Skip to Main Content
Tilasto-oppaat
Tilastokeskuksen etusivulle

Remote Access to Research Data

2 SISU microsimulation model

2.1 Background and purpose of the SISU model

Statistics Finland develops and maintains the SISU microsimulation model, which describes Finland’s personal tax and social security systems. The model is used to calculate the effects of potential legal reforms of personal taxation and social security for the population and public finances.

SISU models illustrate the combined effects of legal reforms

Microsimulation models have been used in Finland for a long time in the preparation of social security benefit and income tax legislation. These models can use individual-level sample data representative of the entire population to calculate the combined effects of legal reforms for both different types of households and the population as a whole. These models are used to evaluate public sector tax revenue, the financial positions of citizens and households, income disparities and the effects of incentives, for example.

The SISU microsimulation model has been developed by Statistics Finland since 2011 in close cooperation with the Research Unit of the Social Insurance Institution of Finland (Kela). Multiple experts from the Ministry of Finance, the Ministry of Social Affairs and Health, the Finnish Institute for Health and Welfare (THL) and the VATT Institute for Economic Research have also been involved in its development.

The SISU model has taken Finnish microsimulation to a new level in usability and calculation accuracy. The accuracy of its calculations has been improved by collecting more representative base data about the population.

2.2 Approved uses of the SISU model

The SISU model has a wide range of uses. You can use it for the following purposes:

  • to plan tax and social security legislation
  • to evaluate and monitor the effects of tax and social security legislation
  • for research and studies of tax and social security developments.

The SISU model is used by ministries, research institutes, universities, political parties and interest groups. You can find research and surveys produced with the model on the Statistics Finland website (only in Finnish).

The SISU model is a static model that is suitable for evaluating the immediate effects of different policy options. Static models disregard any dynamic effects and long-term changes in behaviour, such as labour supply, resulting from legislative reforms.

Microsimulation models are the only tools that can effectively manage the combined effects of the benefits and tax systems. Taxation and social security benefits are based on intricate legislation. Furthermore, social security benefits often overlap and tie into each other, making the income transfer system difficult to manage. Microsimulation models take into account all connections between benefits and taxation.

You can use the SISU model to evaluate:

  • past changes in the system
  • the current situation
  • the future effects of planned legislation reforms.

Finland’s tax and social security legislation has been modelled in the SISU model retroactively for several years, which allows the changes made to the tax and income transfer systems to be analysed. Legislation is planned ahead for several years. For future analysis, the model includes updated data that account for society’s structural changes and the development of income and expenses. Furthermore, the model has been prepared for future development by including index forecasts from the Ministry of Finance for indexed social security benefits.

You can also use the SISU model for data simulation and simulated calculations, which will be described next.

How to use data simulation

In data simulation, you can make calculations based on individual-level sample data representative of the entire population. You can extrapolate the simulation’s results for a wider portion of the population or the entire population. The data include all necessary variables to calculate social security benefits and taxes (income, family relations, etc.). You can calculate social security benefits and taxes for the persons included in the data according to the calculation model’s settings.

For example, you can use data simulation to investigate:

  • the effects of legal reforms on income distribution in the population and the financial standing of different population groups
  • the effects of legal reforms on public finances
  • the effects of economic incentives (income traps, unemployment traps, effective marginal tax rates, etc.)
  • the extent of benefits’ underutilisation.

How to use simulated calculations

You can use example calculations to simulate a certain type of person or household by calculating their taxes, benefits and disposable income. Simulated calculations are a quick way to illustrate how the tax and income transfer systems work. For example, you can simulate how increasing the basic unemployment allowance affects an individual’s taxation, housing allowance and social assistance or the progression of the child benefit’s real value. Unlike data simulations, the results of simulated calculation results cannot be applied to the entire population or a wider portion of the population.

2.3 Possible calculations

The SISU model includes 13 sub-models, each modelling part of the income transfer system. One sub-model typically includes the taxes and benefits in one body of law. 

The bodies of law included in the SISU model are listed below. The figures in brackets indicate the years of modelled legislation. Sub-models of the SISU model:

  • Income taxation (1980/1990–)
  • Real estate taxation of natural persons (2008–)
  • Health insurance daily allowances and parental allowance (1982–)
  • Unemployment benefits (1985–)
  • State pensions and their related benefits (1957/1991–)
  • Children’s home care allowance (1985–)
  • Child benefits, maternity grant and maintenance allowances (1948–)
  • Municipal day care fees (1997–)
  • Student financial aid, housing supplement and study grant (1992–)
  • General housing allowance (1990–)
  • Housing allowance for pensioners (1990–)
  • Social assistance (1989–)
  • Real estate taxation (2009–).

