Globalization in the Food Sector and Poverty

Doerr, Leo M.; Maennig, Wolfgang Article — Published Version Globalization in the Food Sector and Poverty The European Journal of Development Research Provided in Cooperation with: Springer Nature Suggested Citation: Doerr, Leo M.; Maennig, Wolfgang (2025) : Globalization in the Food Sector and Poverty, The European Journal of Development Research, ISSN 1743-9728, Palgrave Macmillan, London, Vol. 37, Iss. 5, pp. 934-964, https://doi.org/10.1057/s41287-025-00711-x This Version is available at: https://hdl.handle.net/10419/330647 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Vol:.(1234567890) The European Journal of Development Research (2025) 37:934–964 https://doi.org/10.1057/s41287-025-00711-x ORIGINAL ARTICLE Globalization intheFood Sector andPoverty LeoM.Doerr1· WolfgangMaennig1 Received: 11 October 2024 / Accepted: 4 August 2025 / Published online: 12 September 2025 © The Author(s) 2025 Abstract This paper provides new evidence on the globalization‒poverty nexus by isolating the effects of the food sector on the prevalence of absolute poverty. Using panel regressions with fixed effects (FE) and Generalized Method of Moment (GMM) estimators for a panel of 12 Latin American countries from 1995 to 2020, we find that increased food trade significantly increases absolute poverty in our sample. Extended analysis suggests that this effect operates primarily through food exports. Using a recently collected indicator of globalization in the food sector, our estimates indicate that the opening of food markets since the mid-1990s may have accounted for approximately two additional percentage points of the population living below the absolute poverty line of $2.15 a day. Our results highlight that, despite its modest share of total trade, food sector globalization should be accompanied by policy measures aimed at mitigating potential poverty traps associated with structural transformations following increased food exports. Keywords Trade policy· Empirical studies of trade· Economic impacts of globalization· Agriculture in international trade· Food policy Résumé Cet article apporte de nouvelles preuves sur le lien entre mondialisation et pauvreté en isolant les effets du secteur alimentaire sur la prévalence de la pauvreté absolue. À l’aide de régressions sur données de panel avec effets fixes (FE) et d’estimateurs par méthode des moments généralisés (GMM) pour un panel de 12 pays d’Amérique latine de 1995 à 2020, nous constatons qu’une augmentation des échanges alimentaires accroît significativement la pauvreté absolue dans notre échantillon. Une analyse approfondie suggère que cet effet s’exerce principalement par le biais des exportations alimentaires. En utilisant un indicateur récemment collecté de la mondialisation dans le * Wolfgang Maennig [email protected] Leo M. Doerr [email protected] 1 Department ofEconomics, University ofHamburg, Hamburg, Von-Melle-Park 5, 20146Hamburg, Germany 935 Globalization intheFood Sector andPoverty secteur alimentaire, nos estimations indiquent que l’ouverture des marchés alimentaires depuis le milieu des années 1990 pourrait avoir entraîné environ deux points de pourcentage supplémentaires de la population vivant sous le seuil de pauvreté absolue de 2,15 $ par jour. Nos résultats soulignent que, malgré sa part modeste dans le commerce total, la mondialisation du secteur alimentaire devrait s’accompagner de mesures politiques visant à atténuer les risques de pièges à pauvreté potentiels liés aux transformations structurelles consécutives à l’augmentation des exportations alimentaires. Resumen Este artículo aporta nueva evidencia sobre la relación entre globalización y pobreza al aislar los efectos del sector alimentario en la prevalencia de la pobreza absoluta. Utilizando regresiones de panel con efectos fijos (FE) y estimadores de Método Generalizado de Momentos (GMM) para un panel de 12 países latinoamericanos entre 1995 y 2020, encontramos que el aumento del comercio de alimentos incrementa significativamente la pobreza absoluta en nuestra muestra. Un análisis ampliado sugiere que este efecto opera principalmente a través de las exportaciones de alimentos. Utilizando un indicador recientemente recopilado de globalización en el sector alimentario, nuestras estimaciones indican que la apertura de los mercados alimentarios desde mediados de la década de 1990 podría haber representado aproximadamente dos puntos porcentuales adicionales de la población viviendo por debajo del umbral de pobreza absoluta de $2.15 diarios. Nuestros resultados destacan que, a pesar de su modesta participación en el comercio total, la globalización del sector alimentario debe ir acompañada de medidas de política orientadas a mitigar los posibles círculos de pobreza asociados a las transformaciones estructurales derivadas del