Effects on Territorial Dynamics and Labor Commuting During the Pre- and Post-Pandemic Period in Chile

Abstract

Labor commuting plays a crucial role in the economic and productive development of territories. Additionally, external shocks, such as the COVID-19 health crisis, may induce shifts in regional labor dynamics. This study analyzes changes in the relationship between labor commuting and productive sectors in Chile during the years before and after the COVID-19 crisis, delineating trends by geographic zones. The research is based on microdata from the National Employment and Supplementary Income Survey spanning 2018 to 2022. The main findings indicate that the decline in commuting recorded at the national level in 2020 exhibited distinct patterns across geographic regions and economic sectors. Notably, employment in the mining sector is associated with a higher probability of commuting compared to other sectors; however, this characteristic diminished during the crisis in most of the analyzed zones. Conversely, the traditionally negative correlation between commuting and employment in the agriculture sector weakened in 2020 across all zones, except for Patagonia (the southernmost region of the country). These results demonstrate that external shocks alter labor commuting dynamics across territories, emphasizing the need for public policies that anticipate diverse mobility trends.

1. Introduction

The health crisis caused by COVID-19 had unprecedented economic consequences. The economic shock had a significant impact on employment, particularly work-related commuting, which is common in modern societies. Indeed, many people travel to areas other than their place of residence for work.

The decision to work in a different territory from one’s place of residence is a relatively recent phenomenon in Chile. This process, known as labor commuting, has rapidly replaced traditional migration [1] and is considered a modern trend driven by technological advancements, economic diversification, and increased access to transportation.

One of the key benefits of labor commuting is its ability to expand the labor market for both job seekers and employers. However, during the crisis caused by the COVID-19 virus outbreak in 2020, significant changes in commuting patterns emerged. Initially, these changes were driven by mobility restrictions, but the impact on commuting persisted beyond the immediate crisis, intertwining with other factors that affect various aspects of the labor market. This study aims to address the following questions: What was the impact of the health crisis on labor commuting in different regions of Chile? How persistent was this impact after the pandemic was considered to have ended? How did the economic sector in which a person is employed influence their likelihood of commuting? Given the spatial heterogeneity present in Chile’s productive structure, diverse effects across territories are expected.

The primary objective of this study is to analyze the relationship between the productive sector in which an individual is employed, and their decision to work in a municipality different from their place of residence (commuting), as well as how this relationship changed before and after the COVID-19 crisis across the different regions of Chile. In this study, an individual is considered a commuter if they engage in their primary work activity in a municipality other than where they reside.

The specific objectives that will enable the achievement of the study’s overall purpose, presented in sequential order, are as follows:

• (SO1) Analyze the probability of commuting in Chile between the years 2018 and 2022.
• (SO2) Compare commuting patterns and their relationship with socio-productive characteristics in different regions of Chile.
• (SO3) Examine the impact of the health crisis on commuting decisions across various regions in Chile.

In the following sections, the study presents a literature review and contextual analysis which serve as the foundation for the proposed methodology which aims to address the research objectives and questions. The paper then presents the results, followed by the main conclusions.

1.1. Theoretical Framework

1.1.1. Context

The COVID-19 health crisis was officially declared in Chile in March 2020, and one of the primary measures to address it involved restricting and regulating human mobility. This, in turn, had a significant impact on employment and production, particularly affecting the employment rate of private-sector wage earners, self-employed workers (who experienced a decline), and those in the informal sector. The crisis had a particularly severe effect on small businesses and enterprises linked to the service sector [2]. Furthermore, the global repercussions of the crisis, combined with international armed conflicts, led to severe supply constraints for tradable goods. This constituted an unprecedented situation in Chile.

During the second quarter of 2020, economic activity contracted by 14.1% compared to the same period of the previous year. Aside from the mining sector, financial services and public administration, all economic sectors were negatively impacted. The most significant contractions occurred in transportation (34.4% decline), personal services (27.9% decline), and commerce (20.6% decline). This was followed by the manufacturing industry, which experienced a 10.6% reduction, business services with a 9.5% decrease, and the agriculture sector, which contracted by 6.3%. Among the affected activities, the least impacted were electricity, gas, water, and waste management (EGA), which declined by 3.1%, and communications and information services, which saw a reduction of 4.1% [3].

In this context, to mitigate the economic hardships faced by individuals, significant resources were allocated in the form of monetary subsidies and withdrawals from pension savings. These measures led to increased demand, creating inflationary pressures and consequently driving up interest rates. Regarding the labor market, there was a notable decline in its dynamism, reflected in reduced labor force participation and employment rates, which fell to 51.8% and 45%, respectively, during the May–July 2020 quarter. Additionally, unemployment surged sharply, reaching 13.1% during the same period [4].

1.1.2. Geographical Zones

For the purposes of this study, Chile was divided into six geographical zones. As illustrated in Figure 1, from north to south, these zones are Great North, Small North, Center, Metropolitan, South, and Patagonia [5]. Each of these zones exhibit certain similarities in terms of their productive structures.

