Basic Income in Ireland

Basic Income in Ireland:

A Study for the Working Group on Basic Income

FINAL REPORT

Tim Callan

Brian Nolan

John Walsh

James McBride

Richard Nestor

6 June 2000

9

This study draws extensively on the 1994 Wave of the Living in Ireland Survey, the Irish element of the European Community Household Panel. Brendan Whelan and James Williams of the ESRI’s Survey Unit were responsible for the survey design, data collection and database creation.

9

Chapter 1

Introduction

1.1 Background to the Study

Partnership 2000 for Inclusion, Employment and Competitiveness gave a commitment that

A further independent appraisal of the concept of, and full implications of introducing a basic income payment for all citizens will be undertaken, taking into account the work of the ESRI, CORI, and the Expert Working Group on the Integration of Tax and Social Welfare and the international research.

In line with this commitment, a Working Group on Basic Income was established, which agreed terms of reference for a study on basic income, divided into two phases:

1. An evaluation of the cost and distributional implications of the introduction of a basic income scheme similar to that proposed by CORI in Pathways to a Basic Income (Clark and Healy, 1997).
2. An examination of the dynamic effects of such a system from a broad economic and social perspective.

This report represents a part of the first phase of this overall study. It deals with the cost and distributional implications of the introduction of a basic income scheme. This phase of the study has been further subdivided into two main elements:

(a)  An aggregate analysis of the costing of the basic income scheme, using macroeconomic statistics, and an analysis of the distributional effects on illustrative households, along the lines of Clark and Healy, 1997.

(b)  A microsimulation-based analysis of the costs and distributional impact of a basic income on actual families, using a tax-benefit model based on relevant survey data.

This report deals centrally with the second set of analyses. It is structured as follows.

Chapter 2 describes the basic approach underlying SWITCH, the ESRI tax-benefit model, and describes the adjustments made to the model’s database to ensure full representation of the income tax base – a key factor in costing the basic income proposal. In order that this microsimulation analysis can be compared with that undertaken using the alternative aggregate costing and illustrative households (item (a) above), a great deal of work has gone into

·  careful specification of the baseline demographic and economic scenarios underlying the analysis

·  specification of baseline policies under the conventional tax and social welfare systems

·  specification of the precise parameters of the basic income scheme to be examined, and

·  specification of the framework for the analysis.

This work is described in Chapter 3 of the present report. Chapter 4 describes the costing of the basic income proposal, and shows how the tax rate required to finance it has been derived. The sensitivity of the basic income tax rate to alternative assumptions on the tax base, potential savings on current income supports, and additional items of expenditure associated with the basic income proposal is considered. Chapter 5 outlines alternative perspectives on the distributional impact, including an assessment of the differential impact on men and women. Chapter 6 looks at the impact on poverty as measured by relative income poverty lines, and under an alternative measurement approach (linked to the definition in the National Anti Poverty Strategy) using information on which households were measured as suffering “basic deprivation”. The main findings are drawn together in the concluding chapter.

1.2 Evaluating Proposals for a Basic Income

There is widespread agreement on some of the problem areas within the current income tax and social welfare structures. Transitions from welfare to work can be impeded by the fact that for some individuals, the net financial reward from taking up employment may be small. For some other individuals, receiving income support under the Family Income Supplement scheme, increases in hours of work or in pay rates may lead to a very limited increase in disposable income – even after the recent change to a net income basis of assessment for FIS.

A key issue for policy is whether policy changes operating broadly within the existing structure of the income tax and social welfare codes, or a radical reform of the tax/transfer system known as basic income, offer a better structure in which choices about income maintenance and income taxation can be made in the future. This is an issue which has attracted considerable international interest (see, for example, Atkinson, 1995; Brittan and Webb, 1990; Gelauff and Graafland, 1994; and van Parijs, 1992). In the Irish context, earlier studies of this issue include Honohan, 1987; Callan, O’Donoghue and O’Neill, 1994; Ward, 1994; Integrating Tax and Social Welfare by the Expert Working Group on the Integration of the Income Tax and Social Welfare Systems, 1996; and Clark and Healy, 1997. The latter study contains a proposal which (suitably amended to deal with changes in the most recent budget) forms part of the terms of reference for the current study.

Our main report focuses on four key issues concerning the introduction of a basic income:

1. What would the introduction of a basic income scheme cost, and what taxation provisions would be necessary to finance it?
2. Who would gain and who would lose from the introduction of the basic income scheme, and by how much?
3. What would be the impact on the extent and depth of income poverty?
4. What would be the impact of the proposal on the distribution of income as between men and women?

No attempt is made at this stage to take into account behavioural responses to alternative reforms. This phase of the study concentrates instead on “cash” or “first-round” effects, which will form a basis on which phase 2 of the study – dealing explicitly with dynamic effects – can build.

9

Chapter 2

The Microsimulation Approach

2.1 Introduction

Much of the debate on tax and welfare reforms tends to focus on the effects of tax and social welfare policy changes on a small number of supposedly "typical" families. While this approach can help to understand the nature of a policy change, it can also be highly misleading. The most commonly analysed "typical" family at Budget time is a one-earner couple, with 2 children, taxed under PAYE. Less than 1 family in 20 actually falls into this category, and those who do differ widely in terms of income, housing tenure and other characteristics relevant to their social welfare entitlements and income tax liabilities.

