Executive summary
Recent years have seen countries increasingly use their budgeting frameworks to achieve broader social and economic outcomes. Such cross-cutting challenges affecting various groups in society mean that an understanding of the underlying distributive implications of budgets is critical to ensuring that expenditure can be targeted and mobilised in the most effective way to achieve economic and social goals simultaneously in a context of severe fiscal constraints.
While the current report does not seek to analyse income inequality as such, it does try to address its implications from a public expenditure standpoint. As inequality has been increasing, many countries have been experimenting with budgeting approaches to address the implications of such an increase for public expenditure. This is also particularly important at a time when countries are considering moving away from untargeted fiscal support to ensure that expenditure is as effectively focused as possible.
As the evidence points to a greater role of transfers than taxes in impacting disposable income inequality, there is a compelling need to understand the distributional implications of public expenditure. This report reviews how distributional considerations are incorporated into the public results-based budgeting frameworks of eight countries, namely Canada, France, Ireland, Italy, Korea, Netherlands, New Zealand and Sweden. The report offers a general overview followed by in-depth case studies. These offer a brief overview of country-specific trends in inequality, before examining how distributional considerations are integrated into budgeting systems, and what tools and data resources are used to do so.
In many of the countries in the sample, the budget office in the Ministry of Finance (or equivalent), is responsible for distributional impact analysis work. However, whether this involves budget offices carrying out the analytical work themselves or co-ordinating the analysis of other units within government varies. In many cases, several organisations carry out supplementary distributional impact analysis. While in some cases this analysis is conducted in tandem with the budget process, in other cases it is done on a more ad-hoc basis.
Many of the countries carry out their work at multiple stages during the budget process. At the beginning, this involves estimating the impacts of proposed policies to aid in decisions on budget allocations. Some countries also provide a formal statement of the budget’s redistributive implications. This is often complemented by ex post evaluative measures, ad-hoc studies on significant policy measures, and independent analyses conducted by Parliamentary research services, statistical or other research institutions.
Countries employ two types of approach to addressing distributional concerns in the budget process – use of microsimulation models and use of results-based budgeting frameworks (and in many cases, both). While all almost all countries use microsimulation models to consider distributional issues, the extent to which these models are used to inform the budget process varies, ranging from being the basis of any distributional impact analysis conducted to being more ad-hoc. In a few cases, macroeconomic and labour modelling is used alongside microsimulation modelling to examine second-round effects, particularly in terms of labour supply. Ownership of microsimulation models also varies – in some countries, their development and management lie with a statistical institution or independent analytical body, while in others different models are owned and managed by different ministries.
Countries that use results-based budgeting frameworks as the basis for conducting distributional impact analysis tend to take a multidimensional approach to this analysis, highlighting the fact the inequality can take many forms beyond income. In most cases, these frameworks serve as an aid to those formulating policy proposals for the budget, allowing them to ensure that the predicted impacts of their proposals are in line with the overall aims of the government. The indicators used for the frameworks tend to be developed by the countries’ statistics institution. While many of these are specific to the country in question and thus vary, most frameworks include measures on income inequality, education, sex and gender, health and wellbeing.
The underlying data used for distributional impact analysis is generally a mix of tax and expenditure administrative data, as well as survey data. Tax and expenditure administrative data is collected from across government and combined when technical circumstances allow, while survey data tends to be collected by the country’s statistics institution. Both types of data are mobilised to address the complex policy issues at hand. The extent to which these data, and the models they are used for, are available to the public varies – in some countries they tend to be completely publicly available, in others only certain portions are available, and in others still they are only available to those within government. The extent of data disaggregation also differs – while all countries tend to disaggregate data by different income segments, gender and age, only a few provide data on other social characteristics such as race, disability and sexual orientation. The capacity to access and link data across surveys and administrative data is a prerequisite for sophisticated modelling approaches.
The seven best practices below are drawn from the experiences of the case study countries and are further elaborated in this report. They can assist countries in improving their capacity to address distributional issues in government spending:
1. Conduct distributional impact analysis as early as possible to inform the choice of spending decisions and policy options.
2. Encourage integration of distributional impact analysis or of broader considerations of inequality into the budget process.
3. Ensure transparency in the distributional impact analysis process and its underlying data to maintain confidence in spending decisions.
4. Maintain independence in the development of analytical models.
5. Ensure that results-based budgeting frameworks and microsimulations models are complementary and promote co-ordinated approaches.
6. Complement microsimulation modelling approaches with economic models that help to take into account the effects on labour supply.
7. Make full use of administrative data as a complement to survey data to inform distributional analysis and disaggregate data by socioeconomic characteristics as fully as is possible while ensuring data confidentiality.