In Latin America, of every 10 dollars of possible collection, 5.7 are paid; 2 are “forgiven” (tax expenditures); and the rest, 2.3 dollars, is lost, mainly due to evasion and management shortcomings.

Increasingly we have more and better data sources to analyze the tax reality of Latin America. In a new CIAT Working Paper (DT-05-2017: El Impuesto sobre el Valor Agregado: Recaudación, Eficiencia, Gastos Tributarios e Ineficiencias en América Latina), our intention is to illustrate the possible benefits of the joint use of the information provided through CiatData regarding the Collection, Tax Rates and Tax Expenditures of the countries of the region. For this we have been analyzing the tax collection and its possible determinants. Later we analyzed the efficiency in this collection by means of the indicators of productivity and C-efficiency and, finally, we have introduced a new methodology that completes this analysis with the use of the information regarding the tax expenditures. The joint consideration of: C-efficiency (the collected) + G-inefficiency (what is not collected by Tax Expenditures) + X-inefficiency (loss of collection for fraud and management, plus the net result of errors in the measurement of the previous components), provides us with a holistic framework for the analysis of taxation that can contribute to the improvement of our understanding of the collection figures.

If the figures obtained demonstrate something, it is the diversity of VAT situations in Latin America. Whatever the indicator we analyze (collection from GDP, rates, efficiency, tax expenditure, inefficiencies …) the most extreme differences between countries will reach 300 or 400 per cent.

On the other hand, it is important to note that the relative wealth level of each country (measured by its per capita GDP corrected by price differences) is not significant in explaining the regulatory tax options adopted (tax rates, tax expenditures), the levels of collection reached, or the varying degrees of efficiency/inefficiency.

The collection obtained by VAT (in relation to the GDP of each country) is mainly explained by the political will specified in the choice of higher rates and lower tax costs, independent of the per capita income level and consistent with the overall election of a higher level of fiscal pressure on the whole of the taxes. And a “curious” result is the high correlation of this election with latitude. The further south, the higher the revenue collected, the higher the tax rates and the lower the tax expenditures (although it does not exist, and it is worth reiterating it, a significant relationship in the same sense with GDP per capita).

In terms of efficiency/inefficiency in VAT collection, C-efficiency figures, G-inefficiency and X-inefficiency are not explained, again, by the level of GDP per capita or by the level of rates. With respect to latitude, although the correlation is never as significant as in the case of the collection, the data -once we isolate the extreme cases- point in the same direction in terms of the sign of the estimated coefficients: the farther south the greater efficiency in the collection (C-efficiency) and the lesser the loss of resources by tax expenditure (G-inefficiency); while X-inefficiency does not keep any relation to this dimension (nor, apparently, with labor market informality).

With respect to the results, in average, in the selected countries:

• 56.9% of the potential VAT is collected (C-efficiency, with maximum values in Paraguay -91.1- and minimum in Mexico -31.5-),
• Due to tax expenditures, reduced rates or special regimes granting exemptions, 19.9% of the potential collection is sacrificed (the so-called “G-inefficiency“, with maximums in Panama -62.6– and minimum in Chile  -5.1-),
• The remainder of this lost potential collection, 23.2%, we call it X-inefficiency -its maximum value in Mexico with 57.9 and minimum in Bolivia with 11%, although three other countries (Paraguay, Ecuador and Panama) even reach negative values of this parameter-, attributable to tax fraud, management distortions and errors of quantification of the components directly measured (C-efficiency and G-inefficiency)

The proposed methodology (C-efficiency + G-inefficiency + X-inefficiency) is theoretically consistent and easy to apply, both to identify possible areas of improvement in VAT collection and to improve, where appropriate, the sources of information regarding collection, tax expenditure or National Accounting figures.

In short, as we have seen, it is not easy to generalize simple explanatory answers to the differences in collection, choice of tax rates and tax expenditures and levels of efficiency in the management. This is an evidence of our ignorance, but perhaps also of an incipient process to overcome it by rejecting simplistic responses without empirical support. Much remains to be learned as to the specific details of the management and the regulations of each country, the improvement of the databases and the influence of factors not strictly economic but political, cultural and social in the tax policy decisions. We hope that the new CIAT databases would contribute to keep progressing in this direction.

727 total views, 1 views today