aumento de las exportaciones de alimentos. JEL F13· F14· F60· Q17· Q18 Introduction The effects of globalization on poverty are a much debated topic in economics, with no consensus on the sign and magnitude of the effects. A summary of the empirical results of the last three decades is given by Winters etal. (2004) and Winters and Martuscelli (2014). Most commonly proxied with trade openness, globalization is unlikely to reduce poverty on a large scale unless it stimulates economic growth, which itself is poverty reducing (Dollar and Kraay 2002; Ravallion and Datt 2002). Most papers that link trade openness directly to poverty prevalence find no significant effects (Dollar and Kraay 2002, 2004; Ravallion 2006). The effects of globalization have also been analyzed in other dimensions, and some studies have explicitly tested globalization in the food sector. It has been reported to aggravate inequality (Artuc etal. 2021), hunger (Mary 2019), employment (Porto 2008), and food security (Chikhuri 2013). On the other hand, food prices rise with inflation, often followed by decreasing unemployment rates (Berentsen et al. 2011; Ball et al. 2013); market opening can stimulate agricultural 936 L.M.Doerr, W.Maennig exports, which can promote growth and therefore reduce poverty (Porto 2008; Sanjuán-López and Dawson 2010). Adding to previous studies, we provide a first cross-country model that estimates the effect of food sector globalization, rather than general globalization, on absolute poverty. We disaggregate trade openness by sector and incorporate a new proxy for food trade globalization from Mary (2019) as an explanatory variable within a poverty model based on Polloni-Silva etal. (2021). Using food trade openness rather than general globalization measures is reasonable since the majority of people experiencing poverty live in rural areas and depend on agriculture, which is dominated by small-scale family farmers (Gollin etal. 2005). For people who live in poverty but do not directly depend on agriculture for their livelihoods, staple foods make up a large proportion of their daily expenses, and fluctuations in food prices can put them in financial distress (Winters and Martuscelli 2014). The increase in the globalization of Latin American agricultural markets is well documented in the empirical literature. Since the early 1990s, the region has experienced a pronounced agro-export boom driven by rising global demand, technological advances in agribusiness, and liberalized trade policies. Countries such as Brazil, Argentina, Chile, and Peru significantly expanded their agricultural export portfolios, mostly in high-value commodities such as soybeans, beef, fresh fruits, and specialty crops (Weisskoff 1992; Carter and Mesbah 1993; Barham etal. 1995). Simultaneously, many countries have scaled back the domestic production of staple foods such as wheat, rice, maize, and dairy, increasing import dependence on these basic goods. For example, Mexico, although a major agroexporter, imports large volumes of corn from the U.S., whereas wheat and dairy imports have surged in Peru, with substantial economic impacts (King 2006; Gonzales and Varona 2024). This pattern is also reflected in our sample: as illustrated in Fig.1, the combined share of food exports and imports in GDP increased markedly, from approximately 90% in 1995 to an average of over 200% by 2020.1 Fig. 1 Trade openness: all sectors vs. the food sector 1 To illustrate in-sample variation, Fig.2 in the appendix also includes confidence intervals, and Fig.5 displays food trade for selected countries. 937 Globalization intheFood Sector andPoverty However, the benefits of this structural transformation have been unevenly distributed. Poverty rates, particularly in rural areas, have remained high in countries with highly unequal land ownership and monoculture-based export economies where the increase in large-scale agribusiness has marginalized smallholder farmers and rural laborers, restricting their access to gains from the agro-export boom (Carter etal. 1996; Pronti etal. 2024). Two main mechanisms through which food sector globalization affects absolute poverty have been identified. On the consumer side, while more market integration can initially lower prices, it increases exposure to volatile international food markets (Litchfield etal. 2003; Huang etal. 2007), with volatility shown to significantly affect poverty rates (Ivanic and Martin 2008; De Hoyos and Medvedev 2011).2 For example, in Latin America for the period of the global food price crisis from 2007 to 2008, greater trade openness amplified food CPI spikes (Flachsbarth and Garrido 2014), and these shocks translated directly to higher poverty rates (Ivanic etal. 2012). On the producer side, integration into global food markets has been associated with increasing land concentration, as