Immigrants’ consumption choices: evidence from Korea’s food service industry
Introduction
Human migration has been a persistent phenomenon in nearly every society worldwide (IOM 2022). As migration increases, understanding the behaviors and impacts of immigrants has become an important area of study. Among these behaviors, food consumption stands out as both a fundamental human activity and a vital expression of cultural heritage (Bach-Faig et al. 2011). This paper examines the impacts of immigrants on the food service industry in South Korea, focusing on their restaurant preferences and the broader implications for the sector. South Korea presents an intriguing case for study. Unlike many traditional immigrant-receiving nations, South Korea has only recently experienced significant immigration, with the number of immigrants nearly doubling between 2009 and 2019 due to labor shortages. This trend, briefly disrupted by the COVID-19 pandemic, underscores the nation’s shifting demographic profile. Historically perceived as a monoethnic society, the growing immigrant population (see Fig. 1) has the potential to reshape both cultural and economic sectors, particularly by influencing the demand for diverse dining options, including ethnic cuisines. Notably, the visible increase in the number of restaurants—especially foreign-food establishments—suggests an evolving food service landscape, which may be partly driven by these demographic changes.

Darker shades indicate higher immigrant density. Left panel (A) shows 2010 data; right panel (B) shows increased density in 2019, reflecting immigrant influx.
The novelty of this study lies in three key aspects.
First, we use comprehensive data. We exploit a district-level dataset covering all 229 districts in South Korea from 2010 to 2019, which includes over 6.5 million restaurant observations. This study provides unprecedented detail and scale in examining immigrant influences on the food service industry. Second, we integrate an analytical framework that combines two commonly distinct questions: whether immigrants are positively associated with restaurant numbers and what their food consumption patterns are. Third, we employ a robust methodological approach. We utilize shift-share designs to address endogeneity concerns. While this methodology is widely used in migration studies, this study is the first to apply it in the context of the food service industry.
Our findings challenge conventional assumptions. Results from the two-stage least squares (2SLS) analysis indicate that immigrants do not have a statistically significant positive impact on the total number of restaurants, including both Korean-food and foreign-food categories. However, when examined at a more granular level, immigrants have a statistically significant positive impact on the number of restaurants offering lower customer unit prices and faster service times, such as cafeterias, confectioners’ shops, and noodle houses. In contrast, the lack of a positive association with foreign-food restaurants may be attributed to their relatively higher cost, making them less accessible to immigrant populations. These findings provide actionable insights for policymakers and urban planners. To promote inclusive urban development, strategies should focus on infrastructure that supports affordable, time-efficient dining options in areas with high immigrant populations. Such measures would not only address the immediate needs of immigrant communities but also create economic opportunities within the food service sector.
Literature review
The literature is reviewed from two distinct perspectives. The first examines how immigrants are associated with or influence the food service industry. The second focuses on the factors shaping immigrants’ food consumption choices and behaviors. Studies in the first domain often employ case studies, interviews, and macro-level data analyses to understand the broader impact of immigration on the food service sector. In contrast, research in the second domain primarily relies on surveys and micro-level data to explore individual and household consumption patterns among immigrant populations.
The relationship between immigrants and the food service industry
The role of immigrants in a country’s industries is often discussed in terms of their contributions as employees. However, in the context of the food service industry, immigrants are examined through additional lenses: as consumers and entrepreneurs. Their employment role is relatively straightforward, with firms often responding to immigration by increasing the number of establishments, particularly in low-skill-intensive sectors (Olney 2013). As entrepreneurs, immigrants frequently excel in operating ethnic restaurants, leveraging their cultural knowledge and unique networks. Globalization reshapes resource structures and market opportunities, creating new avenues for immigrant entrepreneurship (Gurău et al. 2020; Nazareno et al. 2019). As consumers, immigrants contribute to cultural diversity within societies, especially through their demand for culturally specific foods, which drives the growth of diverse culinary offerings.
Regardless of their role—whether as employees, entrepreneurs, or consumers—immigrants are widely recognized for their positive influence on the restaurant industry, particularly in increasing the number of establishments and enhancing the ethnic diversity of food options. For example, Nash (2009) analyzes ethnic restaurants in Montreal using Yellow Pages listings and census data from 1951 to 2001, concluding that immigration, rather than shifting food trends, best explained the growth in the number and variety of ethnic restaurants. Similarly, Mazzolari and Neumark (2012) find that immigrant inflows boost employment in California’s retail and food service sectors and increase ethnic diversity in restaurants, driven by immigrants’ comparative advantage in producing ethnic goods and their consumption demands.
