Global Tax Expenditures Database (GTED)

The GTED collects all publicly available data on tax expenditures (TEs) published by national governments worldwide from 1990 onwards, covering a total of 218 jurisdictions. Based on a step-by-step search process, 121 jurisdictions are currently classified as <em>Non-reporting Jurisdictions.</em> The remaining 97 ones do provide some type of TE data, which was gathered by the GTED team. Wherever available, the GTED gathers revenue forgone estimates and number of beneficiaries of individual TE provisions. It also gathers metadata including the definition of the TE provision, its legal basis and duration. Each record in the GTED is classified in four main categories: Tax Base, Policy Objective, Beneficiaries and Type of TE used. In some cases, second- or third-level categories have been introduced. For instance, <em>Fuel Tax</em> data is categorised at the third level within <em>Tax Base: Taxes on Good and Services Excise Taxes Fuel Tax</em>. If the information for a record is not available or unclear, the respective category is classified as <em>Not stated/unclear</em>. When governments do not publish provision-level data but rather some kind of aggregated information, the GTED gathers this aggregate data. Likewise, if governments report on specific areas of TE only (such as tax incentives for investments, or TEs on income taxes) the GTED presents data on these areas alone. The terms <em>TE reporting</em> or <em>TE report</em> are used broadly, and refer to a large variety of public documents, ranging from annual, comprehensive reports on TEs that are part of governmental budget documentation to individual documents issued by a public body and providing some aggregate information on some specific TE mechanisms. As a minimum requirement, reports must contain some kind of information on the actual use of TE provisions. For instance, a list of available tax deductions for investments, provided by a governmental investment promotion agency, would not be considered a TE report unless they provide revenue forgone estimates or any other data that would allow users of the GTED to obtain information about the actual use of the respective TEs. The GTED distinguishes <em>regular</em> and <em>irregular</em> reporting. A sequence of reports from 1995 to 2005 would not be considered regular reporting in the GTED, since the country had reported on a yearly basis, but not anymore. Likewise, <em>regular</em> is not necessarily related to annual reporting. Germany, for instance, publishes federal subsidy reports including TE data every two years since 1967. A total of 15 such reports have been issued since 1990, containing data on 29 budget years (until 2018). The GTED counts this as 29 years reported, because data is provided on a year-by-year basis and can be consulted and analysed as such. The data is processed in a consistent format seeking to increase the level of longitudinal and cross-country comparability. Whereas revenue forgone estimates are provided as reported by governments (in local currency units, current prices), the GTED also provides figures converted into US dollars as well as indicators providing the revenue forgone through TE provisions as shares both of <em>GDP</em> and <em>Tax Revenue</em> – to compute these two indicators, data from the UNU-WIDER Government Revenue Dataset is used as input. The share of revenue forgone as a percentage of Tax Revenue is computed using figures of total tax revenue collected by countries’ central governments. The share of revenue forgone as a percentage of Tax Revenue is computed using figures