large-scale agribusiness displaces traditional smallholder farming (Bourguignon and Morrisson 1990; Pronti etal. 2024). This shift undermines rural livelihoods and consequently raises poverty rates, particularly where alternative income sources are scarce (Griffin 1976; Griffin etal. 2002). For example, Chile’s agro-export growth led to land concentration and job instability (Carter and Mesbah 1993), whereas Guatemala’s vegetable export boom curtailed peasant land access (Barham etal. 1995). Researchers early on have acknowledged that estimating cross-country poverty models is prone to errors, as the assumption that poverty in Angola and Austria, for example, is comparable, does not hold (Winters etal. 2004). Artuc etal. (2021) attempt to unify the evidence at the household level by pooling household-level data from different countries, but their database is limited to 1year, as the surveys were conducted using different methods and the indicators are not uniform. Data quality and availability do not yet allow the extension of cross-country data at the household level over several years. Following Neaime and Gaysset (2018) and Polloni-Silva etal. (2021), we take a middle path and use aggregated poverty data for countries within a continent that share similar socioeconomic and geographical conditions. In this way, we maintain comparability between countries while providing answers at a larger scale. Using our newly constructed database and poverty equation, our results suggest that globalization in the food sector, in contrast to overall globalization, has a significantly aggravated impact on poverty prevalence. The orientation of the food sector in Latin America toward international markets described by Weisskoff (1992) and Carter etal. (1996), not yet analyzed by its impacts, may account for approximately 2% points of the population living below the absolute poverty line of $2 a day in 2 This channel has received public and policy attention in the wake of recent global shocks, notably the COVID-19 pandemic and the Russia–Ukraine conflict, both of which severely disrupted international food markets and amplified food price volatility (Laborde etal. 2021; Glauber etal. 2022). Note that COVID-19 had multiple implications, which may also affect poverty (Cuesta and Pico 2020; Jara etal. 2022). 938 L.M.Doerr, W.Maennig our sample. Section“Data” presents our data, and Sect.“Empirical Strategy and Results” introduces the empirical model and presents the results. Section“Robustness” presents the robustness analysis, and Sect.“Conclusion” concludes. Data We construct a dataset with 12 Latin American countries covering approximately 84% of Latin America’s total population, resulting in a balanced panel dataset with observations from 1995 to 2020 similar to that of Polloni-Silva etal. (2021) but covering a much larger time span of 26years.3 In line with the previous literature, we define absolute poverty as the percentage of people living below the absolute poverty line (poverty), which is currently set at USD 2.15 a day at the 2017 Purchasing Power Parity (PPP) (World Bank 2022). As mentioned in the introduction, the economic literature commonly proxies globalization by a country’s openness to international trade, defined as the share of imports and exports relative to GDP (Trade). Both indicators are taken from the World Bank’s World Development indicators database (World Bank 2005). To assess the extent of globalization in the food sector, we follow Mary (2019). We begin by collecting data from the United Nations Food and Agriculture Organization statistics database on total agricultural exports and imports, as well as food exports and imports, which include both crop and livestock products (FAOSTAT 1998). Second, we calculate the ratios of agricultural exports (imports) to the total volume of exports (imports). These ratios are then multiplied by the total value of exports (imports) in constant local currency units, resulting in agricultural imports and exports. We then calculate agricultural GDP by multiplying the percentage of agriculture to total value added, obtained from the WDI database, by total GDP in constant local currency units. This allows us to construct a measure of (non)agricultural trade openness by dividing the sum of agricultural exports and imports by (non)agricultural GDP. Finally, we further decompose agricultural trade openness into trade openness in the food sector and trade openness in the remaining sectors. FAOSTAT and WDI data can then be used to calculate the share of food exports (imports) in the agricultural sector and the share of food GDP in agricultural GDP. Thus, we constructed trade openness in the food sector (Trade Food) and trade openness in the remaining sectors (Trade Other) that target non-food nonagricultural goods as subcomponents of overall trade openness. For clarity, Fig.6 of the Appendix provides a flowchart detailing the construction of our key trade openness indices (Total Trade, Food Trade, and Trade Other). While the specific impact of openness in the food sector on absolute poverty has not been the subject of research, the literature is rich on the determinants of (absolute) poverty, and we draw our control variables from it. Using data from the World Bank, we include (real) per capita GDP in USD (gdppc) as the measure of economic 3 A list of countries can be found in Table8 in the Appendix. The only Latin American country with a population of more than 30 million missing from the dataset is Venezuela, for which reliable data are not available. 