Why do the poor travel less? Urban structure,commuting and economic informality in Mexico City

Abstract

While in Mexico City formal jobs are concentrated in the city centre, affordable housing is generally only available in its outer rings. This reduced accessibility to formal employment would suggest that the poor have longer commutes. However, observed travel times show that low-income workers actually have the shortest commutes. Using two linear programming transportation models we found that this is due to the location of informal work activities, which seems to be a function of the residential location of workers involved in the informal sector as a response to the disadvantages of the formal urban structure of jobs and housing that affect the poor.

Keywords

commuting excess commute informality urban structure

Introduction

Contrary to what location theory would predict, in the case of Mexico City affordable housing is generally only available in the outer rings of the urban area. Due to the fact that one-third of all formal jobs are located in the city centre, low-income workers have the least accessibility to such employment (Suárez and Delgado, 2007). This would suggest that the poor have a harder time finding jobs and are forced to make longer commutes and spend more on transportation, but this conclusion is intuitive, with no solid empirical evidence to support it. Although extensive research has been carried out on urban structure and commuting in developed countries, the presence of a large informal economic sector in nations in the developing world may have a significant effect on urban structure and transportation dynamics that could reveal important differences between the patterns observed there and those found in first-world metropolitan areas.

Thus, we first measured the commute times and sensitivity to urban structure of different income groups in Mexico City using a linear programming model. ‘Sensitivity to urban structure’ refers to the extent to which different income groups optimise their work trips. Do workers in different income categories find the closest job available to their place of residence, and does this vary depending on income? This first analysis is similar to the one performed by White (1988) to measure excess commute in a sample of American cities, and by Giuliano and Small (1993) for Los Angeles, although we use a different job-matching criterion. We found that while much commuting is explained by urban structure, lower income groups have both shorter commutes and a higher optimisation of commute times than higher income groups. Data suggest, as would be expected, that this optimisation of travel time is due to the higher percentage of income that those sectors must spend on transportation. The question that remains is why, when living in less-accessible locations, low-income workers have shorter travel times. We found a plausible answer in the location of informal work activities.

Using a second linear programming model, which modifies the assumptions and constraints of the first one, we tested whether there is a relationship between economic informality and sensitivity to urban structure. In other words, what spatial relationship, if any, exists between the residential locations of workers dedicated to informal economic activities and accessibility to formal job markets? Do travel times to formal jobs and sensitivity to urban structure influence the size and location of informal work activities, and the place of residence of workers in this sector? Results suggest that while the extent of informal employment may be a response to the low availability of formal jobs, and formal workers’ place of residence a function of job location, reduced accessibility to formal jobs, coupled with a higher sensitivity to urban structure in lower income groups, generates areas of informal work in places that are a function of the residence of the workers in this sector.

This paper is structured in five sections, the first of which presents a literature review on the relevant discussion of the relationship between residence and workplace location, together with a review of different perspectives on informality. The second section comprises an overview of descriptive statistics that illustrate the spatial structure of Mexico City. The third discusses data sources and methodology, and details the limitations of the analysis due to methodological and data constraints. The fourth section reports the results of our two analyses, and the final section presents our conclusions and their potential implications for policy.

Literature review

Two sets of literature were taken into account for the purposes of this paper: on the one hand, research on residential and employment location produced almost exclusively for developed countries, from which we take the fundamental discussion for the United States; on the other, literature on informality that has been generated for developing countries, especially in Latin America. It is important to note that finding common ground between these two bodies of research is not an easy task, given that they start out from substantially different methodological and disciplinary perspectives.

Employment and residential location: The US approach

Location theory asserts that decisions regarding location are taken in reference to economic activity in the centre. Households locate as close to the centre as their budget allows, at a point where utility is maximised, but where they are outbid by higher-priced economic land uses (Alonso, 1964). As distance to the centre increases, land rent decreases inversely to transport costs. However, poor households locate in the inner city because housing lots are smaller and structures older, and thus more affordable. With the shift from monocentric to polycentric cities, the theoretical approach to urban structure was altered, since the monocentric model seems insufficient to explain the spatial organisation of today’s cities (Small and Song, 1992)

Kain (1968) suggested that job suburbanisation created a ‘spatial mismatch’; that is, as land-intensive, low-paying jobs moved towards the suburbs, low-income workers were either left behind in the inner city, forced to make longer commutes, or unemployed. Other research suggests that job suburbanisation and sub-centring allow the co-location of jobs and housing, thus allowing residents to minimise commute times (Levinson and Kumar, 1997). However, job accessibility studies have provided evidence that low-income populations benefit least from the urban structure (Cervero, 1995a; Suárez and Delgado, 2007), and suggest that a planned balance between jobs and housing could efficiently allow co-location, thus reducing commute times and traffic congestion (Cervero, 1989, 1996; Cervero and Wu, 1997), while having a positive effect on economic performance (Cervero, 1995b).

An opposing argument holds that the link between land use and transportation is weakening (Giuliano, 1995), and that residential location choice is only partially influenced by place of work (Giuliano and Small, 1993). This approach argues that as the number of multiple-earner households grows, finding a place of residence located near the jobs of all household members becomes increasingly difficult (Kim, 1995); thus residential location choice is influenced by other factors, such as access to amenities, services, shopping and schools (Horner and O’Kelly, 2007).