The geographical zones are characterized as follows:

(a).

Great North, comprising the regions of Arica, Tarapacá, and Antofagasta, is the area of Chile where mining contributes the most to the regional gross domestic product (GDP), accounting for an average share of 49.1% of the regional GDP between 2018 and 2022.

(b).

Small North, which includes the regions of Atacama and Coquimbo, although mining is also its main productive activity, generated on average 24.4% of the area’s GDP between 2018 and 2022. Construction and personal services are tied as the second most economically important sectors in the region, with 11.6% of regional GDP each [3].

(c).

Center, in this classification, consists of the regions of Valparaíso and O’Higgins. Unlike the classification by Mancilla et al. [5], the Metropolitan zone is studied separately due to its distinct intercommunal commuting patterns, likely resulting from the region’s high connectivity, which leads to higher commuting rates (Figure 1). In the Center zone, the sectors contributing the most to the regional GDP are financial and business services, housing and real estate services, and personal services, with average shares of 11.0%, 10.3%, and 13.8%, respectively.

(d).

Metropolitan, where financial and business services account for 23.4% of the regional GDP, doubled its contribution in the Center zone, making it the region with the most significant economic activity. The construction sector ranks second, contributing an average of 4.4% over the study period.

(e).

South comprises the regions of Maule, Biobío, Ñuble, and Araucanía, where the GDP is primarily driven by manufacturing industries and personal services, which contribute 16.3% and 17.2%, respectively, to the combined regional GDP. The third most significant activities are financial and business services and housing and real estate services, each accounting for 9.6% of the regional GDP.

(f).

Patagonia includes the regions of Los Ríos, Los Lagos, Aysén, and Magallanes, where the GDP is characterized by manufacturing industries and personal services, which contribute 17.2% and 15.4%, respectively, to the regional GDP. Commerce constitutes the third most significant sector, with an average contribution of 8.7% to the regional GDP.

Overall, despite regional differences in economic specialization, mining remains the most important economic activity at the national level, contributing 26.2% to Chile’s GDP between 2018 and 2022, followed by manufacturing industries (20.2%) and personal services (13.5%) [3].

Regarding labor commuting, as illustrated in Figure 2, the percentage of spatial mobility decreased in all regions of Chile in 2020 compared to 2019, with the most significant decline occurring in the central regions of the country. Additionally, by 2022, three regions (Metropolitan, Center, and Patagonia) continued to exhibit commuting rates lower than those recorded in the pre-pandemic period (2018 or 2019).

1.1.3. Labor Commuting and Associated Variables

Labor commuting has been driven by technological advancements. The diversification of transportation methods and significant improvements in access routes have reduced travel times and costs for individuals, and consequently increased the accessibility of various geographic areas. As a result, among other factors, the Chilean population now has greater spatial mobility, allowing more people to consider employment opportunities in distant locations without changing their place of residence. Additionally, it has enabled companies to access a broader geographic labor pool when seeking workers, with positive local effects, particularly for more remote areas [6]. In summary, the increased mobility of factors can be classified as a positive effect of commuting.

However, it should be noted that commuting impacts various aspects of territorial economies. In this regard, some authors highlight negative economic effects, particularly for the locality where the commuter works—as observed by Aroca and Atienza [7] in the case of the Antofagasta region in Chile. Furthermore, it has been observed that the elimination of spatial friction can also affect the distribution of labor among companies [8].

Chile exhibits high spatial heterogeneity and a distinct geographic shape (Figure 1), which explains particular and diverse commuting behaviors. It has been observed that people residing in northern Chile experience higher levels of commuting, demanding higher wages to compensate for extended travel times [9,10]. In contrast, in southern Chile, particularly in Patagonia, people commute less, with the southernmost areas requiring a higher wage premium for commuting compared to the northern parts of the same region [11].

This study explores differences in commuting patterns among individuals residing in different regions of Chile, with a particular focus on understanding variations based on the productive characteristics of each geographical area, as well as distinguishing the changes observed during and after the 2020 crisis.

The literature suggests that an external shock can have a significant impact on the economy, although these effects tend to be heterogeneous across sectors and regions. Therefore, addressing vulnerabilities to such shocks is a key challenge. For example, disruptions in foreign trade can affect employment in sectors such as mining and, through economic linkages, also affect trade and hospitality, raising the challenge of promoting productive diversification [12].

Another response is labor mobility to other sectors of the economy. Evidence shows that labor flows between manufacturing, general business services, and knowledge-intensive business services are bidirectional and more frequent in metropolitan areas [13]. Consequently, in the face of a shock like the one in Chile in 2020, one possible response is reallocating employment to sectors that do not require territorial displacement.

The creative industry, including arts, entertainment, and recreation, was severely affected by the economic crisis during COVID-19. However, other sectors, such as human health and social work, experienced growth [14].

Nevertheless, the distribution of effects across subsectors was heterogeneous, depending on the specific characteristics of the activity, the prevailing constraints, and the ability to adapt to information technologies. Those who were able to adapt experienced increased demand [15], giving rise to new forms of work such as teleworking and digitalization.

