by
Jørgen Aasness, John K. Dagsvik and Thor O. Thoresen,
Research department, Statistics Norway
The Norwegian tax-benefit model LOTTE is a microsimulation model developed to predict distributional and revenue effects of changes in the taxation of individuals in Norway. The model is static in the sense that it is developed to simulate short-term effects, in contrast to the dynamic microsimulation model MOSART, which is capable of performing simulations into the distant future.
The LOTTE model system (currently) consists of three modules: a module for simulations of personal income taxes, LOTTE-Skatt, a module for indirect taxation, LOTTE-Konsum, and a module describing the labor supply effects of changes in the personal income tax, LOTTE-Arbeid. The latter module has just been developed. In the following we will describe the three parts of the LOTTE model.
The first version of the LOTTE-Skatt model was developed early in the 70'ies. Since then Norwegian policy-makers have extensively used it, and it is fair to say that the model has been a key tax policy tool over the last decades. The model is non-behavioral; we return to this issue shortly when presenting the LOTTE-Arbeid module.
The Ministry of Finance use the model throughout the year. For instance, they now (February 2006) have started preparing the budget proposal for 2007, to be put forward in early October 2006. After the budget proposal is made public, the other political parties in Parliament use the model extensively when developing alternative budget proposals.
As the model is physically located to Statistics Norway, a standard process of running the model often starts with contact between the Ministry of Finance or members of Parliament and the dedicated model operator at Statistics Norway. After running the model, the results are reported back to the "customer" rather promptly, often within hours. Usually results are presented in contrast to a baseline simulation. For instance, when the policy-makers now start preparing the 2007 budget, they will usually compare various tax-law changes to a simulation based on the 2006 tax-law projected to 2007. The Ministry of Finance has had access to a version of the model on a machine in the ministry, but it has rarely been used. Early in 2006, we make arrangements to let the staff at the Ministry of Finance run the model at Statistics Norway through the Internet. For security reasons they have access to a version of the model that is placed on a separate machine in the bureau.
LOTTE is written in SAS and basically consists of four parts: individual income data, a program that projects data to the year of interest, a set of tax-laws, and a simulation routine that applies tax rules to individual records. Standard model results include aggregate revenue effects and various descriptions of distributional effects.
The main data source of the model is individual income tax returns. The income tax returns are now electronically available for all citizens. However, as many questions concerning income and taxes are best understood on the household level, we need to establish households. For that reason Statistics Norway has developed the Income and property statistics for households, which is a sample survey, where information from various registers is combined with household composition data from interviews. Besides information from income tax returns, the Income and property statistics for households also includes income data from the National insurance administration and other sources. The sample weights to readjust to national level results, reflect that the sample is stratified and that weights have been calibrated to fit information from registers. At the time being, the model is based on data from 2003, which implies that simulations are carried out for about 17,000 households. From 2007 and onwards we will employ the National register of ground addresses and buildings for household composition information, which means that all data are derived from registers. We can then (in principle) simulate tax burdens for the whole population.
The data are projected to the current year and to future years in accordance with macro economic predictions for income growth, income composition, etc. These estimates are derived in close collaboration with the Ministry of Finance, and by contributions of Statistics Norway's macro models.
Detailed tax rules for different years are compiled in the tax-law module of the model. The user can easily make tax rule changes, such as altering statutory tax rates and income deductions. For more advanced tax system changes, for instance changes in tax bases, one usually has to change the simulation routine. The simulation routine joins tax rules and data, in that various tax bases are defined and tax-laws are applied to data in that part of the model.
We have recently seen a few examples of applications of the LOTTE-Skatt model in more academic orientated work. For instance, both in Thoresen (2004) and Lambert and Thoresen (2005) the model is employed in order to describe trends in tax progressivity.
This model is based on theory of consumer behavior and standard of living, and theory of aggregation of consumption and standard of living over households and individuals in a population. There is no observed consumption data in the LOTTE database, so the microsimulation model is calibrated based on microeconometric relations and macro consumption data, giving a synthetic total expenditure, and allocated to a detailed set of commodity groups, of each household in the population. The main commodity group in current use consists of 30 commodities, but it may easily be changed depending on the purpose of the analysis and on the available econometric studies.
The expenditures for each household are derived from a system of Engel functions with demographic variables, satisfying adding-up and non-negativity constraints. This non-linear system, with many corner solutions, is calibrated through a iterative algorithm, such that the sum of expenditures over all households are exactly equal to macro household expenditures, for each commodity, given by an external source for the reference year (baseline simulation). For each household the budget shares are calculated in the reference situation and used to calculate a household specific price index, using Laspeyres formula. This index is approximately equal to any "true cost of living index" for small price changes from the reference situation, for each household. This last approximation is usually satisfactorily, since we change the baseline simulation according to the problem at hand. Otherwise we may use a more advanced price index formula.
The standard of living for each household, and every individual within the household, is defined as the household's total expenditure divided by the household's price index and the household's number of consumer units in the reference period. It can be shown that this can be interpreted as a money metric utility, see Aasness (1995). This measure is aggregated to different inequality and welfare measures and used for policy analysis, see e.g. Aasness (1997) and Aasness, Benedictow and Hussein (2002).