Excess travel is considered a response to suburbanisation coupled with auto-dependency (Horner, 2002), which may change in time with infrastructure modifications, accessibility and the spatial match between housing and jobs (Yang, 2005), as well as the types of transportation systems available (Ma and Banister, 2006). Other scholars would agree that low-income populations benefit least from urban structure transformations (Glaeser and Kahn, 2003), and that the location of affordable housing plays the largest role in defining which population sector will be forced to make longer commutes (Sultana, 2006).

While some studies suggest that the socioeconomic characteristics of residents affect economic location – that is, jobs follow people (Scott and Urry, 1994; Storper and Walker, 1983) – others support the opposite relationship; that is, that people follow jobs (Simpson and van der Veen, 1992; Suárez and Delgado, 2010). In Mexico City, it seems that both positions may be true, depending on: (a) the type of economic activity; and (b) workers’ socioeconomic status.

Gender has also been reported to play an important role in work-trip decisions. Women normally have shorter trips (Hanson and Pratt, 1995), and in many cases – especially low-income, single mothers (Chapple, 2001) – seek informal employment close to their place of residence. In Mexico, women’s work trips have been reported to be 10% shorter than men’s (Casado, 2013).

For the case of Mexico City, only limited empirical knowledge exists concerning the variables that affect residential location and the importance of job location in relation to it. Until recently, research on commuting was limited to general descriptions of the process (Cervera, 1995; Navarro, 1988), analyses of transportation infrastructure (e.g. Islas, 2000), and measuring the magnitude of the mobility process (e.g. Graizbord, 2008; Sobrino, 2008). More recent studies, however, have shown that although residential location decisions in Mexico City are influenced by place of work, residential mobility rates are very low for low-income groups whose opportunities for mobility are limited to the outer rings of the city (Suárez and Delgado, 2010). Recent work by Guerra (2013) has shown travel time differences between modes and distance to the centre, as well as mode choice across the city and among income groups.

Economic informality

Economic informality has been studied from various perspectives: legal/regulatory (de Soto, 1989), political manipulation, economic exploitation (Castells, 1986; Castells and Portes, 1989; Perlman, 1976), and economic production (de Soto, 2000), but the spatial component of this research is limited, in many cases analogous to the informal housing process, or focused on acquiring tenure and the development of local economic activities to provide livelihoods for households (de Soto, 1989). Through qualitative inferences, a ‘given’ centre-periphery relationship has been assumed, one in which low-income residents are believed to travel more because they usually live far away from the city centre. Research shows that some family members are willing to travel further for formal jobs in order to acquire such benefits as healthcare for the whole family (Maloney, 2004). As a result, these factors must also be monetised in order to fully understand place-of-work decisions.

From a global perspective, the restructuring of economic relationships has led some researchers to suggest the existence of ‘dual cities’, where unequal access to education makes high-skilled jobs unobtainable to low-wage workers, who are then left behind in dead-end jobs in the informal sector (Mollenkopf and Castells, 1991; Sassen, 1991). But this research is highly theoretical, and empirical spatial evidence of this urban duality is scarce at best.

In the US, informality has been found to depend on the mismatch between poor inner city residents and job suburbanisation. Gender has also been shown to greatly influence the type of jobs that workers are able to find (Chapple, 2001), as women have fewer opportunities for well-paid jobs because their search radii are smaller due to household responsibilities. Race differences have also been identified as a factor that reduces accessibility to employment due to the spatial segregation of quality jobs (Chapple, 2006; Stoll and Raphael, 2000).

In Mexico City, informality has been studied mainly in relation to housing (Cruz Rodríguez, 2001; Ward, 1998), and has been linked to both the social division of space (Schteingart, 2010) and social segregation (Rubalcava and Schteingart, 2012). But accounts of informal economic activity there have been limited to microanalyses of commercial specialisation and localisation in specific street-vending areas (e.g. Méndez, 2006; Williams, 2005), despite the fact that the informal sector spans much more than just this commercial activity, and has been estimated to represent between 40 and 60% of all employment in the capital (INEGI, 2014; Maloney, 2004). Indeed, research shows that informality exists in practically every economic sector and across all income groups (Castells and Portes, 1989), and that there is an intrinsic relationship between the formal and informal sectors, as well as voluntary shifts from one to the other (Thomas, 1995) that may imply total welfare benefits for families (Maloney, 1999, 2004). Again, there is a limited spatial component to this research that needs to be made explicit, especially in relation to residential and workplace location.

In this paper we attempt to fill in some of the gaps in the literature by identifying characteristics of the spatial structure of formal and informal jobs and its relation to residential location and travel decisions. In this way, we hope to add to the knowledge of urban structure and its implications for transportation.

Study area

The population of Mexico City reached 20 million inhabitants in 2010, with an employed economically-active population (OEAP) of close to seven million (INEGI, 2010b).¹ However, employment figures for 2009 showed only 4.7 million jobs in the metropolitan area, of which 34% were concentrated in just four central municipalities out of a total of 75 such demarcations (INEGI, 2009). Those four municipalities, however, account for only 10% of working residents. Because of the spatial distribution of employment and housing, Mexico City has been considered to have a primarily monocentric structure with a dispersed population (Suárez and Delgado, 2009), though other research proposes that the city is undergoing a transformation towards polycentricity. (Aguilar and Alvarado, 2005; Graizbord, 2008).

