The siena microsimulation model (SM2) for net-gross conversion of EU-silc income variables

Microsimulation models are widely used as an integral part of the policy-making process in tax and social policy areas, especially in the US, Canada, the UK and in several Northern European countries (Martini & Trivellato, 1997; Atkinson, 2005). over the past three decades, microsimulation has moved from a description of the distributional impact of the existing tax and transfer systems to a more complex tool for assessing the different impacts of alternatives proposal for changing existing systems. Nevertheless, most microsimulation models are what may be termed as ‘static’. Such models are used to measure the immediate impact of policy changes concerning the tax-benefit system, without taking into account longer-term changes in the composition, characteristics and behavioural relationships of the population. In such models, changing the rules of eligibility or liability produces outcomes showing the gains or losses from the policy change in a given situation. By contrast, ‘dynamic’ models are used to compare the effects of alternatives policies in the medium to long term, such as studying the evolution of retirement systems. They aim to analyse the development of the socio-demographic structure of the population. Such models age the original unit records on the basis of probabilities of different real life events (e.g. birth, death, marriage, separation, unemployment, retirement).

Reviews of static models may be found in Atkinson and Sutherland (1988) and Sutherland (1995), and of dynamic models in Harding (1990) and Zaidi and Rake (2001). For a comprehensive list of the main microsimulation models, see the website of the International Microsimulation Association (http://www.microsimulation.org).

The choice in using a static or dynamic microsimulation model largely depends on the institutional context and also on the quality and suitability of micro data (Mitton et al., 2000). Well-known examples of static tax-benefit models include: TAXBEN, realized by the Institute for Fiscal Studies of UK; STINMOD, a microsimulation model of Australia’s income tax and transfer system built up by the National Centre for Social and Economic Modelling; TRIM (Transfer Income Model), a comprehensive microsimulation model at the Urban Institute in Washington DC, USA; the Canadian SPSD/M (Social Policy Simulation Database and Model) developed by Statistics Canada for evaluating the financial interactions of governments and individuals; and Euromod (Euromod, 2001; Sutherland et al., 2008), representing an integrated multi-country model for the European Union.

Dynamic microsimulation models have been largely developed from 1990s. DYNASIM was the pioneering model for the US developed at the Urban Institute in Washington, D.C., with DYNASIM2 as its current version; another famous example is DYNAMOD implemented by NATSEM, with DYNAMOD-2 as the present working version. There is also the Canadian DYNACAN for public pension analysis; the UK model PENSIM for evaluating income security in old age, and the French DESTINIE for studying pension projections (Zaidi & Rake, 2001).

This paper concerns an aspect of purely static microsimulation involving the modelling of the income of private households and persons. Many static microsimulation models have been developed to simulate taxes, social insurance contributions, benefits and other transfers received to affect the transformation between gross and net forms of income, mediated through complexities of the national fiscal systems. In the EU, important examples are Euromod, already mentioned, and similar national microsimulation models. The primary objective of these models is to provide, on the basis of specific micro-datasets incorporated into the system, a comprehensive facility for simulation of the effect of varying parameters of the tax-benefit system on the income received by various segments of the population. Simulation of taxes and benefits under different regimes (fiscal policy options) forms the output of the system. Such models involving tax-benefit simulation require very detailed and standardised information on household and personal income by component. The logic of such modelling essentially requires household income components in the gross form as inputs which are used to produce corresponding net amounts under the assumed tax-benefit system. In practice, however, generally the required input data are not available in a homogeneous gross form, especially when they come from interview surveys. The required data transformation can of course be done on an ad hoc basis, but it is more efficient, convenient and comparable to a develop systematic procedures and tools for the purpose.

The Siena Microsimulation Model (SM2) described in this paper has been developed as a versatile tool for transforming detailed information on income collected from surveys or other sources into standardised forms required for diverse analyses, including tax-benefit simulation (Verma et al., 2003). The microsimulation involved in SM2 has certain special aspects. On the one hand, it is limited to a fixed tax-benefit regime – the one that actually exists, under which the available income amounts in different forms have been generated. On the other hand, it does not expect inputs in a homogeneous form but generate income amounts in both gross and net forms as outputs. Furthermore, at the outset, SM2 is designed for multi-country application, as a flexible tool which is portable to the maximum extent possible across (at least the European) countries despite great differences in fiscal systems. A distinguishing feature of SM2 is that it can handle diverse tax-benefit regimes using a common logic and a standard set of procedures making it particularly useful for multi-country comparative application.

