What Does the Minimum Wage Do?

Appendix A: Data Sources and Variables

413 Appendix A Data Sources and Variables Several data sets are used in many analyses of the minimum wage and its effects. The same variables appear in much of this work. For the reader’s convenience, we discuss these below rather than discuss them in detail either repeatedly or only the first time each is encountered. DATA SOURCES Governments are the source of the most frequently used data in minimum wage research.1 There are two types of government data sets that are most commonly used in studies of the minimum wage: household surveys and establishment -based data. Other types of data follow individuals or families over long periods of time. Household Surveys In many countries, the government conducts a large-scale survey of households at regular intervals, and it provides the data used to calculate unemployment rates, as well as other information about the labor market. Studies of the minimum wage that focus on demographic groups—for instance, teenagers, minorities, young adults, single mothers, or married women—most commonly rely on household surveys. In the United States, the household survey that typically fills these roles is the CPS. Every 10 years, the U.S. census generates a complete listing of all extant residences in the United States. Every month, the U.S. Census Bureau generates a sample from this list of households to interview.2 Within each household, one person answers questions about all the people in the household. Once selected, a household is interviewed for four consecutive months, is not interviewed for eight months, and is then interviewed over four additional consecutive months. This is known as a 4-8-4 rotation. In the last month of each of the four-month interview cycles, the usual monthly interview is supplemented with questions about hours, earnings, and other economic matters. The households in the last four months of interviews are collectively referred to as the Outgoing Rotation Group (ORG), and it is information from the ORGs—not those from households in any of the other months of their cycle—that is used in most minimum wage studies. Belman and Wolfson.indb 413 Belman and Wolfson.indb 413 6/16/2014 12:20:57 PM 6/16/2014 12:20:57 PM 414 Belman and Wolfson About 50,000 households are interviewed for the CPS each month. Roughly 25 percent of these are in the ORG for either the first or second time and will leave the CPS after this month, either for eight months or permanently. Each month, roughly 12.5 percent of the households are added to the CPS for the first time, and another 12.5 percent start their second cycle of four months. The answers of each household refer not to the whole month but to a reference week, which is defined as the week (Sunday through Saturday) that includes the 12th of the month. Other countries have similar surveys. In Canada, Statistics Canada conducts the monthly Labor Force Survey. It comprises about 56,000 households each month, each household remains in the survey for six consecutive months, and each month one-sixth of the households are in the survey for the first time and another one-sixth are in for the last time. The reference week is the one that contains the 15th of the month. In the European Union, countries also conduct Labour Force Surveys, often quarterly. In the United Kingdom, it consists of about 60,000 households that remain in the survey for five consecutive quarters , also with a rotation structure similar to the Canadian one. The survey is conducted throughout the quarter, and the reference week is distributed uniformly throughout. New Zealand has the quarterly Household Labor Force Survey, consisting of about 15,000 households, each of which remains in the sample for eight quarters. The survey is conducted throughout the quarter and the questions asked refer to the week before the interview. In the United States and Canada, these surveys can be aggregated to the level either of states or provinces, respectively, or the entire nation. In the United States, most studies that aggregate to the state level combine the ORGs for an entire calendar year to minimize problems of high variability associated with small samples in the less-populated states. Canadian studies follow the U.S. convention of aggregating the data into calendar years.3 The studies that use European data, whether individual or more aggregated, are more evenly split between quarterly and annual frequencies. Establishment-Based Data An establishment...