Close to 57% of all jobs in Mexico City are informal (see the section ‘Defining informal sector jobs and workers’ in the third section). The economic sectors with the most informality are retail trade, which is also one of the largest in the city, construction and transportation, food services, and recreation. Women account for 38% of the income-earning labour force, with a presence only slightly lower in the informal economy than in the population as a whole. At least at the two-digit level, women seem to follow the general pattern of informality across all economic sectors (Table 1).

Table 1.

Mexico City: Employment characteristics.

Economic sector	Jobs	% jobs	% informal	% women	% women informal
Mining	5838	0.1	35.0	6.3	14.6
Utilities^a	23,968	0.4	n/a	20.7	n/a
Construction	373,434	5.9	86.3	4.6	2.8
Manufacturing	919,613	14.5	59.1	32.8	34.5
Wholesale trade	171,184	2.7	44.8	27.1	24.4
Retail trade	1,181,693	18.7	79.3	45.3	45.9
Transportation	404,529	6.4	77.6	20.4	19.3
Information	109,195	1.7	24.4	34.2	37.0
Finance and insurance^a	116,901	1.8	n/a	44.8	n/a
Real estate	27,715	0.4	61.4	35.4	0.0
Professional services^a	302,612	4.8	n/a	37.9	n/a
Management services^a	3541	0.1	n/a	38.7	n/a
Administration support and waste services	277,095	4.4	56.0	34.8	0.0
Education^a	398,836	6.3	n/a	65.2	n/a
Healthcare^a	267,210	4.2	n/a	65.8	n/a
Arts, entertainment, recreation	63,608	1.0	75.9	31.4	29.8
Accommodation and restaurants	359,646	5.7	79.1	51.6	53.6
Other services	591,400	9.3	90.1	51.9	55.1
Government^b	374,658	5.9	0.0	36.6	0.0
Total	6,330,760	100.0	56.6	37.6	34.0

Notes: ^aNo informal jobs are shown because the criteria used to identify informality explicitly exclude these sectors.

The government sector is not used in the linear programming models presented in the paper.

Source: Authors’ calculations based on INEGI (2010a).

Table 2 presents a summary of descriptive statistics, grouped according to Delgado’s urban ring configuration (Suárez and Delgado, 2007), shown in the map in Figure 1. Job densities decrease as distance from the centre increases, as do land prices, while population densities increase in the first ring and then decrease towards the fringe. The percentage of informal jobs increases towards the fringe, reaching more than 90% of available jobs in some municipalities (Figure 2). Income, education and rental costs all decrease with distance from the centre, thus presenting a counter-intuitive pattern, compared to US cities (Weinberger et al., 2007).

Table 2.

Mexico City: Urban structure and socioeconomic indicators.^a

Urban ring	Centre city	First ring	Second ring	Outer ring	Fringe	Total
Population density (p/ha)	123	135	109	81	31	88
Job density (jobs/ha)	114	34	19	10	4	21
% jobs (formal)	34%	27%	20%	15%	4%	100%
% working residents	10%	28%	26%	28%	8%	100%
Urban land area (ha)	13,949	38,146	50,715	69,149	50,135	222,094
% urban land	100%	95%	55%	35%	12%	29%
Land price (US$/m²)	735	512	128	111	40	n/a
Average household size	3.1	3.7	3.8	4	4.2	3.78
Average school years (resident workers)	12.6	11.08	10.3	9.8	9.2	10.5
Annual work income (US$)	10,862	6801	6320	5636	4798	6507
Average travel time to work	42.3	54.3	60.9	71.7	56.9	59.6
Average annual rent per dwelling unit (US$)	5298	2222	2034	1443	1192	2475

Note: ^aPopulation density in urban areas; job density in urban areas; established jobs as a percentage of the metropolitan area; occupied economically active population as a percentage of the metropolitan area; land area of urban tracts; percentage of land area of urban tracts to total municipal land; commercial land value (unconstructed).

Source: Authors’ calculations based on INEGI (2009, 2010a, 2010b); Metros Cúbicos (2012) and INEGI (2007).

Figure 1.

Income distribution and urban ring configuration.

Figure 2.

Mexico City: Distribution of formal and informal jobs.

Average time to work was around one hour in 2007, according to calculations based on the origin-destination (O-D) survey (INEGI, 2007). However, there are important differences in travel times depending on residential location and income (Guerra, 2013; Suárez and Delgado, 2010). Travel time to work increases towards the outer ring of the city and then drops significantly in the fringe. Some reports indicate that 20% of workers spend more than three hours travelling to-and-from work each day (UN-HABITAT, 2013; World Bank, 2009), findings that could be interpreted as showing that it is mostly low-income workers who do extreme commutes. However, this is not necessarily true. Low-income workers have slightly higher, but similar, travel times to those of high-income workers in the city centre and the first ring, but lower travel times than all other income groups in the rest of the city, especially in municipalities in the fringe (Figure 3).

Figure 3.

Average travel times by urban ring and income category.

Women’s travel times are on average five minutes shorter than those of men. This difference increases with distance to the centre. Across income groups, the difference between women and the rest of the population becomes larger as income rises.