In this regard, it has been noted that the tendency to commute underwent significant changes due to the crisis, with a reduction in the number of people commuting over long distances. For instance, Castillo and Méndez [16] found that in the Araucanía region, commuting decreased, with mobility becoming concentrated in adjacent areas. However, Kim and Kwan [17], who studied the impact of COVID-19 on mobility during the first seven months of the crisis, concluded that the impact was limited to the first two months.

In Chile, long-distance labor commuting is mainly linked to mining and construction activities [9,10,18]. In contrast, in the agriculture sector, commuting is less frequent and is more commonly associated with temporary work [5].

On the other hand, commuting can also be viewed as a mechanism that allows individuals to take home higher wages compared to what they could earn in their place of residence. In fact, it has been shown to reduce wage disparities between urban and rural areas [19]. Similarly, commuting is positively associated with education, as highly skilled workers tend to commute greater distances, in contrast to those with lower qualifications. In this regard, Cui et al. [20] confirm that low-income individuals travel shorter distances, explained by the fact that lower wages are associated with lower human capital, as well as by the costs involved in commuting [21].

The link between precarious work and geographic mobility has been examined, with studies confirming that low-income individuals must absorb a larger proportion of the costs of commuting to and from work [22]. This suggests that more contiguous territories, with dynamic productive activities, are more likely to encourage people to commute, leading to more complex labor markets, and could be the case for the situation in the Metropolitan zone in Chile.

Therefore, not all segments of the population have the same opportunities to commute. For example, in general, women exhibit less spatial mobility, covering shorter distances and spending less time traveling between their place of residence and work. Halimah and Chotib [23] introduced the concept of spatial entrapment, meaning that the mobility gap favoring men could negatively impact gender wage disparities, further deepening their disparity. Similarly, it has also been observed that wage differences between men and women influence differences in commuting patterns [24].

In Chile, research has shown that agricultural workers are more likely to commute if they have a higher level of education; however, significant differences are observed between different regions of Chile. The positive effect of having tertiary education on commuting is greater in the Central zone, while in the South zone, mobility is primarily associated with full-time jobs, and in the Great North, it is linked to formal sector jobs and younger individuals [5]. However, in Chilean Patagonia, the positive effect of age on commuting occurs primarily among individuals aged 45 and older [11].

Consequently, these differences in the relationship between commuting and various socio-economic strata of the population and geographic zones lead us to expect different responses regarding the spatial labor mobility experienced following the crisis caused by COVID-19.

In summary, the evidence suggests that the factors associated with commuting differ between productive sectors. Given the heterogeneity of Chile’s productive structure, this may significantly impact the territory. These characteristics allow us to identify the variables that will be included in the estimates.

In addition, an economic shock, in this case caused by COVID-2019, may trigger other phenomena. In particular, there have been reports of labor shifts between sectors of the economy, reinforced by the emergence of new practices that affect employment, such as teleworking and technological development. These contributions will be the starting point for possible explanations of the main findings.

2. Materials and Methods

For the analysis, logistic models with a binary dependent variable were used to estimate the probability of commuting. Subsequently, the propensity score matching (PSM) method was applied to calculate the impact of working in a specific economic sector on the probability of commuting in each of the years. The data were obtained from the National Employment Survey (ENE) and the Supplementary Income Surveys (ESI), collected during the October–November–December (OND) quarter from 2018 to 2022. The database is only statistically significant at the regional level. However, considering only this quarter, certain limitations must be acknowledged, such as not capturing seasonal variations in employment. We do not know if the shock was greater in other quarters of 2020. Nevertheless, this criterion is adopted to include income information since the ESI is conducted only during these months.

2.1. Logistic Models

The probability that an individual commutes to work is calculated as

$$\mathrm{P}\mathrm{r}\mathrm{o}\mathrm{b}\left({\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{m}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g}}_{\mathrm{i}}=1\right)={\displaystyle \frac{{\mathrm{e}}^{{\mathsf{\beta }}^{\prime }{\mathrm{Z}}_{\mathrm{i}}}}{1+{\mathrm{e}}^{{\mathsf{\beta }}^{\prime }{\mathrm{Z}}_{\mathrm{i}}}}}$$

(1)

where commuting is the dependent variable that takes a value of 1 if the individual works in a different city from where they reside, and 0 otherwise. ${\mathrm{Z}}_{\mathrm{i}}$ is a matrix of variables that explain the probability of commuting, which includes sociodemographic characteristics, human capital endowment, and the economic sector in which individual i is employed. The model also incorporates fixed effects for the year in which the data are available. The probability of commuting for individuals who work in the service sector in 2018 is characterized by being women, not heads of household, having an educational level equivalent to primary education, not working in the formal sector, having part-time work schedules, and having income and age corresponding to the average value.

Equations were applied for the whole country, allowing for an analysis of the characteristics associated with an individual’s decision to work in a city different from their place of residence at the national level (SO1). Subsequently, the model was applied to each identified region to compare territorial behaviors regarding commuting (SO2).

