Skip to main content

The FRED® Blog

Some educational effects on employment

Since the end of the Great Recession in June 2009, labor markets have improved dramatically. National unemployment rates have fallen from 10 percent to 4.9 percent. In no small part, this has been driven by the roughly 10 million new jobs created. Even so, labor market improvements haven’t been evenly distributed across the U.S. population: Those with higher levels of education have done much better than those with lower levels.

The graph shows the total, post-recession change in employment for workers over 25 years of age grouped by level of education. In the first year after the recession, few if any new jobs were created for anyone, regardless of education. After the first year, jobs steadily increased for those with a bachelor’s degree or higher. And it took another year for labor markets to improve for those with some college or an associate’s degree.

Unfortunately, the same cannot be said for those with, at most, a high school education. Since the end of the recession, these individuals continue to experience no net job growth. Labor markets have not improved for everyone.

How this graph was created: Search for “Employment Level” and select the following tags (in the left sidebar): “education,” “25 years +,” and “sa.” Select the four series and click “Add to Graph.” Edit the range of the graph to start in June 2009 using the controls in the top right-hand corner or the sliding bar below the graph. Because we want to see how employment has changed since the end of the recession, we need to change employment levels to cumulative changes in employment since June 2009. Here’s how we do that: For each series, find the June 2009 value and subtract it by using the “Create your own data transformation” field: For example, for “Employment Level: Bachelor’s Degree and Higher, 25 years and over,” the June 2009 value is 43,362; so you will apply the formula a-43362. After transforming each series, if the y-axis title and y-axis labels overlap, reduce the general font size in the “Graph Settings” menu.

Suggested by Michael McCracken and Joseph McGillicuddy.

View on FRED, series used in this post: LNS12027659, LNS12027660, LNS12027662, LNS12027689

St Louis adds 15,600…no, wait…22,400 jobs in 2015: Be aware of data revisions

When the Bureau of Labor Statistics (BLS) released the latest state and local employment data on March 14, 2016, the story of recent job growth changed for many parts of the country. Here in St. Louis, the local economy has about 20,000 more jobs than previously estimated.

Data revisions occur because counting new jobs is a difficult process that relies on samples and advanced statistical techniques. As more information becomes available, data are revised. Estimating smaller geographies is especially difficult: Revisions are less frequent, but their magnitude can be more substantial than for larger areas.

The BLS uses the monthly Current Employment Statistics (CES) survey to estimate local employment for nonagricultural industries, but the best source of local employment statistics comes from their Quarterly Census of Employment and Wages (QCEW). The QCEW includes data derived from establishments’ reports to the various unemployment insurance programs that are released with about a 6-month lag. Every March, the BLS reconciles the CES estimates with the data from the QCEW, which can result in significant revisions, as we’ve seen here in St. Louis.

The top graph shows the St. Louis MSA’s total nonfarm employment before and after the BLS completed its revision. Job growth had been underestimated by close to 20,000 jobs over the two-year period. Perhaps surprisingly, this upward revision was predictable: The QCEW had been growing at a much faster pace than the CES for much of the period, and stronger growth was reported across many industries as the BLS revised employment upward for the majority of industries in the MSA.

Not every industry’s revision was positive. Employment in transportation and utilities (shown in the bottom graph) seemed to be growing at a rapid pace from late 2014 through the end of 2015. But the revision reduced reported employment in the industry by close to 10%: from 53,300 down to 49,600. It’s common for revisions to have a significant effect on industries in a region, as the initial data simply don’t allow sound employment estimates. Knowing whether or not the data have been revised is important when deciding if you should take the number at face value or with a grain of salt.

How these graphs were created: The St. Louis Fed maintains records of all data revisions in its ALFRED® database, which allows you to retrieve vintage versions of data that were available on specific dates in history. On the “All Employees: Total Nonfarm in St. Louis, MO-IL (MSA)” page on FRED, click on “Vintage Series in ALFRED” on the left sidebar to retrieve the two most recent releases, which currently include the revision. Under the Graph / Graph Settings tab, change the graph from bar to line and select other release dates. This will create the top graph. Follow the same general process to create the bottom graph.

Suggested by Charles Gascon and Paul Morris.

View on FRED, series used in this post: SMU29411804300000001SA, STLNA

Tracking more Fed policy tools

Those outside the Fed often cite the federal funds rate as the only tool in the FOMC’s monetary policy toolbox. But there are more—a fact first demonstrated when the FOMC employed “non-traditional” policy instruments in its successive quantitative easing programs, all of which involved purchasing some assets. As the FOMC has started to increase the federal funds rate target from near zero, it has also made clear that it can also use two other interest rates to set monetary policy: the interest rate on required reserves and the interest rate on excess reserves. FRED has recently added data on these two rates so users can track how these policy instruments are evolving.

