Puerto Rico
Labor Market

No composite index. Earlier versions of this page combined many series into a single index. In the fixed January 2014–August 2026 evaluation, no composite passed all the checks. The level composite was unstable: re-estimated each month with only the data available up to that month, its recent readings typically shifted (root mean square) by 2.57 times the index's own standard deviation, against a limit of 0.50. The momentum composite was stable but moved almost in step with the three-month change in private payrolls (correlation 0.963, above a 0.95 cutoff), so it added little to it. The page now shows the unemployment rate and payroll change directly, plus a count of how widespread improvement is.

Unemployment rate. The seasonally adjusted Local Area Unemployment Statistics estimate, in percent, as published.

Private payrolls. The percent change over three months in seasonally adjusted private employment from the Current Employment Statistics survey.

Breadth. Of the 24 series, the share of those with a reading whose three-month change points to a stronger labor market. Ten describe the level of the market (the unemployment rate, participation, the employment-population ratio, a proxy for how many of the unemployed leave unemployment within a month, three household-survey shares, two unemployment insurance claims rates and the manufacturing PMI's employment index); for these the change is the difference from three months earlier. Fourteen are already three-month changes (payrolls in the 10 industry supersectors, private payrolls, administrative and support services, hourly earnings, aggregate weekly hours). Each is signed so that higher means stronger. An unchanged series counts as not improving. Because series arrive on different dates, the latest month can rest on fewer series than earlier ones.

Job losers' share. This household-survey share counts people on temporary layoff, permanent job losers and people whose temporary job ended. When the survey has too few people on temporary layoff for a reliable estimate, the monthly release leaves that count blank, and the share has no reading for the month. This has become common: the share has a reading in 1 of the 12 months to September 2026. In the other months, breadth rests on the remaining series.

What these readings are not. They describe where the market is and which way it moved. They are not forecasts. Over the 53 months from January 2022 to July 2026 with data, equations predicting next month's change in the unemployment rate from the unemployment rate, the payroll change or both had root mean square errors 1.009, 1.012 and 1.017 times those of assuming no change; none was significantly more accurate.

Sources. U.S. Bureau of Labor Statistics (Local Area Unemployment Statistics; Current Employment Statistics); PR Department of Labor and Human Resources household survey; U.S. Department of Labor, ETA 539 weekly claims; Instituto de Estadísticas de Puerto Rico manufacturing PMI.