With the logit model, 120 sector–time interactions were evaluated in this research, but only 8 were statistically significant (p-value < 0.05). This result led us to include the following method, propensity score matching, as a more appropriate method.

2.2. Impact Evaluation

To measure the effect of economic sectors on commuting before, during, and after the health crisis caused by the 2020 pandemic (SO3), the PSM method is used. This technique allows for the estimation of causal effects through a treated group and a control group, thus reducing bias associated with pre-treatment differences between the groups [25]. It is assumed that the selection can only be explained in terms of the observed characteristics of each individual in the treated group, with the goal of finding a comparable individual in the non-treated group for the comparison [26].

In this study, the treatment is defined by belonging to a specific economic sector. Individuals working in that sector are considered treated, and the control group consists of all those working in other economic sectors. Next, the possibility of commuting for individuals in the analyzed sector is studied and compared to the commuting possibilities of individuals from other sectors, as if they had worked in the same sector. For this purpose, following Rosembaum and Rubin [25], a matching process is conducted based on the probability of belonging to the treatment group:

$$\mathrm{P}\left(\mathrm{X}\right)=\mathrm{P}\mathrm{r}(\mathrm{D}=1│\mathrm{X})$$

(2)

where X is a set of observable characteristics, and D is a variable indicating whether the person has worked in a specific sector (received the treatment), which entails the condition of common support necessary to identify the treatment effect. Thus, with Equation (2), individuals in the control group are matched with those in the treatment group. For comparison, an appropriate pair for each individual is sought, i.e., one from the treated group and one from the control group with similar probabilities of commuting. After all treated units are matched with control units, the difference in commuting between individuals in the sector and those working in other sectors is calculated, and finally, the average treatment effect on the treated units (ATT) is computed [27].

For the matching process, the nearest neighbor method was used, where for each individual in the treatment group, the closest individuals from the control group (in this study, five observations) with the most similar probability were selected.

2.3. Data

The data for this study come from the National Employment Survey (ENE) and the Supplementary Income Survey (ESI), both conducted by the National Institute of Statistics (INE) of Chile [4,28]. The ESI is conducted during the OND quarter of each year and complements the ENE, applied to the same subjects in both surveys. This allows the merging of ENE and ESI data for the OND quarter, providing employment and income data. A pooled dataset is created to cover information from two years prior to the onset of the COVID-19 health crisis (2018 and 2019) and two years following (2021 and 2022).

The dataset consists of all individuals who report being employed and provide information on all relevant variables.

Table 1. Description of variables used, 2018 to 2022 (number of individuals).

	Mean	Standard Error	Number of Observations		Mean	Standard Error	Number of Observations
Agriculture	0.0633	0.0006	182,922	Income	783,007.6	2236.0	1,789,770
Mining	0.0151	0.0003	182,922	Working hours	40.4618	0.0353	171,739
Industry	0.1053	0.0007	182,748	t2018	0.2047	0.0009	44,189
Construction	0.1527	0.0008	182,748	t2019	0.2091	0.0010	40,523
Commerce	0.2829	0.0011	182,748	t2020	0.1816	0.0009	25,093
Service	0.3806	0.0011	182,748	t2021	0.1983	0.0009	37,124
Male	0.5802	0.0012	182,922	t2022	0.2063	0.0009	35,993
Head of household	0.4557	0.0012	182,922	Great North	0.1018	0.0005	16,922
Primary education	0.1370	0.0008	182,922	Small North	0.0902	0.0005	13,746
Secondary education	0.4195	0.0012	182,720	Center	0.1857	0.0008	24,396
Tertiary education	0.4433	0.0012	182,720	Metropolitan	0.2080	0.0011	16,789
Full-time work	0.8072	0.0010	172,036	South	0.2519	0.0009	35,481
Informal employment	0.2616	0.0010	182,922	Patagonia	0.1624	0.0006	27,929
Age	42.5062	0.0312	182,922

Source: elaborated based on ESI and ENE OND, 2018–2022.

lists the variables considered, along with the number of observations, the mean, and the standard deviation, including the distribution of the sample by year and zone.

The study includes six dummy variables, one for each economic sector of the main occupation. The distribution of the sample across these sectors was as follows: 6.3% in the agriculture sector; 1.5% in the mining sector; 10.5% in the manufacturing industry, electricity, gas, steam, and water supply sector; 15.3% in the construction, transportation, and storage sector; 28.3% in the commerce, lodging activities, financial and real estate, information, and communication sector; and 38.1% in services, including professional, scientific, administrative services, public administration, defense, education, healthcare, arts, entertainment, domestic, and extraterritorial organizations.

