Informe Nº: 20/07/2017
The destruction of the national statistical system was extremely grave and its reconstruction will require long time and lot of effort. During the transition, the data produced by INDEC has limitations. A specific example is the large share of people who do not report their income in the Permanent Household Survey. The high and heterogeneous number of incomplete surveys requires extreme care in the analysis of important and sensitive issues such as the incidence of poverty.
The latest data on poverty published by INDEC showed that approximately one-third of the urban population in Argentina is poor. More recently, the statistic agency made available the databases with which the general public can make their own calculations in order to reproduce INDEC’s numbers and analyze the different dimensions of poverty. Traditionally this allowed researchers to analyze characteristics of poor households with more disaggregation and formulate evidence based conclusions and public policy advice.
However, during the transition to the INDEC reconstruction, one of the main uses of this databases is to look for the inconsistences that arise from official statistics. Examples are that the incidence of poverty –according to INDEC– in the north of the country (traditionally poorer) is similar to that of Rosario; or that Córdoba has much more poor people than Rosario, when they historically had similar characteristics.
Because poverty is measured in terms of declared income, a key point is the quality with which INDEC collects income data. Processing INDEC databases, it can be seen that:
These examples illustrate the very different criteria that it is being applied by the provinces when collecting information. Averaging all the localities that INDEC collects information in, it can be seen that 1 out of 4 registries do not have income date and the dispersion between cities is extremely high. This cast doubts on the reliability of the information and the indicators that are extracted from it, limiting the comparisons among cities and localities.
It is highly likely that this under-declaration of income is one of the factors that explain the inconsistency of the indicators. For example, Córdoba’s lower-income families have a total income (ARS$ 8,750) similar to Jujuy’s (ARS$ 8,650) and lower than Formosa’s (ARS$ 9.960) and the average size of low income families (4.46 members per household) is much greater than that of Rosario (4.04 members per household). Problems are aggravated when analyzing specific groups –for example the incidence of poverty in children– as the quality of information worsens when working with one part of the survey.
There is no reason to suspect that the origin of these distortions is the intentional manipulation of statistics, as was the case with the previous government. The opposite, it is highly likely that it is a result of the institutional degradation generated by that situation. Appling household surveys is a very complex process that requires the joint effort of INDEC (setting the collection tools) and the Provinces (executing the field work with their own statistical agencies). The high variability of non-collection of family income data between localities suggest that the criteria applied by the statistical agencies of the provinces are far from homogeneous, given the many and varied difficulties that arise when the questionnaire is applied on terrain. This introduces biases in the samples that negatively impact the quality of the data.
INDEC faces the enormous challenge of rebuilding the official statistical system. It may not be fair to request high quality in such a short term for such complex surveys that also entail the coordination with 24 jurisdictions. Unfortunately, in the short term, only approximate information can be obtained, and it is necessary to appeal to the user´s responsibility in order to avoid the use of partial information that leads to misleading conclusions.