SciELO - Scientific Electronic Library Online

vol.67 issue3Theoretical proposal of a methodology for the analysis of business risks using processes and fuzzy logic in the Cuban tourism sectorUnconventional monetary policy of the ECB and unemployment author indexsubject indexsearch form
Home Pagealphabetic serial listing  

Services on Demand




Related links

  • Have no similar articlesSimilars in SciELO


Contaduría y administración

Print version ISSN 0186-1042

Contad. Adm vol.67 n.3 Ciudad de México Jul./Sep. 2022  Epub June 06, 2023 


The diversity of the local economic structure and its relation with occupational demand and poverty rates in Mexico's metropolitan areas

Francisco Javier Segura Mojica1  * 

1Tecnológico Nacional de México/Instituto Tecnológico de San Luis Potosí, México


This research explores how the diversity of the economic structure in metropolitan areas is related to occupational demand and to moderate and extreme poverty rates. Data from a sample of 29 metropolitan areas in Mexico were analyzed. The variables Simpson Diversity Index (IDS), Informality Rate, State GDP Variation and Years of Schooling were incorporated into the analysis as predictors of employment and poverty. Descriptive and inferential statistics were used as analysis tools, and linear regression and PLS models were formulated. As a relevant result, it was found that there is a significant positive correlation between the IDS and the number of Permanent Insured; as well as significant negative correlations between the IDS and the percentage of the population in a situation of moderate poverty and extreme poverty

JEL Code: C29; 017; R12

Keywords: Simpson diversity index; metropolitan zones; employment predictors; poverty predictors


Esta investigación explora la forma en que la diversidad de la estrucura económica en las zonas metropolitanas se relaciona con la demanda ocupacional y con los índices de pobreza moderada y extrema. Se analizaron datos de una muestra de 29 zonas metropolitanas de México. Al análisis se incorporaron las variables Índice de Diversidad de Simpson (IDS), Tasa de Informalidad, Variación del PIB Estatal y Años de Escolaridad como predictores de empleo y pobreza. Como herramientas de análisis se utilizó estadística descriptiva e inferencial y se formularon modelos de regresión lineal y PLS. Como resultado relevante se encontró que existe una correlación significativa de signo positivo entre el IDS y el número de Asegurados Permanentes; así como correlaciones significativas de signo negativo entre el IDS y el porcentaje de población en situación de pobreza moderada y pobreza extrema.

Código JEL: C29; O17; R12

Palabras clave: índice de diversidad de Simpson; zonas metropolitanas; predictores de empleo; predictores de pobreza


It is difficult to imagine that a company can emerge and remain in the market without being part of a business ecosystem or outside its influence. Aspects such as the manufacturing specialization of a city or region, the local and national business environment, the number of companies participating in the different economic sectors and subsectors, the interdependence of sectors and subsectors, the formation of supply chains, and the size of the market seem to be related to the emergence and closure of businesses, and therefore to companies' life expectancy.

Metropolitan areas are complex ecosystems with high concentrations of businesses. Nevertheless, each metropolitan area possesses characteristics that differentiate it from its peers. The question that gives rise to this research is whether and to what extent the characteristics of metropolitan areas, understood as business ecosystems, exert any influence on variables associated with social and business development, such as poverty and employment.

Literature review

According to the National Institute of Geography and Statistics (Spanish: Instituto Nacional de Geografía y Estadística, INEGI) (2014), there are 59 metropolitan areas in Mexico where 73% of the population resides and 77% of the GDP is produced. A metropolitan area is formed when a city "exceeds its politicaladministrative territorial limit to form an urban area located in two or more municipalities" (Sobrino, 2003). The formation of metropolitan areas is a socio-spatial process whose dynamics are influenced by economic and social phenomena such as the social and spatial division of labor, the integration of regions into the flow of local and global economic value, supply chains, and globalization or selective specialization of local economies (Bernardes and Castillo, 2007).

Different authors (Santos, 1977; Sanchez, 1991; Bernardes and Castillo, 2007) agree that these are complex places, composed of different subspaces, whose administration is subject to the variability of public policies and private investment decisions. There is also the perception that, although each metropolitan area has unique characteristics, in countries such as Mexico and Brazil-still located on the "periphery" of the economic giants-their development faces major limitations in the absence of longterm territorial planning.

Iracheta (2010) considers that metropolitan areas concentrate location advantages (economies) for social and economic actors that are superior to most neighboring cities, and offer better living conditions, provision of services, and equipment. He also notes that the disadvantages of metropolitan areas are a tendency toward disorganized and unsustainable growth, unequal provision of services- especially for poor social sectors-, lack of resources to address social needs, weak institutions, and lack of a framework for intergovernmental coordination.

Following Iracheta's (2010) rationale, the locational advantages offered by metropolitan agglomerations-derived from productive, sectoral, and spatial concentration processes, as pointed out by Garza and Schteingart (2010)-would presumably exert some kind of influence on phenomena such as poverty and economic informality. Lezama (2014) mentions that these are socially constructed spaces where social relations evolve, but inequality and inequity do so as well.

One of the ways in which the complexity of urban entrepreneurial ecosystems seems to relate to the phenomenon of poverty is how the configuration of labor markets means that access to employment would be one of the determinants for the population to improve their quality of life progressively, but that they can also quickly fall into poverty in situations of economic contraction, given their dependence on income from work (Aguilar-Zurita, Martínez and Armenta, 2018).

A key variable for understanding the interactions between the economic and social subsystems is the phenomenon of informality. According to Robles, Sánchez, and Beltrán (2019), it is the result, on the one hand, of the low development and productivity of an economy, and on the other, of the heavy demographic concentration in urban areas. Income inequality, lack of social security, and deficient tax collection would be among its consequences.

Robles et al. (2019) state that the population increase in urban areas has a positive relation with informality. Consequently, a priority of public policies in these areas should be the generation of employment and not only mobility, public services and housing, which are the predominant concerns of metropolitan administrations.

