A New Toolkit on Tax Expenditure Estimation by the Asian Development Bank (ADB)

By Sandeep Bhattacharya & Selcuk Caner

This blog was first published in Development Asia.

A Guide on Estimating the Revenue Losses from Various Tax Concessions

Tax systems are designed to collect revenues to finance public expenditures in the most efficient manner. Since taxes constitute a price on economic activity, the tax structure should aim to collect taxes fairly and efficiently. In addition to raising revenues for financing public expenditures, governments use the tax system to pursue social and economic objectives. These objectives may include correcting market failures, increasing employment, improving income distribution, promoting economic growth, attracting investments, and lowering the cost of tax administration.

For example, instead of making direct payments to low-income households, the government may choose to exempt their income from personal taxes, provide a tax credit, allow for an expense deduction, offer preferential tax rates, or defer tax liability. Similarly, the government may exempt a portion of income allocated to savings accounts to increase the overall savings rate or stimulate investments that might not occur otherwise. In such cases, the tax system forgoes a part of the tax revenues that would have been collected. This set of policy measures is referred to as tax expenditures. They are expenditures undertaken instead of direct expenditures to achieve social and economic policies.

While all tax incentives are also typically tax expenditures, not all tax expenditures are meant to incentivize economic behavior.

Tax expenditures encompass benefits provided to specific population groups or sectors through preferential tax treatments in the tax system. Tax incentives are designed to alter production and consumption behavior, with the anticipation that the revenues forgone will ultimately be recovered from increased tax revenues generated by future economic activities stimulated by the incentives.

Tax expenditures can be costly (the global average over the 1990-2020 period is 3.8 percent of GDP and 23 percent of tax revenue). While high-income countries use tax expenditures to pursue social and welfare policies benefiting households, low-income countries employ tax incentives to stimulate growth and attract investment.

To assist developing member countries (DMCs) in the APAC region in properly accounting for and reporting the revenue foregone from tax concessions, the Asian Development Bank (ADB) has published a toolkit on tax expenditure estimation. This toolkit provides a step-by-step guide for policy makers and their advisors on estimating the revenue losses from various tax concessions. Additionally, it includes references and hyperlinks to further readings and studies of relevance, making it a must-read for government officials, members of parliament, and many others.

This article serves as a summary of this toolkit, which aims to provide more informed debate on the size and efficacy of fiscal policies.

A necessary condition for estimating tax expenditures is the definition of the benchmark against which they should be measured. Ideally, the benchmark should reflect the broadest possible tax base. However, many countries opt for a benchmark that closely aligns with their actual tax system, despite the possibility of adopting a conceptual system. Adopting a universal benchmark system would facilitate a uniform international comparison of tax expenditures and help assess fiscal policy measures aimed at supporting industry or a section of the population. However, in practice, this is not done, and benchmark tax systems vary by country, making international comparisons of tax expenditures challenging.

Estimation Methodologies
Several methods and models are employed in estimating the revenues forgone resulting from a tax expenditure item that benefits the taxpayer. However, it should be noted that the revenue forgone method is static and does not consider the revenue impact of intertemporal benefits, nor does it account for the interactions of several tax expenditures simultaneously and the resulting behavioral changes.

Tax expenditures can be grouped into two broad categories: those on income and those on goods and services. Methods and models for estimating tax expenditures vary, including simulation models based on aggregated taxpayer data, indirect methods using data from external data sources, such as national accounts or surveys, and micro-data obtained from individual taxpayer sources. These models calculate revenues forgone by determining the difference between the estimated tax liability with and without the tax expenditures.

The ADB’s tax expenditure estimation toolkit provides sample models for estimating the revenue cost estimation of tax expenditures in the following areas: 1) personal income taxes (PIT), 2) corporate income taxes (CIT), 3) value-added taxes, and 4) customs duties and excises.

Personal income taxes
Personal income tax microsimulation models, based on individual taxpayer data, include information on income and deductions obtained from individual tax returns. These models provide the revenue impact of all tax expenditures related to individual taxpayer’s income. Furthermore, they enable the user to estimate the distributional impact of tax expenditures. The taxpayer data can be further expanded using data from household income and expenditure surveys for the average household, allowing for estimates not otherwise possible, such as those for households that do not file tax returns. In countries with substantial number of tax expenditures for income support, the model should be expanded to capture all tax expenditures and households.

Corporate income taxes
The toolkit includes two types of corporate income tax models that demonstrate the use of aggregate and microdata in estimating tax expenditures. The CIT microsimulation model uses taxpayer data from corporate tax returns, providing an outcome that reveals the distributional impact of tax expenditures by size and industry. Consequently, it becomes possible to determine whether the targeted taxpayers indeed receive the intended benefits. In countries where tax expenditures aim to stimulate production and investment, detailed corporate-level data is necessary for accurately estimating the revenues forgone. Additionally, alongside a microsimulation model, a marginal effective tax rate model (METR) can be employed to determine the impact of certain incentives on investments, such as accelerated depreciation, that only affect the timing of the tax liability without altering the overall amount of tax payable.

Value added taxes
Tax expenditures on goods and services are mainly provided within the VAT system. The toolkit includes a VAT model that mimics the mechanics of the VAT system using an input-output table. The model allows for the estimation of revenues forgone resulting from preferential tax rates and exemptions by type of industry.

Custom duties and excises
The other type of tax expenditure model is the customs and excises model. This model utilizes customs data classified according to the harmonized system of nomenclature (HSN) codes. Since customs data collected through electronic IT systems normally includes all exemptions and preferential tax rates, it is easy to use.

In developing their own models of tax expenditures, the modeling effort invested by DMCs should focus on the significance of the tax expenditures. For example, in cases where most of the tax expenditures granted in a DMC are under the PIT, a detailed model of PIT would be necessary. In other countries where tax expenditures on goods and services constitute a large share of the total tax expenditures, comprehensive models of VAT and excises would be more useful in estimating the revenue and distributional impacts of the tax benefits.

Resource: Asian Development Bank. 2023. Tax Expenditure Estimation Tool Kit. Manila.


About the authors:

Sandeep Bhattacharya
Senior Public Management Specialist (Tax), Public Sector Management and Governance Sector Office, Sectors Group, Asian Development Bank.

Sandeep has more than 28 years of experience in tax policy and administration, consulting, and academia. Prior to joining ADB, he taught classes in taxation, public economics, statistics, and econometrics, as well as supervised student research at Duke University. He has a PhD in Economics from Georgia State University and has degrees from Duke University (Master of Public Policy), Delhi School of Economics (MA in Economics), and St. Stephen’s College, Delhi University (BA Honors in Economics).

Selcuk Caner

Selcuk Caner has more than 25 years of expertise in economic, fiscal, trade, and financial policy gained through collaborations with prominent organizations such as the International Monetary Fund, Asian Development Bank, US Treasury, and KPMG. He holds a PhD in Economics from the State University of New York, an MSc in Applied Mathematics and a BSc in Business Administration from the Middle East Technical University.