric form of linear ARDL: where Δ is defined as the operator of differences; i = 1, 2, …, N refers to each country's numbers; t = 1, 2, …, T denotes the time periods; p and q represent the optimum lag for dependent and independent variables, respectively; and u i is the groupspecific effect. The residuals & e_x1D700; i are assumed to be white noise. β 1i , β 2i , β 3i , β 4i and β 5i are the parameters for the short term, while α 1 , α 2 , α 3 , α 4 and α 5 are the parameters for the long term.
To estimate short-term dynamic coefficients (e.g. Kun et al., 2015), Eq. (1) has been re-specificated by using the model of panel error correction model (ECM) as: where the co-integration term refers to the error correction term (ECT). λ i is the coefficient of speed adjustment in the error correction model towards a long-term equilibrium and is required to be significant and negative. β 1i , β 2i , β 3i , β 4i , and β 5i are shortrun coefficients as shown in Eq. (1).

The panel nonlinear ARDL model
In contrast to the symmetric linear ARDL Model, asymmetries are calculated in this case to investigate food price asymmetric reactions to oil prices by decomposing the sum of negative and positive partial following Salisu and Isah (2017) and Shin et al. (2014). According to this model, positive and negative oil price shocks are not assumed to affect the price of food similarly. The ARDL model can therefore be expressed nonlinearly in the following form: where LOLP − and LOLP + represent the logarithm of partial sums for changes in negative and positive oil prices, respectively, indicating negative and positive oil price shocks.
Since we added the error correction term of the linear model in Eq. (1), we likewise utilize this term for the nonlinear version as follows: The error correction term ECT t-1 captures the long-term equilibrium of the nonlinear panel. ARDL is captured by the error correction term ECT t − 1 as Eq. (2). λ i is an error correction parameter that determines the speed at which the independent variable adjusts to reach its long-term equilibrium as a result of shocks in the dependent variable.

Data stationary test result
As a prerequisite for selecting an econometric model, panel unit root tests are applied on each variable to ensure that the data used are stationary at levels or first-order differences. With the chosen samples being countries, we suspend the cross-sectional dependence across them in our model. We utilized the cross-sectional dependence Pesaran CD test to assure this assumption. The CD test findings indicate the    (Pesaran 2007) which is among the secondgeneration panel unit root tests for checking the stationarity levels of the variables in the sample countries. The findings of the CADF tests indicated in Table 5 that the variables considered in our study are not integrated at an order greater than I(1). These levels of integration confirm the convenience of the panel ARDL approach.

Food prices symmetrical response to oil prices
The estimated symmetric impact is summarized in Table 6 using Eq. (1). Here, all the equations are estimated with a pooled mean group (PMG) of dynamic heterogeneous panels (Pesaran et al. 1999), Pesaran and Smith's (1995) mean group estimator (MG) and the dynamic panel model with fixed effects (DFE). The Hausman test p values for MG and DFE are not significant which indicates that the PMG is the adequate estimator in all sample countries for modelling the symmetric nexus among food prices and oil prices. Thus, we accept the estimated results for PMG methods.
As shown in Table 6, the long-run finding shows that oil prices impact food prices positively and significantly for all samples. This corresponds to several empirical studies (e.g. Alghalith 2010; Baumeister and Kilian 2014; Olayungbo and Hassan 2016; Taghizadeh-hesary et al. 2019), suggesting that a rising oil price will induce higher food prices in the long term. Further, we observe that the coefficients of LOLP of oil exporters are greater than those for oil importers which indicates that the food prices of oil exporters are rising more than that of oil importers with an increasing oil price. The reason may be that the revenue gains from high oil prices for oil exporters will spur more food demand and increase energy consumption costs in the food production process and ultimately lead to higher local food prices. It is also important to note that the total social welfare of oil exporters may decrease because high food import costs offset the benefits of oil exports. Another phenomenon is that oil prices have a significantly negative short-term influence on oil exporters' food prices, unlike for oil importers where this is not significant. The reason for this is that oil exporters can take short-term measures to mitigate the negative effects on the food system from high oil prices, while oil importers have greater difficulty in taking effective short-run measures.   The rate of inflation (INF) is similarly affected by food prices for both oil exporters and oil importers positively and significantly in the long term, while the short-term effect is insignificant. This finding is in line with Furceri et al. (2016), indicating that inflation has boosted food prices. However, this effect is larger for oil exporters than oil importers. Specifically, a 1% increase in the inflation rate increases food prices in the long term by 0.049% for oil exporters, while this change is 0.037% for the oil-importing group in the long term. This indicates that prices of food for oil exporters are more sensitive than for importers.
In addition, the trade openness index (TO) affected negatively and significantly the long-run food prices for all samples. This effect can be explained by the competitiveness of food commodities when MENA countries open up to foreign trade, leading to further long-term declines in food prices. The short-term coefficient of trade openness was found to be negative and insignificant in the MENA region. The reason may be that it takes a long time for the welfare effect of trade freedom to impact food prices.
The long-run results show that urban population (URB) also affects food prices positively and is statistically significant for the sampled countries. This implies that rural migration to cities within the same country-or between MENA countries-has resulted in a substantial increase in the urban population, thereby depriving labour of the agricultural sector in rural areas. As a consequence, the agricultural sector's performance in these areas has deteriorated, leading to higher food prices.
For all sample countries, the error correction coefficients (ECTs) terms are statistically significant, negative and less than one, demonstrating that short-run fluctuations in the system will converge in a long-run relationship. For oil exporters, the distortions due to food price shocks can be corrected at a speed of 31.4%. Alternatively, for oil importers, the speed of   long-run equilibrium adjustment is 44.2%-which is faster than for the oil-exporting group.
To test how robust the estimated outcomes are, this study re-estimated the PMG model by gradually adding variables, and the panel OLS results are reported (see Appendix  Tables 11, 12 and 13). The former results show that as the number of variables increases, the coefficients of most variables fluctuated in a small range, verifying that the estimation results are robust. The latter results indicate that, as expected, most of the variables are correctly signed, proving that the long-run result of PMG estimates is reasonable.