939 Globalization intheFood Sector andPoverty performance. To control for effects from the labor market, we include national unemployment rates as a percentage of the total labor force (Unemployment) and the female labor force participation rate as a percentage of the overall female population older than 15years (Female workforce). The female labor force participation rate serves as a proxy for labor market dynamics but also captures important gender dimensions of poverty alleviation, as increased participation is often linked to improved household welfare and reduced vulnerability (Seguino 2000; Gaddis and Klasen 2014). Moreover, we add consumer prices (CPI, 2010 = 100) to account for poverty, which is driven by inflation. Finally, we include the word governance indicator rule of law (rule of law) and the WDI’s percentage of the urban population (urbanization) of the total population to account for general political stability and rural exodus. Table1 presents comprehensive summary statistics. Empirical Strategy andResults Replication: Determinants of(Absolute) Poverty The determinants of poverty have been modeled in different frameworks (Ravallion 2006; Nikoloski 2011; Kwon and Kim 2014; Awaworyi Churchill and Smyth 2017; Omar and Inaba 2020; Polloni-Silva etal. 2021).4 As a starting point, we draw on Polloni-Silva etal. (2021) as one of the most recent empirical models that takes into account the current state of the literature; they also model absolute poverty for a sample of Latin American countries similar to ours: where Poverty is the percentage of people living below the poverty line defined by the World Bank, GDPPC is (real) per capita GDP, Inflation is Consumer Prices, Trade is the share of imports and exports relative to GDP, Unemployment is the national unemployment rate, Female workforce is the female participation rate in national labor markets, Urbanization is the urbanization rate, Rule of law is the rule of law index, and ci is a full set of country fixed effects. Table2 Column 2.1 presents the results of estimating Eq.(1) for the same years as in on Polloni-Silva etal. (2021) (2004–2017) and shows that we were able to replicate their results in general. The per capita GDP has a significant negative effect on poverty prevalence; a 1% increase in initial levels of the GDPPC is followed by a decrease of 0.18% of people living below the absolute poverty line of $2.15 per day. Our results show weaker (negative) effects than those of on Polloni-Silva etal. (2021) but are in line with most of the relevant literature (Nikoloski 2011; Kwon and Kim 2014; Awaworyi Churchill and Smyth 2017). Consumer prices also have a significant negative effect on poverty, in line with Polloni-Silva (1) 𝑃 𝑜𝑣𝑒𝑟𝑡𝑦 𝑖𝑡 = 𝛼 0+ 𝛽 1 GDPPC i+ 𝛽 2 Inflation it + 𝛽 3 Unemployment it + 𝛽4FemaleWorkforceit + 𝛽5Urbanizationit + 𝛽6Ruleoflawit + ci + 𝜀it, 4 A comprehensive summary of studies analyzing the determinants of poverty is given in Table10 in the Appendix. 940 L.M.Doerr, W.Maennig Table 1 Variable descriptions and summary statistics Variables Definition NMean SD Min Max Poverty hr $2.15 Poverty headcount ratio at $2.15 a day (2017 PPP) (%) 273 8.092 6.309 0.400 28.60 Poverty gap $2.15 Poverty gap at $2.15 a day (2017 PPP) (%) 273 3.320 3.098 0.200 15.10 Poverty hr $3.65 Poverty headcount ratio at $3.65 a day (2017 PPP) (%) 273 17.32 10.35 2.100 48.90 Poverty gap $3.65 Poverty gap at $2.15 a day (2017 PPP) (%) 273 7.130 5.164 0.700 23.50 MPI Multidimensional poverty headcount ratio (World Bank) (% of population) 105 6.467 4.959 0.600 21.90 Trade Log (imports + Exports)/ GDP 312 0.669 0.349 0.146 1.728 Trade Food Log (Food imports + Food Exports)/Food GDP 312 1.411 0.880 0.306 4.318 Trade other Log (Non-agricultural imports + Non-agricultural Exports)/Nonagricultural GDP 312 0.640 0.361 0.130 1.735 Food Imports Import value of total food/ Food GDP 312 0.249 0.512 0.000 2.896 Food Exports Export value of total food/ Food GDP 312 0.258 0.332 0.000 1.521 GDPPC Initial GDP per capita in 100 US$ (constant 2015) 312 63.48 33.85 16.94 151.2 941 Globalization intheFood Sector andPoverty Table 1 (continued) Variables Definition NMean SD Min Max CPI Consumer price index (2010 = 100) 307 88.31 33.94 20.59 172.8 Unemployment Unemployment, total (% of total labor force) 291 7.118 3.702 2.021 20.52 Female workforce Labor force participation rate, female (% of female population ages 15+) 312 49.71 7.785 33.86 72.07 Urbanization Urban population (% of total population) 312 70.24 11.50 42.94 92.11 Food exporter Dummy = 1 if country is a net food exporter 312 0.670 0.471 0 1 948 L.M.Doerr, W.Maennig Factors Influencing theRelationship Between Poverty andFood Trade Openness Several factors may condition the relationship between food trade and poverty. First, the effect of food trade openness on hunger—and, by extension, on the very poor—may vary depending on a country’s net food trade position. As noted by Mary (2019), welfare implications differ for food-importing versus food-exporting countries. This is particularly relevant for our sample, where most countries are net food exporters (see Table1). We re-estimate the baseline equation with a dummy for net food exporters, interacting it with food trade openness (Column 5.1). The results show no evidence that food trade openness depends on a country’s net export status, consistent with Thirlwall (2013) and Gacitia and Bello (1991). However, further models estimated in robustness Sect.“Food Exports Vs. Food Imports” suggest that food exports and imports may have distinct effects on the very poor. Second, the food trade‒poverty relationship may differ between urban and rural areas. Urban populations are more quickly affected by changes in the food market, benefitting from short-term price drops or facing long-term price sensitivity (Cohen and Garrett 2010). In contrast, rural areas may gain export opportunities or lose income because of cheaper imports (Jaffee and Henson 2005; Akanle etal. 2013). To explore this, we add an urbanization–food trade interaction to Eq.(2) in Table5Column (5.2); the interaction term is insignificant, indicating no difference in poverty responses between urban and rural areas.11 Finally, the effect of food market liberalization may depend on the level of absolute poverty. To investigate this, we construct three binary indicators, poverty low (below 5%), poverty medium (5–10%), and poverty high (above 10%), and interact these with food trade openness in columns (5.3)–(5.5). The results show that the food trade‒poverty relationship varies by poverty level. In countries with poverty rates below 5%, the coefficient of food trade is similar to that of the baseline specification, with no significant interaction term. The largest impact is observed in countries with moderate poverty (5–10%), where a 1% increase in food trade openness is associated with a 0.04% point rise in poverty. In contrast, countries with high-poverty rates appear unaffected by the food trade. This may be explained by the socioeconomic structure and (food) trade barriers in different country groups. For example, countries with low poverty, such as Costa Rica and Argentina, have limited subsistence farming and industrialized, export-oriented sectors, mitigating trade-related shocks. In high-poverty countries, such as Colombia and Bolivia, food market protection shields vulnerable populations from trade volatility (see Fig.5 in the Appendix). 11 In this specification, the food trade coefficient cannot be interpreted meaningfully, as it reflects a country with zero urban share (Aiken etal. 1991); in our data sample urbanization ranges from 43 to 92%, cf. Table1). cerns about country-specific unobserved heterogeneity 𝜃i in our level equation in (3) (Roodman 2009). Footnote 10 (continued) 949 Globalization intheFood Sector andPoverty Table 5 Poverty and food trade openness: interactions Variables Dependent: PHR $2.15 (5.1) TWFE (5.2) TWFE (5.3) TWFE (5.4) TWFE (5.5) TWFE Trade Food 0.0191** − 0.0132 0.0140** 0.0446*** 0.0017 (0.0088) (0.0355) (0.0055) (0.0049) (0.0043) Trade other 0.0149 0.0071 − 0.0021 − 0.0033 − 0.0019 (0.0154) (0.0174) (0.0051) (0.0048) (0.0046) Initial GDPPC − 0.198*** − 0.214*** − 0.147*** − 0.160*** − 0.128*** (0.0306) (0.0279) (0.0157) (0.0134) (0.0115) CPI 0.0619*** 0.0634*** 0.0874*** 0.0964*** 0.0986*** (0.0171) (0.0177) (0.0217) (0.0186) (0.0198) Unemployment 0.108 0.0860 0.128* 0.153** 0.150* (0.110) (0.0993) (0.0745) (0.0759) (0.0806) Female workforce − 0.0521 − 0.0570 − 0.00843 0.00524 0.0639** (0.0512) (0.0514) (0.0323) (0.0276) (0.0309) Urbanization 0.156* 0.169* − 5.06e− 05 0.0396 0.0601* (0.0909) (0.0879) (0.0402) (0.0366) (0.0350) Food exporter − 0.572 (0.765) Food exporter * Trade Food − 0.0018 (0.0108) Urbanization * Trade Food 0.0005 (0.0005) 950 L.M.Doerr, W.Maennig *p < 0.1, **p < 0.05, ***p < 0.01, Cluster-robust Std. errors are in parentheses. The dependent variable is poverty headcount ratio at $2.15 a day (2017 