Research approach and methodology

Our research approach consisted of measuring commute times by income group, and comparing them to optimal commute times, which we calculated through linear programming transportation models. These models estimate the optimal number of trips between all origins and destinations, so that workers are assigned to the closest job available that matches their sector of activity and income category, while minimising average commute time. This involved two steps. In the first, we used the residential and employment location data of all residents and all jobs, without distinguishing between formal and informal employment. This analysis was designed to test which income groups had longer commutes, but also which income groups optimised their commute times. Comparing observed (real) and optimal commute times allowed us to assess the sensitivity to the urban structure for each income category. This is the same approach used by Giuliano and Small (1993) for Los Angeles, though that model used occupation as a matching proxy for income. Instead, we used income data and economic sector of employment.²

In the second step, we once again looked at optimal commute times by income category; however, we used only the location of formal jobs as the demand side of the analysis, while the supply side consisted of all workers in their respective residential locations, regardless of whether they held formal or informal occupations. Under the hypothesis that residents compete for close-by jobs that match their sector of activity and income category, the analysis generated the optimal commute time to formal jobs per income group, as well as an estimated surplus of residents in each municipality who would be left without formal jobs and, therefore, would be forced to find work in the informal sector. We then compared this estimated surplus of informal working residents with actual figures per municipality to see if accessibility to formal jobs and urban structure could, in fact, determine where informal workers live.

Description of data sources

Data sources include the 10%-sample database from Mexico’s 2010 population census, which contains individual-level information on place of residence and place of work at the municipal level, as well as the sector of occupation, type of employment, income, and legal worker benefits received. Mean commute times between municipalities were calculated on the basis of the 2007 O-D survey (INEGI, 2007), while travel times between pairs of municipalities that had no travel data were estimated using a log-log regression with travel time as a function of distance by road between municipality centroids (R2 = 0.69). Travel times within 16 municipalities in the fringe not included in the O-D survey were estimated as the average travel time from the peripheral municipalities included in the O-D survey geography. Expenditures on transportation were calculated using the 2010 household income and expenditure survey (INEGI, 2010a). All analyses were performed at the municipal level, the smallest geographical area for which personal census records are released, and considered the entire working population living in urban areas with three conditions: (1) people employed in agriculture or government were not included; (2) only those who declared that they were working at the time of the census were counted; and (3) the economic activity had to be remunerated monetarily. Although the agricultural sector constitutes a significant part of the informal work market, it is not inherently an urban function, and while government jobs belong to the formal employment market, the military (which does not share the same spatial relationship between jobs and housing as other sectors) cannot be identified with the rest of that sector, so including it in the analysis could skew the results.

Defining informal sector jobs and workers

Based on a prior definition of informality (Thomas, 1995), informal residents were identified as those whose work amounted to self-employment or day-labour, but were not professionals or involved in healthcare, education, finance, telecommunications, government-owned industries or services, or economic sub-sectors heavily-regulated by government.³ Employees who were not receiving healthcare benefits or a pension fund (both legal rights granted to formal workers in Mexico), were also considered as informal. Thus, the location and size of employment and place of residence for both the formal and informal sectors were determined by adding the number of workers and residents in each sector per municipality.

Measuring observed and optimal commute times

First, we calculated mean optimal commute times with a linear programming transportation model. Conceptually, the result is the average time that working residents would spend on daily work trips if every worker chose, in combination with the rest of the working population in the city, the job nearest their place of residence, so as to minimise average commute times. This model requires three types of data: (1) the demand for workers in each destination, which is the number of jobs in each municipality obtained from the place of work variable in the census database; (2) the supply of workers in each area of origin, which is the number of working residents in each municipality obtained from the place of residence variable in the census database; and (3) an average travel time matrix between each origin and destination, obtained from the O-D survey (travel times are not reported in the Mexican census).

To avoid underestimating optimal commute times, we matched workers and jobs in nine economic sectors (q = [1, 2, …, 9]: manufacturing; commerce and transportation; information; finance and real estate; technical services; professional services; restaurants, hotels and entertainment; and other services, using as well four income categories within each sector (quartiles). According to these matching criteria, someone working in a low-income job in manufacturing could not be assigned to a job in a different sector and different level of income, but only to a job in his or her income category and work sector. Thus, the procedure was run 36 times, once for each income category in each economic sector.

Equation 1 yields the number of optimal trips (n*) between pairs of municipalities (ij) for each work sector (q) divided into the income categories (k) considering the matrix of average travel times (t) between origins (i) and destinations (j). One constraint of the linear programme is that it can assign each resident only one job. This is achievable because there is exactly the same number of jobs (d) in each economic sector and income category as there are workers (s). The optimal number of trips is then multiplied by the average commute times between origins and destinations, and added up to calculate total optimal commute time (T*) (Equation 1.1). Finally, this result is divided by the total number of trips to obtain the overall average optimal commute time matched by income category and work sector (γ*) (Equation 2).

