A short guide to measures of employment, wages and income

Over the last few years I’ve spent a lot of time discussing employment with various other market participants, commentators, reporters and other assorted interested observers and, in my experience, one of the biggest obstacles to productive exchange is an incomplete understanding of the various measures of income & employment, what they attempt to estimate and how the estimates are arrived at. I put together this short description, which I hope to keep updating in the future, in an attempt to create a quick reference to various measures of employment, wages and income both for myself and colleagues.

To avoid any confusion related to nomenclature, it’s important to define some of the terms that I will use and how they relate to each other. The labor force is the portion of the population that is employed, unemployed, or discouraged (1). Estimates of the labor force, (un)employment levels, and demographic characteristics of the labor force come from the Current Population Survey (CPS), commonly referred to as the “household survey”, collected from a sample of 60,000 households by the Bureau of Census for the Bureau of Labor Statistics (BLS). Some key definitions are reproduced below:

Employed persons consist of: persons who did any work for pay or profit during the survey reference week; persons who did at least 15 hours of unpaid work in a family-operated enterprise; and persons who were temporarily absent from their regular jobs because of illness, vacation, bad weather, industrial dispute, or various personal reasons.

Persons are classified as unemployed if they do not have a job, have actively looked for work in the prior 4 weeks, and are currently available for work. Persons who were not working and were waiting to be recalled to a job from which they had been temporarily laid off are also included as unemployed.

Discouraged workers are a subset of persons marginally attached to the labor force. The marginally attached are those persons not in the labor force who want and are available for work, and who have looked for a job sometime in the prior 12 months, but were not counted as unemployed because they had not searched for work in the 4 weeks preceding the survey.

In addition to the household survey, the BLS also conducts the Current Employment Statistics (CES) survey, commonly referred to as the “establishment survey” because it surveys establishments (employers). It covers about 143,000 employers which, in aggregate, employ over 588,000 persons (2). The establishment survey includes any full- or part-time worker that was paid during the pay period that includes the 12^th of the month excludes self-employed workers, farm workers, domestic workers and non-civilian government workers. In addition to estimates of payroll employment, the establishment survey provides estimates of hours worked and hourly earnings for the estimated 77.2 million workers paid at hourly rates (3).

I will use the word “worker” to describe a member of the labor force and the word “employee” for a person on an establishment’s payroll. It is important to note that because an employed worker may be employed by zero (e.g. in the case of self-employed) or more establishments, these are not interchangeable. The household survey counts as “employed” workers who may or may not be on a payroll, while an employee is a worker that is currently on a payroll. The sum of all the remuneration employees receive for their work is compensation.

An estimate of Compensation of Employees is released quarterly by the Bureau of Economic Analysis (BEA) as part of the National Income accounts. Compensation of employees is the sum of wages and salaries and noncash benefits (e.g. retirement plan contributions and employer-provided health care plans) paid or provided to US residents by US or foreign employers. (4) This measure can effectively be thought of as the “labor share” of income. It is important to note that this measure excludes income from self-employment, including what is commonly referred to as “1099 income,” non-employee compensation paid to independent contractors, which is recorded as Proprietor’s Income.

Wage & Salary Disbursals, a monthly estimate which is lagged by one month is an almost perfect proxy for the wages & salaries portion of compensation of employees. This measure represents the actual aggregate compensation that employers paid both wage and salary employees and is a function of employment levels, salaries, hours worked and the rate at which those hours were paid, including overtime and paid time off. This measure deviates slightly from the quarterly series in that it is based on disbursements while the quarterly series is based on accruals.

The two most popular measures of employment are the establishment survey’s non-farm payrolls (NFP), an estimate for the number of jobs in the universe covered by the establishment survey, and the employment level estimate from the household survey. As previously mentioned, these measures are not equivalent, but the closest measure to NFP in the household survey is probably Employment Level - Nonagriculture, Wage and Salary Workers. As can be seen clearly here, the difference between the two is explained almost entirely by multiple jobholders.

Average Weekly Hours and Average Weekly of Production and Nonsupervisory Employees are measures of hours worked (not scheduled). It is calculated by dividing the total weekly hours by the number of employees paid for those hours for a cell and then calculating a weighted average across industries (2).

Average Hourly Earnings and Average Hourly Earnings of Production and Nonsupervisory Employees are not measures of wage rates. They are aggregate measures of the amounts paid to workers which may include distortions such as overtime or holiday pay or fluctuations in the proportion of hours worked by higher or lower waged workers. They are calculated by dividing the sum of the payroll by the total hours worked for a cell and then calculating a weighted average across industries. It’s important to note that these series cannot be used as a measure of wage inflation as they do not control for overtime or the composition of the workers. For example, in a rising wage environment that coincides with an increase in low wage employees and an increasing number of hours worked by those employees, average hourly earnings could fall even though a measure that controlled for occupation would rise. Special bonuses and non-cash compensation are not reflected in this measure.

