The term "sales volume" applied to the housing market simply refers to the quantity of homes sold in a certain area during a certain time period.  As such, the volume of sales need not directly affect the decision of whether and when to purchase a house.  However, for those interested in following the housing market, sales volumes are a key indicator of the "health' of the housing market, where declining sales volumes tend to be associated with declining prices.

There are two main sources of data on sales volumes, depending on whether one is interested in existing homes or new homes.

For existing homes, the main data source is the National Association of Realtors.  NAR publishes homes sales data monthly.  Note that the data refer to homes sales entering "escrow", which is 6-8 weeks prior to the final closing of a sale.  So the July data, for example, publishes in late August, really reflect activity around May.

If we plot existing sales over time, we see that the housing market peaked (in terms of volume) in 2005, and through July 2007 is off approximately 20% so far.  However we can also see that sales volumes remain far above "typical" levels, and so may still have a long way to fall.

Why do existing home sales matter?  Unlike new homes sales, they do not contribute directly to GDP. But clearly the rate of existing homes sales matters a great deal to those who make money when people move house - especially realtors, but also mortgage bankers and brokers, furniture stores, not to mention the owners themselves.

For new homes, the main source of data is the Census Bureau.  Note that unlike the NAR data on existing home sales, which includes all categories of housing, the Census Bureau data only includes single-family homes (and not condos, etc.). As a result, the data can skew a lot of big metropolitan areas like New York City and Chicago, where the vast majority of residences are condos and apartments. The data are, however, more current than the data on existing home sales, since they reflecting contract signings from the previous month.

The size of the recent housing bubble is more obvious in the data for new home sales than it is for existing home sales - sales activity peaked at around 1.4M per month, whereas previous cycles peaked at under 900K per month.  And just as the peak is higher, the fall has been greater - around 40% so far, vs under 20% so far in existing home sales.

The real picture is probably worse than shown in the graph, because the data do not include cancelations (which is when buyers sign up for a home and then "cancel" their order). Cancelations have increased dramatically in the current housing downturn, to levels of 40% for many home builders, so that the decline in the home builders' fortunes is even worse than even the precipitous drop shown in the graph would imply.   

Why do new home sales matter?  First, unlike exsisting home sales, they do contribute directly to GDP.  When a new $1M home is sold, $1M of economic outout has been generated; that is not the case when an existing home changes hands at a price of $1M. 

Second, new home sales have been an uncannily good predictor of recessions.  4 of the last 5 recessions (generally defined as two or more succesive quarters of negative growth) were preceeded by periods of sharply falling new home sales, the exception was the 2001-2003 recession, which was associated with the ending of the internet bubble and a saw a sharp fall in business investment.

Additional Resources:

NAR Data

U.S. Census Bureau

Housing Starts Data