Food prices asymmetrical response to oil prices
Food prices in MENA countries may have different impact mechanisms in the face of fluctuating oil prices. To explore this asymmetric response, we use panel data samples for country groups and time series for each country to analyse the heterogeneous response of different samples.

Estimation results using panel data
With the null hypothesis that positive and negative changes are not jointly significant, Wald tests are conducted to investigate the existence of an asymmetry relationship with panel data used for both groups of oil-exporter and oil-importer nations. Table 7 shows that for the entire sample of MENA countries, the null hypothesis is rejected for both short and long terms, which supports the asymmetrical relationship and implies that oil prices do not have an identical impact on food prices as oil prices rise and decrease. From the perspective of country groups, however, the F statistics of the Wald test are not significant. This result predicts the heterogeneous response within the oil-importing and -exporting country groups.
The estimated asymmetric impacts of changes in oil price on food price sample groups are presented in Table 8 using Eq.
(3). First, all equations are estimated using PMG, MG and DFE estimators and then choose the most appropriate estimator based on the Hausman test in the last line of Table 8. The findings show that the null hypothesis is appropriate, which suggests that all groups are consistent with the PMG estimator.
For the MENA sample, the short-term asymmetrical parameters of ΔLOLP þ t−1 and ΔLOLP − t−1 are negative at a 5% level of significance, implying that food prices will decline over the short term, whether oil prices increase or decrease. One possible reason is that in the short term, governments provide timely food subsidies and other measures to stabilize prices. However, the long-term asymmetric parameters (0.296 and 0.302) are positively significant at the 1% level, which indicates that, whether oil prices rise or fall, food prices always rise. This confirms that continuing food price rises are inevitable over the long term even though it will decline in the short term. Another observed phenomenon is that when oil prices rise, food prices rise faster than the corresponding decline when oil prices fall. This finding indicates that when the international oil price rises, it exerts more pressure on the domestic food price.
Error correction coefficients (ECTs) terms are negatively significant for the samples of oil-exporting and MENA countries that support a long-term convergence. When a short-term deviation is caused by shocks in food prices, the adjustment to the long-run equilibrium in the long term is 26.2% and 25% in oil-exporting and MENA countries, respectively. Alternatively, the coefficient of ECT for oil-importing countries is shown to be negative but statistically insignificant, indicating no convergence in the long-run relationship.

Estimation results using time series
We estimate the nonlinear effects of oil price on each country's food price by employing time series for the two groups (oil exporters and oil importers) following the nonlinear ARDL model. The Wald test results for asymmetries are summarized in Table 9.
The asymmetry test findings for the oil-exporting groupincluding Algeria, Bahrain, Iraq, Oman and Qatar-indicated that the Wald test F statistics are significant in the long term, suggesting that asymmetrical influence exists in those countries. Alternatively, economies such as Kuwait, Saudi Arabia and the United Arab Emirates have rejected the null hypothesis regarding the existence of asymmetric linkage in the short and long terms, which indicates that the oil-food prices relationship is not asymmetric. This evidence supports our observations of the cumulative effects of oil price on food price (see Appendix Figure 1), where the asymmetry line shows that food prices in Kuwait, Saudi Arabia and the UAE do not react differently to shocks in the oil price-whether increasing or decreasing.
Except for Tunisia, all oil importers pass the long-term Wald test at a 5% significance level, indicating how asymmetric the changes in the price of oil are influencing long-term food prices. The asymmetry line implies that the impact of oil price on food price in Egypt, Jordan, Lebanon, Morocco, the West Bank and Gaza is not identical for either a rise or decline in oil price, as seen in the plots of the cumulative effects of oil price on food price (see Appendix Figure 2).
The long-run asymmetric effects for each MENA country are shown in Table 10. For the oil-exporting group, the results show that the coefficients associated with increases in the oil price positively affect food prices. Specifically, an increase of 1% in positive oil price changes causes an increase of between 0.24 and 0.59% in food prices. Nevertheless, the coefficients related to decreases in the oil price negatively affect food prices in Iraq, whereas in Algeria, Bahrain, Oman and Qatar, these are insignificant. For oil-importing nations, the outcomes also indicate that food prices are positively influenced by the increasing oil price changes. A 1% rise in oil prices results in a 0.12 to 0.40% increase in food prices. However, the coefficients of oil price reductions have been found to be mixed. Egypt and Lebanon are positively affected, while Jordan is negatively affected, and the impact on Morocco, the West Bank and Gaza is insignificant.
By comparing our time series results with estimates of the panel data, we concluded that due to the countries' different economic statuses, the asymmetric impacts of fluctuations in oil prices on the food price vary. In general, our conclusion is that when oil prices rise, the food prices also rise; but when oil prices drop, food prices do not always decrease. This also proves that, as a necessary commodity, the price of food is sticky-it does not fall easily.

