Revisiting the Purchase Mortgage Generation Rate

One of the key drivers of our forecasting process is what we call the purchase mortgage generation rate, or PMGR.  It is the annual rate at which a market generates purchase mortgages.  It is, literally, the percentage of households in a market that gets a purchase mortgage in a year.

A fundamental principle of our purchase forecast is that the PMGR for a market is a “sticky” metric that typically changes very slowly.  At the national level, we believe that there is a range around 4% where the PMGR is stable and sustainable.

During the 90s, the national PMGR hovered in a sustainable range around 4%.  Then, during the reckless easy credit period of the early 00s, home sales boomed and the national PMGR rose to over 6%, an unsustainable level.  The housing market crashed in 2007 and didn’t begin turning around until 2012.  From then until now, the PMGR has slowly risen every year and is only now approaching its former sustainable range.

The national market is comprised of hundreds of smaller markets, each with its own demographic and economic trends, and each with its own PMGR trend.  High PMGR markets may or may not be characterized by rapid household growth or by rapid economic growth.  They are simply markets where a higher share of households is buying homes every year.

The chart above shows the variability of PMGR trends across different markets.  The Tucson and Indianapolis CBSAs were among the top PMGR markets pre-boom.  Tucson had a bigger housing boom than Indianapolis, but both have PMGR trends that are again rising toward sustainable levels higher than the national average.   Cleveland had less of a run-up during the housing boom, but its sustainable PMGR level now appears to be about the national average, near 4%.  San Jose’s sustainable PMGR level is now significantly lower than the national average, about 2.5%.

These different PMGR patterns reflect the specific housing market conditions in each market.  They reflect demographic trends, including in- and out-migration.  They reflect homeownership rates, housing turnover rates, affordability, and even construction trends.  All these factors combine into this one, slow-moving metric that drives our forecast on a market-by-market basis.

In our previous blogs from last year, we discussed our purchase forecast process in the framework of households “ready, willing and able” to buy homes.  We split that framework into a two-part calculation.  The first part was the “able” component – an adjusted household count that reflected the reduced number of households able to buy a home due to the economic disruption of the pandemic.  The second part was the “ready & willing” component – the share of the adjusted household count that would actually make a home purchase.  This latter component was, in fact, an adjusted PMGR. 

As it turned out in 2020, the “able” component fell as many households dropped out of the pool of potential homebuyers.  Initially, we had also expected the pandemic to reduce or at least flatten the “ready & willing” component.  However, as the year progressed, it became clear that the “ready & willing” component (the PMGR part) rose more than enough to offset the declining homebuyer pool.  With more people working from home, those with the means were motivated to upgrade their housing situation, many of them to move up from renting to owning.  That led to the biggest purchase market the U.S. has ever seen. 

As we move through 2021, we expect the PMGR to continue to revert toward its historically sustainable range around 4 percent.  We believe purchase originations this year will come in about 1.6% higher than last year, over 4.9 million loans.  In dollars, we are forecasting $1.509 trillion, another new purchase volume record.

It’s worth noting how this compares to the previous purchase boom of 2003-2006.  In those years, purchase counts were all over 5 million and peaked at over 7.1 million loans in 2005.  However, in the fifteen years since then, average loan sizes have risen enough that this year will have the highest dollar amount ever.