Federal Reserve Economic Data

The FRED® Blog

Foreign exchange intervention

Foreign exchange intervention is the buying and selling of foreign currencies by central banks and finance ministries to influence the value of their own currencies. It has become rare for developed countries to intervene in foreign exchange markets. The last U.S. intervention in the foreign exchange market was in March 2011, as part of the G3 intervention with Japan and the European Central Bank (Neely, 2011).

In the bad old days, it was very difficult for researchers to obtain data on foreign exchange intervention, which central banks and finance ministries often treated as confidential. Beginning in the 1990s, however, central banks began to release foreign exchange intervention data to researchers. As intervention became rarer among developed economies, the trend toward releasing intervention data accelerated. Still, researchers needed to contact central banks directly to ask for such data.

About 10 years ago the Federal Reserve Bank of St. Louis began contacting central banks and finance ministries to obtain intervention data to post publicly on FRED to facilitate research on foreign exchange intervention. This dataset has been a little-known feature of FRED, but has been very useful to researchers. The countries that have contributed data include Australia, Germany, Japan, Italy, Mexico, Switzerland, Turkey, and the United States.

How this graph was created: Search for “foreign exchange intervention” and choose the series you want to graph. Note that currencies differ. Here, we chose the ones related to the Japanese yen.

Suggested by Christopher Neely

View on FRED, series used in this post: JPINTDDMEJPY, JPINTDEXR, JPINTDUSDJPY, JPINTDUSDRP

The unemployment bathtub

Economists often find a bathtub to be a useful metaphor for the behavior of unemployment. There’s some inflow of newly unemployed workers and some outflow as workers find jobs. A classic way to measure the inflow has been with initial claims of unemployment benefits, the blue line, in which we see spikes at the start of each recession. This inflow of newly unemployed persons initially reduces the mean duration of unemployment, the green line. But the green duration line rises as the blue initial claims line falls—since people who become unemployed early in the recession and remain so are unemployed for a while by the time the recession winds down. Every recession follows this pattern: Claims peak, then unemployment peaks, then duration peaks. The logic is essentially that of the bathtub: First it fills quickly; then, after some time, it begins to drain. But as this is happening, those left in tub have been there longer and longer.

How this graph was created: Search for and select the 4-week moving average of initial claims. Set its units as an index with scaled value of 100 at the 2007 pre-recession peak. Then use the “Add Data Series” option to add the other two series: the seasonally adjusted civilian unemployment rate (with the same units as the first series) and the seasonally adjusted mean duration of unemployment (with the same units as well).

Suggested by David Wiczer.

View on FRED, series used in this post: IC4WSA, UEMPMEAN, UNRATE

Is the PPI going crazy?

The graph above shows the producer price index since 1913. It measures the cost of items used in the production process and is thus different from the consumer price index, which measures the cost of final goods to consumers. Two aspects of the graph are striking: Prices have increased quite a bit since 1913, and prices in recent years seem to be subject to wild fluctuations. There’s no doubt the ups and downs of commodity prices such as oil and metals have an effect here, but are the recent years really as wild as they look?

In part, the second observation is a consequence of the first. Prices now are roughly 18 times greater than those in 1913. So a 1% increase will look 18 times larger now than before. This “optical illusion” can be fixed in two ways. 1. Look at percent changes. The first graph below shows these changes from the same month a year before, which takes care of any potential seasonal effects. Recent fluctuations are indeed somewhat larger than in preceding decades, but they’re nowhere close to the large fluctuations in the first years of the series. 2. Look at natural logarithms. The second graph below includes a transformation so that any change in the series looks the same in relative terms: that is, a 1% increase looks the same in 1913 and 2015. Again, we see that the fluctuations were much larger in the early years.

How these graphs were created: Search for and select the PPI for the first graph. Change the units to “Percent Change from Year Ago” and you have the second graph. For the third graph, start with the first graph, choose “Create your own data transformation,” and select “Natural Log” among the transformations.

Suggested by Christian Zimmermann

View on FRED, series used in this post: PPIACO


Back to Top