Concentration on the effects of a policy change on a small number of hypothetical households cannot provide an overall picture of the gains and losses associated with complex reform packages; and by concentrating on a small number of supposedly "typical" families may lead to the neglect of effects which are important for significant groups.

Microsimulation models simulate the tax and benefit position of a large-scale sample of families, using micro-level data on individual and family incomes and other characteristics. These microsimulation models have a number of advantages. A tax-benefit model based on a large-scale representative sample of the population automatically takes account of the wide diversity of circumstances in the population; can help to identify the overall pattern of gains and losses; and can help to assess the impact of policy changes on financial incentives to work.

The usefulness of microsimulation models in analysing tax and social security policy has been amply demonstrated by such international experience. Taxbenefit models have been constructed for most OECD countries, with the US and the UK having a particularly rich experience in their construction and use[1]. In many instances, models of this type are the only way in which accurate costing of complex changes to taxes and benefits can be derived. But the more fundamental advantage of such models is that they permit a representative picture to be constructed of the overall effects of a policy change, from which it is possible to identify the characteristics of gainers and losers from a policy change, the overall impact of a change on the distribution of income, and the impact on financial incentives to work.

2.2 The SWITCH Model

The tax-benefit model used in the present report is the latest version of SWITCH, the ESRI tax-benefit model. This model was first constructed in 1989, using data from the 1987 Survey of Income Distribution, Poverty and Usage of State Services. It has been tested, developed and used continuously since then in a wide range of analyses. A new version was constructed in more recent years, based on data from the 1994 Living in Ireland Survey. This model has now been specially extended and developed to deal with the range of microsimulation analyses required for the present report. Here we outline the key features of the model which are relevant to the current analysis.[2]

SWITCH combines two key ingredients: detailed micro-level data on the incomes, family composition and labour market participation of a large scale nationally representative survey of households; and a set of computer programs which model, or “simulate”, the social welfare entitlements and income tax liabilities of these households . The 1994 Living in Ireland Survey provides the information required by the model. It contains detailed information on more than 4,000 households with individual interviews covering more than 10,000 adults. The Survey forms part of an EU-wide household panel study managed by EUROSTAT. The first wave of this panel study was undertaken in the latter half of 1994, and the Irish element included many additional questions on incomes, labour market status and receipt of social welfare payments designed to provide a suitable base for the construction of a tax-benefit model. The model-based analysis is not, however, restricted to analysis of the situation in 1994: several procedures have been developed which bring the model database into line with recent developments in employment, unemployment, rates of pay and the size and structure of the population: these are described in Section 2.3 and in Chapter 3.

Overall the Living in Ireland Survey provides a good representation of the population in terms of age, marital status and sex. It also provides a nationally representative picture in terms of the key issues of distribution of employment and unemployment across households and individuals. (Callan, Nolan, Whelan, Whelan and Williams, 1996). Coverage of the income tax base is also quite good, and special adjustments to improve this coverage are described in the following section. The survey- and model-based estimates of the social welfare population show good representation of the broad groups of schemes, and most of the major schemes. While model-based estimates of expenditure on some individual schemes (particularly the smaller schemes) can be higher or lower than actual expenditure, the overall coverage and cost estimates are much closer. Thus, the model database appears to be more than adequate for the estimation of the cost and broad distributional consequences of a basic income scheme.

The structure of the computer routines used to simulate present income tax and social welfare policies is described in Callan, Richardson and Walsh (1997). One underlying assumption is that there is full compliance by taxpayers in respect of tax liabilities arising from income reported to the survey, and full take-up of social welfare entitlements. There are, however, two exceptions to this rule.

First, model-based analysis (Callan, Richardson and Walsh, 1997) suggests that the actual rate of take-up of Family Income Supplement is rather low. Thus an alternative approach is to assign take-up of FIS in a way which ensures that the rate of take-up approximates the latest estimates (about one in three of caseload and expenditure). This alternative – and more realistic - approach, based on a low rate of take-up, is used throughout the present study.

Secondly, there is a divergence between the concept of farm income used in the survey (“family farm income” as used in the National Farm Survey) and income for tax purposes. A precise adjustment for this is not possible; but an approximate adjustment has been made which ensures that the model’s simulation of the tax revenue from farm incomes is brought into line with the actual tax revenue generated from that source. This adjustment suggests that farm incomes as estimated for the survey must be reduced by over half to ensure that the predicted tax take from farm incomes is close to the actual tax take.

2.3 Calibration of the SWITCH Model

The total income tax base is a key element in the costing of the proposed basic income. Furthermore, the distribution of that total income tax base can be of critical importance in determining the distributional effects of changes to tax and social welfare policy, including radical changes such as the introduction of a basic income. For this reason our analysis has paid particular attention to comparisons between estimates of the tax base and its distribution from the SWITCH model and statistical information supplied by the Revenue Commissioners.

In principle, a random sample of households, in which each household had an equal chance of selection, and non-response was randomly distributed, would be “self-weighting” i.e., each household would have an equal probability of selection (e.g., a chance of 1 in 250) and estimates of national totals could be produced from survey totals by simply multiplying the relevant survey totals by 250. Many samples – including the Living in Ireland Survey – are not of this type.[3] The sampling frame for the Living in Ireland Survey is based on the Electoral Register – a list of persons. This gives a higher probability of selection to households containing several adults of voting age. In order to adjust for this factor, a lower weight must be attributed to households with several adults, and an above average weight to households with fewer individuals. Differential weights can also help to ensure that a survey is representative in the face of differential non response rates e.g., as between urban and rural areas.