The need for transformation between gross and net forms of income of course goes well beyond the specific context of microsimulation modelling. In broader terms, therefore, this paper is concerned with the following important practical problem in the collection and use of information on income of households and persons, in particular when it is obtained through personal interviewing.

Income of households is made up of diverse components received by different members. Its elements may be compiled from different types of sources, which may differ in concepts and definitions and may not refer to exactly the same reference time. The different sources may be subject to differing patterns of response and recording errors, sampling errors, inconsistencies and incompleteness etc. We are not concerned here with such conceptual and measurement issues, but with the following additional important problem.

Income can be recorded in various forms – such as gross, or net of taxes and/or other retentions at source, or as the final amounts actually received – differently for different components and for different income earners in the household. Aggregating these elements of income into the household’s total income and its main components requires not only that information is available on all the elements, but also that it exists in a homogeneous form to permit aggregation. The form must also be the same for all households so as to permit aggregation to the sample or the population. Furthermore, the same information in more than one form is often required to meet different analytical objectives. For instance, for poverty and social exclusion research it is necessary to have information on total household disposable income. Total disposable income means gross income less income tax, regular taxes on wealth, compulsory social insurance contributions, employers’ social insurance contributions, and inter household transfers paid. To study the effect on income distribution, the breakdown of disposable household income separating out old-age and survivors’ benefits and other social transfers is needed. On the other hand, for the study of redistributive effect of taxes, for microsimulation and many other research and policy purposes, information is also required on gross income and its detailed breakdown by component and individual income recipient.

Different forms of income (gross, disposable or net, etc.) are related through extremely complex rules of national fiscal systems, often with subnational variations as well. This complexity has many aspects. (i) The relationships or rules vary by income component and according to characteristics and circumstances of the income recipient, in great detail and with many special cases. Some components may be exempt from tax and other deductions, while others may be subject to either or both, fully or in part. (ii) The rules may apply to different types of units – to individual persons, whole households, or some other intermediate ‘tax units’ within households. (iii) Different aggregations across income components may be involved in the application of the rules: some components may be treated individually, while others pooled together. (iv) How income is received can vary: it may for instance be received after retentions at source, or received gross to be taxed later. (v) What form it is reported in may vary from one component and recipient to another in the same data set. (vi) Who receives income may vary: while most income is received by individuals, some parts (e.g. certain transfers) may actually pertain to the whole household. (vii) When the income is received or deductions made and the period to they which relate may differ. All this complexity is increased where individuals have a choice among alternative rule-sets to be applied to their particular case. We hardly need to mention discrepancies between rules and their actual application: individuals failing to claim benefits, other transfers and reimbursements to which they are entitled; or not paying taxes and deductions which are due.

The immediate context for the development of SM2 has been certain specific requirements of EU-SILC (EU Statistics on Income and Living Conditions). EU-SILC is a statistical source, developed by European Commission (Eurostat) and implemented by all EU and also many other European countries, for the generation of comparable and detailed information on living conditions and income of households and persons. The central issue to be addressed is that, while the source, type and form of input (collected) information varies across and even within countries, the output required at the European level has to be comparable and standardised (Eurostat, 2002). Furthermore, while the information which can be collected is limited to particular forms because of limitations of the sources providing it, it is required in both net and gross forms for diverse academic and policy research. We see SM2 as a tool, under continuing development, for meeting these objectives in the international, comparative context. Starting from data on household and personal income given in different forms (including some missing data), and on the basis of the prevailing fiscal system in a country, the model estimates full information on income by component, with breakdown of gross amounts into taxes, social insurance contributions, social transfers, and net and disposable income. Therefore it can be applied to diverse data sets to generate variables (such as the EU-SILC Target Variables) in a standard form. Furthermore, it is designed to be flexible to deal with an annual flux of data in different forms across and within countries and also with periodic changes in the national tax systems, which a longitudinal data source such as EU-SILC must deal with.

In order to meet these objectives, SM2 is designed such that its core consists of a standardised set of routines which can handle a great diversity of input data forms and national tax systems. Country-specific routines are required to convert the input data intox standardised forms, and also to specify parameters of the national tax system in an appropriately standardised form. These, then, form inputs to the central core of the system designed to generate the required standardised outputs. The system has been developed to maintain a clear distinction between the common and the country-specific parts, and even more importantly, to maximise the part which can be standardised. This feature makes the system an appropriate and convenient tool for multi-country application.