of total tax revenue collected by countries’ central governments. Besides all the effort put into ensuring comparability, cross-country analysis of TE data needs to be done cautiously. The main issue, which is inherent to TE data, regards <em>benchmarking</em>. TEs are defined as departures from – usually country-specific – normal tax structures or benchmarks. On this note, the GTED uses the data published by official governmental institutions, sticking to their own definitions of benchmarks, without trying to complement official figures or challenge what different countries consider as the standard tax system or the benchmark. When it comes to the methodology used by governments to compute the fiscal cost of TE provisions, the vast majority of countries report on TEs based on the <em>revenue forgone approach</em> that estimates the amount by which taxpayers have their tax liabilities reduced as a result of a TE based on their actual current economic behaviour. Since the revenue forgone methodology is static, the potential interconnections between different TE provisions are not taken into account when computing the fiscal cost of TEs based on it. Hence, aggregating revenue forgone estimates of the individual provisions computed separately and without taking behavioural changes into account would not result in a figure that represents the total cost of all TEs. While providing users of the database with the opportunity to draw comparisons across countries or country groups, we want to be clear that any such comparison should be mindful of different levels of reporting, differences in national benchmark systems and methodological shortcomings of revenue forgone estimations. Country Income Groups and Regional Classifications are based on the latest World Bank classifications.全球税收支出数据集（Global Tax Expenditures Database，简称GTED）收录了1990年以来全球各国政府公开发布的所有税收支出（Tax Expenditures，简称TEs）相关数据，涵盖共计218个管辖区。通过分步检索流程，目前已有121个管辖区被归类为<em>未申报管辖区</em>。剩余97个管辖区均提供了一定类型的税收支出数据，均由GTED团队采集。只要可行，GTED都会收集各项税收支出条款的收入损失估算值与受益人数数据。此外，该数据集还会采集相关元数据，包括税收支出条款的定义、法律依据与实施期限。 GTED中的每条记录均分为四大核心类别：税基（Tax Base）、政策目标（Policy Objective）、受益对象（Beneficiaries）与税收支出类型（Type of TE）。部分场景下还增设了二级或三级分类。例如，<em>燃料税</em>数据被归类于<em>税基：货物与服务税→消费税→燃料税</em>三级分类下。若某条记录的相关信息缺失或模糊，则对应类别将被归类为<em>未说明/模糊不清</em>。 若政府未发布条款级别的数据，仅提供某种汇总信息，GTED也会采集此类汇总数据。同理，若政府仅披露特定领域的税收支出信息（例如投资税收优惠或所得税相关税收支出），GTED也仅收录该领域的相关数据。<em>税收支出申报</em>或<em>税收支出申报文件</em>的表述范围广泛，涵盖各类公开文件：从作为政府预算文件组成部分的年度综合税收支出报告，到公共机构发布的、针对特定税收支出机制提供汇总信息的单独文件均属此类。 作为最低要求，申报文件必须包含与税收支出条款实际使用情况相关的信息。例如，由政府投资促进机构发布的投资类税收减免清单，若未提供收入损失估算值或其他可让GTED用户获取对应税收支出实际使用情况的数据，则不被视为合格的税收支出申报文件。 GTED区分<em>定期申报</em>与<em>不定期申报</em>。例如，某国在1995年至2005年间每年发布申报，但后续不再发布，则该时间段的申报序列不会被GTED认定为定期申报。同理，<em>定期申报</em>并不必然等同于年度申报。以德国为例，该国自1967年起每两年发布一次包含税收支出数据的联邦补贴报告，自1990年以来已累计发布15份此类报告，涵盖29个预算年度（截至2018年）。GTED将此计为29个已报告年度，因为数据按年度维度提供，可按年度进行查阅与分析。 数据集采用统一格式进行处理，旨在提升纵向与跨国可比性。尽管收入损失估算值均按政府公布的本地货币（当期价格）呈现，但GTED同时提供换算为美元的估算值，以及以<em>国内生产总值（Gross Domestic Product，简称GDP）</em>和<em>税收收入（Tax Revenue）</em>占比形式呈现的税收支出收入损失指标——计算这两项指标时，输入数据取自UNU-WIDER政府收入数据集（UNU-WIDER Government Revenue Dataset）。 税收收入占比的计算采用各国中央政府征收的总税收收入数据。税收收入占比的计算采用各国中央政府征收的总税收收入数据。 尽管已尽力提升数据可比性，但对税收支出数据进行跨国分析仍需谨慎。税收支出数据固有的核心问题在于<em>基准对标（benchmarking）</em>。税收支出被定义为偏离（通常为各国特有的）标准税制或基准的条款。据此，GTED仅采用官方政府机构发布的数据，严格遵循各国自身的基准定义，不会尝试补充官方数据或质疑不同国家认定的标准税制或基准。 关于各国政府计算税收支出条款财政成本的方法，绝大多数国家采用<em>收入损失法（revenue forgone approach）</em>：该方法基于纳税人当前的实际经济行为，估算其因税收支出而减少的纳税义务。由于收入损失法属于静态方法，在计算税收支出的财政成本时，未考虑不同税收支出条款之间的潜在相互影响。因此，对单独计算且未考虑行为变化的各项税收支出条款的收入损失估算值进行加总，无法得出覆盖所有税收支出的总财政成本。 尽管本数据库可为用户提供跨国或国家群组间的比较机会，但我们需明确：任何此类比较都应考虑到申报水平的差异、各国基准体系的差异以及收入损失估算方法的固有缺陷。国家收入组别与区域分类均基于世界银行最新发布的分类标准。