PPP) in %. Models 5.1–5.5 correspond to the baseline specification with different interactions Table 5 (continued) Variables Dependent: PHR $2.15 (5.1) TWFE (5.2) TWFE (5.3) TWFE (5.4) TWFE (5.5) TWFE Poverty low * Trade Food 0.0172 (0.0106) Poverty medium * Trade Food − 0.053*** (0.0059) Poverty high * Trade Food 0.0625*** (0.0065) Year FEs Yes Yes Yes Yes Yes Country FES Yes Yes Yes Yes Yes Observations 260 260 260 260 260 R20.909 0.909 0.753 0.810 0.812 951 Globalization intheFood Sector andPoverty Table 6 Poverty and food trade openness: alternative indicators *p < 0.1, **p < 0.05, ***p < 0.01, Cluster-robust Std. errors are in parentheses. Models (6.1) and (6.2) correspond to the baseline specification with the poverty headcount ratio at $3.65 a day (2017 PPP) as the dependent variable. Models (6.3) and (6.4) correspond to the baseline specification with the poverty gap at $2.15 a day (2017 PPP) as the dependent variable. Models (6.5) and (6.6) correspond to the baseline specification with the poverty gap at $3.65 a day (2017 PPP) as the dependent variable. Models (6.7) and (6.8) correspond to the baseline specification with the Multidimensional Poverty Index from the World Bank as the dependent variable Variables Dependent: PHR $3.65 Dependent: PG $2.15 Dependent: PG $3.65 Dependent: MPI (6.1) TWFE (6.2) TWFE (6.3) TWFE (6.4) TWFE (6.5) TWFE (6.6) TWFE (6.7) TWFE (6.8) TWFE Trade Food 0.008 0.020*** 0.017*** 0.025*** (0.010) (0.005) (0.006) (0.006) L. Trade Food 0.017** 0.020*** 0.019*** 0.020*** (0.008) (0.004) (0.005) (0.006) Trade other − 0.032 − 0.042 0.019** 0.012 0.007 − 0.001 0.067*** 0.071*** (0.023) (0.026) (0.009) (0.009) (0.012) (0.013) (0.013) (0.014) Initial GDPPC − 0.43*** − 0.441*** − 0.076*** − 0.072*** − 0.174*** − 0.175*** − 0.31*** − 0.30*** (0.0386) (0.0371) (0.0186) (0.0180) (0.0227) (0.0219) (0.0403) (0.0426) CPI 0.0591** 0.0597*** 0.0320*** 0.0282*** 0.0466*** 0.0445*** 0.143*** 0.129*** (0.0236) (0.0221) (0.0103) (0.00991) (0.0130) (0.0124) (0.0214) (0.0225) Unemployment 0.187 0.306** 0.108* 0.116** 0.103 0.144* 0.412*** 0.367*** (0.157) (0.143) (0.0553) (0.0522) (0.0807) (0.0736) (0.0742) (0.0740) Female workforce − 0.154** − 0.183** − 0.00868 0.00231 − 0.0439 − 0.0444 − 0.15*** − 0.16*** (0.0731) (0.0722) (0.0327) (0.0327) (0.0403) (0.0404) (0.0501) (0.0538) Urbanization 0.296** 0.254** 0.0505 0.0273 0.137** 0.106 0.543*** 0.516*** (0.123) (0.127) (0.0442) (0.0463) (0.0625) (0.0647) (0.127) (0.132) Year FEs Yes Yes Yes Yes Yes Yes Yes Yes Country FES Yes Yes Yes Yes Yes Yes Yes Yes Observations 260 254 260 254 260 254 100 100 R20.929 0.933 0.863 0.862 0.913 0.915 0.986 0.985 952 L.M.Doerr, W.Maennig Alternative Measures ofPoverty To further increase the robustness of our estimations, we follow Awaworyi Churchill and Smyth (2017) and regress a full set of alternative poverty indicators in our base Eq.(2). First, as the results in Table4 show evidence of potential delayed poverty effects of globalization, we follow Ravallion (2006) and include first lags of our trade proxy into the robustness specifications in the next steps. We first use the poverty headcount ratio at $3.65 to address people who live in less severe poverty (Columns 6.1 and 6.2 in Table6). As expected, the agricultural and food market effects estimated for our sample are less strong than those estimated for people living in less severe poverty, which is consistent with the findings of Artuc etal. (2021). However, the underlying pattern remains unchanged; we obtain evidence of a significant negative effect of globalization in the food sector on poverty, in contrast to the lack of effect of overall globalization. The poverty headcount ratio has been criticized as a measure of absolute poverty because it does not consider how far people’s incomes are below the absolute poverty line (Nikoloski 2011; Awaworyi Churchill and Smyth 2017). To address this shortcoming, the World Bank Research Group developed the poverty gap index, defined as the ratio by which the average income of the poor falls below the absolute poverty line. To determine whether the dept of poverty reacts differently to increased food market globalization, we re-estimate our baseline equation with the poverty gap at $2.15 (Columns 6.3 and 6.4) and $3.65 (Columns 6.5 and 6.6) as the dependent variable. We observe some evidence of an aggravating effect of globalization in all other sectors on the poverty gap at $2.15, which is consistent with Winters etal. (2004), who find that overall trade openness may only have an effect on the very poorest. Finally, we use the multidimensional poverty index (MPI) proposed by Alkire and Santos (2014) as a complete indicator that assesses more dimensions of poverty than just available income, such as Infrastructural constraints (e.g., access to clean water, electricity and housing), as well as education, health care, and good nutrition (Columns 6.7 and 6.8). Despite a much smaller data coverage, and similar to Awaworyi Churchill and Smyth (2017), the MPI responds strongly to all covariates, including general trade.12 Apart from that, and similar to the other poverty indicators used, the results remain stable over both specifications, especially concerning the significant effect of food trade openness. Food Exports Vs. Food Imports The poverty implications of trade liberalization may differ between import and export dynamics, particularly in agrarian sectors (Davis 2002; King 2006). We thus re-estimate Eq.