Skip to main content

The FRED® Blog

Housing starts

A “housing start” is a new housing unit for which construction has begun. The graph above shows monthly housings starts in the U.S. for the past 20 years (Nov 1998 to Nov 2018) separated into three groups: single-family houses, houses with two to four units, and houses with five or more units. Note that this statistic pertains to the number of housing units and not number of houses/buildings.

The graph shows that housing starts dropped during the past recession and then increased again. That’s no surprise. However, single-unit starts have not reached their pre-recession levels. Initially, the reason was thought to be that households were either finding it more difficult to access mortgages or getting cold feet when considering the potential pitfalls of homeownership, such as the large number of foreclosures during the recession. Now, ten years after the recession, we may have to find another explanation for this change, which appears to be more than just transitory.*

Even if it doesn’t provide a definitive explanation, the graph below makes it easier to see this change by showing the percentages of the total number of units started.

*Maybe the new generation is less interested in single-family homes in the suburbs, which would be consistent with the decline in driving that we observed in a recent blog post.

How these graphs were created: For the first graph, search for “housing starts” and you should find all the seasonally adjusted series on the first page of results. Select them and click “Add to Graph.” For the second graph, take the first, go to the “Edit Graph” panel’s “Format” tab, and select graph type “Area” with stacking “Percentage.”

Suggested by Christian Zimmermann.

View on FRED, series used in this post: HOUST1F, HOUST2F, HOUST5F

How to measure inflation expectations

What level of inflation do people expect over the next several years? We could look at some surveys to try to answer this question, but nothing beats market measures where participants have some skin in the game. One such method of measuring inflation expectations is to compare how Treasury markets price two types of bonds: “normal” bonds—with a constant nominal interest rate—and “inflation-indexed” bonds—with a yield that includes realized inflation. One can tease out inflation expectations by subtracting the real bond yield from the nominal yield. This is the so-called break-even inflation that we show in the graph above for all available maturities.

The graph shows that these expected inflation rates fan out at particular times, typically downward. And, every time, the shorter maturities seem to have the strongest reactions. This is simple arithmetic. For example, a 10-year expectation also contains the 5-year expectation; and, as long as expectations average out in the long run, the shorter-term expectation will be more variable. An exception would occur if the market expects “normal” inflation in the next five years, but “abnormal” inflation during the five years thereafter. That’s very unlikely to happen, at least in terms of expectations.

How this graph was created: Search for “break-even inflation,” select the series, and click “Add to Graph.” From the “Edit Graph” panel, open the “Format” tab and move the series up or down to order them chronologically in the legend.

Suggested by Christian Zimmermann.

View on FRED, series used in this post: T10YIEM, T20YIEM, T30YIEM, T5YIEM, T7YIEM

Saving for Christmas

Back in the day, banks offered Christmas savings accounts, which allowed folks to regularly set aside some funds that would become available in time for Christmas purchases. The scheme is similar to certain types of education savings or retirement savings accounts that encourage saving for a particular purpose and impose penalties when one deviates from the goal (like withdrawing money early). These Christmas accounts have disappeared, as they were costly to banks and credit cards have clearly become popular substitutes.

FRED has some data on these Christmas savings accounts. The data points are a bit scattered throughout the years, though, much like ornaments on a tree. Along with the bright colors, this makes for quite a display! But each year has data points for June and December (at least), so we can see how the account holdings increase linearly throughout the year and reset at Christmas.

How this graph was created: Search for “Christmas savings,” select the series, and click “Add to Graph.” Then go to “Edit Graph”/”Format” to use FRED’s new palette, which lets you customize graphs with all your favorite festive colors.

Suggested by Christian Zimmermann.

View on FRED, series used in this post: TIIPCCSSA

Staying up to speed on U.S. driving trends

The graph above shows how much Americans are driving. Because there’s a very strong seasonal pattern, which spikes in the summer, we use this 12-month “moving” series to achieve a smoother line. (Just one of the many options in FRED that helps you choose how to display the data!) We see that mileage has steadily increased over the years, with three exceptions in this sample period: Two were the massive gas price hikes—in the 1970s and 1980s—and the third is the aftermath of the Great Recession. In fact, never has a driving slump been as long and pronounced as this recent one. Does this indicate that something has changed?

The second graph looks at the same series, but this time it’s divided by a measure of population. Now we can see that yearly miles per person peaked around June 2005 at about 13,200 and then dipped all the way down to about 12,000 in March 2014. As of August 2018, it’s a bit higher, at almost 12,500 miles. But it’s been leaning downward again and may decrease even further. Are we seeing a change in commuting and traveling habits? As always, FRED will keep compiling the data so you can stay up to speed on these trends.

How these graphs were created: For the first, search for “miles traveled,” select the moving 12-month series, and click “Add to Graph.” For the second, take the first and go to the “Edit Graph” panel: Search for and add the “civilian population” series, and then apply formula a/b*1000. (Multiplying by 1000 achieves the correct units.)

Suggested by Christian Zimmermann.

View on FRED, series used in this post: CNP16OV, M12MTVUSM227NFWA

The cost of servicing public debt: An international comparison

In a previous blog post, we looked at the cost of servicing U.S. debt. The metric we used is the gap between the real interest rate on debt and the growth rate of real GDP. We perform a similar exercise here, but we add a selected sample of OECD countries: Germany, Italy, Japan, and the U.K. This mix is interesting because Italy and Japan have high ratios of government debt to GDP, while Germany and the U.K. have more moderate ratios, which are all shown in the graph above.

The dotted line represents the 100 percent mark. As of 2017, three countries in our sample have debt-to-GDP ratios greater than 100 percent of GDP:  Italy, Japan, and the U.S. The U.S. debt-to-GDP ratio started to rise with the onset of the Great Recession in 2007, while the ratios for Japan and Italy started to rise in the 1990s.

As noted above, we calculate the cost of servicing debt for these countries as the difference between the real interest rate (measured as the difference between the interest rate on 10-year government bonds and the CPI inflation rate) and the growth rate of real GDP (measured as the sum of real GDP per capita growth and population growth). The second graph shows that, in 2017, Italy had the highest cost of servicing its debt, followed by Japan and the U.S. However, all of these countries have a negative cost of servicing their debt, which implies that they have a low burden of debt, since the growth rate of the economy is greater than the real interest rate for each of these countries.


It’s also worth noting that in the recovery period after the Great Recession, only Italy and Japan had positive costs of servicing their debt. Population growth in these countries is very low or even negative, which increases the cost of servicing the debt according to this measure.

How these graphs were created: For the first graph, search for and select the non-seasonally adjusted series “General Government Debt for Italy.” From the “Edit Graph” panel, select the “Add Line” option and repeat the above step for Japan, Germany, the U.K., and the U.S.

For the second graph, search for and select the series “Long-Term Government Bond Yields: 10-year: Main (Including Benchmark) for the United States” and set the units to be “Percent” and frequency to be “Annual” (average). Then add three more series to this line: “Consumer Price Index: All Items for the United States” (with units set to “Percent Change”), “Constant GDP per Capita for the United States” (with units set to “Percent Change”), “Population Growth for the United States” (with units set to “Percent Change at Annual Rate”). Then, in the Formula bar, enter the formula a-b-c-d. In the “Add Line” tab, repeat the above steps for Germany, Italy, Japan, and the U.K.

Suggested by Asha Bharadwaj and Maximiliano Dvorkin.


Subscribe to the FRED newsletter

Follow us

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