Massidda et al. (2017) explore the impact of immigration on Italy’s tourism sector, including hotels and restaurants, and find a positive relationship between immigration and the number of tourism establishments and their employees. Kim (2023) notes a significant increase in Korean restaurants in Frankfurt since the mid-2010s, emphasizing that small businesses, especially restaurants, become a primary source of income and a means of societal integration for Korean immigrants. Similarly, Liang (2023) observes that immigration’s influence on the restaurant industry extends beyond major urban centers, with Chinese restaurants increasingly appearing in emerging immigrant destinations across the United States.
Immigrants’ food consumption behavior
Research on immigrants’ food consumption behaviors generally focuses on two primary objectives: addressing health concerns and examining the acculturation process through food choices. Daily food consumption is influenced by a variety of factors, including cultural background, religious practices, taste preferences, product availability, and socio-economic conditions (Ergin and Kaufman-Scarborough 2010; Popovic-Lipovac and Strasser 2015).
Koçtürk (2004) finds that the high cost of healthy foods often leads immigrants to choose cheaper, less nutritious options, such as snacks high in fat and sugar or sweetened beverages. Similarly, Edmonds (2005) highlights convenience and affordability as key factors driving immigrants’ reliance on fast food and pre-packaged meals. Children’s preferences also play a significant role, as immigrant mothers frequently prioritize their children’s food choices over their own (Satia-Abouta 2003).
Barriers such as uncertainty about unfamiliar foods or preparation methods—often exacerbated by language difficulties—further shape immigrants’ food consumption patterns (Bayanzadeh 2008). Additionally, demanding work schedules, stress, loneliness, exclusion, boredom, and unemployment contribute to a reliance on convenient but unhealthy food choices (Mellin-Olsen and Wandel 2005; Bayanzadeh 2008).
Ethnic grocery stores and restaurants serve as important cultural bridges, helping immigrants maintain connections to their traditional cuisines while adapting to the food culture of the host country. For instance, Mellin-Olsen and Wandel (2005) document how Pakistani immigrant women in Norway often face difficulties accessing traditional foods and ingredients, such as specific vegetables or spices. Njomo (2013) emphasizes the pivotal role of ethnic grocery stores and restaurants in shaping food choices by offering culturally specific options.
Immigrants’ dining-out behaviors are influenced by factors including the length of stay in the host country, age, visa status, and the presence of children in the household (Rajagopal et al. 2009). Acculturation plays a central role in this process, with immigrants becoming more open to exploring new foods during the early stages of their integration (Choe et al. 1994). However, time constraints and acculturation pressures often result in significant changes in dietary habits (Badanta et al. 2021).
Proposing the hypotheses
While existing literature has provided valuable insights into the relationship between immigration and the food service industry, several notable limitations existed.
Firstly, within the two analytical frameworks—macro and microdata perspectives—the findings often appear contradictory. Macro-level analyses and researchers’ general observations suggest that the number of restaurants increases as the immigrant population grows. However, micro-level analyses reveal that immigrants have distinct preferences and food consumption patterns. This suggests that immigrants may not uniformly benefit the entire food service industry but rather have a more concentrated positive impact on specific restaurant categories.
Secondly, most studies heavily rely on survey data and make limited use of advanced econometric models. Previous papers do not employ empirical strategies to mitigate potential endogeneity bias in the estimates of the effect of immigration on the food service industry.
Thirdly, existing research has yet to address two key questions within a unified analytical framework: How does immigration affect the food service industry? And what are immigrants’ specific consumption choices regarding different restaurant types? This study aims to bridge these gaps by integrating macro and micro analytical frameworks to address these questions simultaneously. To achieve this, we propose the following hypotheses:
H1: Immigrants may not exhibit a clear positive correlation with the overall number of restaurants. In contrast, native Koreans are likely to drive an increase in the total number of restaurants.
H2: Immigrants are positively associated with the number of foreign restaurants, as they are likely to have significant demand for cuisines familiar to their home countries.
H3: Immigrants are likely to contribute to an increase in the number of low-cost establishments, reflecting preferences influenced by affordability and convenience.
