The impacts of trade freedom on food insecurity from countries with different profiles: an empirical analysis

Os impactos da liberdade comercial na insegurança alimentar de países com diferentes perfis: uma análise empírica

ABSTRACT

This research considered data from 124 countries (cross-section) to assess the effects of trade freedom (TF) on food insecurity (FI). The TF proxies, defined by Extreme Bounds Analysis, were tested in linear/non-linear formats, for different income-levels and using quantile-regressions. We found that a higher TF could be useful in lower-income countries and/or those with critical FI cases, but it would be ineffective where TF is very low/high. Moreover, FI would be worse in poor/recessive, inflationary and landlocked economies, rural areas, with overvalued exchange rates, high population growth, low agricultural productivity, with armed conflicts and/or natural disasters and a hot-dry climate.

KEYWORDS:
Public health; food security; trade freedom; quantitative methods

RESUMO

Este artigo considerou dados cross-section de 124 países para avaliar os efeitos da liberdade comercial (LC) na insegurança alimentar (IA). As proxies de LC, definidas via Extreme Bounds Analysis, foram testadas em formatos lineares/não lineares, para diferentes níveis de renda e utilizando regressões quantílicas. Estimou-se que um maior nível de LC poderia beneficiar países de baixa renda e/ou com IA crítica, mas seria ineficaz onde a LC fosse muito baixa/alta. Ademais, a IA seria pior em economias pobres/recessivas, inflacionárias e sem litoral, em zonas rurais, com câmbio sobrevalorizado, elevado crescimento populacional, baixa produtividade agrícola, conflitos armados, catástrofes naturais e clima quente/seco.

PALAVRAS-CHAVE:
Saúde pública; segurança alimentar; liberdade comercial; métodos quantitativos

1. INTRODUCTION

Despite advances in world agriculture (Paarlberg and Philip, 2008), food insecurity has gained strength due to climate change, the COVID-19 pandemic and the recent Russia-Ukraine conflict (Hassen and Bilali, 2022). According to Sun and Zhang (2021), approximately 720 to 811 million people starved in the world in 2020, and more than 2.37 billion did not have access to enough food due to high prices. The authors also state that those numbers will increase in the coming years due to the harmful effects of COVID-19 on the food production chain.

Among developed countries, the food security requires securing the people’s access to the food they want without compromising their other basic needs, such as housing, transportation, and leisure. However, in poorer countries, the lack of food or the difficulty of obtaining it can become a public health problem, leading thousands of people to chronic undernutrition (Long et al., 2020). It is no coincidence that 98% of undernutrition cases are concentrated in developing countries (Bezuneh and Yiheyis, 2014).

Based on this scenario, several authors started to investigate whether trade liberalization could minimize food insecurity (Abdullateef and Ijaiya, 2010; Pyakuryal, Roy and Thapa, 2010; Mccorriston et al., 2013; Bezuneh and Yiheyis, 2014; Dithmer and Abdulai, 2017; Fusco, Coluccia and Leo, 2020; Sun and Zhang, 2021; Fathelrahman, Davies and Muhammad, 2021; Montolalu et al., 2022). Such authors were based on theories from International Economics (Krugman, Obstfeld and Melitz, 2022), which suggest that curbing trade barriers would stimulate both the local comparative advantages and the productive specialization, allowing consumers to enjoy greater food availability and variety (Montolalu et al., 2022).

Although some of these studies have indicated that trade freedom could benefit food security (Dithmer and Abdulai, 2017; Fusco, Coluccia and Leo, 2020; Fathelrahman, Davies and Muhammad, 2021; Montolalu et al., 2022), this topic remains controversial. When considering 34 studies on the subject, McCorriston et al. (2013) noted that 13 pointed to favorable results for lowering trade barriers in developing countries, 10 indicated that such a practice would harm the food security, and 11 had controversial effects. Moreover, Bouët et al. (2005) and Bezuneh and Yiheyis (2014) reveal that trade liberalization could harm some poorer areas and would mainly favor developed countries and large agricultural exporters.

Such divergences may be associated with the multiple ways of measuring food security (FI) and the different methodologies employed (McCorriston et al., 2013; Sun and Zhang, 2021). In this sense, this research differs from the others by assessing the FI level from 124 countries (Food and Agriculture Organization, FAO et al., 2022), using the Extreme Bounds Analysis (EBA) technique (Levine and Renelt, 1992) to define the appropriate proxies for trade freedom (TF), and models with interactive terms (Greene, 2002) and quantile regressions (Cameron and Trivedi, 2010) to estimate the effect of TF in countries with different income and FI levels, respectively.

This paper’s remainder is organized as follows: the second section addresses the possible effects of trade liberalization on food security and other factors associated with this problem. The third section contains the methodology and the database description. Finally, the results, conclusions, References and Appendix are presented sequentially.

2. LITERATURE REVIEW

2.1. The debate on trade liberalization and its effects on food security

In general, policies on international trade tend to divide economists. On the one hand, the liberal current suggests that each country should focus on what it produces more efficiently in a barrier-free environment (Smith, 1776; Ricardo, 1821; Heckscher, 1919; Ohlin, 1933; Krugman, 1997; Hayek, 2020; Friedman, 2020). Thus, competition would allow such countries to specialize in goods with some (absolute or comparative) advantage, increasing the potential production, consumption and the general well-being level (Marlin-Bennett and Johnson, 2010).

Alternatively, List (1841) believes that underdeveloped countries (with “Infant Industries”) would be unable to compete, in the foreign market, without a government intervention. The author affirms that each nation’s advantages (absolute and relative) would be built over time (and not inherited). Therefore, it would be up to the government to create the necessary conditions, through some incentive, for such advantages to stand out. Since then, other authors have begun to criticize unbridled liberalism (Marx and Keynes¹; Prebisch, 1951; Kaldor, 1981; Stiglitz, 2006; Chang, 2002).

These arguments can explain both the different results of trade liberalization on food security (McCorriston et al., 2013) as the world’s fear of abandoning some protectionist policies. After all, if curbing trade barriers can improve the access to cheaper and more diversified foods (Tanaka and Hosoe, 2011; Chikhuri, 2013; Dithmer and Abdulai, 2017; FAO et al., 2022), it would also make economies more exposed to external fluctuations that could compromise food stability (Paarlberg, 2000; Tanaka and Hosoe, 2011). Moreover, Moon (2011) states that agriculture needs to balance some traditional economic objectives (e.g., efficiency gains, cost reduction, and higher yields) with other ethical and ecological aspects that could not achieved by just following a free market rule. Finally, in developing countries, trade freedom could worsen food insecurity before improving it (∩-shaped effect), which would raise some ethical questions about how to address the initial impacts of this measure (Kang, 2015).

2.2. Different ways of measuring food security and trade freedom

There are two usual indices to measure food (in)security (Allee, Lynd and Vaze, 2021). The first one, from the Economist Impact Group (EIG, 2023), called the Global Food Security Index (GFSI), started in 2012 (annual frequency), and adopts 68 indicators to measure the food security in 113 countries (Maricic et al., 2016, Chen et al., 2019; Izraelov and Silber, 2019).² The second one is based on the Food Insecurity Experience Scale (FIES) questionnaire, applied by FAO et al. (2022), since 2014. The FIES allow calculating the prevalence of moderate/severe (FI _m+s ) and severe (FI _s ) food insecurity and serve as a benchmark for the United Nations (UN) goals to combat hunger (Allee, Lynd and Vaze, 2021). The latest FAO report contained data on food insecurity for nearly 150 countries in the 2014-16 and 2019-21 periods (FAO et al., 2022).³ Figure 1 contains the latest information on the GFSI, FI _m+s and FI _s .

