Federal Reserve Economic Data

Our FRED graphs and maps can be customized to allow you to tell the story behind the numbers. In an earlier post, we described the differences between using fractile and equal interval data legends. Today we use those customization options to create binary maps.

A binary map is a data visualization format where the range of data is sorted into two categories. The FRED map above uses that format to color the states where population increased between 2021 and 2022 (the darker areas) and where it decreased (the lighter areas). The data are reported by the US Census Bureau; in addition to conducting its decennial census, it also provides annual population counts for states and counties.

The map shows many states, including Texas, Florida, and North Carolina, that gained residents and several states, including New York, California, and Illinois, that lost residents. However, for all but three states, these population changes were unevenly distributed within the state. To show that, we can tap into the same US Census data, but at the county level.

Our second FRED graph uses the same binary format described above to identify population changes between 2021 and 2022 in each county. Notice that in every state where overall population decreased, there is at least one county where population grew. Similarly, in almost all states where overall population increased, at least one county lost residents. Only in Maine, New Hampshire, Delaware, and the District of Columbia did population increase in all counties or county-equivalent areas.*

Let’s wrap up with one reflection on the data visualizations we created. The binary maps are best suited to easily show the prevalence of increases and decreases in population across regions; they don’t allow us to visually compare the magnitude of those changes. For example, while Vermont gained 92 residents and Arizona gained slightly more than 94,000, both state areas are shaded the same color. But, if we’re interested in visually comparing data ranges using a map, FRED’s default of five fractile data intervals is a reliable starting point.

*In this dataset, the District of Columbia is a single county.

How these binary maps were created: Search FRED for “Resident Population by State” and select any of the states listed. Click the “View Map” option. Click “Edit Map” and change the units to “Change, Thousands of Persons”. Next, under “Format,” change the “Number of color groups” to 2 and “Data grouped by” to “User Defined Method.” Change the interval values to 0 for the first entry. Last, click on the colored boxes to customize the colors for each data interval.

Suggested by Patrick Wade and Diego Mendez-Carbajo.

The FRED Blog has examined the impact of the COVID-19 pandemic on the overall level of economic activity and employment in the healthcare sector. Today we explore a related topic: how households changed their healthcare spending during that time.

The FRED graph above plots data reported by the US Bureau of Economic Analysis—specifically, the percent change from a year ago in the annual inflation-adjusted value of three types of household healthcare expenditures. Thanks to the graph, we can visualize the impact the pandemic had on household spending on medical products, appliances, and equipment (blue bars); hospital and nursing home services (red bars); and outpatient services (green bars).

During the outbreak of the pandemic in 2020, annual household spending on health services (both inpatient and outpatient) decreased. This was a first since 2002, when data are initially available. However, annual household spending in health goods increased. What gives? The combination of broad mandatory social distancing and strict epidemiological protocols in medical facilities can likely explain the decreased demand for face-to-face health services while demand for health goods didn’t wane.

FRED has data on household expenditures by type of health good, and we will tap into those to gain additional insights into consumer spending.

Our second FRED graph shows the breakdown of total household spending on health goods into its two subcategories: durable goods, therapeutic appliances and equipment (purple bars) and nondurable goods, pharmaceutical and other medical products (orange bars).

During 2020, annual household spending on medical drugs and other products increased at a pace similar to the pace recorded in previous years. And annual household spending on therapeutic products decreased, not unlike it did during the 2007-2009 recession. Both recessions are marked by gray shaded areas in the graph. So, perhaps, broad economic conditions impacting the employment status and income level of households, rather than a pandemic, can best help explain the cyclical decrease in spending in that particular type of goods.

How this graph was created: Search FRED for and select “Real personal consumption expenditures: Medical products, appliances, and equipment.” Next, click on the “Edit Graph” button and use the “Add Line” tab to search for and add the other series. Next, click on the “Edit Line 1” tab, change the units to “Percent change from year ago,” and click on “Copy to all.” Use the “Format” tab to select “Graph type: Bar.”

Suggested by Mickenzie Bass and Diego Mendez-Carbajo.

Over the past few years, job quits has been one of the more closely watched labor market metrics. But not everyone has been monitoring the same exact data series. FRED has two versions of quits: one seasonally adjusted, one not seasonally adjusted. Both measure the number of times workers left their job (excluding retirements), but different audiences may find one version more suitable to their needs.

The unadjusted series shows how late fall into winter seems to be a particularly unpopular time to leave one’s job, while activity picks up during the summer. Consider how a desire to receive a holiday bonus might prevent people from leaving their jobs during November and December, how students heading back to school can lead to a wave of turnover in August, or how the flip to a new year could spur some workers to want a fresh start at a new job in January.

For some users of FRED, these data can provide valuable insights to inform action planning. If company leaders know that the winter likely won’t see much change, they can use that time to prepare for the potential attrition at other points in the year. Or a savvy employee could leverage these data to negotiate a retention bonus, threatening to join the summer exodus if their employment needs aren’t met.

But for other users, the unadjusted data series is unnecessarily messy. The frequent movement from month to month comes at the expense of being able to clearly see longer-term trends. The seasonally adjusted series removes the impact of factors specific to certain times of the year, resulting in a much smoother line.

For economists and policymakers, the adjusted data make it much easier to analyze how underlying factors impact worker movement. When the seasonally adjusted number of quits changes over an extended period, it signals that economic conditions, policies, and other factors have led to changes in worker behavior. This perspective is behind past FRED blog posts (such as here, here, and here) that have reported on seasonally adjusted quits.

One series isn’t better than the other; but, depending on what you’re analyzing, one may be a better fit for your specific purpose.

How this graph was created: Search FRED for “quits” and click on “Quits: Total Nonfarm” to create a graph with the seasonally adjusted version of the data. Then, from the “Edit Graph” panel, use the “Add Line” tab to search for and select the non-seasonally adjusted version of the same name (you may find it easiest to search using the series code “JTUQUL”).

Suggested by Andrew Spewak.