Given the specific context and objectives of its development, SM2 is fully ‘data based’. It is taken as given that information on all income components has been collected, compiled or imputed in some form, and that the objective is to convert it, under a specified national tax system applicable at the time, to the standard form. It incorporates generally the same or similar level of detail as other major microsimulation models – a little less detailed on some points but also more complete on some others.

The paper is structured as follows. Section 2 describes the basic relationships between different forms of income and presents the simple model for the gross-to-net conversion. This basic model is elaborated in the following sections by introducing complexities step-by-step. Section 3 describes the net-to-gross conversion through the iterative procedure on which SM2 is based. This also describes how the problems of local convergence and non-convergence are addressed. Section 4 introduces the additional complexity resulting from differences among tax regimes in how particular components of income are treated. An outstanding feature of SM2 is that these special features of the different tax systems can be captured within the general structure of the model simply by appropriately defining special types of ‘deductions’ and ‘tax credits’ for the component concerned. A number of examples are provided in Section 4.

Detailed applications have already been developed for France, Italy and Spain using European Community Household Panel (ECHP) data as the input (Eurostat, 2004) in order to test the country-specific routines. Recently, the National Statistical Service for Greece (2007) has applied the SM2 routines to the first wave of EU-SILC data for Greece. We have developed an equivalent EU-SILC based application for Italy. This provides the basis for a more fully-worked illustration of the application of SM2, which is described in Sections 5 and 6. Section 7 concludes by summarising the main points in the paper and expanding upon the potential utility of SM2 for wider application.

Figure 1

Download asset Open asset

Figure 1 shows the basic relationship between gross and net forms of income when more than one income component and possibly more than one individual in the tax unit are involved. The relationships between gross taxable income for a particular component, H_i and quantities like gross income G_i and income after retentions at source XST_i are usually simple, dependent only on i, the income component concerned and determined independently of other components and other persons in the tax unit. The same applies to the relationship between H_i and net N_i for components which are taxed separately at a flat rate or a rate determined by the level of income from that component alone, and of course also for tax exempt components. Sometimes, dependence of the relationship on other sources of income may also be involved, but mostly these are simply in the form of upper limits which may apply to certain quantities pooled over more than one component.

Generally, however, all or most taxable income is pooled together over components and over persons in the tax units for the purpose of determining the amount of tax due. The relationship between H_i and N_i for components in the pool is more complex than shown above. In any case, going from known H_i to N_i is less problematic since the relationships (the tax rules) are a function of the former. These relationships are specified in more detail in Table 1. Going from given N_i to H, required iterative solutions, and are described in the next section. Entries in Table 1 have the following interpretation. The last two columns of the table define the various income measures in terms of measures defined in the preceding rows; those in the first column concern total income, in the second they concern income components.

The form in which data on income by component are available may vary from one country (tax regime) to another, and also among individuals and households within the same country. There are two dimensions of this variation:

(A) Whether and how a particular component is subject to social insurance contributions and to income tax. Income tax may apply in various forms. (i) Some components may be pooled together, across components and also across individuals in some appropriately defined tax unit. (ii) Some may be subject to tax separately, each at a certain flat rate. (iii) Some components in the ‘pool’ may be tax-exempt up to a certain flat rate but taxed beyond that if a higher rate applies. (iv) Some may be subject to double taxation, perhaps representing some combination of the other forms.² (v) And of course, many types of incomes, in particular social transfers, may be tax exempt. Mostly, the form applicable to each type of income is determined by the national tax regime, normally uniform for all respondents in a country. Hence this information can be compiled at the aggregate level and need not be collected at the micro level. There can be exceptions, however, for persons in special circumstances. There can also be other complications, such as more than one components, otherwise treated separately, being subject to common ceilings. In some systems, individuals have a choice among the various options.

(B) The form in which the information has been collected. This may generally vary from one individual to another in the same survey, though a uniform reporting form for the whole sample may prevail for some components. In any case, the information on the form in which the data are available is required at the micro-level. The amount may for instance be reported as gross, or as net of social insurance contributions and/or tax. In the case of tax retentions, details may vary by, for example, whether they are ‘retentions at source’ according to some rules or depend on individual arrangements, and whether they are the ‘final retentions’ of the tax actually due.

Table 2 lists the various reporting forms.