Figure 1
Global distribution of food (in)security

As expected, there is a strong negative correlation between food security, measured via GFSI (Map A), and the moderate/severe (-0.86) and just severe (-0.73) food insecurity (Maps B and C).⁴ Therefore, Maps A, B, and C (Figure 1) tell similar stories, revealing that the issue is critical in Africa, where moderate/severe insecurity (FI _m+s ) affects 55.5% of the population, with 22% of Africans severely affected (FI _s ). The issue is also severe in Latin America and the Caribbean (FI _m+s ≅37.3% and FI _s ≅12.3%) and Asia (FI _m+s ≅23.9% and FI _s ≅9.5%). However, it would be milder in countries from Oceania (FI _m+s ≅12.9% and FI _s ≅3.7%), North America (FI _m+s ≅8.1% and FI _s ≅0.8%) and Europe (FI _m+s ≅7.4% and FI _s ≅1.4%).

As for trade freedom, the literature has considered: a) the trade flow weighted by GDP, TRD/GDP (Dollar and Kraay, 2004; Chang, Kaltani and Loayza, 2009; Dithmer and Abdulai, 2017; Sun and Zhang, 2021); b) the import tariff, TAR _IMP (Dithmer and Abdulai, 2017; Fusco, Coluccia and Leo, 2020; Montolalu et al., 2022); c) the trade-freedom’s indices (Lawson, Murphy and Powell, 2020), from the “Economic Freedom of the World”, TF _EFW (Fraser Institute, 2023) and from “Index of Economic Freedom”, TF _IEF (Heritage Foundation, 2023).

Table 1 contains the correlations between the proxies of food (in)security (GFSI, FI _m+s , and FI _s ) and trade freedom (TRD/GDP, TAR _IMP , TF _EFW , and TF _IEF ), according to the income-level of each country (World Bank, 2023b). The global analysis (with all income-levels) suggests that trade liberalization would always benefit food security. However, it may differ among the income-groups. As for GFSI, the trade flow (TRD/GDP) would only favor food security in low/middle-income countries (correlation of 0.45) and could worsen the situation in high-income countries (-0.08) and medium/high (-0.02). Also, the correlations for low-income countries are often lower than the other income-groups (regardless the proxy), indicating a weaker effect (or even harmful - see the hatched cells) of trade freedom on food security.

Table 1
Correlation between food (in)security and trade freedom by income level

		Number of Countries	Income Level	Trade Freedom (Trd free )
				TRD/GDP	TAR IMP	TF EFW	TF IEF
Food Security Index	GFSI	33	High	-0.08	-0.17	0.28	0.30
		18	Upper-Middle	-0.02	-0.41	0.26	0.22
		25	Lower-Middle	0.45	-0.63	0.44	0.76
		10	Low	0.01	-0.09	0.14	0.23
		86	All groups	0.35	-0.80	0.79	0.82
Food Insecurity Prevalence	FI m+s	44	High	-0.38	0.55	-0.44	-0.59
		31	Upper-Middle	-0.19	0.05	-0.11	-0.02
		37	Lower-Middle	-0.11	0.24	-0.21	-0.34
		14	Low	0.15	-0.16	0.11	0.22
		126	All groups	-0.33	0.58	-0.62	-0.61
	FI s	44	High	-0.34	0.50	-0.39	-0.54
		31	Upper-Middle	-0.23	0.01	-0.12	0.08
		37	Lower-Middle	-0.02	0.21	-0.22	-0.25
		14	Low	0.17	0.04	-0.16	-0.10
		126	All groups	-0.24	0.48	-0.53	-0.49

• Notes: a) As the most recent data from TRD/GDP, TAR _IMP and TF _EFW refer to 2020, the average between 2018 and 2020 was used for all trade openness indicators; b) Hatched cells indicate unexpected results.
• Source: Own elaboration.

2.3. Other food security-associated factors

Although Table 1 may suggest some relationship between trade openness and food security, we must consider other control variables to validate that such results. Thus, based on Dithmer and Abdulai (2017), Fusco, Coluccia and Leo (2020), Sun and Zhang (2021), and Montolalu et al. (2022), we can infer that food security would depend on:

1. Agricultural propensity: This axis includes agricultural production per cultivated area (prod _agr ), the proportion of the population living in rural areas (pop _rur ), and the amount of arable land per inhabitant (land _ara ). These variables would favor food security by the “availability” criterion. However, higher levels of pop _rur could hamper “access” to food since this is a typical feature of more impoverished places.

2. Socioeconomic condition: this group considers the per capita wealth (GDP _pc ), the economic growth (∆GDP _pc ), and the population expansion (∆pop). In general, rich countries would have fewer problems related to food “access” and “quality”. Places with higher economic growth and lower population expansion would have more “access/stability” and “availability” of food, respectively.

3. Exogenous Factors: it includes non-economic elements, such as temperature (temp) and precipitation (prec), maritime access (mar), propensity to natural disasters (dis _nat ), and internal armed conflicts (conf _arm ). Conflicts/disasters could interrupt/destroy crops (Paarlberg, 2000). In general, agricultural success requires mild temperatures and some reasonable rainfall incidence. Moreover, countries without maritime access would struggle more in accessing the foreign market (Dithmer and Abdulai, 2017).

4. Macroeconomic Stability: it includes the domestic inflation (inf), under the hypothesis that it would measure changes in the cost of food and that inflationary economies would be more unstable (Dithmer and Abdulai, 2017; Fusco, Coluccia and Leo, 2020). Furthermore, since maintaining a devalued exchange rate (exc _r ) could deteriorate a country’s terms of trade, imposing additional costs on food imports (Abdullateef and Ijaiya, 2010), this factor was also tested.

5. Trade freedom (Trd _free ): we attempted to define the appropriate proxy(ies) for Trd _free among those from Table 1 (i.e., trade flow - TRD/GDP; tariff barriers - TAR _IMP ; trade freedom indicators - TF _EFW and TF _IEF ). As suggested by Kang (2015), the impact of Trd _free on food insecurity may be non-linear and, therefore, the quadratic and cubic effect from each proxy (i.e., $Tr{d}_{free}^2$ and $Tr{d}_{free}^3$) should be considered (Table 2):

Table 2
Probable linear/non-linear effect of trade freedom (Trd _free ) on food insecurity (FI)

Function	Expected signal	Behavior	Turning points
Linear	${\beta }_{Tr{d}_{free}}<0$	↑ Trd free reduces FI.	There is not.
Quadratic (∩-shape effect)	$\begin{array}{c}{\beta }_{Tr{d}_{free}}>0\\ {\beta }_{Tr{d}_{free}^2}<0\end{array}$	↑ increases FI before reducing it.	Maximum point: $-{\beta }_{Tr{d}_{free}}/2{\beta }_{Tr{d}_{free}^2}$
Cubic (∩-shape effect)	$\begin{array}{c}{\beta }_{Tr{d}_{free}}>0\\ {\beta }_{Tr{d}_{free}^2}<0\\ {\beta }_{Tr{d}_{free}^3}>0\end{array}$	↑ increases FI before reducing it. But it increases FI again when Trd free is very high.	Maximum/Minimum points: $\frac{-\left(2{\beta }_{Tr{d}_{free}^2}\right)\mp \sqrt{{\left(2{\beta }_{Tr{d}_{free}^2}\right)}^2-4\left(3{\beta }_{Tr{d}_{free}^3}\right)\left({\beta }_{Tr{d}_{free}}\right)}}{2\left(3{\beta }_{Tr{d}_{free}^3}\right)}$

• Note: The turning points are obtained by differentiating each function and setting the result equal to zero.
• Source: Own elaboration.