Model setting, variable selection, and data
Model setting and variable selection
Based on the data and our hypotheses, we adopted a reduced two-way linear fixed-effects regression model, as represented in Eq. (1).
where ({R}_{{it}}) represents the number of restaurants for district i at time t, and ({I}_{{it}}) represents the number of immigrants, which is the key variable of interest in this study. ({delta }_{i}) represents the district-specific fixed effect, capturing unobservable characteristics unique to each district in South Korea that remain constant over time. ({gamma }_{t}) denotes the time-specific fixed effect, reflecting unobservable characteristics specific to each time period but consistent across districts. ({varepsilon }_{{it}}) represents the error term for district i at time t. The parameter of interest is (beta), which represents the effect of immigration on the number of restaurants.
({X}_{{it}}) represents a vector of control variables, which are divided into two subgroups. The first subgroup includes primary control variables—key factors that directly influence the number of restaurants. This subgroup consists of the log of GRDP (LogGRDP) and the log of Korean (LogKorean):
-
(i)
LogGRDP refers to the logarithm of the gross regional domestic product. GDP had been identified as a critical factor influencing food service industries (Xia et al. 2006). In this study, GDP was adjusted using South Korea’s GDP deflator to standardize values to the base year 2010.
-
(ii)
LogKorean represents the logarithm of the total South Korean population, another key determinant of the number of restaurants (Mazzolari and Neumark 2009).
The second subgroup includes variables used for robustness checks. These variables account for factors that may influence the decision to open new establishments or reflect demographic shifts affecting dining-out behaviors, thereby impacting the number of restaurants at the district level. The additional controls are as follows:
-
(i)
Employment and labor force participation rates among native adults, specifically individuals aged 20 to 59. Thiede et al. (2016) observe that the population’s age structure affects the availability of service-providing establishments, including food services.
-
(ii)
Educational attainment, measured as the percentage of individuals with a college degree or higher. Some studies (e.g., Banerjee and Zhao 2022; Lund et al. 2017) find that higher education levels are associated with more frequent dining out, while others (e.g., Kim and Chun 2005) suggest a negative correlation.
-
(iii)
Gender distribution, specifically the proportion of South Korean females. This variable is included because home cooking is still largely considered a gendered social norm (Etilé and Plessz 2018).
To eliminate unobserved district fixed effects, we transform Eq. (1) by applying first-differencing, resulting in the following formulation:
However, two key concerns remain in estimating Eq. (2). First, immigrants may choose their residential areas based not only on labor demand or income levels, but also on time-varying unobserved district characteristics, such as housing development (Lee 2020) or access to amenities (Sharpe 2019). These factors may attract significant immigrant inflows, which could also influence the establishment of additional restaurants. Second, reverse causality poses a potential issue. The size of the local immigrant population strongly affects residential decisions (Gould 1994). For example, a high concentration of foreign-food restaurants could indicate a substantial local immigrant community, which, in turn, might attract even more immigrants to the area.
Instrumental variable construction
To avoid endogeneity concerns, this study employs the Bartik shift-share instrumental variable (IV) approach combined with a two-stage least squares (2SLS) estimation strategy. The shift-share design is recognized for its effectiveness in addressing endogeneity concerns and has been widely used in migration and trade research (Bartik 1991; Autor et al. 2013; Kovak 2013; Jaeger et al. 2018; Adão et al. 2019; Borusyak et al. 2022; Carlino and Drautzburg 2020; Goldsmith-Pinkham et al. 2020; Yu et al. 2022). The shift-share IV is designed as follows:
where ({Z}_{{ik}{t}^{0}}) represents the “initial” share of immigrants from source country k in district i, and ({t}^{0}) = 2010. immigrant sources are classified into 17 distinct groups (China, Vietnam, Philippines, Indonesia, Uzbekistan, Cambodia, Thailand, U.S., Japan, Sri Lanka, Nepal, Mongolia, Bangladesh, Pakistan, India, Canada, and Others) to address the issue of zero values at the district level. Additionally, ({m}_{{kt}}) denotes the year-to-year change in the number of immigrants from country k into South Korea as a whole at time t. Consequently, ({I}_{{it}}) represents a weighted average of the national inflow rates into destination i from each source country, with weights determined by the initial distribution of immigrants.