In this section, we describe the standardised ‘core’ of the SM2 system, taking account of complexities of (B), but assuming for the moment that through the information may be reported in diverse forms, all income components over individuals in the tax unit are pooled together and subject to a common tax schedule. A remarkable feature of the system is that by appropriately defining certain ‘deductions’ and tax credits, much more complex variation can be incorporated into this standardised procedures; this will be explained in Section 4.

3.5 Conversion routines

Table 3 shows the procedure for converting the reported amount with any combinations of the above dimensions of variation into a standard form. As noted at the bottom of the table, the income components may be divided into two sets, N and H, depending on whether the amount reported is ‘final net’ (N_i), or is in some other form (G_i, XS_i, XT_i, XTS_i, H_i) more directly convertible to the ‘gross taxable’ form H_i. For all forms other than ‘final net’ N_i, it is convenient to take ‘gross taxable income’ H_i as the standard target of the conversion:

(3) $[G_i,H_i,X{S}_i,XT{S}_i,X{T}_i]\Rightarrow H_i.$

This conversion involves the component and country-specific functional relationships or schedules, namely S_i = S_i(G_i), for social insurance contributions, and T_i = T_i(H_i), for tax retention at source.

As noted, tax retentions at source may be according to fixed schedules, or according to arrangements determined at the individual (micro) level. In a majority of the cases, H_i can be determined directly from the collected amount, for instance from gross amount (G_i) reported for an income component i subject to social insurance contributions, we have: H_i = G_i − S_i(G_i). In other cases, an iterative procedure may be required. However, generally the iteration is very simple and converges quickly. This is because by and large component-specific schedules apply to each component separately. There are no other parameters to be estimated. The need for numerical iteration arises simply from the fact that the unknown quantity to be determined (H_i) appears in an implicit equation.

The second panel of Table 3 shows the relationship between H_i and the reported amount in the form ‘final net’ N_i. Going from N_i to H_i in fact involves a double iterative loop. The inner loop of iteration is applied with an assumed value of the parameter ‘tax rate’ (R, as defined in Table 1). Once this has been done for every income component in the group (including over all individuals in the same tax unit), an outer iterative loop obtains a convergent value of this parameter which is common to all those components. The N_i to H_i conversion process is therefore considerably more complex. Furthermore, this complexity is substantially increased in the presence of missing data, where the modelling and imputation procedures will have to be applied interactively.

Table 4

Download asset Open asset

A remarkable feature of SM2 is that by appropriately defining certain ‘special deductions’ and tax credits, many special features and complexities of different tax regimes can be incorporated into the standardised procedures described in the previous section without altering them in any way.

Deductions refer to the part of gross taxable income which is tax exempt. As noted, these deductions may be component-specific, or may be common deductions which apply to taxable income as a whole. Initial tax due is computed as a function of total net taxable income. This tax liability is normally reduced by tax credits. Again, these may be component-specific, or may be common credits which apply to the initial tax due as a whole. In addition to these ‘normal’ deductions and tax credits, we can define ‘special’ component-specific deductions and tax credits to accommodate variations in the form in which the component is taxed without altering any other aspect of model specification. Table 5 lists a number of such possibilities. In fact, it covers all the situations we have encountered in the diverse fiscal systems of various EU countries, notably those of France, Italy and Spain for which we have developed detailed applications of SM2.

Consider for instance the common situation with one component (such as family benefits) tax exempt, and the remaining components pooled together and subject to a common tax regime (row 1 of Table 5). By simply specifying ‘special deduction’ for the tax exempt component as D_i = H_i, i.e. the same as its gross taxable amount, we automatically retain its tax-exempt nature and it is no more necessary to separate it from rest of the pool. It makes no contribution to the total net taxable income, and its original gross taxable income appears automatically as a part of the final net income. Similarly, if a component is taxed at a flat rate (say f_i) separately from the pool (row 2), we can simply specify its ‘special deduction’ as D_i = H_i, and its ‘special tax credit’ as a negative quantity C_i = –f_i × H_i. It makes no contribution to the tax liability of the pool, but the final tax liability is automatically increased by the appropriate amount. Again, no other treatment separate from the pool is required for this component.

The situation in the case of a component tax-exempt at a flat rate is just the opposite (row 3). Deductions for expenses can be specified as common deductions applicable to the total income i. e. not associated with any specific income component (row 4), and similarly for tax credits for expenses (row 5). Sometimes components are subject to ‘double’ taxation. For example, in Italy self-employment income is liable to income tax as a part of the total income in the normal way, and also to an additional tax (‘IRAP’) at a rate depending only on the component concerned. This complexity is easily handled as shown in row 7.