3. METHODOLOGY AND DATABASE

As food insecurity (FI) was only recognized as a global challenge in the 1990s (FAO, 2003) and it was just recently accepted as a public health issue (Murthy, 2016), its literature is still recent/scarce. In general, the trade freedom (TF) effect, on FI, is measured by partial (Fathelrahman, Davies and Muhammad, 2021) and general computable equilibrium models (Abdullateef and Ijaiya, 2010; Chikhuri, 2013) and simple (Bezuneh and Yiheyis, 2014; Kang, 2015; Sun and Zhang, 2021; Montolalu et al., 2022) or dynamic panel data regressions (Kang, 2015; Dithmer and Abdulai, 2017; Fusco, Coluccia and Leo, 2020; Sun and Zhang, 2021).

However, given the lack of regular measures of FI (recently resolved by FAO), the authors often build some (heterogeneous) proxies to assess the FI in a particular country (Pyakuryal, Roy and Thapa, 2010; Abdullateef and Ijaiya, 2010; Montolalu et al., 2022) or in groups of economies (Chikhuri, 2013; Bezuneh and Yiheyis, 2014; Kang, 2015; Fusco, Coluccia and Leo, 2020; Fathelrahman, Davies and Muhammad, 2021; Sun and Zhang, 2021). Only Dithmer and Abdulai (2017) analyzed the FI problem on a global scale, for 151 countries.

Thus, the present research differs from the others by assessing food insecurity (FI), measured by FAO et al. (2022), in 124 countries.⁵ The FAO’s indices (Figure 1 and Table 1) are worldwide accepted, disseminated and updated. However, they hamper the use of dynamic panels, since there is only triennial information for 2014-16 and 2019-21 and there would be no intertemporal variation to justify a simple panel approach. Therefore, like Allee, Lynd and Vaze (2021), this study relies on cross-section data to regress the following equation:

$FI={\beta }_0+{\beta }_{1}pro{d}_{agr}+{\beta }_{2}po{p}_{rur}+{\beta }_{3}lan{d}_{ara}+{\beta }_{4}GD{P}_{pc}+{\beta }_5\Delta GD{P}_{pc}+{\beta }_6\Delta pop+{\beta }_{7}temp+{\beta }_{8}prec+{\beta }_{9}mar+{\beta }_{10}di{s}_{nat}+{\beta }_{11}con{f}_{arm}+{\beta }_{12}inf+{\beta }_{13}ex{c}_r+{\beta }_{14}Tr{d}_{free}+{\beta }_{15}Tr{d}_{free}^2+{\beta }_{16}Tr{d}_{free}^3+\epsilon$ (1)

Where: β₀ is the constant; β₁...β₁₆ are the coefficients associated to each variable described in section 2.3; ε is an error term.

Equation 1 can be estimated via linear regression, including interactive terms (Greene, 2002), which relate the income-groups (Table 1) with the trade freedom (Trd _free ). In this case:

$FI={\beta }_0+{\beta }_{1}pro{d}_{agr}+{\beta }_{2}po{p}_{rur}+{\beta }_{3}lan{d}_{ara}+{\beta }_{4}GD{P}_{pc}+{\beta }_5\Delta GD{P}_{pc}+{\beta }_6\Delta pop+{\beta }_{7}temp+{\beta }_{8}prec+{\beta }_{9}mar+{\beta }_{10}di{s}_{nat}+{\beta }_{11}con{f}_{arm}+{\beta }_{12}inf+{\beta }_{13}ex{c}_r+{\beta }_{14}\left(D{y}_l*Tr{d}_{free}\right)+{\beta }_{15}\left(D{y}_{ml}*Tr{d}_{free}\right)+{\beta }_{16}\left(D{y}_{mh}*Tr{d}_{free}\right)+{\beta }_{17}\left(D{y}_h*Tr{d}_{free}\right)+{\beta }_{18}\left(D{y}_l*Tr{d}_{free}^2\right)+{\beta }_{19}\left(D{y}_{ml}*Tr{d}_{free}^2\right)+{\beta }_{20}\left(D{y}_{mh}*Tr{d}_{free}^2\right)+{\beta }_{21}\left(D{y}_h*Tr{d}_{free}^2\right)+{\beta }_{22}\left(D{y}_l*Tr{d}_{free}^3\right)+{\beta }_{23}\left(D{R}_{ml}*Tr{d}_{free}^3\right)+{\beta }_{24}\left(D{R}_{mh}*Tr{d}_{free}^3\right)+{\beta }_{25}\left(D{y}_h*Tr{d}_{free}^3\right)+\epsilon$ (2)

Where: Dy _l , Dy _lm , Dy _hm , and Dy _h are (binary) dummies that assess the effect of trade freedom in low (l), medium-low (ml), medium-high (mh), and high (h) income countries. Although simple, just Khan et al. (2019) and Montolalu et al. (2022) used interactive terms in this literature and none of them considered different income groups.

Furthermore, the current study proposes some additional approaches. Initially, the Extreme Bounds Analysis (EBA) technique (Levine and Renelt, 1992) is applied to Equation 1 (without $Tr{d}_{free}^2$ and $Tr{d}_{free}^3$) to select suitable proxies for trade freedom (Trd _free ). In sequence, the effect of trade freedom is estimated via Ordinary Least Squares (OLS), based on Equations 1 (global effect) and 2 (effect by income group). Finally, quantile regressions (Cameron and Trivedi, 2010) are adopted in Equation 1 to identify the impact of Trd _free in countries with different food insecurity levels.

The EBA technique is useful when “different studies reach different conclusions depending on what combination of regressors the investigator chooses to put into his regression” (Hoover and Perez, 2004, p. 766). Formally, the test consists of making several estimates by OLS based on Equation 3:

$FI={\alpha }_0+F{\alpha }_f+{\alpha }_{free}Tr{d}_{free}+S{\alpha }_s+\epsilon$ (3)

Where: FI is food insecurity; Trd _free is the proxy for trade freedom tested; F is a fixed group of regressors⁶ and S is a subset of three variables extracted from the matrix $X_{nxk}^{*}$, which contains the k explanatory variables from Equation 1 (except the constant, the tested variable, and its two exponential terms). Thus, k*=(17-4)=13 and n=124 countries.

Therefore, taking combinations 3 to 3 of S, {k*!/[(k*-3)!3!]}=286 regressions were performed for each Trd _free proxy and those ones which α_free remains significant at 10% (relevance criterion) and with the same sign (stability criterion) were considered “adequate”. The EBA is available on STATA software (Impávido, 1998).

Devised by Koenker and Bassett (1978), quantile regressions (QR) use the median (not the mean) as a measure of central tendency and allow estimating specific parameters for different sample quantiles. After defining a particular quantile (q), the QR estimator will minimize the β of Equation 4 (Cameron and Trivedi, 2010):

${\hat{\beta }}_q=\sum _{i:y_i\ge X\beta }^{N}q\left|y_i-X\beta \right|+\sum _{i:y_i<X\beta }^N\left(1-q\right)\left|y_i-X\beta \right|$ (4)

Where y _i is the dependent variable (FI) of i...N countries and X is the matrix of explanatory variables (see Equation 1). In QR, choosing the quantile (q) implies defining weights for positive $\left(\sum _{i:y_i\ge X\beta }^{N}q\left|y_i-X\beta \right|\right)$ and negative $\left(\sum _{i:y_i<X\beta }^N\left(1-q\right)\left|y_i-X\beta \right|\right)$ deviations from Equation 4.⁷ In this research, we tested the effect of trade freedom in countries with lower (q=0.1 and q=0.25) and higher (q=0.75 and q=0.9) food insecurity levels.

