ReadingResearchRadarInvestment frameworkAI Company Primary Sources
Sign in / Sign up
← Back to reading
NewsMaterial3 min readDisconfirmAI Editorial

New York Fed Research: LinkedIn’s Two Labor Market Indicators Predict U.S. Inflation Better Than Traditional V/U Ratio

Three researchers at the Federal Reserve Bank of New York systematically evaluated three LinkedIn labor market tightness indicators using JOLTS and Bureau of Labor Statistics (BLS) public data, finding that the Jobs per Jobseeker Index (JPJI) outperforms the traditional vacancies-to-unemployment (V/U) ratio in both real-time nowcasting and forward forecasting of CPI and Employment Cost Index (ECI) inflation. The LinkedIn Hiring Rate (LHR) also outperformed the V/U ratio in most combinations, with both datasets released approximately four weeks faster than JOLTS data.

2026-10-30 22:00 GMT+8

Labor market tightness measures how difficult it is for firms to find workers: when there are fewer job seekers relative to job openings, employers face greater pressure to raise wages as they compete for scarce labor, which in turn pushes up overall price levels. Therefore, good labor market tightness indicators are crucial for predicting wage inflation and calibrating monetary policy. The most commonly used proxy indicator in the United States is the vacancies-to-unemployment (V/U) ratio, calculated by dividing the number of job vacancies from the Job Openings and Labor Turnover Survey (JOLTS) by the number of unemployed people. However, JOLTS data suffers from a release lag of about five weeks.

In research published on October 30, 2026, Elizabeth B. Li, Giorgio Topa, and Jacob P. Weber of the Research and Statistics Group at the Federal Reserve Bank of New York systematically evaluated whether three alternative indicators constructed by LinkedIn based on user behavior data could fill this timeliness gap, and tested whether their actual accuracy in predicting inflation exceeds that of the V/U ratio. The three indicators are: the Jobs per Application Index (JPAI), defined as the number of attractive job openings in a given month divided by the total number of job applications submitted on LinkedIn; the Jobs per Jobseeker Index (JPJI), defined as the number of attractive job openings divided by the number of jobseekers who submitted at least one application on LinkedIn during that month; and the LinkedIn Hiring Rate (LHR), defined as the proportion of LinkedIn members who added a new employer to their profile during the month relative to the relevant member base, indexed to the average monthly level in 2016.

Sign in / Sign up
ReadingResearchRadarInvestment frameworkAI Company Primary Sources

The research team limited the analysis window to July 2019 through April 2026, using the JOLTS V/U ratio as a benchmark for comparison. The results show that the correlation coefficient between JPJI and V/U is 0.733, indicating high synchronization between the two time series; the correlation coefficient between LHR and the JOLTS hires rate is as high as 0.95 (calculated from the beginning of 2021, excluding the severe fluctuations at the onset of the pandemic). This means that LinkedIn indicators perform reliably in tracking traditional metrics.

More critically, regarding predictive capability. Researchers conducted rolling regressions separately for real-time nowcasting (using LinkedIn indicators available up to the current period to predict current, unpublished CPI and ECI inflation) and forward forecasting (using indicators available up to the current period to predict next-period inflation), measuring accuracy with root mean squared error (RMSE) and comparing the average RMSE of each indicator against the average RMSE of V/U (with V/U set at 100). In nowcasting, JPJI had the lowest error, outperforming V/U for both CPI and ECI inflation; LHR outperformed V/U for CPI but performed slightly worse for ECI; while JPAI and the JOLTS hires rate consistently underperformed V/U. In forward forecasting, JPJI again performed best, and LHR outperformed V/U for both inflation measures, though still not as well as JPJI.

For investors, the practical significance of these findings lies in timeliness. LinkedIn’s monthly data can be obtained in the first week of the following month, whereas JOLTS data requires about five weeks. If JPJI and LHR can indeed predict inflation with lower error rates, macro hedge fund managers and interest rate traders can obtain a more reliable reading of labor market tightness before official data are released, thereby adjusting their expectations for the Federal Reserve's policy path earlier.

The researchers explicitly point out a fundamental limitation: these indicators may not represent the U.S. economy as a whole, because they only capture jobs posted and job applications submitted on LinkedIn. Furthermore, the 0.733 correlation between JPJI and V/U suggests that JPJI largely tracks information contained in V/U rather than providing a completely independent incremental signal. Whether the improvement in RMSE is sufficient to produce actionable differences in actual investment decisions remains to be continuously verified over longer sample windows and across different stages of the economic cycle.

Sources

  • libertystreeteconomics.newyorkfed.orgLiberty Street Economics↗
Continue researching →