(2), separating the food trade variable into imports and exports. We 12 We find somewhat stronger effects of food trade openness on the MPI than on income-based headcount ratios. As the MPI captures multidimensional deprivations beyond income—including health, education, and living standards—it is likely more sensitive to structural changes induced by food sector globalization (Alkire and Santos 2014). 953 Globalization intheFood Sector andPoverty gradually introduce fixed effects to check for multicollinearity (Table7, columns 7.1–7.3) and use once-lagged values of imports and exports to address endogeneity concerns (columns 7.4–7.6). The results suggest that increasing shares of food imports are weakly associated with poverty reduction (with significance levels ranging from 5 to 10%, depending on the fixed effects applied), whereas increasing shares of food exports are associated with more poverty (with significance levels ranging from 1 to 5%, depending on the fixed effects applied). These findings align with Ng and Aksoy (2008), who find that increasing food imports may improve trade balances and food security in lowermiddle-income countries such as those in Latin America. More recent research also supports the stabilizing role of imports in national food systems (Porkka etal. 2017; Subramaniam etal. 2024). In contrast, food exports appear to be linked to higher poverty rates, especially in Latin America, due to land concentration and the shift Table 7 Poverty and food imports and exports *p < 0.1, **p < 0.05, ***p < 0.01, Cluster-robust Std. errors are in parentheses. The dependent variable is poverty headcount ratio at $2.15 a day (2017 PPP) in %; Model (4.1) corresponds to the baseline specification in Eq.(2), with Food Trade divided into food imports and food exports estimated via OLS Variables Dependent: PHR $2.15 (7.1) (7.2) (7.3) (7.4) (7.5) (7.6) OLS FE TWFE OLS FE TWFE Food Imports − 0.0055* − 0.0090*** − 0.0046 (0.0030) (0.331) (0.0081) Food Exports 0.0077** 0.0108*** − 0.00085 (0.0031) (0.330) (0.0048) L. Food Imports − 0.0050* − 0.00810** 0.00101 (0.0030) (0.319) (0.0076) L. Food Exports 0.0075** 0.0101*** 0.0019 (0.0031) (0.319) (0.0053) Trade other − 0.0047 0.0041 0.00443 − 0.0059 0.0027 − 0.00208 (0.0071) (0.0083) (0.0161) (0.0069) (0.0080) (0.0189) Initial GDPPC − 0.151*** − 0.159*** − 0.203*** − 0.147*** − 0.155*** − 0.199*** (0.0179) (0.0189) (0.0359) (0.0178) (0.0183) (0.039) CPI 0.269*** 0.154* 0.0639 0.306*** 0.193** 0.137 (0.0794) (0.0804) (0.110) (0.0782) (0.0824) (0.107) Unemployment − 0.0301 − 0.0417 − 0.0354 − 0.0282 − 0.0384 − 0.0385 (0.0338) (0.0343) (0.0504) (0.0333) (0.0339) (0.0513) Female workforce − 0.0647 − 0.0115 0.220*** − 0.0740* − 0.0185 0.165* (0.0470) (0.0522) (0.0820) (0.0447) (0.0504) (0.0842) Year FEs No Yes Yes No Yes Yes Country FES No No Yes No No Yes Observations 260 260 260 254 254 254 R20.723 0.750 0.905 0.731 0.753 0.905 954 L.M.Doerr, W.Maennig to export-oriented, high-value crop production (Carter and Mesbah 1993; Barham etal. 1995; Carter etal. 1996; Pronti etal. 2024). However, controlling for crosscountry variation in export and import structures (columns 7.3 and 7.6) implies that the relationship is sensitive to national-level unobserved factors. Conclusion We first replicate the findings of Polloni-Silva etal. (2021) on the determinants of poverty for a dataset of 12 Latin American countries covering the years 1995 to 2020. Second, we add our (overall) globalization proxy to the model and find no significant poverty effects, in line with Ravallion (2006), Kwon and Kim (2014), Anser etal. (2020), and Omar and Inaba (2020). In contrast, when we use globalization in the food sector instead of overall globalization in the third step, we find a significant and robust positive effect on the prevalence of absolute poverty. Our findings suggest that despite its small share of overall trade, the increased integration of the Latin American food sector into international markets since the 1990s may account for up to 2% points of the population living below the absolute poverty line of $2.15 a day in our sample. Extended analysis provides evidence that our poverty effects are carried by countries with moderate absolute poverty