Additionally, 58% of the sample consisted of men (with 1 for the variable sex if male at birth), 45.6% were heads of households (1 for the variable if head of household), and the average age was 42 years. Regarding educational levels, most individuals possessed tertiary education (44.3%) and secondary education (42%), with 13.7% having primary education. Most individuals worked full-time in their main occupation (80.7%) and 26.2% worked informally. On average, the monthly income was CLP 783,007.6 (adjusted for November 2022), and individuals typically worked 40.5 h per week in their main occupation. In terms of residence, the sample was distributed as follows: 10.2% in the Great North, 9% in the Small North zone, 18.6% in the Central zone, 20.8% in the Metropolitan zone, 25% in the South zone, and 16.2% in the Patagonia zone (Figure 1). Finally, temporal fixed effects were used for the years 2018, 2019, 2020, 2021, and 2022. As observed in Table 1, the sample is balanced across the years.

3. Results

In this section, the variables associated with the probability of commuting during the period between 2018 and 2022 were first analyzed at a global level. Next, the model was developed to examine trends by geographic zones of Chile. Finally, the impact of economic activity on the commuting decisions across different zones from 2018 to 2022 was evaluated.

3.1. Commuting Probability

The commuting probabilities in Chile were estimated using three logistic models (

Table 2. Coefficients and marginal effects on the probability of commuting in the country.

	Model 1			Model 2			Model 3
	Coef	dy/dx	p > z	Coef	dy/dx	p > z	Coef	dy/dx	p > z
Mining	1.673	0.252	***	1.539	0.127	***	1.268	0.103	***
	(0.064)			(0.066)			(0.071)
Construction	0.286	0.026	***	0.294	0.013	***	0.387	0.021	***
	(0.028)			(0.032)			(0.033)
Industry	−0.008	−0.001		0.022	0.001		0.066	0.003
	(0.034)			(0.035)			(0.035)
Commerce	−0.318	−0.022	***	−0.296	−0.010	***	−0.203	−0.008	***
	(0.025)			(0.026)			(0.027)
Agriculture	−0.707	−0.043	***	−0.473	−0.015	***	−0.414	−0.016	***
	(0.037)			(0.040)			(0.043)
Small North	0.950	0.112	***	1.017	0.065	***	1.079	0.080	***
	(0.046)			(0.046)			(0.049)
Center	1.609	0.238	***	1.637	0.141	***	1.741	0.173	***
	(0.040)			(0.041)			(0.044)
Metropolitan	2.845	0.537	***	2.875	0.393	***	2.980	0.447	***
	(0.040)			(0.041)			(0.043)
South	1.374	0.189	***	1.443	0.114	***	1.547	0.142	***
	(0.040)			(0.040)			(0.043)
Patagonia	−0.092	−0.007		−0.050	−0.002		−0.023	−0.001
	(0.049)			(0.050)			(0.052)
Male				0.361	0.017	***	0.252	0.013	***
				(0.022)			(0.023)
Household Head				−0.001	0.000		−0.135	−0.006	***
				(0.022)			(0.023)
Age				0.026	0.001	***	−0.015	−0.001	**
				(0.004)			(0.005)
Age2				0.000	0.000	***	0.000	0.000	+
				(0.000)			(0.000)
Secondary Education				0.386	0.018	***	0.163	0.008	***
				(0.032)			(0.034)
Tertiary Education				0.855	0.051	***	0.322	0.017	***
				(0.034)			(0.039)
Informal Employment							−0.773	−0.025	***
							(0.031)
Full-time Work							0.456	0.025	***
							(0.031)
Ln(Earning per Hour)							0.371	0.017	***
							(0.017)
t2019	−0.019	−0.002		−0.031	−0.001		−0.026	−0.001
	(0.032)			(0.033)			(0.034)
t2020	−0.173	−0.013	***	−0.222	−0.008	***	−0.226	−0.009	***
	(0.033)			(0.034)			(0.036)
t2021	−0.175	−0.013	***	−0.211	−0.008	***	−0.212	−0.009	***
	(0.029)			(0.030)			(0.031)
t2022	−0.092	−0.007	**	−0.130	−0.005	***	−0.131	−0.006	***
	(0.029)			(0.030)			(0.031)
Constant	−2.335		***	−3.523		***	−6.096		***
	(0.044)			(0.109)			(0.192)
N	182,748.0			182,547.0			168,132.0
Pseudo r2	0.1511			0.1705			0.2030
BIC	46,745,491.9			45,607,067.6			40,708,063.1
#Pr(conmuta)	0.0883			0.04157			0.0473
Correctly classified (%)	69.17			68.77			66.06
Positive predictive (%)	74.02			77.73			82.36
Negative predictive (%)	66.24			63.36			56.22

+ p < 0.10, ** p < 0.01, *** p < 0.001. The standard errors are in parentheses. # Probability of commuting for people working in the service sector in 2018: women, not heads of household, with educational levels equivalent to basic education, who do not work in the formal sector, with working hours that are not full time, and income and age corresponding to the average value (source: own elaboration).

). First, economic sectors and geographic zones were incorporated, followed by demographic characteristics, and finally, information on individuals’ labor conditions. Additionally, year effects were controlled using fixed effects, with 2018 as the base year.