The Sustainable Development Goals (SDGs) defined by the UN in 2015 are the framework of a new development agenda whose objectives are eradicating poverty, protecting the planet, and ensuring prosperity for all. In Mexico, the 2030 Agenda for Sustainable Development was adopted, and the Sustainable Cities Index, in which the original 17 goals were translated into 107 indicators, was established (Citibanamex-Centro de Investigación y Docencia Económica [CIDE] 2018).

The reason why the monitoring of the SDGs in Mexico has an emphasis on metropolitan administration is the degree of population concentration in urban spaces, coupled with the processes and imbalances that generate chronic problems of poverty, vulnerability, inequality, and environmental impact (Citibanamex-CIDE, 2018).

Terraza, Rubio, and Vera (2016) note that in urban growth processes, both the need for capital and the means of production converge, as well as the need to meet the aspirations of society that would otherwise be unrealizable. In this respect, they point out that if fairer growth prospects are sought, it is necessary to resort to new planning methodologies and, above all, to think about urban economies from other perspectives that are simultaneously open to local and global aspects. This way, the urban space favors the interaction of nano and micro enterprises with large companies, which could be achieved through managing and strengthening business ecosystems.

The metropolitan area as a business ecosystem

Each metropolitan area has ecosystemic characteristics and pressures that simultaneously drive business creation and closure. It follows that one way to explain the growing specialization and improved efficiency of companies in metropolitan environments is by using a biological metaphor, comparing organizations to biological beings, which are part of ecosystems, focusing on survival and seeking a balance between their subsystems and the environment (Montoya et al., 2012).

The integration of economic units into business ecosystems is a topic that has been widely explored. Authors such as Williamson (1975) state that reducing transaction costs is one of the main motivations for companies to integrate, which also leads to improved productivity since collectivity allows companies to improve their efficiency (Montoya et al., 2012) by favoring their degree of specialization. Among the business groupings that have been widely studied are clusters (Kothandaraman and Wilson, 2001) and productive agglomerations (Teixeira and Ferraro, 2009).

Along these lines, Moore (2005) proposes the notion of organizational ecology to characterize the economic communities in which leading producers, suppliers, customers, and competitors interact. By developing specialized functions, they carry out co-evolutionary processes; that is, they help beneficial relations between two species to evolve.

Montoya et al. (2012) point out that in this type of ecosystem the agents, i.e., the companies, are related by competition, cooperation, or mutualism, and that in general, there is an intention to access markets, solve problems, or access technology that motivates companies to integrate.

Duranton and Puga (2000) point out several key aspects of diversification and specialization in cities, including the fact that both characteristics coexist. To calculate the degree of specialization, they propose a Specialization Index that quantifies a sector's share in local employment. For diversification, they propose the inverse of the Hirshman-Herfindal index, which consists of the sum of the squares of firms' market shares in a given sector.

Regarding highly specialized cities, Duranton and Puga (2000: 534) point out that behind their economic structure lies a strong dependence on natural resources. They also note that large cities tend to be more diversified, while cities with similar levels of specialization have similar sizes, and that the growth of a city is related to specialization and diversification.

In another paper, Duranton and Puga (2019: 43) propose a model of how cities and urbanization interact with aggregate income and economic growth. They identify patterns in which residents of less productive localities are incentivized to move to more productive localities; however, this trend is constrained because city residents impose regulations limiting the arrival of new residents. Thus, when modeling the heterogeneity of localities, they propose considering both agglomeration economies and urban costs, and warn that limiting the size of cities also limits the benefits o agglomeration.

Associating industrial composition with diversity, Park (2020) warns that industrial diversity tends to increase in technology-intensive industries but decreases in traditional ones. He also found that vertical markets are strongly correlated with diversity in building space and that high land prices hinder agglomeration and may have a negative association with diversity.

Analyzing the formation of clusters of knowledge-intensive companies, Pérez-Campuzano (2021) finds that variables such as educational level influence the location of companies but not necessarily the number of companies. In contrast, the presence of other companies and the transportation and mobility variables influence the number of companies that are set up in these clusters.

Another factor related to economic diversity and poverty levels is the rate of entrepreneurship. In this regard, Lee and Rodríguez-Pose (2020) warn that the effect of entrepreneurship on poverty levels in cities depends on the sectors in which it operates. For example, when it is concentrated in tradable sectors with other cities or regions, it can generate positive multiplier effects and impact poverty levels, while if it is focused on non-tradable sectors, it can saturate local markets, neutralizing its effects on poverty.

Socio-ecological systems and structural coupling

A perspective that can also be used to explain the behavior and survival capacity of organizations is that of socio-ecological systems and structural coupling. This theoretical approach proposes that there is a web of relations around the resources that are necessary for human life where social and environmental variables interact (Ostrom, 2009). Accordingly, interactions do not only occur in the social sphere but are also related to physical space.

Among the concepts incorporated into the socio-ecological system approach are selforganization, which designates mechanisms that respond to the preconditions of the system from which its structure can be modified, and attractors or states of self-organized stability (Gunderson and Holling, 2002; Urquiza and Cadenas, 2015).

Another concept that can help understand a social-ecological system is resilience, which in its general form is understood as the capacity of a system to adapt to changing conditions in its environment and to resist or recover from impacts without losing its integrity. In the case of a social-ecological system, it seems to be related to the diversity of the elements that make it up, in such a way that a greater variety of elements is a major advantage in stressful and risky situations (Urquiza and Cadenas, 2015).

In this context, reviewing the concepts and metrics used to estimate the specific diversity of communities is relevant. It is important to note that, although the subject has been widely debated and has also led to semantic and conceptual problems (Hurblert, 1971), specific diversity is considered an emerging property that is related to the variety of communities and is derived from two components: the variety or richness of species and equitability, which is the distribution of abundance among the number of species.

