Printer Friendly

Comparative cross-national research on income and economic well-being: the Luxembourg income study.

Comparative Cross-National Research on Income and Economic Well-Being: The Luxemburg Income Study

Introduction

SINCE 1970, the use of household survey data for taxation and income transfer policy and analyses within academia and by governments has increased dramatically. These data have been used in microsimulation models to provide us with increasingly detailed insights into the effects of government policy changes on economic behavior (e.g., labor supply and savings), on public budgets, and on family well-being.

However, although the analytical techniques used in Canada, West Germany, and the United States are similar, the analyses are usually limited to one nation. Based on the assumption that the institutional differences between nations provide fertile ground for research and for unique insights into the effects of government policy, it seems reasonable to suppose that a cross-national focus using comparable data would improve policy analysis. The Luxembourg Income Study (LIS) is a data bank of household income surveys designed to accommodate such a focus. This paper describes the LIS, an example of research using the LIS data bank, and the future endeavors of the LIS project.

Under the sponsorship of the Government of Luxembourg, the LIS project began in April 1983. The purpose of the project was to gather, in one central location, sophisticated microdata sets that contain comprehensive measures of income and economic well-being for a group of modern industrialized countries. This location is the Center for Population, Poverty, and Policy Studies and the International Networks for the Studies of Technology, Environment, Alternatives, and Development in Luxembourg.

From its inception, the LIS was to be a researched project with a flexible, nonbureaucratic administrative structure. On January 1, 1989, the LIS data bank contains data sets from 10 countries: Australia, Canada, Israel, the Netherlands, Norway, Sweden, Switzerland, the United Kingdom, the United States, and West Germany. Table 1 presents an overview of these data sets by country, by data set name, size, and income year, by data sampling frame, and by the coverage of the population.(1) The U.S. data set used in the LIS is the Current Population Survey; micro-data sets from other countries have been adjusted to yield comparable variables.

The Contents of the LIS Data Bank

The LIS data bank consists of microdata sets from annual household surveys from the 10 LIS member countries; these data sets contain information for each household that is divided into two categories: Income variables (including taxes) and demographic variables. There are 42 income variables and 28 demographic variables from which a researcher may choose. (See the appendix for a list of these variables.)

In selecting the income variables, we have concentrated particularly on the sources of taxation and of cash and in-kind transfer income in each country.(2) From the list of income variables, a researcher can construct the following: Several definitions of income--including market, or factor, income; several sources of private and public income transfers; several types of pensions--socially provided (e.g., social security in the United States) and employment-related pensions for private sector and public sector workers; and a set of noncash income components--including "near cash" transfer-benefits, which are nominally defined as in kind but which have a cash-equivalent value equal their market value.(3) The several categories of transfer income allow for a detailed breakdown of means-tested benefits, employment-related social insurance entitlements, and universal benefits, such as child allowances. In several countries, realized capital gains are also recorded. The earnings and the average wages (average annual earnings provided by total hours worked) of head of household and of spouse are recorded separately.

The demographic variables for each household or family unit include information--such as age, education, occupation, industry, type of worker, and disability status--for both spouses in married couple households. From this list of variables, a researcher has many choices. For example, the relatedness" variable allows a researcher to choose either the "household" (i.e., all persons living together regardless of relatedness) or the "family" (i.e., all persons living together who are related by blood, marriage, or adoption) as one unit of analysis. Although a completely comparable set of definitions of units are not available across all countries, one differences in the definitions of the income recipient unit are carefully noted for each country's data set.(4) The LIS data bank also includes three sets of weights--for persons, households, and families.

The Comparability of the Data Sets in the

LIS Data Bank

One of the most important issues for a database that is to be used for cross-national analyses is the comparability of the data used in the data sets that compose the data bank. One criterion of comparability requires that the variables on the data set are defined and measured in the same way cross different nations. Therefore, emphasizing the availability of extensive, detailed income measures, especially those of tax and transfer income, the LIS staff have chosen data sets from each country that are, by their nature, comparable. To be eligible for inclusion in the LIS database, data set must contain a set of income and demographic variables that can be rearranged to be consistent with the common definitions agreed upon by the LIS staff and the LIS member countries. The comparability of these variables is necessary to accommodate policy analyses related to income distribution.

