What are Mortgage Purchase Applications?

Today’s news on mortgage purchase applications was disappointing as it fell 8 percent from the prior week.  This data is collected by the Mortgage Bankers Association from its members.  Because it is collected on a weekly basis, it is considered one of the timeliest of economic indicators.  As the name implies, mortgage purchase applications capture the trend in the number of application filings for a home purchase.  We should note that it measures applications only and not approvals.  As with any data measurement there will naturally be statistical noise - sampling and self-selection biases.  It is further possible for those with approved mortgages to cancel the deal for whatever reason – including the frustrating recent appraisal issues.  In addition, the mortgage application completely misses out on the trend of all-cash home purchases, which have risen significantly in the past two years.  About a quarter to a third of all completed transactions were all-cash, compared to the roughly 10 percent that is the historical normal average.  Miami and Las Vegas have experienced upwards of 60 to 70 percent of completed deals being all cash.  Due to these reasons there will never be a tight correlation between mortgage purchase applications and home sales closings.

Still, week-to-week trends are interesting to monitor.  The latest index reading of 167 is the lowest since the May, June, July period of last year, the immediate months after the deadline for the home buyer tax credit eligibility.  Comparisons of now versus one year ago show that the economy added 1.3 million net new jobs, there are about 3 million more people living in the country, home prices are a bit lower, stock market wealth is a bit higher, and mortgage rates are lower.  In other words, all the factors for higher home sales and applications are present.  Unfortunately, consumers are the ones ultimately deciding on applications (with banks deciding on approvals) and it is frustrating to observe such soft activity on mortgage purchase applications.

However, we need to be constantly mindful that all data are not sacrosanct and will contain measurement errors as mentioned above.  In particular, the mortgage purchase data at times has had some very strange readings.  For example, mortgage data implied that there was an approximate quadrupling in home sales from 1990 to 2000 as the index rose from 100 to 400.  Existing home sales over that time period rose by only 3.2 million to 5.2 million, not even a doubling of sales.  New home sales rose from 533,000 to 880,000, again less than doubling sales.  Clearly, there was a huge upward data drift in the mortgage purchase data over that time period.  In recent times, since the financial crisis of 2008, there has been notable mortgage industry consolidation.  It is unclear if this trend had any impact on the data collection.

Separate data on mortgage refinance activity, however, has witnessed a good uptick in the past few months as some homeowners with equity are taking advantage of falling mortgage rates.

The point of this short article is that while weekly mortgage data are useful in helping us gauge the market conditions, there will likely be some degree (small or meaningful) of upward or downward measurement errors.  Therefore, we should not be too tightly hung up on the precise meaning of the reported magnitude changes.  The bias is unknown at the moment, but let’s hope that the latest soft reading is either a one-week fluke or contains an unknown downward measurement error.