The Margalef (1956) index expresses the specific abundance, that is, the relation between the number of species (S) and the total number of individuals observed (n).

The Shannon-Wiener (H) and Simpson (D) indices measure diversity, incorporating specific richness and equitability in a single value. The Shannon-Wiener index derives from information theory and measures the information content per individual in samples obtained at random from an extensive community, so that diversity is understood as the degree of uncertainty in predicting to which species an individual taken at random from a sample of S species and N individuals corresponds (Pla, 2006).

Simpson's diversity index indicates the probability of finding two individuals of different species in two successive random extractions without replacement (Bouza, 2005). It is expressed in the following equation:

Si=1-i=1Spi2 (1)

where pi is the proportional abundance of the ith species and represents the probability that an individual of species i is present in a sample, so the sum of pi equals 1. Therefore,

pi=niN (2)

Values close to 0 in Simpson's Diversity Index would indicate the dominance of a few species, while values close to 1 would indicate greater diversity, less dominance of certain species, and greater ecosystem stability.


The research was quantitative, correlational in scope, and cross-sectional in design.

The research questions are:

  • Is there any statistically significant relation between Simpson's Diversity Index and moderate and extreme poverty rates in metropolitan areas?"

  • Is there any statistically significant relation between Simpson's Diversity Index and job generation in metropolitan areas?"

The research aim is to determine whether the diversity of lines of business in a metropolitan area influences poverty levels and the capacity to generate or retain jobs.

For this research, the subsectors of economic activity will be considered as the equivalent of the species of an ecosystem, and the number of economic units of each subsector per metropolitan area will be the equivalent of the number of individuals of each species per ecosystem. The basis for determining the subsectors (or species) was the Industrial Classification System for North America (Spanish: Sistema de Clasificación Industrial para América del Norte, SCIAN 2018), and the data on the number of units or individuals were taken from INEGI's National Directory of Economic Units (DENUE).

The research consisted of collecting data and calculating and correlating the variables mentioned in the research questions, as shown in Table 1.

Table 1 Operationalization of variables 

Variable Calculation method Source of data
Specific wealth of the business ecosystem Number of different subsectors registered in a metropolitan area. National Directory of Economic Units (DENUE). INEGI, 2018
Simpson's diversity index Si=1-i=1Spi2 National Directory of Economic Units (DENUE). INEGI, 2018
Population living in extreme poverty by metropolitan area Number of people per metropolitan area living with an income insufficient to purchase a basic food basket and lacking at least three of the following basic needs: food, health services, social security, education, basic housing services and housing quality, and total population of the metropolitan area. CONEVAL, Poverty Module at Municipality Level 2010 and 2015
Population in moderate poverty by metropolitan area Number of people living with an income insufficient to satisfy their basic needs and suffering at least one of the following deprivations: food, health services, social security, education, basic housing services and housing quality, and total population of the metropolitan area. CONEVAL, Poverty Module at Municipality Level 2010 and 2015
Variation in the number of people permanently affiliated to the IMSS (Mexican Institute of Social Security) (Number of people permanently affiliated to the IMSS as of December 31 of year t-Number of permanently affiliated people to the IMSS as of December 31 of year t-1)/Number of permanently affiliated people in year t-1 IMSS-Affiliated Workers by State (IMSS-STPS 2020)
Labor informality rate Labor informality rate 1(TIL1)= (Informal employment/Population employed)100 Robles, Sánchez and Beltrán (2019) based on the Socioeconomic Conditions Module of INEGI (2014)

Source: created by the authors

Data analysis

First, a sample of 29 of the 59 metropolitan areas identified by INEGI in Mexico was defined (see Table 4). The main criterion for selecting the ZMs (metropolitan areas) was their demographic importance since they are home to 76.31% of the metropolitan areas' population and 48.5% of Mexico's total population.

Next, using INEGI's National Directory of Economic Units (DENUE), data were extracted corresponding to the number of companies or establishments existing in each metropolitan area for each of the 93 subsectors of economic activity considered in the North American Industrial Classification System (SCIAN) catalog used by INEGI (2018).

Using Microsoft Excel: Mac 2011 software, the Specific Wealth and Simpson indices were calculated for each metropolitan area. The results are shown in Table 2.

Table 2 Demographic characterization and productive diversity of the metropolitan areas 

Metropolitan area Population 2015 (1) Percentage that represents the ZM with respect to the state population Informality rate in the ZM (2) Specific Wealth (Number of sub-sectors with activity in the ZM) (3) Simpson's Index (4)
Aguascalientes 1 056 265 80.47% 0.384933973 86 0.933843611
Tijuana 1 860 704 56.12% 0.393296447 83 0.93985469
Tuxtla Gutiérrez 829 387 15.90% 0.56499469 82 0.906411577
Chihuahua 925 200 26.01% 0.317617044 86 0.944490822
Saltillo 915 536 30.98% 0.337479946 81 0.93624918
Colima 368 270 51.78% 0.465633067 81 0.9332005
Valle de México 21 275 109 84.74% 0.511379701 91 0.916044921
La Laguna 1 374 909 78.35% 0.393296447 84 0.930002072
León 1 773 158 30.29% 0.503896857 85 0.931602144
Acapulco 901 368 25.51% 0.707107578 78 0.904233269
Pachuca 561 422 19.64% 0.585806666 80 0.926231842
Guadalajara 4 943 520 63.02% 0.467748781 89 0.930885656
Toluca 2 207 581 13.64% 0.579309775 87 0.916511249
Morelia 914 644 19.95% 0.520546409 83 0.933495914
Cuernavaca 1 003 174 52.69% 0.61159498 81 0.921543874
Tepic 470 695 39.85% 0.430290702 79 0.924041532
Monterrey 4 749 513 92.77% 0.35357351 87 0.940081264
Oaxaca 667 716 16.83% 0.672828404 84 0.918531184
Puebla-Tlaxcala 2 994 147 48.54% 0.595826714 87 0.917153386
Querétaro 1 334 231 65.46% 0.496654103 88 0.938889391
Cancún 763 310 50.83% 0.464400055 83 0.936386029
SLP 1 164 798 42.86% 0.394783397 84 0.93459691
Guaymas 218 258 7.66% 0.335983197 81 0.927144579
Villahermosa 827 692 34.56% 0.55691054 86 0.930932296
Tampico 936 004 27.20% 0.46106418 89 0.928060424
Tlaxcala Apizaco 552 620 43.42% 0.709054623 80 0.916367174
Veracruz 910 399 11.22% 0.488289091 89 0.923277428
Mérida 1 158 935 55.26% 0.4524553 87 0.933666528
Zacatecas 374 329 23.70% 0.453933947 78 0.936132763

Source: (1) CONAPO (2015)

(2) Created by the authors based on Robles, Sánchez and Beltrán (2019).