Another criterion of comparability is the timeliness of the data sets. When LIS began in 1983, the only available, acceptable data set for Norway was for 1979, and the 1979 Current Population Survey for the United States contained more extensive income measures than the previous surveys. Hence, 1979 was chosen as the target survey period. West Germany, with a data set only for 1981, and Switzerland and Australia, with data sets for 1982, joined later.(5)

Despite careful efforts to ensure comparability, a few anomalies still exist. For example, the definition of family used in the data set from the United States differs slightly from the one used in the data sets from the Netherlands, Norway, and Sweden. In these countries, unmarried persons living together as if they were married are defined as a married family unit. Moreover, in Sweden, persons 18 years old (the age of legal independence) or older who reside with their parents are defined as one-person families. These definitional differences are not important to many research projects; however, a researcher studying government policies that affect the incomes of college students, for example, would have to consider these differences if Swedish data are to be included in the study. (Smeeding, Schmaus, and Allegrezza (1985) present a more detailed discussion of the comparability of the data sets in the LIS data bank.)

The Quality of the Data Sets in the LIS

Data Bank

Another important issue for a database is the quality of its data sets, especially those composed of survey data. Although there are many factors that determine the quality of a data set, the criteria used by the LIS staff in selecting the data sets include the survey response rate, the sample size, and the population coverage. Two problems that affect the quality of the data in the data sets merit particular attention--item nonresponse and underreporting of income.

The problem of item nonresponse plagues virtually all income survey data, including the data sets that make up the LIS database. Two methods are used to correct for this problem. The first method is "hot deck" imputation, which assigns a value to the nonresponse item based on the value of the record nearest (in terms of age, sex, family size, and other characteristics) to the record with the item nonresponse. The second method is "cold deck" imputation, which assigns a value to the nonresponse item based on average levels from similar records. Using one of these methods, the statistical and research agencies in each member country have adjusted all LIS data sets for item nonresponse.

The underreporting of income is another problem with income survey data, especially for some types of income, such as property or self-employment income. This problem is compounded if inferences are to be drawn from cross-national comparisons in which the relative rates of underreporting differ for the same income variables across nations. Canada, the United Kingdom, and the United States have constructed ratios of survey estimates to aggregate national accounting data. For all three countries, the survey estimates for total income (weighted and summed over all families) are approximately 90 percent of the national income aggregates for overall income. Although some variance in underreporting is apparent across types of income, we would judge these data sets to be of comparable overall quality with respect to their reporting differences. (A more complete description of the quality of the data sets in the LIS data bank is found in Smeeding, Schmaus, and Allegrezza 1985).

Research Using the LIS Data Bank

The LIS data bank has proved to be an extremely useful resource for both basic and applied economic and social science research. For example, Aguilar and Gustaffson (1987) used the LIS data bank as a resource for their comparison of the role of income taxes across nations. They found that the structure of income taxes was quite different in the eight LIS member countries that they studied.(6) In some countries, almost all families pay income taxes. In other countries, a sizable proportion of families do not pay any income taxes (see table 2); in particular, low-income families with elderly heads of household and single-parent families do not pay income taxes in many of the countries studied. They also found that the redistributional effect of income taxes could be as high as 17.7 percent (in terms of the percentage change in the Gini coefficient due to income taxes) and that this redistribution tended to increase with the relative size of income taxes as measured by the average tax rate.

Aguilar and Gustafsson used regression models to compute marginal tax rates evaluated at the average income level for different groups in each country (see table 3). They concluded that marginal income tax rates vary substantially both across countries and across population groups within countries. In general, Sweden, Norway, and Israel had the highest marginal tax rates; Switzerland and the United States constituted the middle group, and the United Kingdom and Canada had the lowest marginal tax rates. In some countries, the marginal tax rates were consistently high across all population groups; for example, in Sweden and, to a lesser extent, in Norway, all forms of money income, including all forms of cash transfer, are subject to income taxes. The U.S. marginal income tax rates were below the group average for couples without children and for small families, but these rates were nearly average for larger families. For the elderly and single-parent families, the U.S. rates were much lower than average. Given the recent reductions in marginal income tax rates in the United States and the United Kingdom, it will be interesting to recompute table 3 when the data from later years become available in the LIS data bank.

