Technical Document: An Updated Methodology for the Distribution of Personal Income
The distribution of personal income is a topic of wide interest, including to politicians, economists and statisticians. Although the literature on the subject has grown dramatically over the past two decades, particularly spearheaded by Piketty and Saez (2003), income inequality has long been a topic of interest (Kuznets 1941, 1953). In fact, the predecessor of the Bureau of Economic Analysis (BEA), the Office of Business Economics, published estimates of the size distribution of income for selected years between the mid-1940s and 1970s (Office of Business Economics 1953).
Recently, there has been a growing interest in the distributed national accounts in particular. Since every household member has a right to the economic resources of the household and influences decisions about economic activities, the household is viewed by the Census Bureau and in the System of National Accounts (2008) as an institutional unit for the compilation of distribution results (Organization for Economic Cooperation and Development 2018). Looking at households as a unit of measurement and national accounts as an aggregate, represents personal income
a natural and appropriate income concept for decision making (Fixler, Gindelsky and Johnson 2019). Personal income is the income received by all persons or on behalf of all persons from all sources – from participation as workers in production, from owning a home or a business, from owning financial assets, and from the state and corporations in Form of transfers. It includes income from domestic sources as well as from the rest of the world; it does not include any realized or unrealized capital gains or losses.
There are many metrics that can be used to measure the distribution of personal income and to quantify income inequality. They are in data files on the BEA websitefor each year from 2000 to 2019 in Tables 1 and 2, additionally in a summarizing table of the inequality indicators for all years (Table 3). Table 1, “Main components of personal income and disposable personal income by decile”, distributes the share of household income in the National Income and Product Accounts (NIPA) Table 2.9, “Personal income and its distribution by households and by non-profit institutions, the households serve “, according to decile in two ways. First, the households are sorted according to the (household size-adjusted) equivalised income and assigned to the deciles (the first 10 numerical columns). Next, households are rearranged according to equivalised disposable income (personal income minus taxes) and assigned to deciles. Table 1 also provides the totals for each item in the first column and summary rows for personal income and disposable personal income.1
By providing income shares according to both the decile of personal income and the available decile of personal income, we can gain additional knowledge from the distribution of income, both in comparison to the generally provided quintiles (Tables A3 and A4) of the Census Bureau and in In terms of pure pre-tax income. Table 2, Inequality Metrics, provides inequality statistics for household income, personal income, and personal disposable income.1 A subset of the summary metrics is available for all years
- Note that Table 2 was Table 3 in the March 2020 and December 2020 releases. The original “Table 2” is no longer part of the
in table 3.
This is a broad statistic that gives a complete picture of income distribution, including the proportion of each quintile, the top 1 percent, the top 5 percent, the Gini index, and the ratio of the 90th percentile to that 10th percentile.2 These statistics are contained in the literature (including Piketty, Saez and Zucman 2018; Atkinson, Piketty and Saez 2011; Auten and Splinter 2019) as well as in the World Inequality Database, the accounts of the Federal Reserve Board (Distributional Financial Accounts), the Census Bureau (Tables A3 and A4 mentioned above) and Bureau of Labor Statistics (BLS) publications (Cunningham 2015).
Below is a description of the methodology for estimating the distribution of personal income as found in the NIPA tables, in particular table 2.1 (“Personal income and its disposition”), line 1. Section 1 describes the overall strategy, including the How external records are used to calculate statistics. Sections 2 to 7 describe the most important components of personal income. Section 8 describes methodological updates from the previous version, and Section 9 summarizes the data sources.
1. Overall strategy
The annual social and economic supplement to the current population survey (a basic micro-data set, hereinafter referred to as “CPS”) was used for the analysis. There are many components of national accounts that need to be assigned to households (that is, an amount is assigned to households for each component). These components can be grouped into four broad categories, “Breakdown of Personal Income for Households,” on the BEA website: (1) Adjusted Money Income (AMI), (2) Financial (F), (3) Health (H) and (4) Other Transfers (net) (T). Some of these components were assigned to households using additional data associated with CPS variables. In order to maximize transparency and benefit to data users, everyone must
Release. The first 12 lines are for total income (without equivalence) and the last 3 lines are for equivalised income.
- An annex table with inequality statistics calculated on the basis of equivalised income is available on request.
The data used is publicly used data that can be accessed via the links given in Section 9 “Data Sources”.
The CPS, jointly sponsored by the Census Bureau and BLS, is a nationwide representative annual household survey of the civil non-institutionalized population (around March and April of each year. According to the Statistics Office, the survey is the primary source of labor force statistics for the United States.2 It is important to note that the income data in the CPS is 1 year behind the data collection year (i.e. the survey year). Our analysis begins in 2000 due to the availability of our component data sources, particularly Medicare data (see Section 9).3
Overall, the estimation strategy involves the following four steps: (1) identify a NIPA grand total to distribute, (2) identify CPS variable (s) and external variables that could be used to assign this total, (3) sum all components Generate NIPA totals on personal income and subtotals of interest and (4) inequality statistics. This process results in a data set in which each CPS household has a value for each component of personal income. After all the components have been added to calculate personal income, equivalised personal income is calculated by dividing personal income by the square root of the number of household members. For example, if the household income is $ 10,000 and the household has four members, the household equivalent income is $ 5,000 (half of $ 10,000). Using equivalence, we can obtain comparable figures (ie an adjustment to the size of the household) for all households. Equivalent income ratings are used for all income inequality metrics.
- A major redesign was made here during the analysis period. We took on the first part of the redesign in 2014 and the second in 2018. This redesign has a huge impact on results and increases inequality.
2. Allocations based on external data sets
Several additional records are used to map the NIPA totals to households.
- Adaptation for very high incomes
Data from the Internal Revenue Service (IRS) Statistics of Income (SOI) program is used to adjust the top (ie, highest incomes) of the income distribution for three main reasons. First, it is assumed that the CPS surveys people with very high incomes without success, which leads to a bias due to non-response in inequality estimates (Bollinger et al. 2018). Second, there is the impression that people with the highest incomes are not adequately reported. Third, the CPS has top codes that vary from year to year for those with the highest incomes so as not to risk identifying these individuals. For example, if a person reports an annual income of $ 10 million, the census may assign them a value of $ 1 million. For these reasons, it is advisable to adjust the CPS income (Armor et al. 2021). The adjustment process is described below and differs from previous exercises (Fixler et al. 2017; Fixler, Gindelsky and Johnson 2018, 2019, 2020; Gindelsky 2020a, 2020b).
Before we start with the SOI aggregation procedure outlined in Gindelsky (2020b, Section 2A), we first take into account the false reports in the reported SOI data. As detailed by DeBacker et al. (2020), Johns and Slemrod (2010), Auten and Splinter (2019) u Existence of a “tax gap”) due to tax violations. In fact, BEA makes a correction in the macro data for the aggregated false reports of owner income and partnership income as seen in NIPA table 7.14, row 2 versus wage income) and varies significantly across the distribution.4th
- DeBacker et al. (2020) Make estimates (2006-2014) Incorrect Reports by Income Category for Wages, Dividends, Appendix C, Appendix D, Appendix E, and “All Others” in Table A4, “Percentage Changes in Income as Result of Examination by Income Source and AGI Group.”
Disclaimer of liability
BEA – Office for Economic Analysis published this content on December 13, 2021 and is solely responsible for the information contained therein. Distributed by public, unedited and unchanged, on December 15, 2021 2:28:09 PM UTC.