(3) Created by the authors based on the National Directory of Economic Units-INEGI (2015).

(4) Created by the authors with data from DENUE_INEGI (2015).

Subsequently, using the Poverty at Municipality Level 2010 and 2015 module of the National Council for the Evaluation of Social Development Policy (CONEVAL, 2015), Table 3 was constructed, which shows the variation of extreme poverty and moderate poverty indicators in the metropolitan areas considered in the sample. As can be seen, in 28 of the 29 Metropolitan Areas, the extreme poverty indicator decreased, while in 17, the moderate poverty indicator decreased.

Table 3 Evolution of the percentages of extreme and moderate poverty in metropolitan areas during the period 2010-2015 

Metropolitan area Extreme Poverty 2010 (1) Moderate Poverty 2010 (2) Extreme poverty 2015 (3) Moderate poverty 2015 (4) Variation in extreme poverty Variation in moderate poverty Years of schooling (5)
Acapulco 2.2 28.1 1.6 24.5 -0.6 -3.6 10.07
Aguascalientes 3.5 27.6 1.8 27.6 -1.7 0 9.68
Cancún 8.9 37.8 6.7 35.2 -2.2 -2.6 10.11
Chihuahua 2 23.8 0.6 19 -1.4 -4.8 10.78
Colima 2 18.8 1.3 16.2 -0.7 -2.6 10.43
Cuernavaca 1 29.2 1.8 25.8 0.8 -3.4 10.36
Guadalajara 1.6 29.6 1.1 27.2 -0.5 -2.4 10.34
Guaymas 5.9 38.3 0.8 29.7 -5.1 -8.6 9.97
La Laguna 4 32.8 2.2 29.3 -1.8 -3.5 9.09
León 13.2 36 12.1 44.5 -1.1 8.5 9.15
Mérida 3.5 28.3 1.9 28.2 -1.6 -0.1 10.55
Monterrey 2.2 23.6 1.4 24 -0.8 0.4 10.01
Morelia 5.9 32.1 6.1 31.4 0.2 -0.7 9.69
Oaxaca 6.8 31.3 5.9 35.8 -0.9 4.5 10.03
Pachuca 2 24.2 3.7 28.2 1.7 4 9.87
Puebla-Tlaxcala 2.5 22.8 1.9 22.3 -0.6 -0.5 10.66
Querétaro 1.6 19.4 1 15.6 -0.6 -3.8 10.53
Saltillo 4.6 30.3 3.5 35.4 -1.1 5.1 10.43
SLP 5.7 31.5 3.8 36.9 -1.9 5.4 10.01
Tampico 3 25.3 1.5 21.5 -1.5 -3.8 10.37
Tepic 3.7 23.2 2.6 25.1 -1.1 1.9 9.97
Tijuana 3 25.7 1.9 22.1 -1.1 -3.6 10.48
Tlaxcala Apizaco 5.4 24 4.3 26.4 -1.1 2.4 9.58
Toluca 7.9 42.6 5.6 30.5 -2.3 -12.1 10.69
Tuxtla Gutiérrez 2.4 25.5 1.8 26.9 -0.6 1.4 10.02
Valle de México 3.4 31.7 1.5 32.1 -1.9 0.4 10.22
Veracruz 4.5 26.8 3.6 29.7 -0.9 2.9 10
Villahermosa 2.7 24.3 1.8 21.1 -0.9 -3.2 10.29
Zacatecas 2.4 28.8 1.8 24.1 -0.6 -4.7 11.05

Source: (1) (2) (3) (4) CONEVAL, Poverty Module at Municipality level; (5) Sustainable Cities Index, 2018.

Access was also gained to the consultation module of workers permanently affiliated to the Mexican Social Security Institute [IMSS] by state (Ministry of Labor and Social Security, Spanish: Secretaría del Trabajo y Previsión Social [STPS], 2020), and the data corresponding to December 31, 2010, and December 31, 2015, were extracted. They were compared to determine the net variation of affiliated workers during that period, which corresponds to the period of comparison of poverty indicators provided by CONEVAL.

Table 4 Variation in the number of permanently affiliated workers by state 

State Permanent workers with the IMSS 2010 Permanent workers with the IMSS 2015 Variation in permanent workers 2010-2015
Aguascalientes 186 894 246 114 0.317
Baja California 557 218 694 849 0.247
Chiapas 175 140 194 949 0.113
Chihuahua 588 412 729 766 0.240
Coahuila 494 461 613 955 0.242
Colima 88 411 96 325 0.090
Distrito Federal 2 239 625 2 727 787 0.218
Durango 162 509 199 879 0.230
Guanajuato 560 289 727 292 0.298
Guerrero 116 567 123 225 0.057
Hidalgo 135 696 156 294 0.152
Jalisco 1 123 635 1 335 131 0.188
México 994 753 1 169 621 0.176
Michocacán 286 732 314 761 0.098
Morelos 153 411 173 345 0.130
Nayarit 91 184 102 366 0.123
Nuevo León 1 050 359 1 282 413 0.221
Oaxaca 145 194 172 008 0.185
Puebla 379 947 456 609 0.202
Querétaro 279 316 380 249 0.361
Quintana Roo 215 671 260 446 0.208
San Luis Potosí 255 445 312 647 0.224
Sonora 405 258 474 292 0.170
Tabasco 134 116 156 958 0.170
Tamaulipas 480 704 536 105 0.115
Tlaxcala 56 838 65 528 0.153
Veracruz 572 400 619 226 0.082
Yucatán 253 866 300 288 0.183
Zacatecas 112 111 138 137 0.232