Aguilar and Gustaffsson also computed average income tax elasticities by family type (see table 4). Families with elderly heads of household have the greatest tax elasticities, followed by single-parent families and couples with more than two children. The United States was singled out as the country with the highest income tax elasticities, particularly for the elderly, in 1979. This overall steep elasticity was largely the product of the high inflation of the 1970's that pushed the middle-income population group into higher tax brackets as nominal incomes rose faster than the corresponding tax exemptions and deductions. The Economic Recovery Act of 1982 and the subsequent changes in marginal tax rates in 1986 have reduced the tax elasticity in the United States for all groups, except possibly for the elderly. Among LIS member countries, the United States is the most lenient country in taxing the social security benefits of retirees; at present, only half of these benefits are taxed and then only if the adjusted income is above $25,000 for single retirees and above $32,000 for couples. Tables 2, 3, and 4 indicate that, compared with the other countries, the U.S. pre-income-tax treatment of social security benefits has kept the average and marginal tax rates for the elderly low but the tax elasticity for this group high.

The LIS data bank has also been used for research on human resource issues, including the following studies: . The distribution of household income and the relative income positions of the old and the young, of urban and rural residents, and of single parents (O'Higgins, Stephenson, and Schmaus 1989; Ringen and Hedstrom 1989; Hauser and Fischer 1989; and Buhmann, Rainwater, Schmaus, and Smeeding 1988); . The distribution of earnings for both men and women and the change in earnings over a worker's life cycle, including the transition to retirement (Achdut and Tamir 1989, Phipps 1988, and Wagner and Lorenz 1988); and . Comparative studies of the welfare state and its policies toward the elderly and children (Smeeding, Torrey, and Rein 1988; Smeeding and Torrey 1988; and Kohl 1987).

These examples of research present only a glimpse at the breadth of the analyses undertaken using the LIS data bank. Although the data bank is still in its infancy, it has already become a valuable resource for research into both the normative and the positive aspects of tax and transfer policy in modern Western nations. 1. The reader will note in table 1 that the data sets from some countries are expenditure surveys and tax record files. The choice of which data set to include was determined by several criteria, for example, depth and comprehensivity of income reporting on expenditure surveys and the definition of tax units and income sources used for tax surveys. 2. A project to add several sources of noncash income for most LIS countries is currently under way and should be completed by October 1989. See Buhmann et al. (1987) for additional information on this project. 3. For instance, near cash benefits include food stamps in the United States and Sweden. 4. For instance, the data sets for only six countries use the definition of household defined above. However, except for some minor differences that are discussed later, the data sets for all 10 countries use a reasonably consistent definition of family. 5. Providing a 1983 data set, the Netherlands was the most recent country to win the LIS. Because of business cycle considerations, this 1983 data set must be compared with the data sets for other years with some degree of caution. 6. Their study excluded data from the Netherlands and Australia.

Table 1.--An Overview of the LIS Data Sets

Table 2.--Income Taxes, Income Tax Rates, and Redistributive Effects in Eight Countries

Table : 3.--Marginal Income Tax Rates in Eight Countries by Type of Family

Gaining Access to the LIS Data Bank

The LIS project and the data bank are permanently housed at the Center for Population, Poverty, and Policy Studies and the International Networks for the Studies of Technology, Environment, Alternatives, and Development (CEPS/INSTEAD) in Luxembourg. The data are stored in the central computers of the Government of Luxembourg; access to these computers is via several computer terminals at the Center and is restricted by the Government of Luxembourg's data access and privacy laws. Due to these restrictions and to the assurances of confidentiality under which the central statistical offices of some member countries have loaned copies of their data sets to the LIS, public use tapes are not available. However, there are three ways that researchers may gain access to the LIS database: (1) Visit the Center in Luxembourg, (2) communicate with the Center by traditional postal services, or (3) use the BITNET-EARN-NETNORTH computer telecommunications network for access to the data bank.(7)

Although the first two ways are either expensive or time consuming, the third has proven to be a very efficient and inexpensive alternative. Almost all of the major educational institutions in North America and Europe have access to the BITNET-EARN-NETNORTH system, and they can therefore easily access the LIS data bank.(8)