All 13 sub-models are connected to the main model, which accounts for the interactions between the sub-models in its calculations. In addition to the sub-models and the main model, extensions for indirect taxation, dividend taxation and employer contributions are available for the SISU model.

All sub-models include a section for simulated calculations and a data simulation program. You can perform calculations by sub-model or with the main model, which combines the sub-models to describe the income transfer system as a whole.

2.4 SISU model technology and user instructions

The SISU microsimulation model was coded with the SAS statistical software suite. A user-friendly interface has been created for the model with the SAS EG software (Add-In solution). You can use the model with the SAS suite as code or through the user interface.

The SISU model includes modifiable ready-made outputs. You also have the SAS suite’s versatile tabulation and graphical tools at your disposal to analyse results.

You can access the SISU model by using the Statistics Finland remote access service or your own workstation. Both operating environments use the same model code.

How to use the model with remote access

The extensive register-based microdata in the SISU model, representative of the entire population (a sample of some 800,000 people), is only accessible by remote connection through the Statistics Finland remote access system. In addition, the remote access service allows you to use the model with a smaller data sample (some 23,000 people from 2018, including variables from interviews). You will need a licence to use the microsimulation data.

To gain remote access, your employer must enter into an agreement with Statistics Finland. A separate agreement that names you as the model’s user is also required for remote access to the SISU model. 

Once the required licences and agreements for using the SISU model in the remote access environment are in order, you will receive login instructions for Statistics Finland’s FIONA remote access system. You must connect to the FIONA remote access server from your workplace. An additional licence to use the remote access system from your home or a similar fixed location over your employer’s secure connection may be granted on a case-by-case basis.

The remote access service includes the necessary software and the SISU model, which includes the necessary model code and data. You will also have a personal work folder on the server. The remote access environment allows you to share code with other users through a monitored process.

You can transfer the sum-level microsimulation results from the remote access environment to your own workstation. As the microsimulation user, you are responsible for excluding unit-level data from the transferred output files and preventing the disclosure of these data. Statistics Finland will later verify that the information in the transferred results has been protected as required. To learn more about how to protect data and the result review procedure, read the Research Services instructions (pdf).

How to use the model on your own workstation

You can make simulated calculations on your own workstation using the SISU microsimulation model. Your workstation must have the SAS statistical software suite to use the model. You can download the model’s code to your workstation for free from the Statistics Finland microsimulation page. The model code includes the necessary programs and parameter files for simulation, as well as extensive documentation about how to use the model and its features. The model code does not include the microdata used for data simulation.

2.5 Model contents

The primary components of the SISU model are as follows:

  • the SAS model code
  • law parameter tables
  • data sets
  • documentation describing the use and features of the model.

The SISU model has up-to-date, individual-level microdata representative of the entire population that include all variables (income, family relations, etc.) needed to calculate taxes and benefits. In addition to the above, the model includes income tax and social benefits legislation modelled as SAS program code.

The formulae for the calculations are primarily programmed functions and macros that use the microdata and parameters. The parameters are constants from legislation (factors, euro amounts, etc.) that have been tabulated in the model for each sub-model.

Microsimulation calculations use simulation programs that combine the data, legislation code and parameter tables needed for the calculations of one legislative block into one program. By changing the model’s legislation (model code and parameters), you can analyse the effects this will have on people’s income and public finances.

The figure below demonstrates the simulation’s logic and process.

Figure. Simulation model logic

Enter the control parameters and retrieve the law parameters and law macros. Apply the law macros and law parameters to the data as per the control parameters. The result is new microdata based on the input data.

The parameters, the SAS model code’s law macros and the simulated calculation programs cover legislation from several years, which allows changes to be viewed over time. By contrast, the data simulation model is built separately for every base year of the data set.  

Model microdata 

The SISU model has two available base sets: a small-sample income distribution service data set and an extensive register-based data set that is more representative of the population. 

The income distribution statistics service data set is household-level and individual-level research data derived from the annually produced database of the survey on income and living conditions. These data are collected from interviews and multiple administrative registers. The register includes data from some 25,000 people.

The microsimulation register data set is research data specifically compiled for the SISU model. These data cover approximately 15 per cent of the households and people in Finland on an annual basis. The data are collected from statistical data, registers and methodologically produced data.

The additional Kela data sets are data sent by the Social Insurance Institution of Finland regarding the benefits it administers. These data supplement the data sets for microsimulation purposes.

The real estate tax data set is microdata maintained by the Finnish Tax Administration. It includes the data collected by the Tax Administration about buildings and land owned by people, enterprises and associations.
.

