CPB uses models when making estimates, scenarios and policy analyses. CPB uses several models that differ considerably from each other and are used for different purposes. On this page you will find an overview of all models currently used by CPB.
Saffier 3.0 is used for analyzes and estimates for the Dutch macro economy in the short and medium term. For a non-technical overview, read: 'Het macromodel van het CPB: Saffier 3.0' (in Dutch), or the extensive Technical documentation. Since December 2021, CPB has been using the macro-econometric model Saffier 3.0. CPB uses Saffier 3.0 for short-term projections such as Central Economic Plan (CEP) and Macro Economic Outlook (MEV) and for medium-term policy analyses, such as calculating the coalition agreement. Saffier 3.0 is an updated and modernized version of its predecessor: Saffier 2.1.
For the foreign exogens we use NiGEM plus postprocessor. See NiGEM website (https://nimodel.niesr.ac.uk/). NiGEM is based on theory, (historical) data and estimates. It is an econometric model that is transparent and peer-reviewed and models more than 60 countries and regions. NiGEM is suitable for making policy analysis and predictions. For more information about the use of Nigem in the estimation process, see the associated CPB Background document (in Dutch).
The CPB World Trade Monitor (World Trade Monitor in English, or WTM for short) is a model for collecting, aggregating and summarizing worldwide monthly data on international trade and industrial production. Its purpose is to map out the monthly development in trade and production at the earliest possible stage, and with the widest possible coverage in terms of countries. The coverage is broad enough to identify monthly developments at a global level, as well as for major economic regions. The World Trade Monitor is a project co-funded by the European Commission, and its results are published monthly on the website. For a detailed discussion of the model, see the accompanying technical description.
We use the gravity model to analyze the impact of changes in trade policy in the medium term. The Dutch economist Tinbergen is one of the founders of this model: the gravitational equation from physics turned out to be very suitable for describing trade flows between countries. At the beginning of the 21st century, the gravity model has been economically theoretically underpinned and the literature in this field has expanded enormously. The gravity model shows that economically large countries trade relatively often with each other. However, this decreases with increasing distance between these countries, whereby distance is not only physical but also, for example, economic, cultural or political. CPB has used the latest insights from the literature to build a gravity model that can be used to analyze changes in trade policy and trade disruptions, such as Brexit. For an extensive discussion of the gravity model, see the accompanying CPB Background Document.
Housing market model
EMU is a time series model that estimates government spending, the government balance (EMU balance) and government debt for the short and medium term. In addition, developments in government expenditure are confronted with Dutch and European fiscal rules. The EMU model incorporates fiscal policy and determines its macroeconomic translation. This is in line with both the classification of the collective sector of the National Accounts (NR) used by Statistics Netherlands and the classification of the Budget Memorandum of the national budget. Budget items related to income from tax and social security contributions and expenditure on social security and care are provided by other CPB models (Taxus, MIMOSI and ZOEM). The EMU model itself estimates all other budget items, the financial transactions that affect the government debt and the interdependence. This information provides government expenditure, which is subdivided into both National Account categories and budget preparation items. Finally, gross public expenditure is broken down by government functions. This classification reflects what is spent on different policy areas. Read more about EMU in this CPB Background Document (in Dutch).
The TAXUS model is used for short- and medium-term estimates of taxes (excluding payroll and income taxes). Read more about TAXUS in a CPB Background document. TAXUS is a time series model used to estimate tax revenues (excluding payroll and income taxes) in the quarterly short-term forecasts and the CPB's medium-term forecasts. The cross-section model MIMOSI is used to estimate the wage and income tax. In terms of level of detail, the model is fully in line with both the tax classification used by Statistics Netherlands in the National Accounts (NR) and the classification of government taxes by the Ministry of Finance. Tax revenues are estimated in interaction with the SAFFIER macro model. TAXUS highlights the many aspects of tax receipts. For example, it distinguishes between tax receipts on an EMU accrual basis and on a cash basis, distinguishes different types of payers (households, companies, abroad) and recipients (central government, local authorities, European Union abroad), calculates the year-on-year progression factor and the total policy change. in the tax burden (blo). Finally, TAXUS contains the administration of the effect of policy on the various aspects of tax revenues (policy development of the burden; blo).
The Gamma model is used for long-term analyzes of ageing, net profit and sustainable public finances. The GAMMA model is managed and used in policy research as an analysis tool in the Aging and Pensions programme. GAMMA is a dynamic general equilibrium model with overlapping generations for the Dutch economy. Because the model distinguishes overlapping generations, it is particularly suitable for analysis of the aging problem. A description of Gamma can be found in CPB Background Document 'Gamma: a long-term model for the sustainability of public finances' (December 2019, in Dutch), CPB Document 147: 'GAMMA, a Simulation Model for Aging, Pensions and Public Finances' (June 2007, in Dutch) and CPB Memorandum 115:'Description Gamma Model; version pension study' (March 2005, in Dutch).