The last case (row 8) is an important one, as it handles a special and complicating factor in the treatment of social insurance contributions, which themselves are subject to tax, as in France for instance. By specifying the taxable part of social insurance contributions as negative deductions from (i.e. in effect as additions to) gross taxable income (H_i = G_i − S_i), the net taxable income (the amount actually subject to tax) is augmented by the taxable part of the social insurance contribution, say ΔS_i: Y_i = H_i + ΔS_i. No further special treatment of this feature of the system is required in the model, despite the complexity and unusual nature of this feature.

By way of illustration, this section describes the use of the SM2 model to construction of EU-SILC target variables on income for elements of the Italian fiscal system, and implemented for the reference year 2003. Applications have also been developed for other countries. The essentials of the national systems in terms of main components of income, tax and social insurance contributions will be described below in Table 6 for Italy. Similar tables for Spain and France are provided in the appendix (Tables A.1 and A.2). There we also note briefly some features of these countries’ fiscal systems that appear particularly difficult to model in principle, but are handled in a relatively straightforward way within SM2 structure. See also comments in Section 4.

In Italy the new EU-SILC survey was conducted for the first time in 2004: 24,270 households and 61,542 individuals were interviewed. The new survey replaced the ECHP (European Community Households Panel) survey as the main reference source at EU level for comparative statistics on income distribution and social exclusion.

For the net-gross conversion of EU-SILC income target variables, ISTAT decided to test the application of the SM2 model using the new survey data and to experiment with some methodological innovations based on the ISTAT experience in using both administrative data and sample survey data. The development of the SM2-EU-SILC for Italy required through revision and update of the existing ECHP-based SM2, developed by Verma, Betti and co-researchers as reported in Eurostat (2004). For one thing, EU-SILC collects additional relevant variables, and the model had to be extended to incorporate them.

Particular attention has been paid in the construction of the tax units for the estimation of tax credits for dependent persons. The tax credits for dependent relatives establish a connection between the household members; as mentioned previously, two types of tax units can be found in a household – a ‘family unit’ and the ‘individual’. To identify these two types of tax units at the household level, the ‘family procedure’ used in ISTAT social surveys was applied. The procedure allows the construction of family relationships in the households the identification of dependent persons.

The Siena Microsimulation Model (SM2) has been developed to meet a very practical need in a systematic and efficient manner. In interview surveys collecting information on household and personal income, the respondents may report income amounts as gross or net of taxes and other deductions. The data must be made homogenous before use for analysis, especially comparisons across population groups and countries. SM2 provides a robust and convenient procedure for this purpose. Starting from data on household and personal income given in different forms, and on the basis of the prevailing tax regime in a country, the SAS routines of the model are designed to estimate full information on income by component, with a breakdown of gross amounts into taxes, social insurance contributions of various types, and net income.

There are at least three main ‘clients’ or potential users of this model.

First potential users are the more general microsimulation models. Such models are widely used as an integral part of the policy-making process in tax and social policy areas, and over the past three decades, microsimulation has moved from the description of the distributional impact of the existing tax and transfer systems to a more complex tool for assessing the different impact of alternatives proposal for changing existing systems. The primary objective of these models is to provide, on the basis of specific micro-datasets incorporated into the system, a comprehensive facility for simulation of the effect of varying parameters of the tax-benefit system on the disposable income received by various segments of the population. Simulation of taxes and benefits under different regimes (fiscal policy options) also forms an output of the system. Such models involving tax-benefit simulation require very detailed and standardised information on household and personal income by component. The logic of such modelling essentially requires household income components in the gross form as inputs which are used to produce corresponding net amounts under the assumed tax-benefit system as outputs. In practice, however, generally the required input data are not available in a homogeneous gross form, especially when they come from interview surveys. The required data transformation can of course be done on an ad hoc basis, but it is more efficient, convenient and comparable to develop systematic procedures and tools for the purpose.

Consequently, given its specific purpose of estimating full information on income by component in both gross and net forms from empirical data, SM2 is not meant to be an alternative to general tax-benefit simulation models, but as a complementary tool which those models can usefully exploit. Of course SM2 itself is a microsimulation model. However, the microsimulation involved in SM2 has certain special aspects, which make SM2 somewhat unique. On the one hand, it is limited to a fixed tax-benefit regime – the one that actually exists, under which the available income amounts in different forms have been generated. On the other hand, it does not expect inputs in a homogeneous form but generates income amounts in both gross and net forms as outputs.