Multiplying the estimates of average hourly earnings and average weekly hours would result in an estimate of average weekly earnings, the average gross weekly pay for an hourly position before any deductions (such as payroll and income taxes, health plan deductions, etc.) which are available on a monthly periodicity; however, a better measure of hourly earnings is released quarterly based on the household survey. The Usual Weekly Earnings of Wage and Salary Workers as part of the provides an estimate of quantile boundaries of weekly earnings before taxes and other deductions (5) which control for various demographic factors as well as weighted aggregates meant to be representative of the population.  In contrast to all the previous data series mentioned, this release holds information as to the distribution of earnings and allows us to see where changes in wages and employment are happening in the context of sex, age, race, occupation, full- or part-time status, and educational attainment.

Together, these measures give us an idea of how money is paid by employers, to how many people, as well as how that income is distributed and the aggregate utilization of hourly workers. However, none of these measures are a true measure of wage inflation, although the median of earnings of full-time employees by occupation can be close. For a closer look at wage inflation we have to look at the National Compensation Survey (NCS).

The NCS is a quarterly establishment-based survey that provides comprehensive measures of employer costs for employee compensation, compensation trends, and the incidence of employer-provided benefits among civilian workers, excluding federal, quasi-federal workers, overseas, and self-employed workers (6). The two main components of the NCS are the Employment Cost Index (ECI) and Employer Costs for Employee Compensation (ECEC). The ECI, one of the better measures of compensation inflation, provides a “measure of the change in the cost of labor independent of employment shifts amongst occupations and industry categories” and its composition by occupational groups and industry supersectors grouped into private sector or state & local employers. The ECI tracks changes in compensation within establishment jobs, not individuals, so an employee being promoted or changing jobs within the establishment would not affect the index. While the ECI measures changes in per employee hour wages paid to employees and is presented as an index level, the ECEC measures changes in the cost of per employee hour cost to employers and is presented in dollar amounts.

Finally, the US Treasury provides a daily record of assorted deposits and withdrawals in the Daily Treasury Statement (DTS) which includes deposits for “Withheld Income and Employment Taxes,” the portion of gross pay that employers withhold for income and FICA tax purposes. In contrast to numbers from the previously mentioned releases, this is not an estimate and is not revised. It is a true, real-time indicator of aggregate wages disbursed and can be used as a proxy until Wage & Salary Disbursals data is ready later-on. It is important to note that, because deposits for withholdings are not necessarily made simultaneously with payroll (the Treasury allows for a short lag between disbursal and deposit) and payroll periodicity varies between establishments, short-term periods will be noisy and this data can’t be used to approximate month-to-month changes in the seasonally-adjusted annual rate of disbursals without significant adjustments. In particular, deposits tend to be elevated on common “paydays” like Friday, Monday, and the turn of the month.

Works Cited

1. Bureau of Labor Statistics. Labor Force Statistics. Current Population Survey. [Online] April 2015. http://www.bls.gov/cps/lfcharacteristics.htm.

2. —. Technical Notes for the Current Employment Statistics Survey. Current Employment Statistics. [Online] April 2015. http://www.bls.gov/web/empsit/cestn.htm.

3. —. Minimum wage workers: Characteristics of minimum wage workers, 2014. Current Population Survey. [Online] April 2015. http://www.bls.gov/opub/reports/cps/characteristics-of-minimum-wage-workers-2014.pdf.

4. Bureau of Economic Analysis. Methodology Papers. Bureau of Economic Analysis. [Online] November 2014. http://www.bea.gov/national/pdf/chapter10.pdf.

5. Bureau of Labor Statistics. Usual Weekly Earnings Technical Note. Current Population Survey. [Online] April 2015. http://www.bls.gov/news.release/wkyeng.tn.htm.

6. —. Handbook of Methods, Chapter 8, National Compensation Measures. National Compensation Survey. [Online] http://www.bls.gov/opub/hom/pdf/homch8.pdf.

Housing Affordability 1971-2009: Chart Roundup (UPDATED)

This post is part of the series Housing Affordability 1971-2009

I have updated some of the charts from the previous posts and put them in one post for easier access. For explanations on the data and commentary, please follow the links to the source posts. Click "read more" to see all updates.

From Interest Rates and Borrowing Capacity:

A Look at Real Household Income 1967-2009

Ruh-roh! Once adjusted for inflation (CPI-U, All Items),  it looks like real gains in household income have been concentrated amongst few over the last four decades. This seems timely considering the debate over the extension of the Bush tax-cuts. Click on the image for a larger version.

Please remember that real incomes trend toward stability, because if everyone's income increased, then prices would also increase, both because labor is an input of goods and services and because rising incomes would shift the demand curve to the right.

Housing Affordability 1971-2009: Chart Roundup

This post is part of the series Housing Affordability 1971-2009

I have collected all of the charts from the previous posts in the series and put them in one post for easier access. For explanations on the data and commentary, please follow the links to the source posts.

Housing Affordability 1971-2009: Data Sources and Collection Methods

This post is part of the series Housing Affordability 1971-2009

This is the last post of the series on housing affordability. At the bottom you will find a link to the spreadsheet so you can download it and play with the data if you want. Before that, I'll just add a couple of comments about the data and where it came from.

