Multidimensional Poverty and Natural Disasters in Argentina (1970–2010)

ABSTRACT

This paper studies the effect of multiple natural disasters occurred in the different districts of Argentina between 1970 and 2010 on their multidimensional poverty, as measured by a Multidimensional Poverty Index (MPI) comprising five core dimensions of well-being: housing, basic services, standard of living, education and employment. The paper uses household microdata of the last census, and natural disasters registry from the DesInventar database. We find that natural disasters significantly increase the MPI and, while the magnitude of the impact found is moderate, effects are persistent, especially in dimensions related to infrastructure and long-term investments. We find that on average, extensive disasters are more harmful than intensive ones, although the latter do have significant impacts on sanitation infrastructure. We also find that hydrological disasters are the ones with significant impact. Finally, natural disasters have greater effect on the poorer districts of the country, corresponding to the northeast region.

Keywords:

• Multidimensional poverty
• MPI
• natural disasters
• Argentina
• regional disparities

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1 Argentina’s regional disparities in terms of multidimensional poverty are extensively studied in González and Santos (2020).

2 For language ease, we use the term district to refer to the Argentinean administrative unit called “departamento” or “partido”, which in some but not all cases coincide with the municipal level of disaggregation.

3 The Great Buenos Aires area includes the City of Buenos Aires and 24 departments surrounding the city, which belong to the Buenos Aires Province (known as Conurbano Bonaerense).

4 This methodological difference reflects a different concept of risk and disaster between the two databases. While in EMDAT the defining feature is the natural phenomenon itself and not the conditions that make the phenomenon to cause damage, in DesInventar, a disaster is a manifestation of a risk -understood as a social construction- which makes the natural event to generate losses (LA RED, 2002).

5 Santos and Villatoro’s (2018) proposed MPI includes an income indicator. However, as census data do not contain information about household income, this MPI only considers non-monetary dimensions. This MPI does not include an indicator on access to social security either, as this information is not available in the census data.

6 For example, if a certain disaster (of type $e$, occurred in district $d$ in year $t$) caused damage in at least one school, then $\mathrm{I}\mathrm{I}\left(L_{edtz}^i>0\right)$, for $i=schools$.

7 Note that the proportion of constructed area also works as a proxy for distinguishing urban areas from rural ones, in absence of a variable that contains that information in the datasource.

8 Admittedly, it would have been relevant to include the initial MPI level in 1970 as a control. In this paper we have only worked with census data of 2010. In future work we intend to expand this analysis into a panel, including estimates for previous census waves.

9 We also checked for multicollinearity computing the inflation factor of variance for each variable and found no evidence of this problem either, as all the variance inflation factors are below 10.

10 As mentioned in Section 3.3, Table A2 in the Appendix reports the Hausman test of this regression indicating that there are not endogeneity issues.

11 As a robustness exercise, we changed the separating threshold for intensive-extensive disasters, considering intensive disasters those that caused at least 8 deaths or 3 damaged homes under one alternative, and considering those that caused at least 4 deaths or 1 damaged home as intensive disasters under another alternative. We still find that only extensive disasters appear to have a significant effect on MPI in districts of Argentina.

Additional information

Funding

This work was supported by ANPCyT-PICT 2015-2079.

Notes on contributors

Fernando Antonio Ignacio González

Fernando Antonio Ignacio González is a Doctoral Fellow at the Instituto de Investigaciones Económicas y Sociales del Sur, (IIESS, UNS-CONICET). His main topics of interest include the economics of disasters and the analysis of regional disparities, with special emphasis on the case of Argentina.

Maria Emma Santos

Maria Emma Santos is an Assistant Professor at Dept. of Economics at Universidad Nacional del Sur and a CONICET Research Fellow at the Instituto de Investigaciones Económicas y Sociales del Sur, Bahia Blanca-Argentina. She is also a Research Associate to the Centro de Estudios para el Desarrollo Humano (CEDH) of Universidad de San Andres in Argentina, and to the Oxford Poverty and Human Development Initiative, at the University of Oxford, UK. Together with Sabina Alkire, she developed the Global Multidimensional Poverty Index, published in the Human Development Report since 2010. She works on measurement and analysis of multidimensional poverty.

Silvia London

Silvia London is Doctor in Economics and Principal Researcher of CONICET (National Scientific and Technical Research Council). She currently works at the Economics Department of the Universidad Nacional del Sur (UNS), Bahia Blanca, Argentina. She is the managing Director of the Instituto de Investigaciones Económicas y Sociales del Sur (IIESS, UNS-CONICET). Dr. London does research in Development Economics, Economy and Environment, and Economic Modelling.