\min {T_{qk}^{*} = \sum_{i} \sum_{j} t_{ij} \cdot n_{ij}^{*}

(1)

subject to

\sum_{j} n_{ij} = s_{ij,} \sum_{i} n_{ij} = d_{ij} and n_{ij}^{*} \geq 0

T^{*} = \sum T_{qk}^{*}

(1.1)

y^{*} = T / N

(2)

In order to estimate optimal commute times for each income category, the final sum totals only the optimal commute time among work sectors within each income category, thus yielding four different optimal commute times, one for each income category. This procedure allows the results to be disaggregated by municipality, work sector, or urban ring. Finally, ‘explained commuting’ (ec) is calculated as the proportion of mean optimal time over mean observed time, either for the entire population, or for each income group (Equation 3).

ec = γ^{*} / γ

(3)

Next, we calculated observed commute times (T) (Equation 4). To assure consistency with the linear programming model, these were estimated as the product of the total number of observed trips between municipalities (n_ij) obtained from census data, multiplied by the average travel time between each pair of municipalities (t_ij) obtained from the O-D survey. These were then added up and divided by the total number of trips (N). To calculate observed commute times for each of the four income categories, the same procedure was used, but taking into account only the trips of each category. This also yielded four separate results.

T = \sum_{i} \sum_{j} t_{ij} \cdot n_{ij} / N

(4)

Average travel times between municipalities, calculated using the O-D survey, take into account the mode split of all work trips between pairs of municipality. However, because 80% of trips in the city are made by public transportation, observed travel times are likely to be overestimated for the high-income group, which has access to cars. Nevertheless, this will be in the same degree as for optimal travel times, so the proportional relationship between observed and optimal commute times is maintained.⁴

Measuring observed and estimated location of informal residents

To determine the relationship between the places of residence of informal sector workers and the structure of formal employment, we first divided the database into formal and informal jobs, and then performed a second optimisation procedure. In this case, however, the demand side of the analysis consisted only of formal jobs, while the supply side encompassed all working residents. Because the supply of all workers exceeds the demand for formal jobs, the constraint that assigned each resident to only one job in the previous model was relaxed, such that all jobs must have one worker assigned, but not all residents are assigned to a job. As a result, there is a surplus of residents in each municipality who cannot obtain formal employment. This analysis was based on the hypothesis that all workers compete for the closest formal job available that matches their work sector and income category, leaving those who are too far away from appropriate formal jobs with no possibility of obtaining one. If they were to secure formal employment, then average optimal commuting time would increase. Thus, the linear programme used is the same as in Equation 2, though subject to:

\sum_{j} n_{ij} \leq s_{ij}, \sum_{i} n_{ij} \geq d_{ij} and x_{ij} \geq 0

Next, we separately calculated optimal commute times for the formal and informal sectors, each divided into four income categories by applying Equation 1. Finally, we compared observed and optimal commute times for formal workers, and the observed number of residents per municipality working in the informal sector with the surplus estimated above, to determine whether the place of residence of informal workers may indeed be a function of urban structure.

Limitations of the analysis

Our analysis may slightly overestimate explained commuting due to the use of highly-aggregated data that reduce the number of origins and destinations (Horner, 2002). However, after exploring several data sources, the 2010 census micro-data was identified as the only database that had all the appropriate variables required to estimate informality, together with place of work and place of residence.

Linear programming models use an average travel time matrix to generate results. Therefore, both observed and optimal travel times are weighted means, which in this case underestimate actual travel times. However, they do so for both optimal and observed travel times in the same manner, thus permitting the same patterns across urban rings and income groups as actual averages. For this reason, the results of the linear programing model should be interpreted as relative, not absolute.

The literature on this topic suggests that gender could be used as a matching criterion. However, the use of two-digit economic sectors would not generate appropriate job-matches for men and women. Hence, a technique distinct from linear programming should be used to control for gender.

Finally, our analysis is limited by the fact that it leaves aside benefits such as healthcare. Unfortunately, at this time we have neither sufficient data to monetise these benefits nor census information on private healthcare. We believe that the aforementioned limitations should be addressed in future research.

Results

Overall sensitivity to urban structure

Our results show that average observed commute times per municipality range from 20 to 70 minutes, with a mean of 44 minutes. As expected, municipalities in the city centre and the first urban ring – where most jobs are located – have below-average commute times, while municipalities in the second and outer urban ring exceed the average. Municipalities in the fringe show shorter, though still above-average commutes, compared to those in the outer ring. Optimal travel times (with the exception of the outer ring) increase towards the fringe, and do not follow the same pattern as observed travel times. This results in a higher explained commute in the city centre and in the fringe, but a lower explained commute in the middle rings (Table 3). Our analysis shows that, in general, urban structure explains 83% of commuting. The overall average observed commute time was close to eight minutes above optimal.

Table 3.

Sensitivity to urban structure by urban ring.

	Observed travel time	Optimal travel time	% explained
Urban ring
Central city	35.3	27.8	79%
First ring	41.9	37.9	91%
Second ring	43.4	38.4	88%
Outer ring	49.2	35.8	73%
Fringe	45.0	34.6	77%
Total	43.8	36.2	83%
Income category
Low	39.0	34.2	88
Mid-low	42.1	35.6	84
Mid-high	46.6	37.9	81
High	48.4	37.3	77
Total	43.8	36.2	83

Source: Authors’ calculations.

