The regional geography of social mobility in Mexico

ABSTRACT

We study intergenerational social mobility in Mexico at the subnational level to determine whether poorer regions have similar rates of social mobility as richer ones. We approximate socioeconomic status with an index that captures the variability across households in durable goods, services and education of household heads for both adult children and their parents. The estimated indicators suggest a clear north–south gradient in social mobility: the children of poor parents in the northern regions experience greater upward mobility than in the central and southern regions.

KEYWORDS:

• social mobility
• inequality
• intergenerational persistence
• regional economics
• Mexico

JEL:

• D63
• J62
• N36
• R10

ACKNOWLEDGEMENTS

We thank Dositeo Graña, Yunoen Badillo and Gerardo Castillo for excellent research assistance. We are also grateful for the useful comments of the participants at the Seminar on Social Mobility and Preferences for Redistribution at El Colegio de México; at the 6th Annual Sobre México Congress on Economics and Public Policy at the Universidad Iberoamericana; at the 27th Meeting on Public Economics at the Universitat de Barcelona; and at the Agence Française de Développement’s (AFD) internal seminar in Mexico City. We especially thank the editor and anonymous reviewers for their constructive and valuable comments. The opinions expressed in this paper must in no way be considered to reflect the official position of the European Union, AFD or El Colegio de México.

DATA AVAILABILITY

The datasets analysed in the current study are available from the authors upon reasonable request.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1. Previous studies have estimated intergenerational social mobility by gender (Torche, 2015a), skin colour (Campos-Vazquez & Medina-Cortina, 2019; Monroy-Gómez-Franco & Vélez-Grajales, 2020) and region (Orozco-Corona et al., 2019). In general, social mobility in Mexico is found to be low. It is particularly low among women, people with darker skin and people born in the southern region. There have been no studies at the state level or attempts to understand cross-state differences.

2. The results seem to indicate that intergenerational social mobility is lower in Mexico than in advanced countries such as Canada, the United States or other OECD countries (Connolly et al., 2019; Chetty et al., 2014a; Alesina et al., 2018). However, our measure of SES is not necessarily comparable with the indicators of income, earnings or occupational status used to estimate social mobility in those countries.

3. Total poverty (using not only a basket of food but also expenses for clothing and transportation) follows the same pattern as extreme poverty.

4. It is difficult to construct a harmonized series for inequality, because the survey has changed over time. We present here two different measures, one from Campos-Vazquez and Lustig (2017) that refers to labour income inequality at the individual level, and the other from Coneval (2019), which refers to inequality in total income computed at the household level.

5. Although there are other methods to calculate an index of economic well-being, such as count measures, polychoric PCA, and predicted income, the traditional PCA has the clear advantage of simplicity. Moreover, there is no clear superiority of other methods for the creation of such an index (Poirier et al., 2020).

6. The weights in each survey are applied to the final sample. As each survey represents the same population (adults aged 25–64), each survey will account for approximately half that population. INEGI’s sample account for 51.9% and EMOVI’s sample for 48.1%. Because the results between the two surveys are very similar, the results are unaffected if we multiply the weights by the share of observations of each survey.

7. In contrast to the present study, some other studies on intergenerational mobility use direct measures of wealth. For example, Clark and Cummins (2014), in a long-term analysis of five generations in England and Wales, estimate mobility by using estimated wealth at death for a sample of 19,000 individuals who died between 1858 and 2012. Adermon et al. (2018), in a multigenerational mobility analysis in Sweden, collected information on personal wealth and inheritances from official administrative records. In a study covering three generations in the United States, Pfeffer and Killewald (2018) estimate net wealth as the sum of financial assets, real assets and home equity, less financial obligations, with wealth information obtained from the Panel Study of Income Dynamics (PSID). Finally, in a two-generation analysis in 20th-century France, Garbinti and Savignac (2020) obtained detailed information on wealth from the French Wealth Survey, using information on home ownership and real estate. Their results thus relate more to financial and non-financial wealth rather than using it as a proxy, as in the present study.

8. It is important to recall here that the SEI, over which the percentile ranks are computed, includes schooling as one of its components. Given that educational achievement in Mexico varies significantly within and across generations with age and sex, in some of the specifications of the model we control for these variables to obtain estimates of the social mobility indicators that are free from demographic bias. The results from these specifications (see Figures A11 and A12 in the supplemental data online) indicate that the estimates reported in this section, in which we do not control for age and sex, are upper bounds of the extent of social mobility. The ranking of states according to their degree of mobility remains mostly unchanged under all the specifications estimated. In other words, our main estimates are conservative because, after we control for age and gender, we find that social mobility could be even lower.

9. From 1990 to 2016, the southern states did not see an increase in their per capita GDP. In fact, Campeche, Chiapas and Tabasco have a lower real per capita GDP today than in 1990 (economic activity in Campeche and Tabasco relies mostly on oil production, which has been declining in recent years). It is thus likely that individuals in the rest of the country benefited from growth, while those in the south did not: their relative standard of living worsened.

10. We implement three robustness checks: median regression, an estimation for subsamples according to the socioeconomic rank of parents, and excluding migrants. The results are similar (see Figures A7–A10 in the supplemental data online).

11. The authors thank an anonymous reviewer for suggesting this extension of the analysis.

12. Table A6 in the supplemental data online shows the results for all regions and ventiles.

13. The observations from the southern and northern regions in our data represent 24.5% and 15.4% of the total number of observations, respectively. Thus, the pattern found here is not driven by the relative size of the data samples.

14. For the results of rank–rank regressions using the rank position in the regional distribution of economic status, see Table A7 in the supplemental data online. The main changes relative to the estimates of social mobility presented in Table 1 are the following. First, the estimated degree of intergenerational persistence changes more in the south than in the rest of the regions (0.57 versus 0.63). Second, this indicator is now only slightly smaller in regions with a large degree of persistence of inequality. Thus, the reduction observed in the estimate for the south means that intergenerational persistence now varies less between the regions of the country. Third, the use of regional SEI distributions also reduces the range of estimates of upward mobility across regions, but this last result is difficult to interpret since measures of upward mobility as defined in this paper are no longer comparable when they are computed from different SEI distributions (Chetty et al., 2014b).

15. Table A8 in the supplemental data online shows the Pearson and Spearman (rank) correlations between each variable and our indicators of social mobility.

16. Figure A14 in the supplemental data online shows that the ‘Great Gatsby Curve’, as described by Corak (2013) and Chetty et al. (2014b), is also found in Mexico.

Additional information

Funding

This study was prepared with the financial assistance of the European Commission through a research programme on inequalities in developing and emerging countries, coordinated by the Agence Française de Développement (AFD). It is part of the project ‘Shedding Light on the Political Economic Barriers to Fighting Inequality in Mexico’ [project number 60678] at El Colegio de México and the Centro de Estudios Espinosa Yglesias.