Conclusions and policy implications
The linear and nonlinear panel ARDL models are used in this paper to investigate the symmetrical and asymmetrical relationships between world oil prices and local food prices for the MENA region from 2000 to 2020. Findings from the symmetric linear ARDL model show that oil prices have a long-run positive and significant impact on food prices for oil-exporting and importing MENA nations. The positive impact is larger for oil exporters than for oil importers. This outcome concludes oil prices have a greater influence on increases in food prices for those economies that export oil at high prices. This results in increased costs for imported food due to the higher energy cost needed to produce food commodities in the country of origin. Furthermore, we found that the inflation rate and urban population affect positively and significantly the food prices for oil exporters and importers groups, while trade openness is negative and significant in relation to food prices. These results suggest that expanding trade openness can help these countries source cheaper food resources. Additionally, reducing inflation and controlling the scale of urbanization will foster agricultural development and ensure an adequate food supply.
For the entire sample from the MENA region, outcomes indicate that the oil price's effect on food price is asymmetrical in the short and long terms, although this effect is insignificant for the oil exporters and importers groups. This finding follows Meyer et al. (2018) and Ibrahim (2015). However, the short-term asymmetric effect was negative, while this was found to be positive in the long term, indicating that-whether oil prices rise or fall-food prices always rise. These results indicate that the food products price is sticky: once food prices rise, it is harder to reduce them. Therefore, policymakers should take measures to improve agricultural labour productivity, increase the supply capacity of agricultural products and develop renewable energy sources (such as photovoltaic cells) to eliminate the dependence of agriculture on oil. For the MENA region, developing biomass energy is discouraged to prevent the reduction in available agricultural land. Thus, while developing biomass energy, the government should reasonably assess its potential impact on agricultural land and food supply.
Regarding the nonlinear ARDL results for each country using time series, we found an absence of asymmetrical behaviour for nations including Kuwait, Saudi Arabia and the UAE from the oil-exporting group and Tunisia from the oil-importing group. Except for those countries, others have asymmetrical effects in food price responses to oil prices in the long term. Oil prices have a major impact on countries such as Iraq and Jordan, and food prices tracked oil price fluctuations. To maintain the stability of food prices, these countries need to make countercyclic adjustments: when the oil price is expected to rise, they should stockpile a large amount of food, and when the oil price is expected to fall, they should sell food. For countries such as Egypt and Lebanon, the rate of increase in food prices fluctuates with rising oil prices; hence, these governments need to establish long-term food price stability mechanisms-increasing food production, establishing a stable international food trading partner and other policies regardless of how oil prices fluctuate. Additionally, for countries such as Algeria and Bahrain which are seriously affected by increases in the oil price-but not by its decline-need to stockpile more agricultural products before forecasting higher oil prices.
As one of the global regions with insufficient food supply, food price stability in the MENA region is an important guarantee of food security. First, MENA governments should review their agricultural policies by providing incentives and implementing effective mechanisms to increase domestic production to avoid the effects of high food prices that result from energy price fluctuations; second, as most MENA agricultural resources (whether water or arable land) are located in oil-importing countries (characterized by a scarcity of financial resources) while oil exporters have enormous oil wealth (matched by a scarcity of arable land and water), cooperation should be strengthened between countries to stabilize oil and food prices. Third, according to the differentiated symmetrical and asymmetrical mechanisms of food and oil prices, MENA countries should establish specific coping strategies. For countries with a relatively large impact on international oil prices, an early warning mechanism could mitigate price spikes in food and oil.
Due to data limitations, this paper does not use monthly data of food prices for MENA countries to measure the impact accurately of fluctuations in international crude oil prices. Wheat is the primary staple in MENA, and the region is the largest wheatimporting region in the world. Therefore, future researches should examine the interaction between crude oil prices and MENA wheat prices.

Declarations
Ethical approval Not applicable.

Consent to participate Not applicable.
Consent for publication Written informed consent for publication was obtained from all participants.
Competing interests The authors declare no competing interests.