3.1. Database

1. Food Insecurity (dependent variable): this is the prevalence rate of moderate/severe (FI _m+s ) and severe (FI _s ) food insecurity, from 2019-21 (average), measured by FAO et al. (2022) for almost 150 countries.⁸

2. Agricultural propensity: We used cereal yield (tons/hectare)⁹ of each country to capture agricultural productivity (prod _agr ). Furthermore, we included the proportion of the population living in rural areas (pop _rur ) and the amount of arable land in per capita hectares (land _ara ), all available on the World Bank website (2023c).

3. Socioeconomic status: It includes per capita GDP (GDP _pc ) of purchasing power parity (PPP) in thousands of international dollars - US$* (constant/2017),¹⁰ the annual variation of per capita GDP (∆GDP _pc ), in constant values of local currency, and the annual rate of population expansion (∆pop) (World Bank, 2023c).

4. Exogenous Factors: It is the annual average of temperature (temp) in degrees Celsius (Trading Economics, 2023) and precipitation (prec) in millimeters/month (World Bank, 2023c). A dummy was used to capture maritime access (mar), whose value 1 indicates that the country has a coastline (Centre d’Études Prospectives et d’Informations Internationales - CEPII, 2023).¹¹ As for the propensity to natural disasters (dis _nat ), the total number of homeless, injured, and dead (Emergency event database, EM-DAT, 2023), in these disasters, was divided by the population of the country (International Monetary Fund - IMF, 2023) and the result was multiplied by 100. Regarding armed conflicts (conf _arm ), we considered the number of deaths per million inhabitants (World Bank, 2023c).

5. Macroeconomic Stability: It includes annual domestic inflation (inf), measured at consumer prices (CPI), and the purchasing power parity exchange rate conversion factor (exc _ppc ), based on the relationship between the USA’s CPI and that of any other country, which shows how many dollars would be needed to buy the same amount of goods in country “x” compared to the USA¹² (World Bank, 2023c).

6. Trade Freedom (Trd _free ): the proxies include the trade flow (exports+imports) weighted by GDP (TRD/GDP) and the amount (simple average) of tariff barriers on imports (TAR _IMP ), both from WITS (2023).¹³ Furthermore, the trade freedom indices from the Economic Freedom of the World - TF _EFW (Fraser Institute, 2023) and from the Index of Economic Freedom - TF _IEF (Heritage Foundation, 2023) were tested. The TF _EFW ranges from 0 to 10 and includes tariff/non-tariff measures, differences between the official/parallel exchange rate and capital/people flow restrictions. The TF _IEF ranges from 0 to 100 and includes only tariff and non-tariff barriers. In both cases, higher scores indicate greater trade freedom.

As the most recent , TAR _IMP , and TF _EFW data refer to 2020, the 2018-2020 average was adopted to all explanatory variables. Table 3 brings the main descriptive statistics from database and, based on the correlations of each explanatory variable with food insecurity (FI _m+s and FI _s ), suggests the data is quite consistent with the literature consulted.

Table 3
Expected signs and descriptive statistics of the variables

	Description	Measure	Acronym	Expected Signal	Mean	Std. Dev.	Min.	Max.	Correlation
									FI m+s	FI s
(a)	Insec. mod/severe	% of population	FI m+s	n. a.	30.53	23.86	2.20	88.70	1.00	0.89
	Insec. severe	% of population	FI s	n. a.	10.68	12.48	0.50	61.80	0.89	1.00
(b)	Agric. productivity	Ton/Hectare	prod agr	-	3.89	2.91	0.02	23.56	-0.59	-0.51
	Rural population	% of population	pop rur	+	38.42	20.63	0.00	82.82	0.58	0.48
	Arable land	Hectare/Per capita	land ara	-	0.23	0.25	0.00	1.60	-0.18	-0.13
(c)	GDP per capita	Thousand US$*	GDP pc	-	21.84	20.47	0.85	113.49	-0.72	-0.57
	Economic growth	% per year	ΔGDP pc	-	-0.58	2.58	-8.15	5.43	-0.10	-0.11
	Population growth	% per year	Δpop	+	1.12	1.15	-2.53	3.38	0.63	0.56
(d)	Temperature	ºC degrees	temp	-	18.08	8.37	-3.71	30.01	0.64	0.48
	Precipitation	Mm/month	prec	+	88.64	54.59	1.51	243.83	0.13	0.11
	Maritime access	binary (1=access)	mar	-	0.80	0.40	0.00	1.00	-0.07	-0.15
	Natural disaster	% affected	dis nat	+	0.57	1.24	0.00	6.50	0.37	0.27
	Armed conflict	Deaths/Million Pop.	conf arm	+	9.09	68.90	0.00	764.95	0.31	0.38
(e)	Inflation	% per year	inf	+	7.06	26.20	-0.31	274.38	0.23	0.20
	Exchange rate: PPP	Internal/external CPI	exc ppp	-	0.54	0.24	0.17	1.19	-0.45	-0.31
(f)	Trade flow	% of GDP	TRD/GDP	-	86.28	54.30	0.00	386.32	-0.32	-0.23
	Tariff barrier	% of imports	TAR IMP	+	7.18	5.16	0.27	23.66	0.58	0.47
	Trade FreedomEFW	Index: 0 a 10	TF EFW	-	6.97	1.31	2.91	8.82	-0.62	-0.53
	Trade FreedomIEF	Index: 0 a 100	TF IEF	-	76.42	10.53	49.43	90.67	-0.62	-0.51

• Note: The conf _arm correlation was calculated without Azerbaijan (outlier with lowest FI _s and highest conf _arm ).
• Source: Own elaboration.

4. EMPIRICAL RESULTS

The Extreme Bounds Analysis - EBA (Table 4) indicate that the trade flow (TRD/GDP) would not be stable (different signs) nor relevant (without significant coefficients at 10%) to explain food insecurity (FI _m+s or FI _s ). Furthermore, tariff barriers (TAR _IMP ) would not be relevant to explain FI _s . Thus, only the trade freedom indices (TF _EFW and TF _IEF ) proved fully “adequate” in the test. Strictly speaking, it would be possible to use TAR _IMP to analyze FI _m+s . However, as TF _EFW and TF _IEF already include it implicitly, just the referred indicators remained in subsequent analyses.

Table 4
EBA test associated with trade freedom proxies (Trd _free )

Variable	Tests	Moderate/severe insecurity FI m+s		Severe Insecurity (FI s )
		Minimum	Maximum	Minimum	Maximum
Trade flow TRD/GDP	Signal	-0.167#	0.031	-0.067	0.024
	Significant	NS	NS	NS	NS
Tariff barrier TAR IMP	Signal	0.157	2.737*	0.105	1.155#
	Significant	2.238*	2.737*	NS	NS
Trade Freedom TF EFW	Signal	-11.294*	-1.541	-5.041*	-1.043
	Significant	-11.294*	-9.157*	-5.041*	-4.731*
Trade Freedom TF IEF	Signal	-1.415*	-0.148	-0.614*	-0.092
	Significant	-1.415*	-0.982*	-0.614*	-0.585*

• Notes: a) p-value: #<0,15; *<0,10; **<0,05; ***<0,01; b) Hatched cells=failed EBA.
• Source: Own elaboration based on STATA software.

The unconditional regressions (A-L models) respected all signs from Table 2, although some are not significant (Table 5). In general, the linear estimates (models A, D, G, and J) suggest that Trd _free would always be significant at 1% (p-value: ***<0.01) and that higher trade freedom levels (either via TF _EFW or TF _IEF ) could reduce food insecurity (both FI _m+s and FI _s ).

