Lucila Berniell
Economista Principal, CAF -banco de desarrollo de América Latina y el Caribe-
The most recent Economy and Development Report (EDR 2018) by CAF shows that the sluggish economic growth of Latin American countries is due to the low productivity of their economies. This puts the productivity issue at the heart of the debate about how to boost economic and social development of the countries in our region.
Regardless of the strategy used to boost productivity, either from the private sector or public policies, it is becoming increasingly clear we need to use a large data flow for successful implementation. For example, as part of the diagnoses used to design interventions aimed at reducing barriers to financing or to access to other resources for innovation, there is a need to survey which companies are most affected and how to reach them through different targeting criteria. In addition, good data describing business activity is vital in any effort to assess policies that aim to boost productivity.
Traditional statistical sources for productivity and employment analysis include economic censuses and different types of surveys that aim to characterize production, investment and innovation decisions made by businesses. However, these sources are not without their problems: They are expensive, not always have the coverage or granularity required for certain analyses, and their updating frequency is low or irregular. In addition, building longitudinal data (which allow us to see the full “movie” and not just a snapshot of businesses) based on censuses and surveys is not always possible. These limitations have rendered traditional data sources irrelevant, in favor of data that are cheaper, have greater coverage and allow longitudinal monitoring of production units: administrative data with information on production and employment. Furthermore, the advent of e-governments has given momentum to this trend.
Administrative data or records generated by governments on an ongoing basis and as a result of theirinternal management processes that support their relationship with citizens and taxpayers, offer a great opportunity to meet an important portion of the growing demand for information on productivity issues at low cost. These data include, for example, those generated in social security systems, which infer creation and loss of registered (formal) jobs, as well as qualification levels required by different economic sectors. They also include foreign trade data, which help analyze the potential for businesses to tap into international markets. There are also transactional data from tax authorities, which help infer aspects related to billing (and therefore, production of goods and services), as well as the way in which different companies relate to one another as suppliers and buyers of supplies. In addition, records generated in financial transactions (in institutions regulated by public authorities) can help understand the extent to which companies use different lending tools.
Challenges in using administrative data for productivity analysis
Despite the great potential of administrative data for the design, implementation, monitoring and assessment of policies to boost productivity, these data have some limitations, as they do not come from sources designed under statistical criteria, but from other administrative processes. In other words, while data from surveys are built with methodologies to reflect accurately a particular economic or statistical concept, administrative records are built on the footprints left by processes used to support transactions or legal proceedings involving some public agency and companies and individuals in the private sector. Therefore, before using them for statistical purposes, they must be purged and specially treated, which often depends on the analytical or public policy objectives pursued.
Another major challenge in leveraging administrative data for productivity analysis in different Latin American countries is that public agencies must reach institutional agreements to exchange records, which may enrich the flow and quality of the information that describes business dynamics. In particular, these agreements should condition and regulate the circumstances (data protection and security) under which agencies holding administrative records allow access to tax-related data for other third party agencies (e.g. academia). For these agreements to be reached and implemented, the custodian of records and statistics producers need to have a legal, technological and capacity-building infrastructure for statistical analysis that ensure that the administrative data will receive an appropriate statistical usage.
These topics and other details associated with the need to protect data quality (due analysis of internal consistency, cross-checking variables from different sources or over time, or externally, comparing data from administrative sources with those from traditional sources), are addressed in detail in a paper recently published as part of the Working Documents Series of the Socioeconomic Research Directorate of CAF. The guidelines established in the research serve as a practical guide to reliably and securely expand the statistical use of administrative data, which will help delve into productivity analysis in Latin American countries.
