About
Frequently asked questions - Government Revenue Dataset (GRD)


Q. What are the various options for accessing the GRD?

The GRD can be accessed in several ways. The GRD Explorer presents the dataset in a user-friendly, intuitive online tool that allows users to visualize or download custom selections of the data. The Explorer tool is the quickest and easiest way to access the data and allows users to download the data in both % of GDP and Local Currency (LCU). For users wishing to access to the entire dataset, this can be downloaded in Excel or Stata format from within the Explorer tool, or from here (we will ask you to fill in a short form first).

Q. What type of data (time series, panel, etc.) does the GRD provide?

The GRD provides cross-country data on various revenue and tax aggregates. The researcher determines the type of data to be downloaded from the database. For example, if there is interest at looking at tax revenue mobilization in a specific country (e.g. Mozambique), then data on that specific country is obtainable, and so forth.

Q. Should analysis employ data from the merged, general government or central government dataset?

It will depend on the nature of the research question. For studying total revenue collection most users are likely to use the main, or merged dataset, which employs general government data, where available, while inserting central government data if there is evidence that subnational revenue collection is limited. By contrast, relying exclusively on central government data will dramatically underestimate total revenue collection in fiscally decentralized states, while general government data is only available for a limited group of countries. While relying on the merged dataset implies a modest underestimation of revenue collection in countries for which general government data is not available, this will generally be consistent over time, and preferable to the major risks introduced by relying exclusively on central or general government data.

Q. Should analysis include or exclude social contributions from total revenue and total tax revenue?

This is highly dependent on the research question. Social contributions are not the same as other taxes, as they are contributions toward a specific area of public spending. but they nonetheless represent an important area of government involvement. Most users are however advised to rely on revenue and tax figures exclusive of social contributions, owing to problems of completeness and comparability for social contributions figures. Specifically, many countries do not consistently report social contributions, leading to more extensive missing data. What is reported in the ‘social contributions’ category does not appear to be entirely consistent across countries, particularly for developing countries. When relying on IMF Article IV reports, including social contributions in the analysis could produce misleading results.

Q. If I am interested in resource revenues, should I use the “total resource revenue” variable, or an alternative strategy to measure resource dependence?

Relatively few countries report a figure for “total resource revenue”, with some additional countries identifying the resource components of total tax revenue or of total non-tax revenue, but not both. In the absence of such data, users may nonetheless wish to include a proxy for resource wealth in econometric analyses. In these cases, the best option available to most users is likely to be to calculate “total non-tax revenue” as the difference between “total revenue” and “total non-resource tax revenue”. This variable will thus include all types of non-tax revenue, including resource revenue, but acts as a useful proxy for resource revenue because resource revenue explains most of the variation in “total non-tax revenue” across countries.

Q. Why do the subcomponents of tax revenue not sum to the respective totals?

Total tax revenue is generally equal to the sum of the sub-components of tax revenue. However, the total tax figure exceeds the sum of the sub-components in cases where some revenue was not allocated to any specific category in the underlying source, which may occur for a variety of reasons. For example, Total Taxes on Income, Profits and Capital Gains is sometimes larger than the sum of CIT + PIT, as some sources allocate some revenue to “other income taxes”, which cannot be allocated to either personal or corporate income taxes. Very occasionally, the subcomponents of revenues are slightly greater than or smaller than the reported totals in the underlying source. The GRD, as a rule, does not modify data from underlying sources, thus retaining these minor discrepancies to ensure transparency. Users can correct such data for their own analyses, though changes of this kind should be carefully recorded for any analysis.

For most EU countries, the data in the GRD comes from the OECD Revenue Statistics. The total tax figure in such cases is inclusive of ‘Customs duties collected for the EU’, reflecting the ‘Total Tax Revenue’ figure in the OECD. However, this item is not allocated to any subcomponent, meaning that the sum of subcomponents will often be marginally lower than the total tax figure.

Q. Why does revenue data in the GRD not match that in other sources?

The GRD expresses all data in terms of a common underlying GDP figure, taken from the World Economic Outlook (WEO). This helps to ensure comparability of data across different sources, but might be higher, or lower, than tax ratios expressed in other sources, based on different GDP data. The underlying GDP figures will reflect the rebasing of GDP or recent adoption of more modern systems of national accounting. In some cases, this leads large changes in GDP, which might cause users to find lower tax ratios than an older source that does not express tax figures in terms of up-to-date GDP figures. Tax statistics in other sources may be inflated by revenues earned from natural resources. The GRD, where possible, presents both total revenue and total tax figures both inclusive and exclusive of resource revenues.

Q. Is it possible to estimate total subnational revenue by subtracting Central Government revenue from General Government revenue?

The differences between General and Central Government revenue should be equal to total subnational revenue. However, any such calculations should be undertaken with extreme caution, and checked carefully, as definitional differences, or other unusual accounting practices, between central and general government sources could lead to over or under-estimating subnational revenue.

Q. Data that I am interested in using has been flagged as potentially problematic. What does that mean?

We aim to alert users to cases where data might be potentially unreliable or misleading. The flags are defined in each data file, and often notes relating to the need for the individual flags are included. Users are advised to carefully consider the inclusion of any data that is flagged as Caution 1 – Caution 4 from their econometric or cross - country analysis, depending on the variables required. In cases where results do not hold following the exclusion of such data, users are recommended to communicate this to their audience and/or exclude the data all together. Because of well-known imperfections with the collection of government revenue data – and with GDP data used in some countries to calculate tax revenues – best practice for econometric analysis should always be to pursue extensive robustness testing, including sensitivity to the inclusion of data from individual, or small groups, of countries. Caution should also always be employed when comparing tax ratios across countries, as high, or rapidly rising, tax ratios in individual countries may reflect the underestimation and irregular rebasing of GDP.

Q. What are some of the limitations to the GRD?

There are widely recognised concerns over the quality of government revenue data in existing sources. The GRD strives to present the ‘best’ available data consistently and to flag any noticeably problematic or questionable data. This already represents an improvement over any other source. However, the broader accuracy of the numbers contained in the GRD is only as good as the underlying data obtained from each source. In particular, data on natural resource revenues have not historically been collected consistently, or according to strictly enforced definitions. As such, the disaggregation between non-resource taxes and resource taxes is likely to be imperfect at the margin – though still better than the alternative of not making any distinction.

No data on resource revenues is generally available for countries with small levels of resource revenue (generally less than 1% of GDP) and, as such, all revenue is treated as non-resource tax revenue in these cases. There are also potential inconsistencies across countries, or over time, in the disaggregation of tax revenue between taxes on international trade and taxes on goods and services. This reflects potential inconsistencies in the classification of taxes on goods and services collected at the border, by customs officials. While every effort has been made to ensure consistency, users should be aware of this potential issue in individual countries when working with these disaggregated categories.

There may be cases where the data in the GRD appears slightly less complete (in terms of disaggregation) than that in, for example, the Government Finance Statistics (GFS). This would be as a result of using IMF Article IV data that, whilst not as detailed, allows the resource component of tax to be isolated (see Technical Note 1 for more information).

Q. Does GRD provide data in averages?

GRD does not produce data in averages. It only produces annual data. It is the choice of the analyst how the data should be aggregated (averages per country, averages across regions, averages across income groups etc).