It was consistently observed that the probability of commuting was higher in mining in Chile, with a marginal effect between 25.2% and 10.3% higher than in the service sector. Additionally, a positive effect was noted for construction, although smaller, with the marginal variation ranging between 2.6% and 1.3%.

By contrast, the sectors of agriculture and commerce were associated with a lower probability of commuting, a trend that persisted across all estimates—although the negative marginal effect was greater in activities within the agriculture sector. On the other hand, it was observed that working in the industrial sector had no impact on commuting in any of the three estimates.

Regarding geographical zones, when compared to the Great North, the only zone not associated with higher commuting was Patagonia, while the other zones exhibited a positive and significant marginal effect across all three models. This can be explained by the mobility associated with the mining sector, which predominantly operates in the Great North zone. Additionally, the high effect observed in the Metropolitan zone is due to specific characteristics of metropolitan regions, where there is greater intercommunal mobility and transportation conditions that facilitate commuting.

Regarding demographic variables, it was confirmed that commuting is higher among men and individuals with higher education levels. The impact of tertiary education is significantly greater than that of secondary education, with a difference ranging between 0.9 and 3.3 percentage points (Table 2, Models 2 and 3). However, results regarding age and head of household status were not consistent across estimates.

In the third model, additional variables were included to determine whether a person worked under informal conditions, had a full-time work schedule, and their hourly wage. The calculations align with the existing literature, indicating that commuting is associated with formal jobs, full-time employment, and higher wages.

To analyze the temporal aspect, it was essential to consider that Chile experienced the so-called “social uprising” in 2019, an event that may have influenced commuting behavior. Additionally, the effects of COVID-19 became evident in 2020. Therefore, 2018 was established as the baseline for the estimations.

The results indicate that no significant impact was observed in 2019 across any of the three models. However, in 2020, the probability of commuting consistently decreased, with a minimum drop of 0.9% compared with 2018 (Table 2, Model 2). Similarly, all estimates show that the lower probability of commuting persisted in 2021 and 2022. Nonetheless, in 2022, the negative effect was substantially reduced, possibly due to a recovery in certain economic activities and adjustments in public policies that lifted some mobility restrictions.

3.2. Probability of Commuting in Different Zones of Chile

To compare commuting and its relationships with socio-productive characteristics in the different Chilean zones, six equations were estimated—one for each geographical zone. The marginal effects are presented in

Table 3. Marginal effect on the probability of commuting in different zones of Chile.

	Great North		Small North		Center		South		Patagonia		Metropolitan
Minning	0.159	***	0.250	***	0.203	***	0.571	***	0.283	***	0.046
Construction	0.040	***	0.069	***	0.061	***	0.126	***	0.049	***	0.060	***
Industry	0.042	***	0.023	*	0.016	+	0.050	***	0.036	***	−0.021
Commerce	−0.031	***	−0.016	*	−0.025	***	−0.016	*	−0.011	*	−0.056	***
Agriculture	−0.031		0.007		−0.042	***	−0.005		0.053	***	−0.298	***
Male	0.062	***	0.047	***	0.054	***	0.050	***	0.019	***	0.044	***
Household Head	0.001		−0.009	*	−0.020	***	−0.027	***	−0.009	*	−0.032	***
Age	0.005	**	0.001	+	0.000		−0.001		−0.001		−0.006	***
Age2	0.000	***	0.000	*	0.000		0.000		0.000		0.000	**
Secondary Education	−0.016		0.040	***	0.035	***	0.032	***	−0.003		0.040	**
Tertiary Education	−0.024	*	0.051	***	0.080	***	0.057	***	0.003		0.085	***
Informal Employment	−0.039	***	−0.034	***	−0.065	***	−0.095	***	−0.030	***	−0.195	***
Full-time Work	−0.001		0.021	**	0.074	***	0.063	***	0.006		0.138	***
Ln(Earning per Hour)	0.032	**	0.029	***	0.063	***	0.067	***	0.012	***	0.088	***
t2019	−0.005		−0.001		0.028	**	−0.078	***	−0.001		0.028	*
t2020	−0.018	*	−0.014	*	−0.019	*	−0.091	***	−0.015	***	−0.020
t2021	−0.011		−0.012	*	−0.018	*	−0.094	***	−0.007		−0.018
t2022	0.000		0.001		−0.007		−0.080	***	−0.013	*	−0.002
N	17,051		14,836		31,452		41,912		27,543		35,338
r2	0.0986		0.1603		0.0816		0.0737		0.0518		0.0814
Wald chi2(18)	516.22		962.47		1327.83		1687.53		387.74		1763.71
#Pr(conmuta)	0.0825		0.0713		0.1648		0.2244		0.0584		0.4514
Average age	41.192		43.205		43.059		43.217		42.856		41.834
Average Monthly Hourly Income	15,684.3		13,314.4		13,497.8		12,269.9		13,805		16,385.6
Correctly classified (%)	58.37		60.16		59.46		58.72		58.40		46.9
Positive predictive (%)	12.15		30.87		63.04		71.09		12.99		95.09
Negative predictive (%)	86.28		77.84		57.29		51.26		85.82		17.8

Nota. ⁺ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. # Probability of commuting of people working in the service sector in 2018: women with basic education, working part-time and formally, and the zonal average value for the continuous variables of age and income per monthly hour. Source: author’s own creation.

