Federal Reserve Economic Data

While much of the future depends on things that are impossible to forecast or time (for example, a pandemic), particular dynamics in some economic data have allowed some success in predicting a recession in the short run.

The FRED graph above shows data from Marcelle Chauvet and Jeremy Piger; the data set is based on economic data that tend to lead business cycle indicators—that is, they provide insight before the other data do. Judging from the graph, every recession (shaded in gray) has been preceded by a small increase in this computed probability of recession. There have been errors, but those errors have always predicted a recession that did not happen.

At the time of this writing, these data do not seem to exhibit any noticeable increase, which implies the data are not signaling a significant risk of recession. Hence, if a recession occurred soon, that would mark the first time this indicator would fail in this way. Another indicator, the Sahm Rule, is aligned with this assessment. But who knows? Abnormal things have happened in the data in the past few years.

How this graph was created: Search FRED for “recession probability” and select the series.

Suggested by Christian Zimmermann.

The Current Population Survey (CPS) from the US Census is one of the oldest, largest, and most well-recognized surveys in the United States. Every month, the survey is used to collect information from a probability-selected sample of about 60,000 households. The US Bureau of Labor Statistics then uses those data to calculate the median usual weekly earnings of wage and salary workers.

The Federal Reserve Bank of Atlanta also uses CPS data, and the FRED graph above shows monthly values of the wage growth tracker the Atlanta Fed produces with the data. The households sampled in the CPS are interviewed at regular intervals, and the Atlanta Fed measures the growth rate of typical labor earnings by comparing the wages of about 2,000 anonymous individuals who respond to the same survey 12 months apart.

The wage growth tracker series report the median, or typical, percent change from a year ago in all the hourly wages used in this calculation. That’s why it is called “unweighted.” The wage growth tracker reports both a 3-month moving average (the blue line) and a 12-month moving average (the red line) because the survey respondents change from month to month. A moving average smooths out data outliers that are likely to distort the visualization of trends and cycles by adding all the data points over the stated number of months and dividing them by that number. The longer the length of the moving average, the smoother the data are. (Also notice that the 3- and 12-month moving average calculations are labeled by the last month of those time frames. So, while the data start in January 1997, the earliest data points are in March and December 1997.)

Now, what do the data show? First, the tracked wage growth rates have not been smaller than zero between 1997 and the time of this writing. Perhaps this isn’t surprising because the CPS reports non-inflation-adjusted, or nominal, earnings. Second, the downward trend in tracked wage growth recorded between 1997 and 2011 reversed after that later date. Last and most startling, the slowdown in tracked wage growth that followed the 2001 and the 2007-2009 recessions (the shaded areas in the FRED graph) did not materialize after the 2020 recession. Instead, tracked wage growth has accelerated noticeably since 2021 and only recently seems to have plateaued. That reflects the currently resilient conditions of the overall labor market and the upward pressure on nominal wages resulting from the recent bout of above-average inflation.

Don’t lose track of the FRED Blog, as we’ll continue to explore this rich trove of wage growth tracker data over the next several months.

How this graph was created: Search FRED for and select “12-Month Moving Average of Unweighted Median Hourly Wage Growth: Overall.” From the “Edit Graph” panel, use the “Add Line” tab to search for and select “3-Month Moving Average of Unweighted Median Hourly Wage Growth: Overall.”

Suggested by Diego Mendez-Carbajo.

Where is there a higher density of poverty in the US? The Bureau of the Census has several data sets that allow us to map the answer to this question by county.

We previously reported about the Small Area Income and Poverty Estimates (SAIPE) program that takes data from various sources (for example, SNAP benefits and IRS filings) to model and estimate poverty rates in each county. One of the sources is the American Community Survey, which has relatively small samples to work with in many counties and, thus, cannot be used in isolation to estimate poverty at a yearly frequency. But it can be more reliable when looking at five-year averages.

This five-year average is what the FRED map above shows us, with the stated year being the last of those five years. The map shows clearly that there is more poverty in the South, with several states having a majority of counties with over 50% of households living under the poverty level.

Now, to understand such a statement, one has to understand what is meant by “poverty level,” as there are many definitions. The Bureau of the Census determines how much basic needs cost in the US on average for different household configurations. Households with a total income below that cost are considered in poverty. Note that there is no geographic variation in the cost, meaning that poverty in high-cost areas is likely undercounted and poverty in low-cost areas is likely overcounted.

How this map was created: Search FRED for “population below the poverty line,” click on any county, and then click on “View map.”

Suggested by Christian Zimmermann.