The need for gross-net transformation of course goes well beyond the specific context of microsimulation modelling. The second important clients of the model are Eurostat and the countries participating in EU Statistics on Income and Living Conditions (EU-SILC) project. As noted, the immediate context for the development of SM2 has been certain specific requirements of EU-SILC. The need arises from the fact that, while the source, type and form of input information varies across and even within countries, the output required at the European level has to be comparable and standardised. The information which can be collected is limited to particular forms because of limitations of the sources providing it. For instance in some countries, especially those with income registers, all income components are available in the gross forms; by contrast in many other countries, in particular in southern Europe, generally net received amounts only can be provided by the survey respondents. However, the information is required as a standard set of ‘target variables’ involving both gross and net forms from all countries participating in EU-SILC. SM2 is specifically designed to meet this requirement. Furthermore, the model has flexibility to deal with an annual flux of data in different forms across and within countries and also with periodic changes in the national tax systems, which a longitudinal data source such as EU-SILC must deal with. Hitherto, SM2 has been implemented for constructing EU-SILC target variables by National Statistical Institutes of Greece and, as described in this paper, Italy; moreover, Portugal (Rodrigues, 2007) and Spain have utilised procedures on the lines of SM2 (in the sense of using aspects of the SM2 logic and structure but developing country-specific software for implementation). Earlier, implementations using ECHP data were developed for Italy, France and Spain. It can be expected that, with the EU countries having to deliver to Eurostat microdata on household income components in the standard gross form (which became mandatory from 2007, EU-SILC wave 4), the need and opportunities for the use of SM2 are likely to expand further. Moreover, EU-SILC data are becoming available to a very wide body of researchers, including those engaged in tax-benefit simulations. Some of those may wish to use SM2 independently to generate more complete series of household income components in net and gross forms.

Hence the third important client of SM2 is international comparative social research, whether for policy or academic purposes. Diverse academic and policy research may require income components in both net and gross forms in greater detail. As emphasised in this paper, SM2 is designed at the outset for multi-country application, as a flexible tool which is portable to the maximum extent possible across countries despite great differences in fiscal systems. As noted in Section 4, a distinguishing feature of SM2 is that it can handle diverse tax-benefit regimes using a common logic and a standard set of procedures making it particularly useful for multicountry comparative application.⁶ Reflecting this, the system is designed such that its core consists of a standardised set of routines which can handle a great diversity of input data forms and national tax systems. Country-specific routines are required to specify parameters of the national tax system in an appropriately standardised form; they also standardise the input data format. These, then, form inputs to the central core of the system designed to generate the required standardised outputs. The system has been developed to maintain a clear distinction between the common and the country-specific parts, and even more importantly, to maximise the part which can be standardised. This feature makes the system an appropriate and convenient tool for multi-country application.

As an illustration of all these three aspects appearing in conjunction, let us consider in conclusion the potential use of SM2 for Euromod already referred to. At European level, the increasing need for cross-country comparison has led to the development of Euromod, an integrated multi-country model for the European Union implemented by a consortium of teams in 15 countries (currently being extended to 4 of the New Member States) for policy simulation. Euromod is a static tax-benefit microsimulation model that provides estimates of the distributional effects of national or European level changes in social and fiscal policy. EU-SILC now forms the reference source for Euromod at European level. Before the introduction of this reference, microdata from different sources were used, requiring a lot of efforts for harmonising national sources and for transforming available data into the required model database. One of the main advantage of using EU-SILC is that it potentially provides the microdata for policy simulation for all EU and closely related countries, and does so with increased comparability across countries (Figari et al., 2007).⁷ In order to maintain and even enhance this comparability, it is highly desirable that uniform net-to-gross conversion procedures are followed, and if possible to even avoid country-specific development of net-to gross conversion procedures. Clearly, the development and implementation of SM2 for countries where necessary would help in the harmonisation of the conversion procedures, with advantages for national EU-SILC data bases, for tax-benefit modelling such as in the context of Euromod, and for the wider requirements of inter-country comparative research. An added but not-always-obvious practical advantage is that often it would be easier for individual countries to adapt and complete SM2 application than to develop their country-specific net-to-gross conversion procedures ad hoc.