.

Consistent with the findings presented earlier, the lowest probability of commuting is observed among individuals residing in Patagonia (5.8%), while the highest probability is in the Metropolitan zone, where it reaches 45.1% for individuals with reference characteristics.

Regarding economic sectors, it was observed that, except for the Metropolitan zone, mining was the sector most strongly associated with commuting across different zones. The effect was highest in the South zone and lowest in the Small North zone—which aligns with previous findings. Conversely, the construction sector exhibited a similar relationship with commuting across different zones, with an average effect of 6.8%. Along the same lines, but contradicting previous evidence, the industry sector was positively and significantly associated with commuting in all zones except the Metropolitan zone, with an average impact of 3.3% on commuting. Additionally, a negative effect of the commerce sector was observed across all zones of the country. However, the agriculture sector presented varying effects depending on the zone: it had a negative impact in the Center and Metropolitan zones, while in Patagonia, it exhibited a positive effect.

Furthermore, in all zones, higher commuting probabilities were associated with men, formal employment, and higher income. Except for the Great North zone, the probability of commuting decreased when the worker was the head of the household and increased for full-time employment.

The effect of age varied by zones, exerting a positive impact in the two northern zones, while in the Metropolitan zone, the relationship was negative. Regarding human capital, it generally exhibited a positive effect, likely due to the expectation of higher income. However, for individuals residing in the southernmost part of Chile (Patagonia), the effect was null, and by contrast, in the Great North zone, tertiary education was negatively associated with the probability of commuting.

Regarding the temporal effect, the findings at the national level (Table 3) were generally confirmed, but some significant zonal differences emerged. In the Metropolitan and Central zones, contradicting national trends, the probability of commuting increased by 2.8% in 2019. By contrast, in the South zone, commuting decreased with a marginal effect of 7.8%. Additionally, in the Metropolitan zone, no negative impact on commuting was observed in 2020 and subsequent years. In 2021, the negative effect was only confirmed in the Small North, Center, and South zones, while in 2022, it was observed in the South zone and Patagonia.

As for the temporal effect, in general, the findings at the country level are corroborated (Table 3). However, some important differences were observed by territory: in the Metropolitan and Center zones, contrary to what was observed at the country level, in 2019, the probability of commuting increased by 2.8% in both, and in the South, it decreased, with a marginal effect of 7.8%; in turn, in the Metropolitan zone, in 2020 and beyond, there was no negative impact on commuting, in 2021, the negative effect was only confirmed in the Small North, Center, and South zones, and in 2022 in the South and Patagonia zones.

3.3. Impact of Productive Sectors on the Commuting Decision

So far, spatial differences in commuting have been observed, revealing that in 2020 (during the health crisis), there was a significant negative effect compared to 2018 in all zones of the country, except for the Metropolitan zone. Additionally, the relationship between the probability of commuting and the economic sector of employment varied across zones, likely due to the heterogeneous productive specialization of the country.

This section presents the results of assessing the effect of the economic sector on the decision to commute, exploring differences by zones. The underlying hypothesis is that productive behavior and commuting decisions may exhibit different patterns before and after the first year of the health crisis (2020), with variations across zones.

Using PSM, estimations were conducted for the five studied periods (2018, 2019, 2020, 2021, and 2022) to examine the effect of working in each sector (treatment) on the probability of commuting by zone in each period (Appendix A). For better visualization, the results are presented in Figure 3, where each bar represents the average treatment effect on the treated (ATT) for each sector, by year, in the six zones studied. The following analysis focuses on the effect of the year 2020, when a shock occurred in the Chilean economy.

As indicated in the previous section, employment in the agriculture, service, and commerce sectors is generally associated with a lower probability of commuting. However, when evaluating the impact by year and zone, distinct dynamics emerge. Specifically, in the agriculture sector, the negative effects diminished in 2020 and persisted into 2021 and 2022 in some zones. The most notable differences were observed in Patagonia, where a positive impact only occurred in 2019, aligning with previous findings. In the Metropolitan zone, the strongest negative impact on the agriculture sector was confirmed across all years, decreasing in 2020 and 2021 before returning to pre-pandemic levels in 2022—a trend similar to that observed in the Center zone.

Regarding the commerce sector, 2020 saw an increased negative effect on the probability of commuting in the Great North, Metropolitan, and Patagonia zones. However, this effect weakened in the Center zone and disappeared in the Small North zone. Meanwhile, the services sector exhibited clearly distinct effects by zones. Particularly, in the Metropolitan zone, the sector consistently increased commuting across all years, with a further increase in 2020. By contrast, a negative and consistent effect was observed in the Small North, Great North, and Patagonia zones, with the decline intensifying in 2020.

