Navigating in the Dark: Using High-Frequency Private Data to Track the Labor Market

In times of government shutdowns, the flow of critical economic data often comes to a halt, leaving policymakers, researchers and the public in the dark about labor market conditions. This information blackout creates particularly acute challenges for the Federal Reserve, which now finds itself at a pivotal juncture in its monetary policy trajectory. The central bank’s decision on whether to continue its current path of easing interest rates or to pause its rate-cutting cycle hinges crucially on labor market indicators—precisely the data that become unavailable during a shutdown.The media has also commented about this issue. For example, see this Washington Post article, “Shutdown delays key labor and economic data, including jobs report.”

However, the advent of high-frequency data from private sources offers a promising solution to this information gap. In this blog post, we explore how data from platforms like Homebase can provide real-time insights into labor outcomes during periods when government data releases are paused.

Using Labor Data Collected by Private Firms

Homebase is a private company delivering payroll, scheduling and time sheet tools to businesses. In September 2025, the Homebase data had around 100,000 businesses, with over a million employees. Homebase covers many industries, but the bulk of its customers are centered in retail and leisure and hospitality, especially in the food and drink industry. During the COVID-19 pandemic and its aftermath, St. Louis Fed researchers used daily worker-level information on hours worked and wages to track the evolution of employment and wage inflation in real-time.See Maximiliano Dvorkin and Asha Bharadwaj’s 2020 On the Economy blog post, “Reading the Labor Market in Real Time,” and Maximiliano Dvorkin and Maggie Isaacson’s 2022 On the Economy blog post, “Tracking Wage Inflation in Real Time.”

Since December 2000, the Job Openings and Labor Turnover Survey (JOLTS) from the Bureau of Labor Statistics provides monthly data on job openings, hires, separations, quits and layoffs. JOLTS defines hires as all additions to the payroll during the month and separations as all employees removed from the payroll during the month. The difference between hires and separations is the net job creation, which is highly correlated with changes in payrolls. Importantly, the net job creation rate is negative in recessions, which can serve as a critical indicator of labor market conditions.

The following two figures, from FRED, the St. Louis Fed's economic database, show the evolution of the hire and separation rates, as a percentage of employment, and the net job creation rate, which is the difference between the two. In normal times, the net job creation rate is positive, but during recessions and for some months after, job creation is negative.

Comparing Homebase and JOLTS

We now turn to the Homebase data. When creating our measures of hires and separations using this data, we aggregate the daily information to a weekly employee level in which each observation shows the number of hours an employee worked at an establishment.

We define weekly hires and separations at the establishment level as follows: A hire occurs when a worker reports hours worked in the current week but not in the previous week. A separation occurs when a worker reported hours in the previous week but not in the current week. In the Homebase data, employees may temporarily stop working for a few days or weeks. Our definition counts these transitions as separations and hires, which is different from the way JOLTS defines and counts separations and hires.

To calculate the hiring and separation rates using Homebase, we take the total hires and separations in a given week and divide them by the number of employees four weeks before.In our calculations for any week, t, we use only information from firms that are in the Homebase dataset for the four consecutive week period ending with week t. The net employment rate is then determined by subtracting the separation rate from the hiring rate.Due to the volatility caused by outliers around holidays, these anomalies are removed to smooth the data series; a five-week moving average is also applied to stabilize the trends.

Finally, we construct a seasonally adjusted measure by multiplying these rates by the ratio of the seasonally adjusted JOLTS series to the nonseasonally adjusted JOLTS series.

The figures below illustrate the nationwide trends in hirings, separations and net employment change as measured by Homebase (solid line) and JOLTS (dotted line). The left axis corresponds to Homebase data, while the right axis corresponds to JOLTS data. Both series are measured in terms of rates and are seasonally adjusted.

U.S. Hiring Rate

U.S. Job Separation Rate

U.S. Net Job Creation Rate

SOURCES FOR THE THREE FIGURES: Homebase, Bureau of Labor Statistics and authors’ calculations.

NOTE FOR THE THREE FIGURES: The last observations of the JOLTS and Homebase data series are August 2025 and Oct. 24, 2025, respectively.

Our measures using Homebase tend to overstate hirings and separations when compared with those derived from JOLTS. This discrepancy can be attributed to two main factors. First, the composition of the Homebase sample differs significantly from that of JOLTS. While JOLTS provides a representative sample across all economic sectors, Homebase’s data predominantly reflect hourly workers in the service and retail industries, which are sectors characterized by higher turnover rates. Second, our methodology counts each instance when an employee takes a week off and then returns to work as a separation followed by a new hire. This approach can inflate the numbers, as it includes scenarios such as vacations or sick leaves. While the magnitude of the two series is different, the trends are remarkably similar.

Benefits of High-Frequency Data

Data sources like Homebase offer a real-time look at what is going on in the labor market. While this is especially useful now during a data blackout from the government shutdown, it is still invaluable outside of that unique scenario. Official government labor data comes out often with a month lag, so more-current data offers insights into immediate trends and fluctuations. This timeliness allows for a more dynamic and responsive approach to labor market analysis and policymaking.

Net job creation turns negative during recessions, and our real-time measures of hires and separations can provide a useful “early warning” of declining labor market conditions, a key input for monetary policy decisions.

By late October, our real-time measures of hiring and separation rates have continued their declining trend, and net job creation hovered close to zero. While these patterns are consistent with moderating employment, the evidence remains too preliminary to assess whether this represents healthy cooling or the start of a more concerning deterioration in labor market conditions.

Notes

1. The media has also commented about this issue. For example, see this Washington Post article, “Shutdown delays key labor and economic data, including jobs report.”
2. See Maximiliano Dvorkin and Asha Bharadwaj’s 2020 On the Economy blog post, “Reading the Labor Market in Real Time,” and Maximiliano Dvorkin and Maggie Isaacson’s 2022 On the Economy blog post, “Tracking Wage Inflation in Real Time.”
3. In our calculations for any week, t, we use only information from firms that are in the Homebase dataset for the four consecutive week period ending with week t.
4. Due to the volatility caused by outliers around holidays, these anomalies are removed to smooth the data series; a five-week moving average is also applied to stabilize the trends.