Table 5
Estimated impacts of trade freedom on food insecurity

	Moderate/severe Food Insecurity (FI m+s )						Severe Food Insecurity (FI s )
	TF EFW			TF IEF			TF EFW			TF IEF
Models	(A)	(B)	(C)	(D)	(E)	(F)	(G)	(H)	(I)	(J)	(K)	(L)
Constant	108.76***	17.61	-200.99**	137.71***	-28.90	-863.01#	45.75***	12.33	-137.92***	57.14***	20.42	-169.29
Trd free	-11.22***	18.18#	139.03***	-1.40***	3.31	39.91*	-5.03***	5.75	88.81***	-0.61***	0.43	8.75
$Tr{d}_{free}^2$	-	-2.26**	-23.24***	-	-0.03*	-0.56#	-	-0.83#	-15.25***	-	-0.01	-0.13
$Tr{d}_{free}^3$	-	-	1.16**	-	-	0.00#	-	-	0.79***	-	-	0.00
R 2	0.38	0.42	0.45	0.38	0.40	0.42	0.28	0.30	0.35	0.26	0.27	0.27
AIC	1083.13	1077.03	1072.51	1082.81	1080.62	1079.84	941.22	939.86	932.44	944.14	945.53	947.10
Max. Point	-	4.02	4.52	-	55.17	48.12	-	3.46	4.45	-	21.50	NC
Min. Point	-	-	8.83	-	-	138.21	-	-	8.42	-	-	NC
Models	(M)	(N)	(O)	(P)	(Q)	(R)	(S)	(T)	(U)	(V)	(X)	(W)
Constant	18.00	31.18	-126.16#	18.96	94.67	-321.77	9.29	22.03	-98.49**	5.41	54.7	91.8
prod agr	-1.37*	-1.34*	-1.43*	-1.37*	-1.36*	-1.32*	-0.61	-0.58	-0.65#0	-0.64#	-0.63	-0.64
pop rur	0.18*	0.19*	0.16#	0.18*	0.19*	0.18*	0.08	0.08	0.06	0.08	0.09#	0.09
land ara	-4.14	-3.55	-2.62	-4.15	-3.64	-2.94	-1.52	-0.96	-0.24	-1.36	-1.03	-1.09
GDP pc	-0.39***	-0.41***	-0.39***	-0.39***	-0.41***	-0.41***	-0.17**	-0.19**	-0.18**	-0.18**	-0.19**	-0.19**
ΔGDP pc	-0.66	-0.68	-0.77	-0.66	-0.69	-0.77	-0.51	-0.53	-0.60	-0.56	-0.58	-0.57
Δpop	5.76***	5.82***	5.87***	5.74***	5.90***	5.60***	2.80***	2.86***	2.90***	2.89***	2.99***	3.02***
temp	0.52**	0.53**	0.55**	0.52**	0.55**	0.57**	0.11	0.13	0.14	0.13	0.15	0.15
prec	-0.01	-0.01	-0.01	-0.01	-0.01	-0.02	0.00	0.00	0.01	0.00	0.00	0.00
mar	-0.99	-1.25	-2.22	-1.00	-1.10	-1.84	-3.15	-3.41	-4.15#	-3.02	-3.08	-3.02
dis nat	1.44	1.45	1.35	1.44	1.39	1.51	0.28	0.30	0.22	0.31	0.28	0.27
conf arm	0.14	0.14	0.09	0.14	0.11	0.12	0.21#	0.21#	0.18	0.22#	0.20#	0.20#
inf	0.06#	0.05#	0.04	0.06*	0.06#	0.06*	0.01	0.01	0.00	0.02	0.02	0.02
exc ppp	8.60	9.30	6.22	8.57	7.18	6.05	7.91	8.58	6.22	7.48	6.58	6.68
D az	-126.5	-125.3	-90.1	-125.43	-103.8	-109.5	-174.6*	-173.4*	-146.4	-179.6*	-165.6*	-165.1*
Trd free	0.01	-4.75	83.88**	-0.01	-2.21	16.15	-0.47	-5.07	62.82***	0.00	-1.43	-3.07
$Tr{d}_{free}^2$	-	0.38	-14.98**	-	0.02	-0.25	-	0.37	-11.39***	-	0.01	0.03
$Tr{d}_{free}^3$	-	-	0.84**	-	-	0.00	-	-	0.65***	-	-	0.00
R 2	0.76	0.76	0.78	0.76	0.77	0.77	0.57	0.58	0.61	0.57	0.58	0.58
AIC	990.03	991.60	985.78	990.03	990.13	992.66	901.95	903.12	895.71	902.26	902.63	904.6
Max. Point	-	-	4.51	-	55.25	43.82	-	-	4.46	-	71.50	NC
Min. Point	-	6.25	7.38	-	-	122.84	-	6.85	7.22	-	-	NC

• Notes: a) p-value: #<0,15; *<0,10; **<0,05; ***<0,01; b) Hatched columns=best models.
• Source: Own elaboration based on STATA software.

The quadratic form $\left(Tr{d}_{free}^2\right)$ was significant only when using TF _EFW to explain FI _m+s (B model), indicating that trade openness could worsen FI _m+s before improving it (∩-shape effect). However, besides being significant in the C, F, and I cases, the cubic form obtained the highest R ² and the lowest AIC criteria among all estimates (C and I models). Therefore, we may infer that: (a) the TF _EFW proxy is superior to TF _IEF to explain both FI _m+s and FI _s ; (b) a trade freedom policy would generate a ∩-shaped effect (like a sine wave) on food insecurity.¹⁴

The C and I models reveal that countries whose TF _EFW are less than 4.52 or 4.45 would face a worsening of food insecurity (FI) when opening up to the foreign market. However, the FI would decrease if they kept on this opening-trade path (and exceeded the TF _EFW values mentioned above). Finally, FI would increase again if they exceeded values of 8.83 or 8.42 for TF _EFW . The interval between the maximum (4.52 and 4.45) and minimum (8.83 and 8.42) points reveals the descending part of the sine curve, where trade freedom policies would reduce FI. Using I model as example, we note that more than 86% of the countries analyzed would be in this interval (Table 5).

Obviously, trade freedom is not the only one affecting food insecurity, and conditional models (M-W) corroborate this statement by showing greater explanatory capacity (higher R²) and adequacy (lower AIC) than their unconditional versions (A-L). In all cases, countries with higher GDP _pc and lower population growth (∆pop) would have lower food insecurity levels (FI _m+s or FI _s ). In most of them, places with higher agricultural productivity (prod _agr ), smaller rural populations (pop _rur ), mild temperatures (temp), low incidence of armed conflicts (conf _arm ), and reduced inflation (inf), would also suffer less from this issue. Moreover, maritime access (mar) was beneficial in the “U” model. Again, the best R ² and AIC were obtained with FT _EFW in the cubic version (O and U). Although the signals associated with trade remain within expected, including control variables shortened the descending part of the sine curve, which now has maximum points at 4.51-4.46 and minimum between 7.38-7.22. In this case (U model), less than 47% of countries could use trade openness to reduce FI _s (Table 5).

The models with interactive terms improved the adjustment criteria (R ² and AIC) and suggest that data would explain up to 83% of FI _m+s and 69% of FI _s (Table 6 - B and D models). The sinusoidal relationship (∩-shape) between trade freedom (TF) and food insecurity (FI) remained valid in all income groups. However, when including control variables (cases B and D), the turning points of the upper-income groups (y _h and y _mh ) become very close, reducing the possibility of minimizing FI through TF policies. Actually, only 18.6% of the high-income-countries and 7.1% of the medium-high-income ones would be in this situation (model B). These percentages would be 69.4% and 38.7% among the medium-low (y _ml ) and low-income (y _l ) countries. Except for the potential benefits (on food insecurity) arising from economic growth (∆GDP _pc ), the lack of natural disasters (dis _nat ), and by adopting an undervalued exchange rate (exc _ppp ),¹⁵ no significant changes were noted in the control variables (Table 6).