The graph above shows these three rates: the federal funds rate target, which has an upper and lower limit to its range, and the two rates on reserves. At this point, there’s not much to see, as the rates on reserves currently coincide with the lower limit of the federal funds rate target and have done so for some time. But these rates need not follow the same path. In fact, the FOMC may implement policy by adjusting one or more of these rates if necessary.

How this graph was created: Search one by one for the four series and add them to the graph. For a shortcut, search for the series IDs: IORR, IOER, DFEDTARU, DFEDTARL.

Suggested by Christian Zimmermann

View on FRED, series used in this post: DFEDTARL, DFEDTARU, IOER, IORR

Wage paradox at the industry level

There’s a well-known disconnect between the fluctuations of average employment and of average wages: Employment is volatile and dips during recessions, while wages tend to be quite stable. This is a problem for economic models, which have difficulty reconciling the fluctuations in productivity that must justify the changes in employment levels despite the smoothness of wages. (There are exceptions, of course, such as Rudanko (2009) and Lamadon (2014).)

Averages, however, don’t tell the whole story, as we’ve pointed out here before. So we look a little deeper at the occupational level. In the past 15 years and through two business cycles, different occupations have clearly been affected differently by both long-term and cyclical changes. Manufacturing is a notable example of a long-term decline, punctuated by more rapid change during recessions; construction has had a stark rise and fall. On the other hand, white collar service work has been more stable over time. In the graph above, we see large changes in employment among “production occupations” but see much less volatility in “installation and repair occupations” (two sets of occupations with similar skills). Construction and extraction occupations have been subject to well-known fluctuations in demand associated with the housing bubble and resource boom, and these factors show up in the employment figures. “Administrative support”—a different set of skills but at a similar level—has been relatively stable over the period and the cycle.

The employment situations look vastly different for these different occupations, but wages are starkly similar, as shown in the graph below: Wages for each occupation, after normalizing out the difference in levels, follow almost exactly the same pattern. This is strange because economists often assume that wage changes will guide the shifts of workers from one occupation to another; but it seems these shifts are occurring without wages leading them. Or, to explain it from the other direction: If demand is low in an occupation, restricting workers’ ability to work there (i.e., reducing the number of available jobs) should depress wages; but, again, wages do not seem to follow the changes in the levels of workers in these occupations. There are potential explanations, of course, but these facts challenge our initial beliefs.

How these graphs were created: Go to “Browse data by release” and on the final page is “Weekly and Hourly Earnings from the Current Population Survey.” Choose “Classified by occupation and sex.” For the top graph, choose “number of workers” and for the bottom graph choose “median usual earnings.” Finally, choose quarterly data and then the four occupations we’ve shown using data for both sexes. Normalize the data to be 100 at the trough of the Great Recession, June 2009.

Suggested by David Wiczer

View on FRED, series used in this post: LEU0254498800Q, LEU0254505100Q, LEU0254509100Q, LEU0254512900Q, LEU0254552200Q, LEU0254558500Q, LEU0254562500Q, LEU0254566200Q

The composition of federal tax receipts

The government provides public goods that need to be paid for…somehow. Often this is done with tax revenue. In the case of the U.S. federal government, the composition is illustrated above (in honor of Pi Day, we had to show a pie chart). But the composition of tax receipts has changed over time, which is illustrated below. In the graphs, we see there are four main sources of income: 1) Social Security tax, whose share increased over the first half of our sample, as its tax rate was adapted to finance an older population retiring earlier. 2) Personal income tax, whose share has been surprisingly stable in the lower 40% for 70 years. 3) Import and production taxes, whose share has shrunk considerably, especially as many tariffs have been abolished. 4) Corporate income tax, which has a reputation for being very high in international comparisons and yet yields a relatively small and decreasing share of total federal receipts. If you look closely, you will also notice some interesting fluctuations, such as an increase in 2013-14 in income from assets (in red; the sale of the assets accumulated in the previous years to bail out some firms).

How these graphs were created: Start from the Federal Government Current Receipts and Expenditures release, select the series you want to display, and click on “Add to Graph.” For the top graph, change graph type to “Pie.” For the bottom graph, change the graph type to “Area” with stacking set to “Percent.” At the time this post was written, the corporate income tax data weren’t yet available for 2015:Q4, so the last data point was removed.

Suggested by Christian Zimmermann

View on FRED, series used in this post: A074RC1Q027SBEA, B075RC1Q027SBEA, W007RC1Q027SBEA, W008RC1Q027SBEA, W009RC1Q027SBEA, W011RC1Q027SBEA, W780RC1Q027SBEA

Subscribe to our newsletter

Follow us

Twitter logo Google Plus logo Facebook logo YouTube logo LinkedIn logo
Back to Top