April | 2017 | FRED Blog

Diverging forecasts

Different stories from hard and soft data

Posted on April 27, 2017

Leading up to the 2017:Q1 GDP release, the two GDP tracking indicators in FRED have told starkly different stories of expected growth. These indicators, also referred to as Nowcast indicators, combine higher-frequency (e.g., monthly) economic data released before the GDP data to estimate growth in the current quarter. As shown in the graph above, in the beginning of April the GDPNow indicator from the Atlanta Fed forecasted a significant slow-down in growth, predicting 0.638 percent annualized growth in the first quarter. In contrast, the St. Louis Fed Economic News Index forecasted annualized GDP growth to be higher, at 2.89 percent in Q1.

What is driving the difference? An analysis into the data underlying the GDP trackers identifies stark differences between “hard” data and “soft” data in the first months of 2017. The Nowcasts rely on both soft data such as consumer and business surveys and hard data such as retail sales and industrial production. The GDPNow indicator uses more hard data, taking an accounting approach to building a forecast; the St. Louis Fed’s News Index is based more on soft data, which is surveyed from news reports. (For more insight on this topic, see this recent Economic Synopses essay.)

The graph below illustrates the contrast between a soft data series and a few hard data series over the beginning of 2017. The blue line is the University of Michigan Consumer Confidence Index, and the red and green lines are industrial production and retail sales, respectively. Each series is indexed to 100 at October 2016 to show the progression over the end of 2016 and beginning of 2017. Since October 2016, consumer confidence has risen dramatically while retail sales and industrial production have risen steadily but slowly. Soft data in 2017 have so far told a much more positive story of economic growth than hard data. One reasons analysts identify as a possible cause for the divergence between survey and hard economic data is consumer and business optimism after the election of Donald Trump that is not yet reflected in the hard data.

How these graphs were created: Top graph: Search for “GDPNow” in the FRED search box and graph the first series that is returned. Click the “Edit Graph” button and select the middle “Add Line” menu. First, search for “St. Louis Fed Nowcast” and add the St. Louis Economic News Index as a new line. Repeat this process for real GDP, selecting the series with units in percent change from preceding period at an annualized rate. Adjust the date range to 2015:Q3 to 2017:Q1. Bottom graph: Search for “Consumer Sentiment” in the FRED search box and graph the first series that is returned. Repeat the process above to add industrial production and retail and food services as additional lines on the graph. Once all three lines are added, select “Edit Line 1” on the “Edit Graph” menu and change the units to “Index.” Set the date to October 2016 and click “Copy to all.”

Suggested by Maximiliano Dvorkin and Hannah Shell.

View on FRED, series used in this post: A191RL1Q225SBEA, GDPNOW, INDPRO, RSAFS, STLENI, UMCSENT

Not all recessions are created equal

Differences in European unemployment rates

Posted on April 24, 2017

The previous recession was a worldwide phenomenon. It originated with a financial crisis in the United States that resonated in other countries, in particular Europe. The graph above shows the unemployment rate for the U.S. and a few European countries. It is taken from the OECD’s Main Economic Indicators, which goes through the trouble of trying to harmonize the definitions across countries, thus making them comparable. What is striking is how varied the experience has been. The gray area represents the period of the U.S. recession. It is remarkable that Germany’s unemployment rate actually was going down through much of this period. In contrast, unemployment shot up in Spain and, to a lesser degree, in Italy. And the U.K., arguably with the strongest financial ties to the U.S., experienced a relatively minor increase in unemployment. How can such varied experiences be explained? For one, the financial crisis was not the only economic event happening across those countries. Second, the labor market institutions and traditions differ a lot as well. Spain in particular is a poster child of rigid labor laws, and Germany was still in the transitional phase of labor market reforms.

How this graph was created: Search for “harmonized unemployment rate total,” then use the tags in the side bar to limit choices to frequency “monthly” and “seasonally adjusted.” Check the countries you want to display and click on “Add to Graph.” Finally, let the sample period start in 2002.

Suggested by Christian Zimmermann.

View on FRED, series used in this post: LRHUTTTTDEM156S, LRHUTTTTESM156S, LRHUTTTTGBM156S, LRHUTTTTITM156S, LRHUTTTTUSM156S

The burden of housing

Population density means higher living costs

Posted on April 20, 2017

The map shows, for each U.S. county, the percentage of households that are “burdened.” That seems to be a rather vague term, but in this case it has a precise definition from the U.S. Bureau of the Census: A household is considered burdened if it has to dedicate at least 30 percent of its income to rent or mortgage payments. Clearly, there are two components that can make a household burdened: low income and high housing costs. You could probably conceive of stories for the reasons that various parts of the country are more or less burdened. (On the map, the darker the color, the more burdened the county.) Consider that major population basins have higher housing costs, the South is generally poorer, and Florida has a lot of retirees on fixed income. Of course, one shouldn’t be surprised that housing costs are higher where there are more public amenities—which is where population tends to be denser. And, of course, some households may simply choose to spend more on their homes.

But why is 30 percent used as a threshold for burden? This is the maximum that is considered by rental assistance programs as well as guidance by mortgage providers. The concern is that households should have 70 percent available for other necessities. This problem tends to apply only to poorer households, although this map covers all households.

How this map was created: The original post referenced an interactive map from our now discontinued GeoFRED site. The revised post provides a replacement map from FRED’s new mapping tool. To create FRED maps, go to the data series page in question and look for the green “VIEW MAP” button at the top right of the graph. See this post for instructions to edit a FRED map. Only series with a green map button can be mapped.