The LIS data bank--its maintenance and its data renewal--is financed by contributions from the national science foundations of the countries whose data sets make up the database. Although there are no charges for reasonable use to researchers from member countries, there are minimal user charges to researchers from nonmember countries and from international research organizations. All users are required to sign a pledge to respect the confidentiality of the data before they can use the data bank. In addition, after research papers or reports are prepared using the LIS data bank, the researcher is required to make the results available as a Luxembourg Income Study at the Center for Population, Poverty, and Policy Studies (LIS-CEPS) working paper. In this way, we can document previous LIS research for those interested in furthering the use of our network, and we can provide for a statistical review of the results by the central statistical offices of the LIS member countries. Twice a year, the Center publishes a newsletter that reports new working papers, new country members, research conferences, and technical matters relating to the database. A current list of LIS-CEPS working papers, a copy of the biannual LIS newsletter, and other information are available from LIS at CEPS/INSTEAD, B.P. No. 65, L-7201 Walferdange, Luxembourg.

Future Endeavors

Solidified by the stable financial base provided by the international consortium of financial organizations of LIS member countries, the LIS staff is planning several projects. To continue to be useful for cross-national research, the LIS data bank is currently being updated with data sets for 1985, 1986, or 1987, depending on their availability from the member countries. Data sets from several additional countries--such as Poland, Italy, Finland, France, Belgium, Luxembourg, and Ireland--will be added to the more recent data sets from the present LIS member countries; adding data sets from Spain, New Zealand, Japan, Hungary, Argentina, and Denmark remain in the planning stages. If data sets from Poland and Hungary were added, then East-West comparisons would be possible. Moreover, data sets from ongoing longitudinal household panel data sets--such as the Survey of Income and Program Participation and the Panel Study of Income Dynamics from the United States and the equivalent data sets from West Germany, Belgium, and Luxembourg--will be added to the LIS data bank in 1989. In time, these data sets will provide an opportunity for cross-national panel research.

In July 1989, the Center will conduct the second annual 2-week summer workshop. Last year, 26 students from 14 countries were granted admission to the workshop. The objective of this workshop is to introduce young scholars (i.e., those at, or just beyond, the doctoral dissertation stage) to topics in cross-national economic and social science research so that their future research projects will be based, in large part, on the LIS data bank. Several student papers (e.g., Phipps 1988 and Wagner and Lorenz 1988) from the 1988 workshop are already in circulation.

The goal of the LIS project is to provide interested researchers with comparable household income survey data on which cross-national economic and social science research can be based. To the extent that the Center can continue to make these data available relatively quickly and inexpensively, the future of cross-national economic and social science research seems bright indeed.

References

Achdut, L., and Tamir, Y. (1989). "Comparative Economic Status of the Retired and Nonretired Elderly." Poverty, Inequality, and the Distribution of Income in an International Context: Initial Research from the Luxembourg Income Study (LIS). Edited by T.M. Smeeding, L. Rainwater, and M. O'Higgins. London: Wheatsheaf Books. Forthcoming.

Aguilar, R., and Gustafsson, B. (1987). "The Role of Public Sector Transfers and Income Taxes: An International Comparison." Luxembourg Income Study at the Center for Population, Poverty, and Policy Studies Working Paper No. 10. Luxembourg: Center for Population, Poverty, and Policy Studies and International Networks for the Studies of Technology, Environment, Alternatives, and Development. April 1987.

Buhmann, B., Hugenaars, A., Hauser, R., Sounders, P., Smeeding, T., and Wolfson, M. (1987). "Improving the LIS Income Measure: Towards Microdata Estimates of the Size Distribution of Cash and Noncash Income in Eight Countries." Luxembourg Income Study at the Center for Population, Poverty, and Policy Studies Working Paper No. 13. Luxembourg: Center for Population, Poverty, and Policy Studies and International Networks for the Studies of Technology, Environment, Alternatives, and Development. October 1987.

Buhmann, B., Rainwater, L., Schmaus, G., and Smeeding, T. (1988). "Equivalence Scales, Well-Being, Inequality, and Poverty: Sensitivity Estimates Across Ten Countries Using the Luxembourg Income Study (LIS) Database." Review of Income and Wealth (June 1988): 115-42.

Coder, J., Rainwater, L., and Smeeding, T. (1988). "The LIS Information Guide." Luxembourg Income Study at the Center for Population, Poverty, and Policy Studies Working Paper No. 7. Luxembourg: Center for Population, Poverty, and Policy Studies and International Networks for the Studies of Technology, Environment, Alternatives, and Development. November 1988.