Table 6
Impacts of trade freedom on food insecurity by income group

	Moderate/severe Food Insecurity (FI m+s )				Severe Food Insecurity (FI s )
Variables	(A)	Turning Points	(B)	Turning Points	(C)	Turning Points	(D)	Turning Points
Constant	-439.80***		-411.02***		-260.86**		-225.91***
prod agr	-		-1.01*		-		-0.37
pop rur	-		0.17#		-		0.06
land ara	-		-0.74		-		-0.55
GDP pc	-		-0.27**		-		-0.13*
ΔGDP pc	-		-1.18*		-		-0.91*
Δpop	-		4.41***		-		1.87**
temp	-		0.47**		-		0.07
prec	-		-0.02		-		0.00
mar	-		1.23		-		-1.79
dis nat	-		1.94*		-		0.85
conf arm	-		0.01		-		0.11
inf	-		0.06*		-		0.01
exc ppp	-		10.21		-		8.59#
D az	-		-23.92		-		-93.38
Trdfree (y h )	202.94***	5.84 (Max.)	162.01***	7.52 (Max.)	114.77***	6.16 (Max.)	87.84**	7.32 (Max.)
$Tr{d}_{free}^2\left(y_h\right)$	-28.68***	8.99 (Min.)	-20.90**	7.92 (Min.)	-16.06***	8.50 (Min.)	-11.59**	7.85 (Min.)
$Tr{d}_{free}^3\left(y_h\right)$	1.29***	Dif.: 3.15	0.90**	Dif.: 0.40	0.73***	Dif.: 2.34	0.52**	Dif.: 0.53
Trdfree (y mh )	242.55***	5.07 (Max.)	190.92***	6.44 (Max.)	128.60***	5.55 (Max.)	96.01***	6.83 (Max.)
$Tr{d}_{free}^2\left(y_{mh}\right)$	-39.96***	7.55 (Min.)	-28.83***	6.82 (Min.)	-19.83***	7.81 (Min.)	-13.38**	7.55 (Min.)
$Tr{d}_{free}^3\left(y_{mh}\right)$	2.11***	Dif.: 2.48	1.45***	Dif.: 0.38	0.99***	Dif.: 2.26	0.62*	Dif.: 0.72
Trdfree (y ml )	299.20***	4.24 (Max.)	266.73***	4.21 (Max.)	168.81***	4.30 (Max.)	142.52***	4.29 (Max.)
$Tr{d}_{free}^2\left(y_{ml}\right)$	-57.02***	6.87 (Min.)	-51.77***	6.65 (Min.)	-31.89**	6.89 (Min.)	-27.17***	6.76 (Min.)
$Tr{d}_{free}^3\left(y_{ml}\right)$	3.42***	Dif.: 2.63	3.18***	Dif.: 2.44	1.90**	Dif.: 2.59	1.64***	Dif.: 2.47
Trdfree (y l )	283.84***	4.77 (Max.)	231.91***	5.12 (Max.)	163.38***	4.66 (Max.)	131.90**	4.80 (Max.)
$Tr{d}_{free}^2\left(y_l\right)$	-51.43***	6.54 (Min.)	-39.70***	6.80 (Min.)	-29.00**	7.13 (Min.)	-22.60**	7.45 (Min.)
$Tr{d}_{free}^3\left(y_l\right)$	3.03***	Dif.: 1.77	2.22**	Dif.: 1.68	1.64***	Dif.: 2.47	1.23#	Dif.: 2.65
R 2	0.69		0.83		0.58		0.69
AIC	1020.38		968.74		897.65		883.81

• Notes: a) p-value: #<0,15; *<0,10; **<0,05; ***<0,01; b) Trd _free =TF _EFW .
• Source: Own elaboration based on STATA software.

The quantile regressions, already with control variables (Table 7), suggest that trade freedom (Trd _free ) would just minimize the food insecurity (FI _m+s and FI _s ) from countries where this problem is more critical (Q=0.75 and Q=0.90). Fortunately, these regressions obtained larger Pseudo-R ² than Q=0.10 and Q=0.25 models, allowing a better understanding of other policies that could be adopted.

Table 7
Impacts of trade freedom on different food insecurity levels

	Moderate/severe Food Insecurity (FI m+s )				Severe Food Insecurity (FI s )
Variables
Constant	-65.36	-87.22	-155.87*	-85.89#	-13.75	-28.48	-132.26*	-218.90***
prod agr	-0.07	-0.42	-1.97***	-1.24**	-0.52**	-0.41#	-0.59	-0.33
pop rur	0.02	0.09	0.29***	0.27***	0.01	0.02	0.07	0.07
land ara	-0.14	0.22	3.59	-0.70	-0.87	-2.27	0.57	1.30
GDP pc	-0.89***	-0.58***	-0.24#	-0.29**	-0.06	-0.08	-0.12	-0.17
ΔGDP pc	-0.04	-0.20	-1.63**	-1.18**	-0.13	0.01	-0.55	-0.19
Δpop	4.39**	4.44***	5.09***	5.17***	0.35	0.84	2.57*	3.52**
temp	0.11	0.20	0.93***	0.96***	0.08	0.08	0.37	0.44*
prec	-0.06#	-0.01	-0.01	0.01	-0.01	0.00	-0.02	0.02
mar	2.98	2.53	-3.13	-1.67	1.35	-0.78	-1.99	-7.35**
dis nat	2.89*	2.46*	1.43	1.14	0.12	-0.28	1.43	0.13
conf arm	0.36*	0.21	0.07	0.08	0.17***	0.20***	0.30**	0.21#
inf	0.12#	0.11*	-0.02	-0.02	0.09***	0.08***	-0.01	-0.06
exc ppp	32.04**	9.96	1.67	0.01	4.11	3.02	5.76	9.73
D az	-277.11*	-171.67	-81.37	-91.82	-129.53***	-159.65***	-245.63**	-180.57*
Trd free	44.01	53.65	106.34**	65.38**	9.71	19.92	73.73*	129.95***
$Tr{d}_{free}^2$	-8.40	-9.08	-19.61***	-12.05**	-1.84	-3.63	-12.28*	-22.43***
$Tr{d}_{free}^3$	0.51	0.5	1.13***	0.69**	0.11	0.21	0.65*	1.21***
Pseudo R 2	0.35	0.49	0.63	0.68	0.21	0.30	0.51	0.60
Max. Point	4.32	5.12	4.34	4.30	4.29	4.50	4.94	4.64
Min. Point	6.66	6.99	7.23	7.34	6.87	7.02	7.66	7.72

• Notes: a) p-value: #<0,15; *<0,10; **<0,05; ***<0,01; b) Trd _free =FT _EFT ; c) Q(x)=quantile considered.
• Source: Own elaboration based on STATA software.

In critical cases of food insecurity, the fight involves stimulating agricultural productivity (prod _agr ) and economic growth (∆GDP _pc ), having a smaller rural population (pop _rur ), more significant income levels (GDP _pc ), and containing the population growth (∆pop). Although immutable, having mild temperatures (temp) was beneficial and could aid in global policymaking. Furthermore, places with maritime access (mar) and fewer natural disasters (dis_nat ) would suffer less from severe food insecurity (FI _s ). In mild cases (Q=0.10 and Q=0.25), one should encourage agricultural productivity (prod _agr ), keep income high (GDP _pc ), contain population growth (∆pop) and internal inflation (inf), adopt an undervalued exchange rate (exc _ppp ) and avoid armed conflicts (conf _arm ). Places with fewer natural disasters (dis _nat ) and higher precipitation (prec) levels would also have greater food security.