After constructing our instrumental variable, we apply the 2SLS estimator. Specifically, in the first stage, we regress the endogenous variable, ({{immigrants}}_{{it}}), on the Bartik shift-share IV and the remaining control variables. In the second stage, we regress ({R}_{{it}}) on the predicted values from the first-stage regression, denoted as (widehat{{{immigrants}}_{{it}}}). During the 2SLS regression, we select fixed effects over random effects based on the results of a Hausman test, which yields a p-value of less than 0.000.
Data
The dataset used in our analysis covers the years from 2010 to 2019. To account for changes in administrative units, such as name updates and mergers, all variables are standardized to align with the most recent administrative designations. As a result, the dataset includes 229 districts across 17 provincial administrations in South Korea.
The data source for restaurant information is the National Business Survey, accessed via the Microdata Integrated Service (MDIS). This comprehensive dataset contains 4.7 million observations for restaurants and 1.8 million for drinking and non-alcoholic beverage establishments across South Korea during the 2010–2019 period. All restaurant classifications are standardized according to the 9th version of the Korea Standard Industry Code (KSIC), resulting in 16 distinct restaurant categories (see Fig. 2). The data are then aggregated by district and restaurant type. With ten years of data across 229 districts, the final sample consists of 2290 observations per variable (see Table 1).

The original dataset includes 16 five-digit-coded restaurant categories, as data for Mobile Food Services were unavailable.
Results
First-stage results
Table 2 presents the results of the first-stage regression, which involves the stepwise introduction of control variables. Overall, the shift-share IV demonstrates robust predictive power regarding the actual immigrant influx in the region. Across columns (1)–(4), regardless of the addition of extra control variables, the inclusion of an additional IV consistently results in a significant increase of 0.65 units in the observed actual immigrant influx (p < 0.01). The rejection of the underidentification test indicates that the model is properly identified (Ho = the instrument lacks explanatory power to predict the endogenous variables in the model for parameter identification). Both the Cragg–Donald Wald F statistic and the Kleibergen–Paap Wald rk F statistic confirm that the weak instrument problem is absent, supporting the effectiveness of the shift-share design in producing unbiased estimates.
Second-stage results
Table 3 presents the findings from the second stage of the 2SLS analysis with primary controls and additional controls (i.e., all controls), respectively. The results indicate that immigrants do not have a significant impact on the number of restaurants in either scenario (p > 0.1). This lack of statistical power is observed not only for the aggregate number of restaurants but also for both foreign-food and South Korean-food establishments. However, the findings reveal a positive correlation between the number of immigrants and the number of drinking and non-alcoholic beverage establishments (p < 0.01). Specifically, an increase of approximately 100 immigrants is associated with a rise of 2–3 such establishments.Footnote 1
On the contrary, the South Korean population exhibits statistically significant impacts on the food service industry. Although it is not the primary variable of interest in this study, it does exhibit statistically significant impacts on the food service industry. As shown in Table 3, native Koreans have a significantly positive impact on restaurant establishments, drinking places, and non-alcoholic beverage establishments, thereby supporting our initial hypothesis.
Furthermore, Figs. 3 and 4 illustrate the impact of immigrants on 16 different types of restaurants and drinking establishments across South Korea. Figure 3 presents the 2SLS results using primary control variables, while Fig. 4 includes results with all control variables. In both figures, immigrants have a positive effect on only three restaurant categories–cafeterias, confectioners’ shops, and noodle houses–all of which are statistically significant (p < 0.05). Notably, these categories also feature the lowest average customer unit prices, which will be further discussed in subsequent sections. Conversely, a negative association is observed between immigrants and chicken shops, although the p-value exceeds 0.05. In contrast, as shown in Appendix 2, South Koreans have a significant positive impact on most restaurant categories in the 2SLS results, regardless of whether primary control variables or additional control variables are included.

The x-axis represents the measures of the coefficients, while the values in the figure correspond to the p-values ***p < 0.01, **p < 0.05, *p < 0.1.

The x-axis represents the measures of the coefficients, while the values in the figure correspond to the p-values. ***p < 0.01, **p < 0.05, *p < 0.1. The red and blue text in the figure denotes the average customer unit price of categories in 2019.