Upon examining income categories, the variation in optimal commuting across groups seems unimportant. Optimal times increased slightly with income, up to the mid-high group, and were maintained for the high-income category. Still, higher-income groups have optimal times above the average, while lower-income groups have optimal times below it. Observed times, however, vary significantly, with the high-income group travelling close to 11 minutes more than the low-income group. This results in a higher explained commute for low-income workers (88%), and a lower explained commute for the high-income group (77%), which suggests that low-income residents are more sensitive to urban structure than high-income individuals, and so optimise their work trips to a greater extent.

Sensitivity to urban structure in the formal and informal sectors

The second linear programming model used only formal jobs as the demand. Results show that optimal commute time decreased by over four minutes, while observed time increased by three minutes (Table 4). Thus, overall explained commuting is 13% less if only the formal sector is taken into account. Our calculations show that work trips in the informal sector are significantly shorter than in the formal sector across all income categories. As in the first model, sensitivity to urban structure in the formal sector decreases with income. Although the average commute in the informal sector also increased with income, the percentage of workers in this sector decreased in the higher income groups. As would be expected, sensitivity to urban structure is associated with expenditures on transportation. While the high-income group spends, on average, 6% of income on transportation, the low-income group spends 18%.

Table 4.

Sensitivity to urban structure and informality by income category.

	Income category
	Low	Mid-low	Mid-high	High	Total
Observed travel time (formal)	45.4	47.1	50.8	51.3	49.5
Optimal travel time (formal)	33.4	34.1	36.4	36.3	35.4
% explained (formal)	73.6	72.4	71.7	70.8	71.5
Observed travel time (informal)	37.0	39.5	42.1	42.5	39.5
% informal resident workers	76%	66%	48%	33%	57%
% income spent on transportation	18%	13%	11%	6%	12%
Annual work income (US$)	1888	3332	4907	14,143	6002

Source: Authors’ calculations; INEGI (2010a, 2010b).

Across the urban rings, there is a clear pattern between distance to the centre and commuting in the formal sector (Table 5), as both observed and optimal travel times increased with distance to the centre, up to the fringe, where both decreased. However, this relationship is not linear. The highest explained commute occurs in the second ring, but decreased as distance from it increased. The city centre and first and second rings had above-average explained commutes, in contrast to the outer ring and the fringe. Observed travel times followed the same concentric pattern in both the formal and informal sectors.

Table 5.

Sensitivity to urban structure and informality by urban ring.

	Observed travel time (formal)	Optimal travel time (formal)	% explained (formal)	Observed travel time (informal)	% informal resident workers
Central city	38.1	27.4	71.8	31.5	43
First ring	46.0	36.8	80.0	38.1	52
Second ring	49.7	44.8	90.1	38.8	58
Outer ring	58.8	37.9	64.4	42.9	61
Fringe	54.3	30.0	55.2	41.1	70
Total	49.5	35.4	71.5	39.5	57

Source: Authors’ calculations.

Location of informal residents

The second linear programming model produced, as an additional result, an estimated volume of residents per municipality that were left to find informal jobs after losing the spatial competition for formal employment across the city. According to our definition of informality, 56% of all workers were dedicated to informal economic activities. As expected, the municipalities with higher percentages of observed informal working residents are located towards the periphery (Table 5). Our estimated numbers of informal working residents per municipality show a very close match with actual figures. When correlating estimated informal working residents with the observed number of informal working residents across municipalities, we found a high-squared correlation of r² = 0.92 (Figure 4), which points strongly to unequal access to jobs in spatial terms as a determinant of informality, and suggests that low accessibility to formal jobs, due to the characteristics of the spatial structure, determines who will end up working in the informal sector.

Figure 4.

Correlation between observed and estimated informal residents per municipality.

Conclusions

Mexico City’s urban structure explains, on average, 83% of commute times. There is, therefore, a strong relationship between residential location and place of work. Lower-income groups are more sensitive to urban structure, and so strive more than the other groups to optimise their work trips. Our data are thus consistent with the theory which suggests that this may be a result of a higher percentage of income being spent on transportation, although factors such as time and information on job availability may also have an important effect on this optimisation, especially for low-income residents living in the periphery. Whether it is residential location that influences the decision of place of work, or job accessibility that affects residential location, cannot be ascertained from our data; however, it is possible that this relationship varies with income, with higher-income groups being able to decide on the basis of residential location, while lower-income groups – subject to the availability of cheap or informal housing – must decide based solely on (informal) work location.

In line with the literature on ‘dual cities’, our results suggest the existence of a dual spatial structure. Informal work represents close to 57% of the economic activity in the city and is present in all economic sectors and income categories, though concentrated in lower-income groups. In the informal sector, travel time is less than in the formal sector, and optimisation is higher. Under the assumption that workers compete for nearby jobs that match their work activity, it follows that when working residents are left without a formal job, they will develop one in the informal sector, which may well be located in an area appropriate for informal economic activity close to their place of residence to further optimise travel time. This hypothesis is strengthened by the high correlation between the estimated surplus of residents left without formal jobs while running the optimisation procedure of formal work trips, the actual number of residents per municipality working in the informal sector, and by the fact that informal work flows are the ones that are most highly-optimised. Thus, if in fact the size of economic informality is a response to the incapacity of the formal economy to accommodate all workers, then the formal urban economic structure will indeed determine which individuals are dedicated to informal economic activities, and where they live (not where they decide to live). The location of informal work may thus depend, in part, on the distance to the place of residence, so that such jobs will be located at a point where income is maximised and transportation cost minimised.

Although there is no overall spatial pattern of travel time optimisation, generally-speaking the first three urban rings follow concentric patterns that do not appear in the municipalities on the fringe. This may be a function of the level of urbanisation, or of the level of connectivity to the main continuous area of the city.

Compared to the US, the urban structure of Mexico City is inverted in terms of the central location of jobs and the peripheral location of low-income residents. However, it is income coupled with segregation and job-resident mismatches that persist as the underlying variables that influence accessibility to well-paid jobs and work trip decisions.

Additional issues for study on this subject abound. Are there differences in job and housing location among types of informal workers? What are the implications of welfare benefits, especially healthcare, on job location decisions? What role does the regulation of land markets play in enabling the co-location of jobs and housing in the informal sector? And if it does play a role, what are the implications? Are these favourable, and in what ways? Can they be translated into urban policies?

Precise policy implications, we believe, will require further research; however, the evidence presented herein suggests the importance of building affordable housing in job-accessible areas. An urban policy based on a balance between jobs and housing could help ameliorate some of the disparities of job access in Mexico City that affect mostly the poor.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Aguilar

Alvarado

(2005) La reestructuración del espacio urbano de la Ciudad de México. ¿Hacia la metrópoli multinodal? In: Procesos metropolitanos grandes ciudades. Dinámicas recientes en México y otros países. Porrúa, Mexico, pp. 265–308.

Alonso

(1964) Location and Land Use. Boston: MIT Press.

Casado

(2013) Estructura regional de los mercados laborales locales en México. PhD Thesis, Coordinación de Estudios de Posgrado, UNAM, Mexico.

Castells

(1986) La ciudad y las masas: sociología de los movimientos sociales urbanos. Madrid: Alianza.

Castells

Portes

(1989) World underneath: The origins, dynamics and effects of the informal economy. In: Portes

Castells

Benton

(eds) The Informal Economy: Studies in Advanced and Less Developed Countries. Baltimore: Johns Hopkins University Press, pp. 11–37.

Cervera

(1995) La encuesta origen y destino de los viajes de los residentes del AMCM. In: Chías

(ed.) El transporte metropolitano hoy. Mexico City: UNAM, pp. 73–84.

Cervero

(1989) Jobs–housing balance and regional mobility. Journal of the American Planning Association 55: 136–150.

Cervero

(1995a) Paradigm shift: From automobility to accessibility planning. Working paper. Berkeley: Institute of Urban and Regional Development, University of California at Berkeley.

Cervero

(1995b) Polycentrism, commuting, and residential location in the San Francisco Bay Area. Working paper 640. Institute of Urban and Regional Development, University of California at Berkeley, Berkeley, CA, USA.

10.

Cervero

(1996) Jobs–housing balance revisited: Trends and impacts from the San Francisco Bay Area. Journal of the American Planning Association 62(4): 492–511.

11.

Cervero

(1997) Polycentrism, commuting, and residential location in the San Francisco Bay Area. Environment and Planning A 29(5): 865–886.

12.

Chapple

(2001) Time to work: Job search strategies and commute time for women on welfare in San Francisco. Journal of Urban Affairs 23(2): 155–173.

13.

Chapple

(2006) Overcoming mismatch: Beyond dispersal, mobility, and development strategies. Journal of the American Planning Association 72(3): 322–336.

14.

Cruz Rodríguez

(2001) Propiedad, poblamiento y periferia rural en la Zona Metropolitana de la Ciudad de México. Puebla: Red de Investigación Urbana, UAM.

15.

de Soto

(1989) The Other Path: The Economic Answer to Terrorism. New York: Basic Books.

16.

de Soto

(2000) The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else. New York: Basic Books.

17.

Giuliano

(1995) The weakening transportation-land use connection. Access Magazine, 3–11.

18.

Giuliano

Small

(1993) Is the journey to work explained by urban structure? Urban Studies 30: 1485–1500.

19.

Glaeser

Kahn

(2003) Sprawl and urban growth. National Bureau of Economic Research (NBER) working paper series, 9733.

20.

Graizbord

(2008) Geografía del transporte en el área metropolitana de la Ciudad de México. Mexico: El Colegio de México.

21.

Guerra

(2013) The new suburbs: Evolving travel behavior, the built environment, and subway investments in Mexico City. PhD Thesis, University of California at Berkeley, CA, USA.

22.

Hanson

Pratt

(1995) Gender, Work, and Space. London: Psychology Press.

23.

Horner

(2002) Extensions to the concept of excess commuting. Environment and Planning A 34: 543–566.

24.

Horner

O’Kelly

(2007) Is non-work travel excessive? Journal of Transport Geography 15(6): 411–416.

25.

INEGI (2007) Encuesta Origen Destino de los Viajes de los Residentes de la Zona Metropolitana del Valle de México. Mexico: Instituto Nacional de Estadística Geografía e Informática.

26.

INEGI (2009) Censo Económico. Aguascalientes: Instituto Nacional de Estadística Geografía e Informática.

27.

INEGI (2010a) Encuesta Nacional de Ingresos y Gastos de los Hogares. Aguascalientes: Instituto Nacional de Estadística Geografía e Informática.

28.