5. CONCLUSION

This research used cross-section data from 124 countries to evaluate the effect of trade freedom (TF) on food insecurity’s prevalence (FI). The TF proxies were defined by Extreme Bounds Analysis (EBA) and tested in linear/non-linear formats, with interactive terms (for different income-levels) and using quantile-regressions. In order to increase the credibility of the inferences, other control variables (suggested by the literature) were also considered.

Both EBA and other regressions suggest that the trade freedom index, from Fraser Institute (which includes tariff and non-tariff measures, differences between the official/parallel exchange rates and controls over capital/people flow), would be the best option in the context of food insecurity. Furthermore, the TF cubic form was the best choice and revealed a sinusoidal effect (∩-shape effect) on FI, imposing hardships on countries with very low/high level of trade freedom. Despite this, the unconditional models show that almost 90% of the countries could implement trade freedom policies to reduce food insecurity. However, models with interactive terms and additional control variables indicate that only 18.6% and 7.1% of high and upper-middle-income countries, respectively, could use TF to reduce FI. These values would be 69.4% and 38.7% in low-middle and low-income countries. Finally, the quantile regressions indicated that TF would just minimize FI from countries where this problem is more critical (being ineffective in milder cases).

The other control variables were quite consistent with the literature and revealed that FI would be worse in poorer/recessive economies, with higher inflation and population growth, lower agricultural productivity and whose internal prices exceed the foreign ones. Locations with a major rural population, armed conflicts and/or natural disasters, landlocked and a hot-dry climate would also suffer more. We hope this research can stimulate other works in this field and that their results will help the policymakers to propose measures to combat food insecurity worldwide.

STATEMENT ABOUT DATA AVAILABILITY

Research data is only available upon request to the researcher.

REFERENCES

• Abdullateef, U.; Ijaiya, A.T. (2010). Agricultural trade liberalization and food security in Nigeria. Journal of Economics and International Finance, 2(12), 299.
• Allee, A.; Lynd, L.R.; Vaze, V. (2021) Cross-national analysis of food security drivers: comparing results based on the Food Insecurity. Experience Scale and Global Food Security Index. Food Security, 13(5), 1245-1261.
• Bezuneh, M.; Yiheyis, Z. (2014). Has trade liberalization improved food availability in developing countries? An empirical analysis. Journal of economic development, 39(1), 63.
• Bouët, A.; Bureau, J.C.; Decreux, Y.; Jean, S. (2005). Multilateral agricultural trade liberalisation: The contrasting fortunes of developing countries in the Doha round. World Economy, 28(9), pp. 1329-1354.
• Cameron, A.C.; Trivedi, P.K. (2010). Microeconometrics Using Stata. Rev. Ed. College Station, TX: Stata Press, 1641p.
• Centre d’Etudes Prospectives et d’Informations Internationales - CEPII. Available at: Available at: http://www.cepii.fr/CEPII/en/bdd_modele/bdd_modele_item.asp?id=6 Accessed in: June/2023.
  » http://www.cepii.fr/CEPII/en/bdd_modele/bdd_modele_item.asp?id=6
• Chang, H-J. (2002). Kicking away the ladder: development strategy in historical perspective. Anthem Press, 187p.
• Chang, R.; Kaltani, L.; Loayza, N.V. (2009). Openness can be good for growth: the role of policy complementarities. J. Dev. Econ. 90(1), 33-49.
• Chen, P.C.; Yu, M.M.; Shih, J.C.; Chang, C.C.; Hsu, S.H. (2019). A reassessment of the Global Food Security Index by using a hierarchical data envelopment analysis approach. European Journal of Operational Research, 272(2), 687-698.
• Chikhuri, K. (2013). Impact of alternative agricultural trade liberalization strategies on food security in the Sub-Saharan Africa region. Int. J. Soc. Econ. v. 40, p. 188-206.
• Dithmer, J.; Abdulai, A. (2017). Does trade openness contribute to food security? A dynamic panel analysis. Food Policy, 69, 218-230.
• Dollar, D.; Kraay, A. (2004). Trade, growth, and poverty. Econ. J. 114(493), F22-F49.
• Economist Impact Group - EIG. Global Food Security Index 2022. Available at: Available at: https://impact.economist.com/sustainability/project/food-security-index Accessed in: Apr/2023.
  » https://impact.economist.com/sustainability/project/food-security-index
• Emergency event: the international disaster database - EM-DAT. Available at: Available at: https://www.emdat.be/ Accessed in: April/2023.
  » https://www.emdat.be/
• FAO; IFAD; UNICEF; WFP; WHO. (2022). The State of Food Security and Nutrition in the World 2022. Repurposing food and agricultural policies to make healthy diets more affordable. Rome, FAO, 231p. Available at: https://doi.org/10.4060/cc0639en
  » https://doi.org/10.4060/cc0639en
• Fathelrahman, E.; Davies, S.; Muhammad, S. (2021). Food trade openness and enhancement of food security-partial equilibrium model simulations for selected countries. Sustainability, 13(8), 4107.
• Fonseca, P.C.D. (2010). Keynes: o liberalismo econômico como mito. Economia e Sociedade, v. 19, n. 3, p. 425-447.
• Food and Agriculture Organization - FAO (2003). Trade Reforms and Food Security: Conceptualizing the Linkages. FAO: Rome, Italy.
• Food and Agriculture Organization - FAO. (2017). Food Security Policy Brief. Food Secur. 2(1).
• Fraser Institute - Economic Freedom of the World/Freedom to Trade Internationally. Available at: Available at: https://www.fraserinstitute.org/about Accessed in: April/2023.
  » https://www.fraserinstitute.org/about
• Friedman, M. (2020). Capitalism and freedom. University of Chicago press, 272p.
• Fusco, G.; Coluccia, B.; De Leo, F. (2020). Effect of trade openness on food security in the EU: A dynamic panel analysis. International Journal of Environmental Research and Public Health, 17(12), 4311.
• Greene, W. (2002). Econometric analysis 5th ed. Prentice Hall. Upper Saddle River: NJ, 802p.
• Hassen, T.B.; Bilali, H.E. (2022). Impacts of the Russia-Ukraine War on Global Food Security: Towards More Sustainable and Resilient Food Systems? Foods 11, nº. 15: 2301.
• Hayek, F.A. (2020). The constitution of liberty: The definitive edition. Routledge, 583p.
• Heckscher, E. (1919). The effect of foreign trade on the distribution of income. Ekonomisk Tidskriff, 497-512.
• Heritage Foundation - Index of Economic Freedom/Trade Freedom. Available at: Available at: https://www.heritage.org/index/ Accessed in: April/2023.
  » https://www.heritage.org/index/
• Hoover, K.D.; Perez, S.J. (2004). Truth and Robustness in Cross-country Growth Regressions. Oxford Bulletin of Economics and Statistics, 66(5), p. 765-798.
• Impavido, G. (1998). EBA: Stata Module to Perform Extreme Bound Analysis. Statistical Software Components (S347401), Boston College Department of Economics.
• International Monetary Fund - IMF: World Economic Outlook Database. Available at: Available at: https://www.imf.org/en/Publications/WEO/weo-database/2023/April/ Accessed in: April/2023.
  » https://www.imf.org/en/Publications/WEO/weo-database/2023/April/
• Izraelov, M.; Silber, J. (2019). An assessment of the global food security index. Food Security, 11(5), 1135-1152.
• Kaldor, N. (1981). The Role of Increasing Returns, Technical Progress and Cumulative Causation in the Theory of International Trade and Economic Growth. Economie appliquée, 34(4), p. 593-617.
• Kang, H. (2015). A study on the relationship between international trade and food security: Evidence from less developed countries (LDCs). Agric. Econ., v. 61, p. 475-483.
• Khan, A.; Shahid, A.; Syed, A.; Aftab, K.; Raza, U. (2019). Impact of climate change on maize productivity in Khyber Pakhtunkhwa, Pakistan. Sarhad Journal of Agriculture, 35(2): 594-601.
• Koenker, R.; Bassett, G. (1978) Regression quantiles. Econometrica: journal of the Econometric Society, v. 46, n. 1, p. 33-50.
• Krugman, P. (1997). What should trade negotiators negotiate about? Journal of Economic Literature, 35(1), 113-120.
• Krugman, P.R.; Obstfeld, M.; Melitz, M. (2022). International Trade: Theory and Policy. 12. ed. Pearson, 368p.
• Lawson, R.A.; Murphy, R.; Powell, B. (2020). The determinants of economic freedom: A survey. Contemporary economic policy, 38(4), 622-642.
• Levine, R.; Renelt, D. (1992). A sensitivity analysis of cross-country growth regressions. American Economic Review, 82(4): 942-63.
• List, F. (1841). The National System of Political Economy. English ed., 1904: London Longman.
• Long, M.A.; Gonçalves, L.; Stretesky, P.B.; Defeyter, M.A. (2020). Food insecurity in advanced capitalist nations: A review. Sustainability, 12, 3654.
• Maricic, M.; Bulajic, M.; Dobrota, M.; Jeremic, V. (2016). Redesigning the global food security index: A multivariate composite I-distance indicator approach. International journal of food and agricultural economics, 4(1), 69-86.
• Marlin-Bennett, R.; Johnson, D. (2010). International political economy: Overview and conceptualization. Oxford Research Encyclopedia of International Studies, 40p.
• McCorriston, S.; Hemming, D.; Godwin, J.; Parr, M.; Osborn, J.; Roberts, P. (2013). What is the evidence of the impact of agricultural trade liberalisation on food security in developing countries? A systematic review. London: EPPI Centre, Social Science Research Unit, Institute of Education, University of London, 122p.
• Montolalu, M.H.; Ekananda, M.; Dartanto, T.; Widyawati, D.; Panennungi, M. (2022). The Analysis of Trade Liberalization and Nutrition Intake for Improving Food Security across Districts in Indonesia. Sustainability, 14(6): 1-18.
• Moon, W. (2011). Is agriculture compatible with free trade? Ecol. Econ., v. 71, p. 13-24.
• Murthy, V.H. (2016). Food insecurity: a public health issue. Public Health Reports, 131(5), 655-657.
• Ohlin, B.G. (1933). Interregional and International Trade. Boston: Harvard University Press.
• Paarlberg, D.; Philip P. (2008). The agricultural revolution of the 20th century. John Wiley & Sons, 154p.
• Paarlberg, R. (2000). The weak link between world food markets and world food security. Food Policy, 25, p. 317-335.
• Prebisch, R. (1951). Theoretical and Practical Problems of Economic Growth, Mexico City, ECLAC, United Nations Publication. 18 May.
• Pyakuryal, B.; Roy, D.; Thapa, Y.B. (2010). Trade liberalization and food security in Nepal. Food Policy, 35(1), 20-31.
• Ricardo, D. (1821). On the Principles of Political Economy and Taxation. UK: J. Murray. 1821.
• Smith, A. (1776). An inquiry into the nature and causes of the wealth of nations. London. Available at: https://era.ed.ac.uk/handle/1842/1455
  » https://era.ed.ac.uk/handle/1842/1455
• Stiglitz, J. et al. (2006). Stability with growth: macroeconomics, liberalization and development. OUP Oxford, 352p.
• Sun, Z.; Zhang, D. (2021). Impact of Trade Openness on Food Security: Evidence from Panel Data for Central Asian Countries. Foods, 10, 3012.
• Tanaka, T.; Hosoe, N. (2011). Does agricultural trade liberalization increase risks of supply-side uncertainty? Effects of productivity shocks and export restrictions on welfare and food supply in Japan. Food Policy, 36, p. 368-377.
• Trading Economics. Available at: Available at: https://tradingeconomics.com/country-list/temperature Accessed in: June/2023.
  » https://tradingeconomics.com/country-list/temperature
• World Bank. Available at: Available at: https://databank.worldbank.org/metadataglossary/world-development-indicators/series/TM.TAX.MRCH.SM.AR.ZS Accessed in: April/2023a.
  » https://databank.worldbank.org/metadataglossary/world-development-indicators/series/TM.TAX.MRCH.SM.AR.ZS
• World Bank. World Bank classification by income group. Available at: Available at: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups Accessed in: April/2023b.
  » https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups
• World Bank. Available at: Available at: https://data.worldbank.org/indicator Accessed in: June/2023c.
  » https://data.worldbank.org/indicator
• World Integrated Trade Solution - WITS. Available at: Available at: https://wits.worldbank.org/ Accessed in: April/2023.
  » https://wits.worldbank.org/