Source: created by the authors based on STPS (2020). Workers affiliated to the IMSS by federal state

As the first step in the statistical analysis, using XLSAT 2016 software, Pearson correlation coefficients were determined between Simpson's Diversity Index and the variables Informality Rate, Variation of Permanent Workers 2010-2015, and Net Variation of GDP 2010-2015. As can be seen in Table 5, all correlations are significant at a level of 0.05%.

Table 5 Correlation matrix / Simpson's Index / Variation permanently affiliated people 

Variables Simpson's Index Informality Rate Variation in permanent workers 2010-2015 Net change in GDP 2010-2015
Simpson's Index 1 -0.753 0.557 0.547
Informality Rate -0.753 1 -0.426 -0.496
Variation in permanent workers 2010-2015 0.557 -0.426 1 0.636
Net change in GDP 2010-2015 0.547 -0.496 0.636 1

Values in bold are different from 0, with a significance level of alpha=0.05.

Source: created by the authors

Based on the above, a linear regression model was formulated using the variables already correlated in order to obtain an equation to estimate the variation in the number of permanently affiliated workers based on the variations that could be registered by the Simpson's Index, the Informality Rate in the ZM, and the Variation of the GDP in the Period. It should be noted that the variation in GDP could only be obtained with a disaggregation level by State, so for the formulation of the model, the percentage of population represented by the ZM with respect to the state population was used as a weighting variable. The model parameters are shown in Table 6, and the goodness-of-fit statistics are in Table 7.

Table 6 Normality tests of variables 

Variable / Test Shapiro-Wilk Jarque-Bera
Simpson's Index 0.162 0.341
Variation in permanent workers 2010-2015 0.786 0.758
Average annual GDP growth 2010-2015 0.417 0.642
Change in GDP 2016 0.080 0.088
Variation in extreme poverty 0.002 <0.0001
Variation in moderate poverty 0.566 0.882
Years of schooling 0.478 0.557
% Informal employment 0.381 0.580

Interpretation of the test:

H0: The variable from which the sample was drawn follows a Normal Distribution.

Ha: The variable from which the sample was drawn does not follow a Normal Distribution.

Since the calculated p-value is greater than the significance level alpha=0.05, the null hypothesis H0 cannot be rejected for the variables Simpson's Index, Change in Permanent Workers, Average Annual GDP Growth, Change in GDP 2016, Change in Moderate Poverty, Years of Schooling, and Informal Employment.

Table 7.Parameters of the linear regression model (permanently affiliated variation): 

Source Value Standard error t Pr > |t| Lower limit (95%) Upper limit (95%)
Interception -0.952 0.500 -1.902 0.069 -1.982 0.079
Simpson's Index 0.981 0.518 1.893 0.070 -0.087 2.049
Informality Rate 0.121 0.051 2.369 0.026 0.016 0.227
Change in GDP in the period 0.719 0.133 5.412 < 0.0001 0.446 0.993

Source: created by the authors

The model equation is:

Ap=-0.951+0.9811Si+0.121Til+0.719 (Pib) (3)



= Variation in permanently affiliated,


= Simpson's Diversity Index,


= Informality Rate, and


is change in Gross State Domestic Product

In order to find a better fit, another model was formulated using the partial least squares (PLS) regression method. This technique reduces the number of predictors, generating a small set of uncorrelated components on which a least squares regression is performed, from which a model with greater solvency in the face of measurement uncertainty is constructed.

In this case, a model with the same number of components (t) as predictors (variables) was obtained. Table 7 shows the correlations between variables and components, while Table 8 shows the goodness-of-fit statistics.

Table 8 Goodness-of-fit statistics of the linear regression model for the Permanently Affiliated Workers variable.  

Remarks 29.000
Sum of weights 29.000
GL 25.000
Adjusted R² 0.507
MEC 0.000
RMSE 0.018
MAPE 41.466
DW 2.162
Cp 2.300
AIC -229.301
SBC -223.832
PC 0.581

Source: created by the authors

The model equation is:

Ap=-0.979+1.031Si+0.091Til+0.611(Pib) (4)

When comparing the statistics of both models, a better goodness of fit was found for the linear regression model since the R² coefficient, which is the variability explained by the selected predictors, is higher in the first model than in the PLS regression.

Subsequently, the predictors mentioned above were correlated with the Moderate Poverty and Extreme Poverty variables, to which the variable Average Years of Schooling was added. The results are shown in Table 10, where it can be seen that both variables have significant correlations with the selected predictors.

Table 9. Correlations between factors and predictors according to the PLS model  

Variable t1 t2 t3
Simpson's Index 0.545 -0.714 0.439
% Informal employment -0.691 0.685 0.232
Change in GDP 2016 0.901 0.426 -0.088
Variation permanently affiliated 2015/2016. 0.589 0.277 0.209

Source: created by the authors

Table 10 Goodness-of-fit statistics of PLS model fit 

Remarks 29.000
Sum of weights 12.093
GL 25.000
Standard deviation 0.018
MEC 0.000
RMSE 0.017

Source: created by the authors

Table 11 Normality and non-correlation tests of residuals 

Test for normality of residuals (Shapiro-Wilk) (1) Durbin-Watson statistic (no correlation of residuals) (2)
W 0.959 D 2.62965
p-value (bilateral) 0.307 Du 1. 64987
alfa 0.050    

(1) Interpretation of the test:

H0: Residuals follow a Normal distribution

Ha: Residuals do not follow a Normal distribution

Since the calculated p-value is greater than the significance level alpha=0.05, the null hypothesis H0 cannot be rejected.

(2) Test interpretation: As D > 1.64987 (Du for a sample of 29 units with 4 terms), it is concluded that there is no autocorrelation.

Table 12 Correlation matrix Simpson's Index / Moderate Poverty and Extreme Poverty 

Variables Simpson's Index Informality rate Variation in permanent workers 2010-2015 Net change in GDP 2010-2015 Extreme poverty 2015 Moderate poverty 2015
Simpson's Index 1 -0.753 0.557 0.547 -0.651 -0.762
Informality rate -0.753 1 -0.426 -0.496 0.563 0.781
Variation in permanent workers 2010-2015 0.557 -0.426 1 0.636 -0.550 -0.498
Net change in GDP 2010-2015 0.547 -0.496 0.636 1 -0.392 -0.434
Extreme poverty 2015 -0.651 0.563 -0.550 -0.392 1 0.767
Moderate poverty 2015 -0.762 0.781 -0.498 -0.434 0.767 1
Years of schooling 0.365 -0.236 0.120 -0.099 -0.480 -0.514

Values in bold are different from 0 with a significance level of alpha=0.05.

Source: created by the authors

Table 13 Model parameters (Population in extreme poverty) 

Source Value Standard error t Pr > |t| Lower limit (95%) Upper limit (95%)
Interception 104.687 36.129 2.898 0.008 30.278 179.097
Simpson's Index -89.119 41.829 -2.131 0.043 -175.267 -2.972
Informality rate 0.000 0.000
Variation in permanent workers 2010-2015 -10.510 5.493 -1.913 0.067 -21.823 0.804
Net change in GDP 2010-2015 0.000 0.000
Years of schooling -1.682 0.784 -2.146 0.042 -3.296 -0.067

Source: created by the authors

Subsequently, a linear regression model was developed to estimate the determinants of extreme poverty, using the predictors shown in Table 9. It was found that the predictors Simpson's Diversity Index and Years of Schooling are significant at a level of 0.05%, and that, according to the goodness-of-fit statistic (R²), they explain 0.557 of the variability of Extreme Poverty.

The model equation is

Pe=104.68-89.11Si-10.5Ap-1.68(Es) (5)



= Extreme poverty


= Variation in permanently affiliated,


= Simpson's Diversity Index,


= Years of schooling

Finally, a linear regression model was formulated with the predictors mentioned above and the Moderate Poverty variable, revealing that informal employment and years of schooling are the determinants whose relation is statistically significant at 0.05%. Table 14 also shows the goodness-of-fit statistics, in which the R² coefficient indicates that the selected predictors explain 76% of the variability of Moderate Poverty.

Table 14 Goodness-of-fit statistics of the model fit (Extreme poverty) 

Remarks 29.000
Sum of weights 29.000
GL 25.000
Adjusted R² 0.503
MEC 2.951
RMSE 1.718
MAPE 60.886
DW 1.837
Cp 2.869
AIC 35.083
SBC 40.552
PC 0.585

Source: created by the authors

Table 15 Normality and non-correlation tests of residuals 

Test on the normality of the residuals (Shapiro-Wilk) (1). Durbin-Watson statistic (no correlation of residuals) (2)
W 0.960 D 2.08977
p-value (bilateral) 0.334 Du 1.64987
alfa 0.050    

(1) Interpretation of the test:

H0: Residuals follow a Normal distribution

Ha: Residuals do not follow a Normal distribution

Since the calculated p-value is greater than the significance level alpha=0.05, the null hypothesis H0 cannot be rejected.

(2) As D > 1.64987 (Du for a sample of 29 units with 4 terms), it is concluded that there is no autocorrelation.

Table 16 Model parameters (Population in moderate poverty) 

Source Value Standard error t Pr > |t| Lower limit (95%) Upper limit (95%)
Interception 174.168 99.927 1.743 0.094 -32.070 380.407
Simpson's Index -123.319 109.200 -1.129 0.270 -348.696 102.059
Informality rate 29.875 8.864 3.370 0.003 11.580 48.170
Variation in permanent workers 2010-2015 -12.350 10.632 -1.162 0.257 -34.294 9.593
Net change in GDP 2010-2015 0.000 0.000
Years of schooling -4.394 1.520 -2.891 0.008 -7.531 -1.257

Source: created by the authors

Table 17 Goodness-of-fit statistic for the model Population in moderate poverty 

Remarks 29.000
Sum of weights 29.000
GL 24.000
Adjusted R² 0.730
MEC 11.057
RMSE 3.325
MAPE 9.882
DW 2.225
Cp 4.092
AIC 74.200
SBC 81.037
PC 0.328

Source: created by the authors

Table 18 Normality and non -correlation tests of residuals 

Test on the normality of the residuals (Shapiro-Wilk) (1). Durbin-Watson statistic (no correlation of residuals) (2)
W 0.978 D 1.88488
p-value (bilateral) 0.795 Du 1,64987
alfa 0.050    

(1) Interpretation of the test:

H0: Residuals follow a Normal distribution

Ha: Residuals do not follow a Normal distribution

Since the calculated p-value is greater than the significance level alpha=0.05, the null hypothesis H0 cannot be rejected.

(2) As D > 1.64987 (Du for a sample of 29 units with 4 terms), it is concluded that there is no autocorrelation.

The model equation is:

Pm=174.16-123.31Si+29.87Til-12.35 Ap-4.393(Es) (6)



= Moderate poverty


= Variation in permanently affiliated,


= Simpson's Diversity Index,


= Informality Rate, and


= Years of schooling


Fontenla (2018) describes ecosystems as ensembles where diversity can be seen as an emergent phenomenon derived from complexity. In this sense, diversity can be understood as a measure of the complexity of ecosystems and vice versa (Morin, 2008). Now, if diversity is an epiphenomenon of complexity, this suggests that a system where species richness and abundance are combined also represents a greater number of interactions among themselves and with the environment. This is consistent with the notion that the greater the diversity, the better the conditions for maintaining the system's integrity, even if its behavior is difficult to predict (Levin, 1998; Marion et al. 2015).