Hauser, R., and Fischer, I. (1989). "The Relative Economic Status of One-Parent Families in Six Major Industrialized Countries." Poverty, Inequality, and the Distribution of Income in an International Context: Initial Research from the Luxembourg Income Study (LIS). Edited by T.M. Smeeding, L. Rainwater, and M. O'Higgins. London: Wheatsheaf Books. Forthcoming.

Kohl, J. (1987). "Alterssicherung im internationalen Vergleich. Zur Einkommensstruktur und Versorgungssituation alterer Haushalte." Luxembourg Income Study at the Center for Population, Poverty, and Policy Studies Working Paper No. 11. Luxembourg: Center for Population, Poverty, and Policy Studies and International Networks for the Studies of Technology, Environment, Alternatives, and Development. May 1987.

Oberst, D., and Smith, S. (1986). "BITNET: Past, Present, and Future." EDUCOM Bulletin 21 (Summer 1986): 10-17.

O'Higgins, M., Stephenson, G., and Schmaus, G. (1989). "Income Distribution and Redistribution." Poverty, Inequality, and the Distribution of Income in an International Context: Initial Research from the Luxembourg Income Study (LIS). Edited by T.M. Smeeding, L. Rainwater, and M. O'Higgins. London: Wheatsheaf Books. Forthcoming.

Phipps, S. (1988). "Measuring Gender Differences in Wage Distributions for Five Countries." Luxembourg Income Study at the Center for Population, Poverty, and Policy Studies Working Paper No. 25. Luxembourg: Center for Population, Poverty, and Policy Studies and International Networks for the Studies of Technology, Environment, Alternatives, and Development. November 1988.

Ringen, S., and Hedstrom, P. (1989). "Age and Income in Contemporary Society." Poverty, Inequality, and the Distribution of Income in an International Context: Initial Research from the Luxembourg Income Study (LIS). Edited by T.M. Smeeding, L. Rainwater, and M. O'Higgins. London: Wheatsheaf Books. Forthcoming.

Smeeding, T., Schmaus, G., and Allegrezza, S. (1985). "Introduction to LIS." Luxembourg Income Study at the Center for Population, Poverty, and Policy Studies Working Paper No. 1. Luxembourg: Center for Population, Poverty, and Policy Studies and International Networks for the Studies of Technology, Environment, Alternatives, and Development. July 1985.

Smeeding, T., and Torrey, B. (1988). "Poor Children in Rich Countries." Science 242 (November 1988): 873-77.

Smeeding, T., Torrey, B., and Rein, M. (1988). "Levels of Well-Being and Poverty Among the Elderly and Children in the United States and Other Major Countries." The Vulnerable. Edited by J. Palmer, T. Smeeding, and B. Torrey. Washington: Urban Institute Press (1988): 89-120.

Wagner, J., and Lorenz, W. (1988). "An International Comparison of the Rates of Return to Human Capital: Evidence from Five Countries." Luxembourg Income Study at the Center for Population, Poverty, and Policy Studies Working Paper No. 23. Luxembourg: Center for Population, Poverty, and Policy Studies and International Networks for the Studies of Technology, Environment, Alternatives, and Development. November 1988.

Table : 4.--Average Income Tax Elasticities(1) in Eight Countries by Type of Family
COPYRIGHT 1989 U.S. Government Printing Office
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:includes a comment by Katharine G. Abraham.; paper from New Horizons in Data Sets conference
Author:Smeeding, Timothy M.; Simpson, Rick; Rainwater, Lee; Abraham, Katharine G.
Publication:Survey of Current Business
Date:Mar 1, 1989
Words:3926
Previous Article:New measures of nonwage compensation components: are they needed?
Next Article:Constant-dollar inventories, sales, and inventory-sales ratios for manufacturing and trade.
Topics:


Related Articles
Program report: labor studies.
Statistical needs in Eastern Europe.
Fifty Years of Economic Measurement: The Jubilee of the Conference on Research in Income and Wealth.
Labor studies.
Level of economic development and income inequality: evidence from the postwar developed world.
SOI staff participate in NTA's 97th annual conference on taxation.
Luxembourg wealth study workshop.
The conference on research in income and wealth.
Background on the Global Insight Study: 'The Economic Impact of Wal-Mart'.
12th Dubrovnik Economic Conference Symposium.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters