Skip to main content

The FRED® Blog

The puzzle of real median household income

The graph above shows two often-reported series that look at a measure of income adjusted for inflation and population: real median household income and real per capital GDP. They should be similar, but there are quite a few differences. For example, median household income has stagnated for about two decades while per capita GDP has steadily increased. Let’s try to straighten out this puzzle.

The blue line in the middle graph shows that the number of people in each household has decreased. So the number of households in the nation has increased faster than population, which means that any measure divided by population grows faster than one divided by number of households. To see how much this matters quantitatively, we divide both income concepts by the number of households in the bottom graph. Obviously, they still don’t line up, but at least the gap is smaller. What explains the remainder?

First, the income definitions are different: Household income is based on a survey that asks people about only their income, not their employer-provided benefits and retirement contributions. In a previous post, we showed that these benefits have increased relatively more than wages. Real GDP includes all income in the economy. Second, if the distribution of income becomes more unequal, then the median decreases while the mean stays put. How much each of these contribute to the remaining gap can only be determined with a look at the microdata.

How these graphs were create: Top graph: Search for “real median household income,” click on the series, open the “Edit Graph” panel, then select “Index (scale value to 100 for chosen date)” for units, with 1984-01-01 as the date. Then add a line after searching for “real per capita GDP.” Choose the same units. Middle graph: Search for “civilian population,” open the “Edit Graph” panel, then search for “number of households,” and apply the formula a/b. Bottom graph: Repeat the procedure for the top graph for the first line. For the second line, use the “Add Line” feature, search for “real GDP,” then add the “number of households” series, and apply formula a/b. Finally, choose as units “Index (scale value to 100 for chosen date)” with 1984-01-01 in the bottom field, as the units pertain to the result of the formula.

Suggested by Christian Zimmermann.

View on FRED, series used in this post: A939RX0Q048SBEA, CNP16OV, GDPC1, MEHOINUSA672N, TTLHH

Retail instalment* purchase patterns from the 1940s

The graph above traces outstanding instalment accounts from different retail categories: household appliance, furniture, and jewelry stores from 1941 to 1951. The groups follow a similar path. From 1943-1947, the series encounter a deep decline, which can be attributed to a number of factors, such as the 1945 recession, shifts in GDP, and changes in the employment rate. In 1947, the series for household appliance stores rises steeply then continues to rise steadily along with furniture stores.

There are also a few differences among the groups. Jewelry stores hit their highs in December of each year, unlike furniture and household appliance stores. This sharp increase at the end of the year is typically due to winter holidays, when consumers are spending more on gifts. We also notice that the series for household appliances stores lags behind the other categories until 1948, when it rises above furniture stores.

To examine more historical retail data, visit FRASER, where you can view other economic publications and the complete release from the graph above: G.17.2 Retail Instalment Credit.

How this graph was created: Search for “Instalment Accounts,” then select “Instalment Accounts Outstanding Household Appliance Stores,” “Instalment Accounts Outstanding Jewelry Stores,” and “Instalment Accounts Outstanding Furniture Stores.” Then click on “Add to Graph.”

*Editor’s note for careful readers: The spelling of instalment is taken directly from the Board of Governors’ release. The spelling of “installment” vs. “instalment” wasn’t standardized when these data were collected in the 1940s. People may have been occupied with other events at the time.

Suggested by Ebony Mosby.

View on FRED, series used in this post: G172IIFUN01, G172IIHA01, G172IIJA01

Holiday jewels

The holiday season has finally arrived! It’s a time to enjoy moments filled with joy, glee, and (according to historical data from the Board of Governors) jewelry. The graph above traces outstanding instalment* accounts for jewelry stores from 1941 to 1946. Each December there’s a sharp increase, presumably from holiday spending. This kind of consistent increase (every year at the same time) reminds us that it’s wise to take into account seasonality before analyzing data.

*Editor’s note for careful readers: The spelling of instalment is taken directly from the Board of Governors’ release. The spelling of “installment” vs. “instalment” wasn’t standardized when these data were collected in the 1940s. People may have been occupied with other events at the time.

How this graph was created: Search for “Instalment Accounts,” select “Instalment Accounts Outstanding Jewelry Stores,” and add it to the graph.

Suggested by Ebony Mosby.

View on FRED, series used in this post: G172IIJA01

Trends and cycles

Are fluctuations in real gross domestic product driven by transitory or persistent factors? Do recessions reflect temporary declines in GDP from trend, or do they have more to do with a shift in the trend itself?

One way to answer these questions is through the lens of the permanent income hypothesis.* The PIH relates to consumer demand: To the extent that households have reasonable access to credit, their desired spending on consumer goods and services should depend primarily on their wealth (permanent income) and not on their current (or transitory) income. Household consumer spending, especially on basics such as food, shelter, and clothing, isn’t likely to change if there’s a temporary change in household income: You’re not likely to move into a smaller apartment because of a lower-than-expected year-end bonus. Yet, as the PIH tells us, consumer spending will change when changes to income are perceived to be permanent or persistent: You may very well have to move to a smaller apartment if you think you’ll be unemployed for a long time.

At the aggregate level, the PIH suggests that consumption expenditures may make for a reasonably good measure of the perceived trend in GDP. So we plot real GDP and real consumption (nondurables plus services) for the U.S. for 1947:Q1-2016:Q2. The graph uses an index for each series so the consumption series passes through the GDP series.

If we interpret consumption as trend, then the data suggest that the nature of the business cycle has changed from the early postwar period (1947:Q1-1984:Q1) to the present (1984:Q2 – 2016:Q2). Early in the sample, both trend and cycle (deviations from trend) play a visibly important role in shaping the time path of GDP. Later in the sample, however, virtually all movement in GDP is accounted for by shifts in trend GDP.

Viewed through this lens, the decline in consumer spending since the Great Recession seems especially worrisome because it suggests that households largely viewed the decline in real income in 2008-09 to be largely permanent.

* See Friedman (1957) for more on the PIH.

How this graph was created: Search for “real gross domestic product per capita” and select the first series in the results. Use the “Units” option near the top of the “Edit Graph” panel to select “Index (Scale value to 100 for chosen period)” and set the date to 1950-01-01. Then, on the “Add line” tab, search for “real consumption per capita nondurables” and add the series to the graph. Add another series to Line 2 by searching for “real consumption per capita services” in the “Customize Data” section and click “Add.” Making sure that each individual series is still in chained 2009 dollars, apply the formula a+b to Line 2. Then, just below where you set the formula, change the units for the new series to an index, using 1947-01-01 as the date, so that the lines lie on top of each other.

Suggested by David Andolfatto and Andrew Spewak.

View on FRED, series used in this post: A796RX0Q048SBEA, A797RX0Q048SBEA, A939RX0Q048SBEA

The convergence of income across U.S. states

Do poor areas tend to “catch up” to richer ones? After much analysis in the economics literature, the evidence is still mixed. But what do the data in FRED show us?

These two graphs trace the evolution of per capita personal income of several U.S. states over almost a century. Here, we choose the 24 most-populated states in 1930 and rank them by their initial level of per capita income. We divide the state’s per capita income by the U.S. average: A value above 1 means the state’s per capita income is above the national average, and a value below 1 means the state’s income is below average. The top graph shows the “poor” states, and the bottom graph shows the “rich” states in this sample. The graphs suggest that state incomes have gradually converged from 1929 to the early 1980s. However, this convergence seems to have stopped since then. In fact, in some cases, state incomes seem to be diverging again.

A possible explanation for this lack of convergence could be differences in the cost of living. It’s more expensive to live in, say, California and New York; so differences in real income could be more compressed than the differences in nominal income, which these graphs show.

How these graphs were created: From the FRED Release view, search through the “State Personal Income Per Capita” release for the states you want and click “Add to Graph.” Then modify each line as follows: From the “Edit Graph” section, under the “Edit Line” tab, type “Personal Income Per Capita” in the search box in the “Customize Data” section. Select the annual series and add that series to the line; then type a/b in the formula section. To remove the many (many) series titles above the graph, go to the “Format” tab and deselect “Title.”

Suggested by Maximiliano Dvorkin and Hannah Shell.


Subscribe to our newsletter

Follow us

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