If this analogy is transferred to the economic field, diversity could be interpreted as a property that allows business ecosystems to maintain their integrity in the face of external shocks, such as economic recessions. In this regard, a more diverse business ecosystem can be more effective in limiting job losses and business closures since strong interdependence allows companies to survive thanks to their integrated production processes.

In this research, Simpson's Diversity Index has been used to evaluate how the productive diversity of metropolitan areas affects their capacity to generate jobs and reduce poverty.

The correlations found between variables such as Diversity, Employment Supply, Moderate Poverty, and Extreme Poverty seem to be consistent with Iracheta (2010) and Garza and Schteingart (2010), in the sense that productive and spatial concentration translates into good economies and, therefore, advantages for economic actors. This is also consistent with the thesis of Moore (2005), who states that in business concentrations, there is a co-evolution derived from mutually beneficial relations between businesses. An interesting finding would be that diversity impacts not only the generation of wealth but also the progressive reduction of income inequality by being positively related to poverty reduction.

On the other hand, according to the modeling of variables performed, when Simpson's Diversity Index is high, the effect of GDP variations on employment, moderate poverty, and extreme poverty variables is lower than when diversity is low. This is consistent with what Urquiza and Cárdenas (2015) proposed, in the sense that diversity is related to the system's resilience.

The Informality Rate variable requires a separate mention. A finding of the present research is that it has a significant statistic relation of negative sign with Simpson's Diversity Index, which indicates that the less diverse the business ecosystem, the greater the tendency of the population to resort to precarious livelihood options. This is consistent with Robles, Sánchez, and Beltrán (2018), who found heterogeneous features according to the geographic region.


The analysis of the information collected implies the following answers to the research questions:

Simpson's Diversity Index has a significant positive correlation (0.557) with the variation in the number of workers permanently affiliated to the IMSS in the metropolitan areas studied and a significant but negative correlation (-0.753) with the Informality Rate. This indicates that diverse business ecosystems are more effective in creating jobs and reducing informality. It is worth mentioning that the size of a metropolitan area has no relation to diversity or the informality rate, as no significant coefficients are found when correlating these variables.

Simpson's Diversity Index also has a significant negative correlation with the Extreme Poverty Index (-0.651) and the Moderate Poverty Index (-0.762). This shows that diversity is a property of business ecosystems that influences the reduction of inequality in income.

Using Simpson's Diversity Index, Informality Rate, State GDP Variation, and Years of Schooling as predictors, it is possible to model the capacity of a metropolitan area to generate employment and reduce moderate and extreme poverty rates over a certain period. The parameters of the developed models show that the selected predictors explain most of the variability of the mentioned results.

The possible uses of the information gathered in this article can be classified into two parts: those related to public policies and those related to specific investment decisions. In the case of public policies, it is important to highlight the importance of creating government programs that promote integrated production, the diversification of lines of business, the lengthening of local supply chains and, in general, the development of capabilities for entrepreneurs to enter niches that allow them to increase the diversity of business ecosystems. Valuable input for decision-making would be the creation of information systems showing the least developed subsectors in a business ecosystem and the capabilities or competencies that need to be developed to enter them.

Concerning specific investment decisions, entrepreneurs need to have access to information on the lines of business where there are opportunities for entry and access to the training required to venture into these areas.


Aguilar-Zurita, N., Martínez-Prats, G., & Armenta-Ramírez, A. B. (2018). Estudio de la pobreza y desempleo en las zonas metropolitanas del sur sureste mexicano: Una revisión documental. Disponible en y consultado el 22 de junio de 2022 [ Links ]

Bernardes, A., & Castillo, R. (2007). Dinámicas metropolitanas en la era de la globalización: la promoción inmobiliaria para empresas en la ciudad de São Paulo, Brasil. EURE (Santiago), 33(98), 45-56. [ Links ]

Bouza, C. N., & Covarrubias, D. (2005). Estimación del índice de diversidad de Simpson en m sitios de muestreo. Revista investigación operacional, 26(2), 187-197. Disponible en y consultado el 22 de junio de 2022. [ Links ]

Citibanamex-CIDE (2018). Índice de ciudades sostenibles 2018. Reporte completo. Disponible en y consultado el 10 de julio de 2020. [ Links ]

CONEVAL (2015). Pobreza a nivel municipio 2010 y 2015. Anexo estadístico de pobreza a nivel municipio. Disponible en de y consultado el 10 de julio de 2020 [ Links ]

CONAPO (2015). Base de datos con las proyecciones de población de las zonas metropolitanas de México para los años 2015 a 2030. Disponible en y consultado el 09 de julio de 2020. [ Links ]

Duranton, G., & Puga, D. (2000). Diversity and specialisation in cities: why, where and when does it matter?. Urban studies, 37(3), 533-555. [ Links ]

Duranton, G., & Puga, D. (2019). Urban growth and its aggregate implications (No. w26591). National Bureau of Economic Research. Disponible en y consultado el 22 de marzo de 2022 [ Links ]

Fontenla, J. (2018). Entre diversidades ecológicas. Poeyana. Revista Cubana de Zoología. No 507: 23-39. Disponible en y recuperado el 18 de julio de 2020 [ Links ]

Garza, G., & Schteingart, M. (Eds.). (2010). Los grandes problemas de México. Desarrollo urbano y regional. T-II. El Colegio de Mexico AC. [ Links ]

Gunderson, L. H., Holling, C. (2002). Panarchy: understanding transformations in human and natural systems. Island press. [ Links ]