1. 1
   Euromod and the Development of EU Social Policy. DIW Discussion Paper 467

   1. AB Atkinson

   (2005)

   Berlin: DIW.

   • Google Scholar
2. 2
   Tax Benefits Models. STICERD Occasional Paper, No. 10

   1. AB Atkinson
   2. H Sutherland

   (1988)

   London: London School of Economics.

   • Google Scholar
3. 3
   Statistical Imputation in Conjunction with Microsimulation of Income Data

   1. G Betti
   2. V Verma
   3. F Ballini

   et al. (2003)

   Rivista Italiana di Economia, Demografia e Statistica 58:35–43.

   • Google Scholar
4. 4
   Euromod: an integrated European benefits-tax model. Final Report, Euromod Working Paper EM9/01

   1. Euromod

   (2001)

   University of Essex: ISER.

   • Google Scholar
5. 5
   Commission Regulation concerning EU-SILC as regards the list of primary target variables. EU-SILC 107/02

   1. Eurostat

   (2002)

   Luxembourg: European Commission.

   • Google Scholar
6. 6
   Euromod Country Report Italy (2001 Tax-Benefit System)

   1. Euromod

   (2004)

   Euromod Country Report Italy (2001 Tax-Benefit System), September, http://www.iser.essex.ac.uk/files/msu/emod/documentation/countries/italy/cr01it0904.pdf.

   • Google Scholar
7. 7
   Income in EU-SILC: Net-Gross-Net Conversion; Common Structure of the Model; Model Description; and Application to ECHP Data for France, Italy and Spain

   1. Eurostat

   (2004)

   EU-SILC 133/04, Working Group on Statistics on Income and Living Conditions (EU-SILC).

   • Google Scholar
8. 8
   Using the EU-SILC for policy simulation: prospects, some limitations and some suggestions

   1. F Figari
   2. H Levy
   3. H Sutherland

   (2007)

   Comparative EU statistics on Income and Living Conditions: Issues and Challenges – Proceedings of the EU-SILC conference. pp. 345–373.

   • Google Scholar
9. 9
   Dynamic microsimulation models: problems and prospects, Discussion paper

   1. A Harding

   (1990)

   London: STICERD, London School of Economics.

   • Google Scholar
10. 10
    The Role of Survey Data in Microsimulation Models for Social Policy Analysis

    1. A Martini
    2. U Trivellato

    (1997)

    LABOUR 11:83–112.

    • Google Scholar
11. 11
    Microsimulation Modelling for Policy Analysis. Challenges and Innovations

    1. L Mitton
    2. H Sutherland
    3. M Weeks

    (editors) (2000)

    Cambridge: Cambridge University Press.

    • Google Scholar
12. 12
    Statistics on income and living conditions model: net gross income conversion, Report to Eurostat

    1. National Statistical Service of Greece

    (2007)

    Pireus: NSSG.

    • Google Scholar
13. 13
    Income in EU-SILC – Net/Gross Techniques for Building and Using EU-SILC Databases

    1. CF Rodrigues

    (2007)

    Comparative EU Statistics on Income and Living Conditions: Issues and Challenges. Proceedings of the EU-SILC conference.

    • Google Scholar
14. 14
    Improving the Capacity and Usability of Euromod – Final Report. Euromod Working Paper EM4/08

    1. H Sutherland
    2. F Figari
    3. O Lelkes

    et al. (2008)

    University of Essex: ISER.

    • Google Scholar
15. 15
    Static microsimulation models in Europe: a survey. Discussion Paper

    1. H Sutherland

    (1995)

    University of Cambridge.

    • Google Scholar
16. 16
    Personal income in the gross and net forms: applications of the Siena Microsimulation Model (SM2)

    1. V Verma
    2. G Betti
    3. F Ballini

    et al. (2003)

    Universita degli Studi di Padova: Dipartimento di Scienze Statistiche.

    • Google Scholar
17. 17
    Dynamic Microsimulation Models: A Review and Some Lessons for SAGE. SAGE Discussion Paper no. 2, SAGEDP/02

    1. A Zaidi
    2. K Rake

    (2001)

    London School of Economics.

    • Google Scholar

Author details

1. Gianni Betti

   University of Siena, Italy

   For correspondence

   betti2@unisi.it

2. Gabriella Donatiello

   ISTAT, Italy

   For correspondence

   donatiel@istat.it

3. Vijay Verma

   University of Siena, Italy

   For correspondence

   verma@unisi.it