About the data

SAS code and tables for data updates

You may need to update the model’s data, because tax and social security legislation are at least two years ahead of the microsimulation’s base data. “Updating the base file” means updating the base year data with the current year’s data and predicting the data for coming years (the six or seven years following the base year). The SISU model updates the base data (excluding real estate tax) separately from the simulations (dedicated SAS programs, tables and documentation are available).

The data are updated based on the changes in income and expenses following the statistical year, as well as the structural changes of society. The purpose of updating is to make the sample people’s and households’ incomes, expenses, fees, deductions, etc. as accurate as possible at the individual, household and national levels. The changes that have taken place since the statistical year in the population’s level of income, tax system, benefits system, and demographics are reflected in the data of coming years.

Law parameters

In this context, “parameters” refers to the monetary amounts (either euros or Finnish markka), and other constants and factors laid down in legislation. These parameters have been tabulated into SAS tables. Each sub-model has at least one parameter table.

Parameters are used to calculate taxation and income transfers. Parameter data exist for multiple previous years. In addition, indexed parameters have been tabulated for three to four years ahead of current legislation, based on Ministry of Finance forecasts.

Model code

The microsimulation’s model code uses the SAS programming language. You control the model directly with the macro variables in the SAS code.

The model’s SAS code is divided into sub-models so that each sub-model has the following:

  • law macro file
  • helper macro file
  • data simulation file
  • simulated calculation simulation file.

The main model that combines the sub-models has its own data simulation file and simulated calculation file. The model’s general settings are set in a configuration file. Like updates, dedicated SAS programs are available for the parameter tables’ index updates and monetary value conversion factors.

The legislation required by the models is programmed by legislation block as SAS law macros. The macros include legislation from multiple years, which allows changes to be analysed over time. The macros are mathematical formulae that use the law parameters and the macro’s internal parameters. The law macros are used for both data simulation and simulated calculations.

The helper macro files include parameter search macros, which allow the law macros to read the parameter tables and select the data for the desired year and/or month of legislation. The helper macro files are external to the sub-models and include income decile calculation macros, income distribution calculation macros and monetary value conversion macros.

Documentation

The SISU model includes extensive documentation that supports the model’s use. It provides information about the model’s structure and functions, but also about the tax and social security legislation behind the different sub-models. The operating information has been included as PDF and Excel files, as well as in the model code and parameter tables.

The model includes the following documentation:

  • SISU model user guide (only in Finnish)
  • key legislation for sub-models
  • law macros, variables and parameters used in sub-models
  • data variable descriptions
  • parameter descriptions in parameter tables
  • internal SAS code documentation.

2.6 Using the model – limitations and instructions

The data included in the SISU model may be used for scientific research and statistical surveys in accordance with the Statistics Act. You need a user licence to use the model’s data through the remote access system. See here for the user licence application.

Agreement for remote access with Statistics Finland

You can enter into an agreement with Statistics Finland for remote access to research data. Find out more about remote access. The agreement specifies the model’s users and must include an information security statement from your subscriber organisation. Before remote access is granted, you must have user licences for the register data included in the SISU model.

Remote access charge

Remote access to the SISU model is subject to a charge. The model’s pricing can be found in the Statistics Finland price list for research data and services. The charge includes remote use of the model, software licence fees and the data included in the model. The commissioning of the SISU model is further explained in the section on the application process and agreement practices: see 4.2 User licence application process.

Exemptions to user licences

No user licence is required if you only do simulated calculations with the SISU model code on your own workstation. You can download the model code for free from the Statistics Finland website.

The downloadable model code includes the model’s documentation, parameter tables, code files and the necessary files for controlling the user interface. The model code does not include the data needed for data simulations or the data update programs.

Required workstation software

You need a workstation with a web browser to use the SISU model remotely. The remote access service includes the SAS EG software and the SISU model, which includes the necessary model code and data.

To use the free downloadable version of the model on your own workstation, you need the SAS statistical software suite (SAS 9.3 and SA EG 5.1 or newer).

Other things to note

Please note that to use the model, you must have a detailed understanding of Finland’s social benefits and tax systems. You are responsible for all changes you make to the model, as well as the results these changes produce.

Help and support

You can contact the Statistics Finland microsimulation team for questions about the SISU model’s data, how to get started with remote access and the model in general.


Ask and read more

  • For more information about microsimulation, contact the microsimulation team.
  • To learn more about the SISU model and how to use it, read the user guide (only in Finnish). The user guide provides information about the model’s general structure and instructions for getting started with the model, creating simulations and operating the user interface.