Hurlbert, S. H. (1971). The nonconcept of species diversity: a critique and alternative parameters. Ecology, 52(4), 577-586. ]

INEGI (2014a). Las zonas metropolitanas en México. Mini monografía. Disponible en y consultado el 12 de julio de 2020. [ Links ]

INEGI (2014b). Módulo de condiciones socioeconómicas. Encuesta Nacional de Ingresos y Gastos de los Hogares 2014. Disponible en de y consultado el 2 de julio de 2020. [ Links ]

INEGI (2018). Directorio Estadístico Nacional de Unidades Económicas (DENUE). Información para la actualización e incorporación de unidades económicas al DENUE. Datos a noviembre de 2018. Disponible en y consultado el 15 de marzo de 2019 [ Links ]

INEGI (2018). Sistema de Clasificación Industrial de América del Norte 2018 (SCIAN 2018. Disponible en y consultado el 02 de marzo de 2019 [ Links ]

Iracheta, A. (2010). El fenómeno metropolitano en México. Economía y Sociedad, 14(25). Disponible en y consultado el 22 de junio de 2022. [ Links ]

Kothandaraman, P. & Wilson, D. (2001). The future of competition value-creating networks. Industrial Marketing Management Review. [ Links ]

Lee, N., & Rodríguez-Pose, A. (2021). Entrepreneurship and the fight against poverty in US cities. Environment and Planning A: Economy and Space, 53(1), 31-52. [ Links ]

Levin S. A. 1998. Ecosystems and the biosphere as complex adaptive systems. Ecosystems 18: 431-436. Disponible en y consultado el 22 de junio de 2022 [ Links ]

Lezama, J. L. (2014). Teoría social, espacio y ciudad. El Colegio de México AC. Disponible en y consultado el 20 de junio de 2022. [ Links ]

Margalef, R. (1956). Información y diversidad específica en las comunidades de organismos. Inv. Pesq. (3) 99-106. Disponible en y consultado el 30 de abril de 2019. [ Links ]

Marion, Z. H., J. A. Fordyce y B. M. Fitzpatrick. 2015. Extending the concept of diversity partitioning to characterize phenotypic complexity. The American Naturalists 186: 134-145. [ Links ]

Moore, J. (2005). Business ecosystems and the view from the firm. The Antitrust Bulletin/Fal. [ Links ]

Montoya, L.; & Montoya, I. (2012). Aplicación de la metáfora biológica para el desarrollo de formas organizativas en la integración empresarial. Revista Facultad de Ciencias Económicas: Investigación y Reflexión, 20(2), 43-54. [ Links ]

Morin, E. 2008. Complejidad restringida y complejidad generalizada o las complejidades de la complejidad. Utopìa y Praxis Latinoamericana, 12(38), 107-119. Disponible en y consultado el 22 de junio de 2022. [ Links ]

Ostrom, Elinor (2009). “A General Framework for Analyzing Sustainability of Social-Ecological Systems”. Science, vol. 325, n°5939, 419-422. Disponible en y consultado el 12 de junio de 2022. [ Links ]

Park, J. I. (2020). Industrial diversity in building units and factors associated with diversity: A case study of the Seoul Metropolitan Area. Research Policy, 49(5), 103984. [ Links ]

Pérez Campuzano, E. (2021). Localización de empresas de servicios intensivos en conocimiento en la Zona Metropolitana del Valle de México. Investigaciones geográficas, (105). [ Links ]

Pla, L. (2006). Biodiversidad: Inferencia basada en el índice de Shannon y la riqueza. Interciencia, 31(8), 583-590. Disponible en y consultado el 30 de abril de 2019 [ Links ]

Robles Ortiz, D., Sánchez Bárcenas, H., & Beltrán Jaimes, L. D. (2019). La informalidad en las zonas metropolitanas de México: un análisis de sus principales determinantes. Revista Desarrollo y Sociedad, (83), 219-262. [ Links ]

Sánchez, J. E. (1991). Espacio, economía y sociedad. Madrid: Siglo XXI. Disponible en y consultado el 23 de junio de 2021. [ Links ]

Santos, M. (1977). “Society and space. Social formation as theory and method”. Antipode, 9, 1, 3-13. [ Links ]

Shannon C., Weaver W., (1949) The mathematical theory of communication. University of Illinois Press. Urbana, IL, EEUU, 144. Disponible en y consultado el 0 de junio de 2022. [ Links ]

Sobrino, J. (2003). Zonas metropolitanas de México en 2000: conformación territorial y movilidad de la población ocupada. Estudios demográficos y urbanos, 461-507. [ Links ]

STPS (2020). Puestos de trabajo registrados en el IMSS. Modulo de consulta por Entidad Federativa. Disponible en y consultado el 10 de junio de 2020. [ Links ]

Teixeira, F., & Ferraro, C. A. (2009). Aglomeraciones productivas locales en Brasil, formación de recursos humanos y resultados de la experiencia CEPAL-SEBRAE. Cepal. Disponible en y consultado el 22 de junio de 2022. [ Links ]

Terraza, H., Rubio Blanco, D., & Vera, F. (2017). De ciudades emergentes a ciudades sostenibles: comprendiendo y proyectando las metrópolis del siglo XXI. Disponible en Disponible en y consultado el 22 de junio de 2022. [ Links ]

Urquiza, A., Cadenas, H. (2019). Sistemas socio-ecológicos: elementos teóricos y conceptuales para la discusión en torno a vulnerabilidad hídrica. L'Ordinaire des Amériques. [ Links ]

Williamson, O. (1975). Mercados y jerarquías, su análisis e implicaciones. F.C.E, México. [ Links ]

Peer Review under the responsibility of Universidad Nacional Autónoma de México.

Received: March 17, 2021; Accepted: June 23, 2022; Published: June 23, 2022

*Corresponding author. E-mail address: (F. J. Segura Mojica)

Creative Commons License Este es un artículo publicado en acceso abierto bajo una licencia Creative Commons