INEGI (2010b) XIII Censo de Población y Vivienda 2010, Base de datos de la muestra. Aguascalientes: Instituto Nacional de Estadística Geografía e Informática.

29.

INEGI (2014) Encuesta Nacional de Ocupación y Empleo 2014. Aguascalientes: Instituto Nacional de Estadística Geografía e Informática.

30.

Islas

(2000) Llegando tarde al compromiso: la crisis del transporte en la Ciudad de México. Mexico: El Colegio de México.

31.

Kain

(1968) Housing segregation, negro employment and metropolitan decentralization. Quarterly Journal of Economics 82: 175–197.

32.

Kim

(1995) Excess commuting for two-worker households in the Los Angeles metropolitan area. Journal of Urban Economics 38: 166–182.

33.

Levinson

Kumar

(1997) Density and the journey to work. Growth and Change 28: 147–172.

34.

Banister

(2006) Extended excess commuting: A measure of the jobs–housing imbalance in Seoul. Urban Studies 43(11): 2099–2113.

35.

Maloney

(1999) Does informality imply segmentation in urban labor markets? The World Bank Economic Review 13(2): 275–302.

36.

Maloney

(2004) Informality revisited. World Development 32(7): 1159–1178.

37.

Méndez

(2006) Estrategias de localización del comercio en vía pública en el Centro Histórico de la Ciudad de México. In: Cruz

(ed.) Espacios Metropolitanos 2: Población, Planeación y Políticas de Gobierno. Puebla: UAM-RNIU, pp. 83–114.

38.

Mollenkopf

Castells

(eds) (1991) Dual City: Restructuring New York. New York: Russell Sage Foundation.

39.

Navarro

(1988) El transporte de la fuerza laboral. In: Benítez

Morelos

(eds) Grandes problemas de la Ciudad de México. Mexico City: Plaza y Valdez, pp. 243–264.

40.

Perlman

(1976) The Myth of Marginality: Urban Poverty and Politics in Rio de Janeiro. Berkeley: University of California Press.

41.

Rubalcava

Schteingart

(2012) Ciudades divididas, desigualdad y segregación social en México. Mexico: El Colegio de México.

42.

Sassen

(1991) The Global City. New York, London, Tokyo and Princeton: Princeton University Press.

43.

Schteingart

(2010) División social del espacio y segregación en la Ciudad de México. Continuidad y cambios en las ultimas décadas. In: Garza

Schteingart

(eds) Los grandes problemas de México, Desarrollo urbano y regional. Mexico City: El Colegio de México, pp. 345–387.

44.

Scott

Urry

(1994) Economies of Signs and Space. London: Sage.

45.

Simpson

van der Veen

(1992) The economics of commuting and the urban labor market. Journal of Economic Surveys 6(1): 45–62.

46.

Small

Song

(1992) Wasteful commuting: A resolution. Journal of Political Economy 100(4): 888–898.

47.

Sobrino

(2008) Movilidad intrametropolitana en la Ciudad de México. Estudios Demográficos y Urbanos 22(7): 583–617.

48.

Stoll

Raphael

(2000) Racial differences in spatial job search patterns: Exploring the causes and consequences. Economic Geography 76: 201–223.

49.

Storper

Walker

(1983) The theory of labor and the theory of location. International Journal of Urban and Regional Research 7(1): 1–43.

50.

Suárez

Delgado

(2007) Estructura y eficiencia urbanas: Accesibilidad a empleos, localización residencial e ingreso en la ZMCM 1990–2000. Economía, Sociedad y Territorio VI(23): 693–724.

51.

Suárez

Delgado

(2009) Is Mexico City polycentric? A trip attraction capacity approach. Urban Studies 46(10): 2187–2211.

52.

Suárez

Delgado

(2010) Patrones de movilidad residencial en la Ciudad de México como evidencia de co-localización de población y empleos. EURE 36(107): 67–91.

53.

Sultana

(2006) What about dual-earner households in jobs–housing balance research? An essential issue in transport geography. Journal of Transport Geography 14(5): 393–395.

54.

Thomas

(1995) Surviving in the City: The Urban Informal Sector in Latin America. London: Pluto Press.

55.

UN-HABITAT (2013) Planning and design for sustainable urban mobility: Global report on human settlements. Report for UN-HABITAT.

56.

Ward

(1998) Mexico City. Chichester and New York: John Wiley & Sons.

57.

Weinberger

Suárez

Chapple

(2007) It’s not worth the trip: Urban structure and job formality in Los Angeles, Mexico City, and New York. In: Transportation Research Board 86th annual meeting, Washington, DC, 21–25 January 2007. Washington, DC: Transportation Research Board.

58.

White

(1988) Urban commuting journeys are not wasteful. Journal of Political Economy 96(5): 1097–1110.

59.

Williams

(2005) Comercio callejero en la Delegación Coyoacán y micro-territorios; enfoques para el desarrollo de políticas públicas. Mexico: CENVI and Secretaría de Desarrollo Social del Gobierno del Distrito Federal.

60.

World Bank (2009) Reshaping economic geography. World development report, Washington, DC, USA.

61.

Yang

(2005) Commuting impacts of spatial decentralization: A comparison of Atlanta and Boston. Journal of Regional Analysis and Policy 35(6): 9–78.