rates between 5 and 10% and that the transition from more global integration of the food sector to increased poverty rates works through food exports. Fig. 2 Trade Food and Trade overall with Confidence intervals 955 Globalization intheFood Sector andPoverty The results are robust to potential endogeneity issues and time series characteristics and hold across the different estimators and absolute poverty measures used. Regarding other determinants of poverty prevalence, our models identify GDP per capita as the main driver of poverty prevalence, which is in line with the relevant literature. Our estimates may be viewed in addition to the findings of Porto (2008), Chikhuri (2013), and Mary (2019), who find aggravating effects of food and agricultural trade openness on variables such as unemployment, food security, and hunger prevalence. Our results also add to the literature concerning the impact of agricultural globalization at the local level, where authors find negative poverty effects for selected Asian countries (Litchfield etal. 2003; Huang etal. 2007). We further add to the literature on the Agro-Export Boom in Latin America from the early 1990s onwards (Carter and Mesbah 1993; Barham etal. 1995; Carter etal. 1996; Pronti Fig. 3 Averaged PHR $2.15 in our sample Fig. 4 Averaged per capita GDP in our sample for the years 1995–2020 956 L.M.Doerr, W.Maennig etal. 2024). Finally, our findings may add to the literature concerning the general determinants of poverty (Ravallion 2006; Nikoloski 2011; Kwon and Kim 2014; Awaworyi Churchill and Smyth 2017; Anser etal. 2020; Omar and Inaba 2020; Polloni-Silva etal. 2021). With respect to policy implications, our identification of a mechanism of the poverty-increasing effects of food sector globalization in Latin America via the export channel may lead to structural transformations calling for accompanying policies. For example, Barham etal. (1995) emphasize that land market reforms enabling smallholders to purchase or retain land can help shield them from poverty traps linked to trade liberalization, as observed in Guatemala. Bergau etal. (2022) recommend promoting farm diversification and improving market access as effective tools to enhance food security and reduce rural poverty. Similarly, Sawadogo (2024) reported that subsidy programs for agricultural mechanization can lower poverty rates, although their effectiveness depends on financing modalities. There are limitations to our analysis. First, the upscaled effects in Sect.“Empirical Strategy and Results” are specific to a particular time period and region. Additionally, the size of these effects may suggest that food trade globalization is a primary driver of poverty in our sample. To contextualize our results, we scale up the estimated income effect similarly to that of the food trade and find that the poverty-reducing impact of income is three times stronger than that of food trade globalization, highlighting the critical importance of broad-based income growth, as supported by previous studies (Ravallion and Datt 2002; Dollar and Kraay 2002, 2004). Second, our analysis may be subject to data limitations. For example, measurement errors or biases in harmonizing PPP measures may complicate cross-country and temporal comparability. Moreover, the reliance on country-level aggregates obscures distributional effects of farm size, commodity, or labor status. Microlevel data may strengthen tests of mechanisms such as land concentration channels; however, systematic, comparable data on farm structures remain scarce in Latin America (Lowder etal. 2021). Fig. 5 Averaged food trade in selected countries, 1995–2020 957 Globalization intheFood Sector andPoverty Fig. 6 Composition of Trade, Trade Food and Trade other 964 L.M.Doerr, W.Maennig Vermeulen, S., and L. Cotula. 2010. Making the most of agricultural investment: A survey of business models that provide opportunities for smallholders. FAO/IFAD/IFAD/SDC. IIED press. London/ Rome/Bern Weisskoff, R. 1992. The Paraguayan agro-export model of development. World Development 20 (10): 1531–1540. Winters, L. A., and A. Martuscelli. 2014. Trade liberalization and poverty: What have we learned in a decade? Annual Review of Resource Economics 6 (1): 493–512. Winters, L. A., N. McCulloch, and A. McKay. 2004. Trade liberalization and poverty: The evidence so far. Journal of Economic Literature 42 (1): 72–115. World Bank. 2005. World development indicators. The World Bank. Available under:http:// docum ents. world bank. org/ curat ed/ en/ 94795 14681 40975 423 September 2025 World Bank. 2022. Fact Sheet: An Adjustment to Global Poverty Lines. The World Bank. Available under: https:// www. world bank. org/ en/ news/ facts heet/ 2022/ 05/ 02/ factsheetanadjus tmenttoglobalpover tylines, September 2025 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.