Economic data from federal agencies has come under question recently as survey responses lag.
Economic data from federal agencies has come under question recently as survey responses lag.
The delay hurts decision making among businesses and investors.
As a result, many businesses are turning to private label data.
Economic data from federal agencies has come under question recently as the U.S. Bureau of Labor Statistics struggles to generate sufficient responses to its survey requests, resulting in large revisions after the initial reporting.
This decline in data timeliness harms the ability of businesses and investors to make decisions.
As a result, many businesses are turning to private label data to understand employment, inflation and the broader condition of the real economy.
For years, systemically important financial institutions, financial intermediaries and private firms have used private label data to manage their businesses.
Private estimates of employment, inflation and growth are not new and have expanded the universe of information.
Homebase, Indeed and Bloomberg, for example, provide alternative employment data that economists use to forecast payroll growth and unemployment. Unlike the government data, though, the private label data usually comes at a cost.
We think the use of private label data is about to accelerate, as criticism of data from the federal government increases with each revision.
This shift is more than an arcane theory. The Federal Reserve depends on this stream of government data to set policy and achieve its mandates for price stability and full employment.
Should the combination of low response rates to the BLS surveys and criticism of the quality of publicly derived data continue? At some point, the Fed, the private sector and the public will need to look for bespoke estimates of economic activity.
At the center of this discussion is the monthly jobs data, and how to improve it. We have three suggestions:
To measure inflation, there are already several alternatives to the BLS consumer price index, a metric likely to draw more scrutiny should inflation accelerate. These include the Fed’s preferred measure, the personal consumption expenditures index calculated by the U.S. Bureau of Economic Analysis.
Several inflation indexes are also produced by the regional central banks.
As for employment, we find the ADP payroll data to be a suitable alternative to the BLS survey, with the indisputable benefit of originating from private sources.
Wall Street has long relied on the monthly release of nonfarm payroll data by the BLS to assess the state of the economy.
But as the labor market evolves, we think attention will be turning to the estimates of private payrolls from ADP.
Changes in the number of paychecks processed by ADP appear to conform to the BLS’ final revised version of private (nongovernment) nonfarm payrolls. But unlike the survey-based data of the BLS, the ADP figure is based on hard data: the number of paychecks actually processed each month.
As for timing, the BLS survey comes out the first Friday of the following month and is revised twice during that month. There have been revisions in all but three of the months since 1979, which usually have not drawn attention unless they showed a dramatic increase or decrease.
Our research has shown that revisions tend to follow the direction of the economy, with positive revisions made during business cycle upturns and negative revisions made when economic growth is slowing.
The ADP monthly data is available at the end of each month or in the first days of the following month. It’s also subject to frequent revisions as new data becomes available, which needs to be better advertised.
Since 2022, the ADP number of paychecks processed and the final revised BLS change in private sector payrolls have a correlation coefficient of 0.79, where a coefficient of 1.0 implies the measures move in sync—a change in one is matched by a change in the other.
In our opinion, the ADP data has never garnered the same attention from Wall Street economists and traders as the government reports have, perhaps because the BLS data offers a single, easy-to-understand number and because the initial ADP data does not consistently match the initial estimate from the BLS.
While we do not expect it to happen overnight, we think attention should shift to the abilities of the ADP and BLS data to predict changes in the business cycle.
To that end, it might be more useful to incorporate employment levels and percentage changes rather than focusing solely on the number of jobs created.
The 10-year correlation between the ADP total payrolls and the BLS private sector hiring during the 2010−19 recovery from the global financial crisis was nearly perfect.
But while the ADP payrolls flattened out in 2019, the BLS number continued to rise. The ADP number might have been a better measure of what was to come.
In 2020, the BLS surveys indicated greater employee losses compared to the ADP estimates, perhaps because the BLS survey includes off-the-books employment or due to other anomalies during the pandemic.
This year, payrolls in the ADP and BLS reports are flattening, with significant decreases in private hiring and government employment in March, April and May.
That leveling off should be no surprise. Large-scale layoffs in the federal government were taking place, while uncertainty around tariffs led to cautionary hiring practices in the private sector.
From 2010 through 2019, BLS payrolls grew at an average rate of about 1.9% per year.
From 2023 through 2024, BLS payrolls grew at an average yearly rate of 1.2%. In the first seven months of this year, payroll growth slowed to a yearly rate of 0.72%.
If the purpose of analyzing the labor market is to predict the directional growth of gross domestic product, we suggest measuring the health of both indicators in the same manner.
The yearly percentage change in ADP payrolls has a correlation coefficient of 0.72 with real GDP growth. The BLS service sector survey has a slightly higher correlation with real GDP of 0.78.
But the ADP data appeared to be a better predictor of GDP from 2011 until 2019, when it began to drop ahead of the GDP decline during the U.S.-China trade war and the run-up to the pandemic shutdown, and then exceeded the GDP decline in 2020.
The yearly change in the BLS service sector data was flat from 2011 to 2020 but seemed to serve as a better predictor during the pandemic and the postpandemic recovery.
From our point of view, federal government statistics remain the gold standard by which to make important policy and private sector allocative decisions. But the methodologies in the collection of that data are not perfect and are facing increased competition from private sector label data.
In a perfect world, more resources—capital and personnel—would be given to the collection of data, and the data would be provided to the public free of charge.
But the reality is that the reliance on private label data is set to become far more important.
By using a widely available data set from ADP, our research implies that the relative growth of actual payrolls might offer a better perspective on the economy than a single number that can be distorted by survey responses.
The number of paychecks processed by ADP appears to be highly correlated with BLS survey estimates, with ADP data potentially avoiding the criticism directed at federal agencies.
We are confident that as the $30 trillion American economy evolves, private sector data will continue to gain importance and broader public recognition.