• 1
  See Marlin-Bennett and Johnson (2010) and Fonseca (2010).
• 2
  The GFSI ranges from 0 to 100, where 100 represents the best possible food security.
• 3
  In order to minimize the sample-selection’s bias, FAO uses the mean of the last FI _m+s and FI _s .
• 4
  The correlations between GFSI and FI _m+s (or FI _s ) were based on 91 countries (common to both statistics).
• 5
  As the global food security index (GFSI), from EIG (2023), incorporates dozens of variables (including some described in section 2.3), increasing the endogeneity’s propensity, this indicator was disregarded.
• 6
  In this research, we adopted F=ϕ.
• 7
  In general, quantile regressions (QR) do not require a normal (Gaussian) distribution and are more robust than OLS in the presence of outliers. Furthermore, they use all the data (and not sub-samples) while estimating the different quantiles (Cameron and Trivedi, 2010).
• 8
  For the purpose of avoiding missing data, this research worked with 124 countries.
• 9
  They are wheat, rice, maize, barley, oats, rye, millet, sorghum, buckwheat, and mixed grains.
• 10
  The US$* (used in the conversion of GDPs) has the same purchasing power as the US dollar in the US.
• 11
  Available on: http://www.cepii.fr/distance/geo_cepii.xls.
• 12
  Thus, if exc _ppc (country x)=0.6, the 1 US$ good/service in the US would cost only 0.6 US$ in country x.
• 13
  According to the World Bank (2023a), the “simple averages” of import tariffs are always better indicators of protectionism level than “weighted averages”.
• 14
  Figure A1 (Appendix) contains the sinusoidal graphs, based on the cubic estimates, with the food insecurity (FI _m+s e FI _s ) behavior associated to different levels of trade freedom (TF _EFW ).
• 15
  Although exc _ppp negatively correlates with food insecurity (Table 3), its effect becomes positive after controlling the wealth (via GDP _pc ), revealing that having lower domestic prices could reduce food insecurity.
• JEL Classification:
  I18; Q18; F13; C21.

APPENDIX

Figure A1
Estimated relationship between food insecurity and trade freedom

Publication Dates

• Publication in this collection
  10 Nov 2025
• Date of issue
  2025

History

• Received
  15 Feb 2024
• Accepted
  03 Oct 2024