The main measurement innovation of this website is to estimate recent withholding growth and then relate it to growth in overall wages in the economy. The description below provides some details on the measurement of withholding growth. Additional information on the usefulness of withholding in estimating aggregate wage growth and revisions is available here.
Withholding Includes Amounts for Both Individual Income and Payroll Taxes. Employers are required to withhold amounts from paychecks and remit those amounts to the U.S. Treasury for both individual income taxes and payroll (Social Security and Medicare) taxes. The withholding data utilized in the analysis combines the different sources of withholding. Approximately two-thirds of total gross federal receipts are remitted in the form of withholding.
Withholding is Measured in Nominal Dollars and in Aggregate Terms. Withholding growth is calculated from amounts measured in nominal dollars. That is, withholding growth includes the effects of inflation. In addition, withholding growth is measured for total withholding remitted to the Treasury, reflecting total wages paid. That is, withholding growth includes the effects of changes in employment, hours worked per employee, and hourly wage gains per employee. Withholding also typically grows a bit faster than wages in a growing economy as a result of so-called real bracket creep, the phenomenon by which increases in real (inflation-adjusted) income pushes more income into higher tax brackets.
The Percentage Change in Withholding Is Measured on a Year-over-Year Basis. Like total wages and many other economic measures, withholding varies significantly over the course of a year in a somewhat systematic, seasonal pattern. Measuring percentage growth by comparing amounts in a period to those in the comparable period of the prior year is one way to measure growth in a seasonally-adjusted manner. That method is employed for the analysis on this website.
The disadvantage of the method is that it becomes difficult or impossible to identify the month-to-month changes in withholding that generate the measured year-over-year growth. (For relatively small percentage increases as occur with withholding, year-over-year percentage growth is approximately equal to the sum of the month-to-month percentage changes.) The traditional way to identify such month-to-month changes is to employ seasonal adjustment methods to the monthly dollar amounts; however, seasonal adjustment methods do not appear to yield reliable results for withholding given the significant calendar-induced volatility described below. Nonetheless, changes in year-over-year growth from one month to the next do provide some information on the month-to-month changes.
Given the Volatility in the Daily Withholding Data, Our Method Uses Moving Averages over Pre-Defined Representative Periods. Withholding tax collections vary widely from day to day. Thus, smoothing procedures are employed to measure growth appropriately. The day-to-day volatility in withholding results from a combination of payday schedules and the federal rules regarding the remittance of withholding amounts:
- Amounts collected by the Treasury Department on Monday are the biggest day by far, corresponding to Friday paydays and the largest firms being required to remit withholding by the next business day.
- Wednesdays and Fridays are the next biggest withholding days, corresponding to the days that large firms must remit their withholding amounts.
- Withholding jumps at the very beginning of the month, corresponding to workers paid at the very end of previous month, and it jumps mid-month, corresponding to smaller firms that can wait to remit their withholding until the middle of the month following when the withholding accrues.
- On federal holidays no withholding is tabulated by the Treasury Department. Withholding amounts recorded for the day after business-week holidays are therefore larger, effectively capturing two days of collections.
Taking moving averages is a typical way to deal with such volatile data. However, the withholding data are so volatile that moving averages over several months of data is required to generate reasonably smooth data on a day-to-day basis; such lengthy moving averages delay recognition of recent movements in the data.
Instead, the proprietary methodology that is employed for the analysis on this site is to calculate moving averages over only several weeks and to exclude periods that we know are distorted by the calendar. Taking the past three or four weeks of data provides enough information to identify meaningful measures of withholding growth, but only if certain periods are excluded. As a result of the withholding jumps at the beginning, middle, and end of each month, and the effects of holidays, we need to exclude periods that begin or end at such times. In that way we identify representative periods that are not affected by the calendar, or at least minimally so. Given that we are assessing year-over-year growth in withholding, we ensure that the representative periods exist both in the current measurement year and in the comparable period for the prior year. If they don’t, then growth is not measured for that day.
Judgment is not employed in calculating withholding growth for a day, because the algorithm is fixed; we know in advance when representative periods will occur. (See the upcoming schedule page.) The limited exception to such advance knowledge is that occasionally outlier results occur, with the outlier thresholds defined in advance. Those outliers are dropped. Over the past 15 years of data, those unanticipated outliers have occurred about once every two months on average.
Withholding Growth is Adjusted for Tax Law Changes. In order to relate withholding to wages, we need to measure withholding on a constant law basis. Because changes in tax law can affect withholding amounts per dollar of wages, the method adjusts withholding growth rates obtained from the actual reported withholding amounts to remove the estimated effects of law changes. To estimate the effects of enacted legislation on withholding growth, we used publicly-available information from the Joint Committee on Taxation, the Urban-Brookings Tax Policy Center, and the Congressional Budget Office, along with our own withholding calculator and other methods based on public information to capture the effects of the law changes enacted in December 2017 and March 2020. Permanent changes in tax law generally result in just one year’s effect on the growth rate in withholding, given that we are measuring year-over-year growth rates. Unless provisions phase in, that result occurs because once the effects of a law change become incorporated into both the current and prior year, year-over-year changes become largely unaffected.
We have estimated growth rates in withholding on such a constant law basis starting in 2005, and about a third of that time period includes adjustments for tax law changes. The amounts of any historical adjustments to withholding growth reflected in any charts are displayed in the notes to the chart on the home page.
The Daily Growth Rates in Withholding are Averaged in Order to Calculate Monthly and Quarterly Growth Rates. The monthly growth rate in withholding is the average of the daily growth rates in the second half of the month; because the daily estimate of withholding growth uses moving averages of a few weeks of daily data, the growth rates in the second half of the month correspond most closely to the calendar month. The quarterly growth rates are the averages of the growth rates calculated for the three months in that quarter.