The revenue effects of changes in indirect taxes is not calculated by Lotte-Konsum, but by a corresponding microbased macromodel called KONSUM, see e.g. Nygård and Aasness (2003) and the references within, with a corresponding commodity grouping which can be aggregated to different levels. The Ministry of Finance uses a selection of the 60x60 = 3600 price elasticities from KONSUM in their procedures for calculating the revenue effects of indirect taxes in their national budget proposals to the Parliament. We can construct and calibrate the macromodel KONSUM and the microsimulation model Lotte-Konsum in pairs, so they are both calibrated to exactly the same macro data and to the same reference period.
Schroyen and Aasness (2006) performs a marginal indirect tax reform analysis, combining output from the microsimulation model Lotte-Konsum, in terms of expenditure tables for household deciles (according to standard of living), and output from the macromodel KONSUM, in terms of a complete system of demand elasticities, in the reference situation. According to standard optimal tax theory this all information needed to perform a marginal tax reform analysis. This is based on the same type of reasoning as is behind the well known property of Laspeyres index being equal to the true cost of living index for infinitesimal price changes (cf. the envelope theorem). The optimal tax theory assumes that the household data adds exactly up to the macro data, which is fulfilled in our pair on simulation models, but which are usually not the case in other similar empirical tax analysis.
The microsimulation model LOTTE-Konsum can also be used in combination with at full scale general equilibrium model, see Åvitsland and Aasness (2004).
Neither revenue effects that works through changes in consumption and indirect taxes nor changes through labor supply effects are usually included when discussing tax policy changes in Norway. It means that the Parliament usually passes budgets relying on information from deterministic, non-behavioral models, that is, without any detailed information on possible effects of tax scheme changes on behavior. Especially during the preparation of the last Norwegian tax reform, the tax reform of 2006, this practice came under pressure. The reason was that the reform involved substantial tax reductions, and the difference between cost measures with and without behavioral effects was expected to be large. Particularly the political parties that wanted to go further in cutting taxes emphasized that revenue estimates did not include any supply side effects, and in that respect overestimated the revenue loss of tax reductions.
At Statistics Norway we have over the years developed substantial knowledge and experience on labor supply estimation for simulation purposes; see for instance Aaberge, Dagsvik and Strøm (1995), Dagsvik and Strøm (2006), and Kornstad and Thoresen (2004). The latter study is the first attempt to employ an estimated labor supply model in combination with LOTTE-Skatt. However, the discussion of labor supply effects due to means-testing was based on a labor supply model for married female wage earners only.
The module currently being established within the LOTTE system includes separately estimated labor supply models for married/cohabiting, single males and single females, based on the econometric approach laid out in Dagsvik (1994). We still restrict to wage earners, but we work on establishing a labor supply model for the self-employed too. In the near future this module will be made available to the Ministry of Finance and the Parliament as a tool in the budget planning process. However, we expect that this module will be less frequently used, compared to the LOTTE-Skatt module.
References
Aaberge R., J.K. Dagsvik and S. Strøm (1995): Labor Supply Responses and Welfare Effects of Tax Reforms, Scandinavian Journal of Economics 97, 635–659.
Aasness, J. (1995): A microsimulation model of consumer behavior for tax analyses, Paper presented at The Nordic Seminar on Microsimulation Models, Oslo, May 1995.
Aasness, J. (1997): "Effects on poverty, inequality and welfare of child benefit and food subsidies» in N. Keilman, J. Lyngstad, H. Bojer, and I. Thomsen (eds): Poverty and and economic inequality in industrialized western countries, Oslo: Scandinavian University Press, 123-140.
Aasness, J., A. Benedictow and M.F. Hussein (2002): Distributional efficiency of direct and indirect taxes, Report 69 in the series "Forskning om skatteøkonomi". Oslo: Norges forskningsråd.
Dagsvik, J.K. (1994): Discrete and Continuous Choice, Max-Stable Processes, and Independence from Irrelevant Attributes, Econometrica 62, 1179–1205.
Dagsvik, J.K. and S. Strøm (2005): Sectoral Labor Supply, Choice Restrictions and Functional Form, forthcoming in Journal of Applied Econometrics.
Kornstad, T. and T.O. Thoresen (2004): Means-Testing the Child Benefit, Review of Income and Wealth 50, 29–49.
Lambert, P.J. and T.O. Thoresen (2005): Base independence in the analysis of tax policy effects: with an application to Norway 1992-2004, Economics Discussion Paper No. 2005-13, University of Oregon, 2005 (http://econpapers.repec.org/paper/oreuoecwp/)
Nygård, O. E. and J. Aasness (2003): "Effects on revenue and consumption patterns of changes in excise taxes on goods exposed to cross border shopping" (in Norwegian), in NOU (2003:17): Særavgifter og grensehandel (Excise taxes and cross-border shopping), Oslo: Akademika, pp. 113-130.
Schroyen, F. and Aasness, J. (2006): Marginal indirect tax reform analysis with merit good arguments and environmental concerns: Norway 1999, unpublished paper, Statistics Norway (Oslo) and Department of Economics, Norwegian School of Economics and Business Administration (Bergen).
Thoresen, T.O. (2004): Reduced Tax Progressivity in Norway in the Nineties. The Effect from Tax Changes, International Tax and Public Finance 11, 487-506.
Åvitsland, T. and J. Aasness (2004): Combining CGE and microsimulation models: Effects on equality of VAT reforms, Discussion Papers 392, Statistics Norway.