The CRASH (Chaos and Recession After a Shock) model has been developed to calculate shock tests for the Dutch economy and public finances. Shock tests involve extreme financial-economic scenarios. These scenarios differ from relatively mild shocks in two respects, such as a temporary slowdown in world trade growth: the shocks are larger (for example, stock prices fall by 40%) and the shocks are combined (for example, shocks occur simultaneously in world trade, interest rates, stock and house prices). The model was used for the first time in the Budget Memorandum 2021. Characteristic of CRASH is its compact nature (the model has fewer than 40 equations), the incorporation of a financial sector and the role of initial conditions (funding ratios of pension funds, capital reserves of banks and loan to-value ratios in home ownership). Click here for more information on the model and here (in Dutch) for the 2020 shock trials.
Personal payment model
The personal payment model is used to calculate changes in the compulsory deductible in the Health Insurance Act (Zvw). Examples of such changes are increasing, decreasing or even abolishing the compulsory deductible, but also other forms of personal payments such as a percentage or shifted deductible. The model is a microsimulation model that has been estimated on individual health insurance expenditures of all insured persons in the Netherlands. The model simulates two effects: the behavioral effect and the financing slide. The behavioral effect is the effect on the behavior of policyholders. For example, due to a higher personal payment, some insured persons will use less care, which leads to a decrease in health care costs. The financing shift refers to the effect that policyholders themselves have to pay an amount out of their own pocket because of their own payment. These healthcare costs are borne by the insured and therefore no longer have to be paid from the premiums. This background document (in Dutch) explains the simulation model and also explains how the model is used for calculations for Care Choices in Map. This Discussion Paper provides a technical and detailed explanation. The model has been validated on two empirical studies by the CPB into the effects of the compulsory deductible in the Zvw (see here and here).
CPB uses the Mimosi model for estimates and analyses in the related areas of purchasing power, poverty, affordability, distributional effects of climate and income policies, marginal pressure, wage costs, social security, allowances, employer’s contributions and wage and income tax. Mimosi is largely based on microsimulation of disposable income for a representative dataset of 100,000 households. For all these households, the various components of the gross-net trajectory can all be calculated, taking into account policy changes as well as economic conditions (such as wages, inflation and pension indexation). This makes it possible, for example, to simulate changes in the purchasing power of different groups of households and tax revenues.
The MICSIM model is used for the long-term analysis of changes in the tax and social security system. With MICSIM, the long-term effects of measures in the system of taxes, benefits, allowances and subsidies on public finances, income inequality, employment and productivity can be fully simulated. MICSIM is a microsimulation model. The labor participation decision is modeled for more than 100,000 people in households. For households with children younger than 12 years old, the choice for formal childcare is also modelled. The behavioral responses were determined using a large and rich panel dataset for the period 2006-2009. The most recent estimates of the coefficients can be found in a CPB Background Document (in Dutch). The properties of the model have been tested against the results of quasi-experimental studies. The most recent version of MICSIM is described in a CPB Background Document. An extensive set of variants can be found in Promising Labor Market Policy (in Dutch), an updated set in Promising Tax Policy (in Dutch).
Labour supply model
The Labor Supply Model is used for short- and medium-term estimates of labor supply. The development of the labor supply is determined by the development of the population according to age and sex and the corresponding participation rates. These participation rates develop partly as a result of trends and partly as a result of the effects of policy, particularly with regard to social security. In the calculation scheme used to estimate the future development of the structural labor supply, the trend development is estimated on the basis of past developments and the policy effects can be estimated on an ad hoc basis. In addition to the trend development and the influence of policy measures, the economy also plays a role in labor participation. If things are going well, extra people will enter the labor market and if things are not going well, some will withdraw. This cyclical component is estimated in the macro and labor market models of the CPB and therefore outside the calculation scheme. Read more about this calculation scheme in a CPB Background Document (in Dutch).
The Business-cycle indicator is used to signal future economic downturns on the basis of leading indicators. The quarterly model SAFFIER II plays a central role in making short-term estimates for the Dutch economy. The CPB also makes use of other instruments, one of which is the Business-cycle indicator. This indicator tries to identify future economic turning points early on the basis of leading indicators. The last revision dates from 2010 and is used as of 2011. The methodology used and current composition are described in CPB Document 219 (in Dutch).
Our team develops forecast models that form a part of the overall CPB forecasting process. In this regard, we work jointly with other sectors at CPB so as to combine domain knowledge of other sectors with technical expertise of our team. For instance, the unemployment forecasts are supported by two models (publication in Dutch) that have been developed in cooperation with Data Science Team. In addition to forecasts we also assist with general CPB research.