[NOTE: Occasionally, data sources will discontinue a series, and that is the case here. ] 

How does the distribution of house sales change over time? FRED includes median and mean sales price data for single-family homes, which can give us a hint. The difference between the two series increases when a proportionally greater number of more-expensive houses are sold, for example. The graph above shows the two series, and it is difficult to see any changes in the difference between them: The lines look parallel through booms, busts, and seasonal fluctuations. (The higher prices in the summer and lower ones in the winter may come from both actual seasonal price fluctuations for equivalent houses and from the changing composition of the sold houses.)

The graph below shows the ratio of median to mean prices. We see, for example, that the ratio went down during the recent crisis and has more than recovered since. This movement is consistent with proportionally more high-end single-family houses being sold (or fewer low-ends ones), with a reversal around 2011.

[NOTE: Occasionally, data sources will discontinue a series, and that is the case here. ] 

How these graphs were created: For the first graph, go to the National Association of Realtors (under sources) and look for the Existing Home Sales release. Select the median series and then the mean series and add them to a graph with a click. Note that you can select single-family homes or all homes (including co-ops and condos), which will show very similar results. For the second graph, remove the mean series, but then add it back by choosing “Modify existing series 1.” Then use “Create Your Own Data Transformation” with formula a/b.

Suggested by Christian Zimmermann