Owner's equivalent rent, rent of primary residence and CPI less shelter
These came from The BLS CPI website, and you can find the unique ID of the series used at the top of the "CPI-U data" worksheet. The two re-based versions just adjust the numbers to make 1971 and 1984 equal 100. This is done by taking the original value and dividing it by the 1971 or 1984 value respectively.

Monthly and yearly new home prices
These prices come from the Census. (monthly, yearly)

Monthly and yearly mortgage rates
These are posted on the Freddie Mac website. No adjustment was made for points.

Median household income
I cheated a little with this one. The series I found for median household income was limited, so instead I used the third quintile of the entire data set provided by the census.. I compared the overlapping years and the differences were so small that I decided it was OK to do this. There is no reason for the third quintile to be any less accurate than the absolute median.

Median fair-market rents
This is the rent data I decided not to use, it came from a HUD survey. Apart from being noisy and the data collection methods (phone survey with random dialing) being of poor quality, the series only started at 1979. I don't know anyone that participates in phone surveys, and didn't trust the sample to be representative.

I am making the spreadsheet available to save others the pain of having to copy values from PDF files into a spreadsheet. I do not mind if you create derivative works with any of the charts, but I do ask that you give me a mention. A simple "Credit to Morally Bankrupt for data in spreadsheet format" with a link to this post so others can download the source spreadsheet would be sufficient.

Download the spreadsheet in Microsoft Excel format.

Housing Affordability 1971-2009: Payments, Prices and Capacity

This post is part of the series Housing Affordability 1971-2009

In the last post I talked about the growth in prices in percentage terms. Today's post includes the same data, but using a nominal scale. While I think the percent change charts are great for looking at long-term, the nominal charts do a better job of communicating the differences in dollars and cents.

Here we can see the relationship between the median-price for new homes and the purchasing power of a payment equal to 30% of the median-household income. Judging by the gap, my estimate of 30% is close, but not perfect. I discussed my reasons for using this figure in Two Ways of Looking at It.  Once I post the source spreadsheet you will be able to fill-in any values you want to see plotted for the %-of-income and down-payment variables. Please note these are not in log-scale because the actual figures became a harder to read. You can find the log-scale versions at the bottom of this post.

Housing Affordability 1971-2009: Long-Term Trends

This post is part of the series Housing Affordability 1971-2009

In the last post I discussed the comparison I used for this analysis and why I chose certain data series over others. In this post we will look at long-term trends in income and prices and how lower interest rates have allowed prices to rise faster than income. Rents and the CPI less shelter figure are also included to illustrate the divergence of the trend home prices from the trend in consumer goods.

I am excluding shelter from the CPI figure because I want to display how the trend in housing differed from everything else and comparing housing prices to an unadjusted CPI would understate the growth in prices.

For rents, I decided to use the "rent of primary residence" series in the CPI. The BLS does not publish rents in their average price survey, and the only nominal figure I found came from the HUD, and after looking at collection methods, I was not impressed with the quality or coverage of the survey. Since the Census Bureau does not offer a national figure, I am still looking for better rents data¹.

Housing Affordability 1971-2009: Two Ways of Looking at It

This post is part of the series Housing Affordability 1971-2009

In the previous post, I discussed how borrowing capacity changes with respect to interest rates, finishing up with an example of the buying power of a $500 monthly mortgage payment from 1971-2009.  The example is obviously a gross oversimplification; income and price levels can and have changed since then.  To try to make some sense of this all, I decided to look at the data from two sides:

• The change over time in the cost of a median-price new home and the monthly mortgage payment necessary to buy it, a function of the price level and interest rate. 
• The change over time in the median-household income and the borrowing capacity based on it, a function of the income level and interest rate.

I chose to define borrowing capacity by calculating the amortized loan principal that would require a payment equal to 30% of a median-income household's earnings.

A Closer Look at Retail Sales

Well, we got new retail sales numbers from the Census Bureau. Calculated Risk reported an increase but, just like last month, these numbers are a little less bright when adjusted for population changes. Please take in mind this is a census year, and we'll have a new population estimate soon, but if the gap between the 1990 and 2000 estimates is any indication, the difference shouldn't that large. While we continue to see minor weakness in the population-adjusted sales, it's starting to look like we are just moving sideways, which is not good news, but it's better than bad news.

One of the items that I find discouraging is the percentage of sales volume that is gasoline sales. While sales numbers are moving roughly sideways, the amount of money spent on gasoline has been increasing. The increasing sales volume of an item with a low short-run elasticity of demand just means people's income are increasingly tied up in covering the necessities and their disposable incomes are shrinking. That is not good news for the consumer. Where are all those gasoline savings Cash4Clunkers was supposed to bring us?

I also added a chart including per-employee retail sales, which show us some tremendous improvement in the retail sector. It looks like retail was able to adjust employee-levels quickly and the per-employee sales number is actually very close to the peak levels.