Discussion
Our first hypothesis suggests that immigrants may not exhibit a clear positive correlation with the overall number of restaurants, whereas native Koreans are more likely to drive an increase in the total number of establishments. The 2SLS estimation results support this hypothesis, although they differ from findings in previous literature. This discrepancy can be explained by the different roles immigrants play. Earlier studies often focus on immigrants as employees in the food service industry, without considering visa restrictions. In South Korea, the E-9 visa for non-professional employment—the most common among over 30 visa types and held by approximately 20% of immigrants in 2020—restricts employment in the food service sector until 2023. As a result, the observed impact of immigrants in this study is likely due to their role as consumers rather than workers in the food service industry. Another contributing factor may be immigrants’ tendency to cook at home. According to the 2019 Survey of Immigrant Status and Employment, which sampled 18,243 individuals, 81% of immigrants live in independent housing. This living arrangement likely facilitates home cooking, which could reduce the demand for dining out.
Our second hypothesis posits that immigrants are positively associated with the number of foreign-food restaurants, driven by their potential demand for familiar cuisines from their home countries. However, the results reveal that immigrants do not have a statistically significant impact on the number of foreign-food establishments. We cannot conclusively determine whether this is due to these restaurants failing to meet immigrants’ preferences. While literature suggests that the accessibility of ethnic foods or ingredients may influence immigrants’ food consumption choices, this remains unproven in our context. Notably, we observe that foreign-food restaurants tend to fall within the most expensive restaurant categories (see Table 4), which may partially explain the lack of statistical power.
Our third hypothesis posits that immigrants are likely to increase the number of low-priced restaurants. This aligns with findings from previous studies conducted within microdata analytical frameworks. When controlling for all other variables, a significant positive correlation is observed between South Korea’s population and the number of restaurants, with significance at the 5% level across 14 out of 16 categories. However, immigrants do not have a statistically significant impact on most restaurant types, with notable exceptions for cafeterias, confectioners’ shops, and noodle houses. These categories are characterized by relatively lower per-capita consumption expenses compared to other restaurant types in South Korea (see Table 4). Data from the Survey on the Management of the Food Service Industry (2015–2019) support this observation, indicating the customer unit price for noodle houses is 5230 KRW; for cafeterias, 5439 KRW; and for confectioners’ shops, 5854 KRW—the lowest unit prices among all restaurant categories. These findings are robust, given that over 95% of long-term immigrants in South Korea originate from developing nations where per-capita GDP is substantially lower than that of South Korea. This economic disparity highlights differences in consumption power between immigrants and the native population. Immigrants often prioritize saving and preserving financial resources (Amuedo-Dorantes et al. 2005), which influences their preference for lower-priced food options (Glytsos 1997; Ahlburg and Brown 1998).
Moreover, we find that time constraints emerge as a critical factor in immigrants’ restaurant preferences. As shown in Table 4, noodle houses, cafeterias, and confectioners’ shops are known for their quick meal preparation times. Cafeterias and confectioners’ shops, in particular, primarily offer quickly prepared or ready-made meals, facilitating faster purchases and takeaways. Consequently, these establishments experience higher daily customer turnover compared to other restaurant categories. Visa data support this trend. For instance, the 2019 Survey on Immigrant Status and Employment by the Korean Statistical Office shows that among E-9 visa holders (non-professional employment), 44.6% live in factory dormitories, and 34.8% of D-2 visa holders (international students) reside in school dormitories. These living arrangements likely contribute to the popularity of cafeterias, which are strategically located near work or study areas to offer convenience. Previous research also underscores time as a key factor influencing food choices for immigrants (Renne 2007; Garnweidner et al. 2012).
Conclusion and implication
This study provides a macro-level analysis of the causal impact of immigration on the number of restaurants in South Korea. Using 10 years of panel data and employing shift-share design and 2SLS estimation methods, the findings highlight that immigrants in South Korea prioritize affordability and time efficiency when choosing restaurants.
Given South Korea’s low birth rate and aging population, the country expects to see a growing influx of immigrants in the coming years. This demographic shift presents opportunities for policymakers and urban planners to foster inclusive urban development. By ensuring that areas with significant immigrant populations are equipped with adequate infrastructure to support low-cost and time-saving establishments, they could meet the demand for affordable meal options while simultaneously creating economic opportunities in the food service sector.
In addition, policymakers and business owners could adopt strategies to increase immigrants’ participation in higher-income restaurant sectors. For instance, implementing diverse pricing models that attract both immigrant and local customers could broaden the customer base while ensuring balanced financial returns. Addressing employment barriers—such as restrictive visa conditions—and providing targeted skill-development programs tailored to the food service industry would further empower immigrants, enabling them to actively contribute to and thrive within this sector.
Responses