On the other hand, as previously noted, commuting in Chile is mainly associated with employment in the mining and construction sectors. The observed trend indicates that mining experienced a general downturn in 2020, with a delayed effect in Patagonia occurring in 2021 and persisting into 2022. However, as previously reported, no consistent effect was found in the Metropolitan zone. The construction sector, by contrast, generally exhibited positive impacts across all years, except in the Great North zone, where no annual effect was observed. In the Small North zone, there was a clear negative impact on commuting in 2020. Additionally, in Patagonia, the sector had a low impact on commuting, which reversed in 2022.

Finally, the industrial sector generally had a low impact on the commuting decision. A consistently positive effect was only observed in the Patagonia zone following the health crisis.

4. Conclusions

The economic impact of the COVID-19 health crisis in Chile was predominantly observed in 2020. This study analyzes the relationship between the productive sector in which a person works and their decision to commute, and how this relationship changed before, during, and after the health crisis in different geographical zones of Chile, demonstrating that these effects persist in some areas and sectors over time.

Indeed, in 2020, commuting decreased, likely due to the economic shock and the sanitary restriction measures. Although commuting rates have yet to recover, differences in outcomes between the country’s geographical zones can be observed, which are potentially linked to the spatial diversity of productive specializations. It is possible that certain activities require greater spatial mobility of the workforce and, in the short term, may recover to pre-pandemic commuting rates. However, as demonstrated in this article, this will not be the case for all productive sectors, which could have significant territorial impacts.

It is observed that the probability of commuting increases with human capital and is higher among men across all regions, which is consistent with other studies. This lower mobility among women could negatively affect the gender wage gap, as noted in the literature. Additionally, the probability of commuting is primarily associated with the mining and construction sectors, with a greater impact on individuals residing in the South zone. Conversely, the commerce and agriculture sectors have a negative effect, particularly in the Metropolitan zone, likely due to the greater distance from the city within this sector and the lower relative price of labor in these areas [5]. Additionally, working in the services sector generally produces a negative impact on the probability of commuting, except for individuals residing in the Metropolitan zone, where the effect was favorable. Given the differences in results between economic sectors, the lower commuting rates may possibly be explained by growing phenomena such as teleworking and digitalization.

In 2020, the negative impact of the agriculture sector on commuting was reduced, likely because this sector acted as a refuge during a labor demand reduction, leading to a shift towards working in this sector amid the decrease in commuting observed in other activities, mainly those linked to mining. In fact, in 2020, spatial mobility for mining jobs decreased in the Great North, Small North, and Center zones, and from 2021, this decline was also observed in the Patagonia zone. It is important to remember that northern Chile specializes in mining, where long-distance commuting is associated with higher wages [14]. During the pandemic, this commuting likely decreased due to mobility restrictions, which continued to affect the Patagonia zone until 2022. However, a recovery was observed in the North, Center, and South zones beginning in 2021.

Regarding the construction sector, although it is generally associated with higher commuting, a contradictory effect in 2020 was only observed in the Small North zone. However, for individuals living in the Center and Metropolitan zones, commuting increased, with the same effect observed in Patagonia at the end of the period. Despite this, overall mobility in 2022 was lower than in 2019, which is linked to the sector dynamics. Faced with rising raw material prices and lower real estate demand, the sector has experienced a delay, with effects on employment.

On the other hand, the industry sector is generally not associated with commuting, with a consistently positive effect only observed in Patagonia, where commuting is likely driven by intraregional commuting due to the specialization of this sector in the zone.

A similar situation occurs with the service sector in the Metropolitan zone, where its positive impact on commuting is consistent, even as observed during the health crisis. This zone specializes in services and is highly interconnected, so it is possible that during the crisis, intraregional commuting was strengthened, which explains the increase in 2020.

Finally, this study used a balanced dataset extending 2 years prior to and 2 years after the health crisis, and demonstrates that the temporal impact on spatial mobility was greater than expected, not recovering to 2018 levels in some zones, meaning that commuting levels have not returned to pre-crisis rates. However, the post-2020 results are driven by other permanent or transient phenomena that have emerged in parallel, some of which promote mobility while others act contrary to it. These factors help explain the upward trend in mobility that was observed in previous periods before the crisis.

This work demonstrated that, in the face of external shocks, changes and effects on labor commuting occurred in various zones, and these effects continue over time. As these dynamics change, public policies must rapidly adapt in concert (i.e., providing incentives for employment recovery or support for productive sectors). Monitoring these movements would allow for the proposal of more precise policies, such as those related to transportation networks and labor subsidies.

However, in geographical terms, territories require differentiated policies, as a generalized shock affects different labor and productive structures differently. This implies that, for example, incentives should be linked to sectors associated with the productive vocations of the territories and their dynamics. Furthermore, in light of the results of this study, it seems appropriate to suggest that incentives and subsidies in response to external shocks also consider the fact that people not only commute between territories but also change economic sectors. Therefore, public policy responses to these shocks may include programs to facilitate job retraining.

Therefore, labor observatories and research centers should focus future research on regional employment analysis studying and simulating generalized shocks, such as new health crises, energy crises, and climate change effects, among others.