In August 2012, using our second quarter data, we introduced a new approach to make the rent versus buy decision and computed a metric called the Breakeven Horizon at the city and metro levels. The breakeven horizon is the number of years after which buying is more financially advantageous than renting (at the precise breakeven horizon one can be indifferent between buying and renting). We computed the breakeven horizon for each household by comparing the costs of owning a home versus renting a home at the end of each year for 30 years (assuming the house is purchased using a 30 year fixed mortgage). Our buy versus rent analysis incorporated all possible costs incurred when purchasing a home as well as those incurred when renting a home to make the comparison between these costs as realistic as possible. The original methodology used to compute this metric is explained in detail here.

We decided to make further improvements to the breakeven methodology for 2012 Q3 based on our own introspection of the previous methodology and feedback provided by Leonard Baron, MBA, CPA; LPB Services LLC and Tim Ellis, Founder of Seattle Bubble. These updates enable a more realistic comparison of the net costs of buying versus renting and hence generate a breakeven number that is most meaningful to the concerned audience. Among others, they address itemizing annual interest payments and property taxes (AIPT) which is the norm for American homeowners and accounting for the fact that homes in different metros appreciate at different rates over the years. Following are the changes introduced in version 2 of the breakeven methodology:

Inflation Rate

We use the 2011-2012 percent change in the overall Consumer Price Index (CPI) from Bureau of Labor Statistics (BLS) for the first year. For years five to thirty, we use the historical median of the annual percent changes in CPI using data from the years 1980 to 2011. For the years two, three and four; we spline interpolate between the current and historical rates of inflation.

Property taxes

We used the most recent property taxes paid by each household in the costs of owning a home as well as to compute tax benefits in version 1. We use the same approach in version 2; however, use a constant property tax percentage of 1% of the current Zestimate to compute property taxes for properties in California to account for Proposition 13 in California.

Tax benefit

In version 1, we computed the yearly tax benefit as the amount of annual interest payments and property taxes multiplied by the marginal tax rate which we assumed to be 25% (average federal tax bracket as indicated by the National Bureau of Economic Research). However, only households whose annual interest payments and property taxes (AIPT) are greater than their standard deductions for a particular tax year can write off a portion of their AIPT and avail of tax benefits. To account for this, we now subtract the AIPT from the average standard deductions[i] and compute the tax benefit by multiplying the delta (only if it is greater than zero) with the marginal tax rate. For subsequent years, we inflate the average standard deductions using the inflation rate before subtracting the AIPT.

Home Appreciation Rates

In version 1, we used a global home appreciation rate of 2% for all the households for which the breakeven horizon was computed. To account for varying rates of home appreciation across metros in the United States and to address Tim Ellis’s concern about assumptions on rates of home appreciation over a long time period, we compute the metro level annual rate of home appreciation for the 30 year horizon as follows-

• For the first year, we use the September, 2012 ZHVI Forecasts produced for 253 metros in the United States.
• From the fifth year onwards, we use the historical median of annual rate of home appreciation. We compute the historical median rate of appreciation using metro level annual percent change in FHFA Home Price Index (HPI) data for the years 1980 to 2000. We exclude data after 2000 to avoid bias in our calculation of the historical appreciation rate due to the run up in home prices during the housing bubble.
• For the years two, three and four; we spline interpolate the rates of home appreciation between the ZHVI forecasts and the historical rates of home appreciation. We do this because some of the metros have a negative ZHVI forecast and it is reasonable to assume that these metros may experience a gradual shift to their positive historical home appreciation rate over the span of 5 years.
• For metros that don’t have a ZHVI forecast, we use the historical rate of home appreciation for all 30 years and for metros that don’t have a historical rate of appreciation we use the historical rate of inflation at the national level.

Rental Appreciation Rates

We used a global rental rate of appreciation of 3% for all the households in version 1. In version 2, we compute the rental rate of appreciation at the metro level as follows:

• For the first year, we use the September, 2012 Zillow Rent Index computed at the metro level.
• From the fifth year onwards, we use the historical annual rate of rent appreciation. We compute the historical rate of rent appreciation using CPI for rent and owners rent for the years 1980 to 2000 from the BLS. This data is available only for 26 metros, so for the rest of the metros we use the region and population level CPI for rent breakdown provided by BLS. We compute the historical rate as the median of the annual percent change in the CPI for rent and owners rent at the metro level.
• For the years two, three and four; we spline interpolate the rates of rent appreciation between the ZRI and the historical rate of rents appreciation.
  

Opportunity Costs

In the previous version, we used a constant annual rate of return of 5% to compute the yearly opportunity costs from purchasing and owning a home for the 30 year time horizon. In version 2, we have an increasing rate of return over the 30 year horizon with the implicit assumption that investments in financial assets over a longer time period would yield higher returns. Hence, we use rates of return from a combination of financial assets such as the 12 and 24 month term CD for the first two years, 5 and 7 year U.S. Treasury notes for year five and seven, Moody’s Aaa and Baa bond yields for year ten and year twenty respectively and for year 30, we use the historical average rate of return from the stock market. To compute the annual rate of return for the in-between years, we use spline interpolation.

Comparing 2012 Q3 with 2012 Q2

Despite having updated the methodology, we compare the Q2 and Q3 numbers as a useful exercise to see the partial impact (some of the impact is caused by the changing data inputs over time) of the methodological changes we have made. In a general setting, however, the breakeven horizon numbers for Q2 and Q3 are not comparable. We find that there are 15 metros with current average breakeven numbers less than what they were in Q2 and 35 metros with current breakeven numbers more than Q2 numbers (only counting metros where the absolute difference between the current and Q2 breakeven numbers is 1 or more years). Although all the changes incorporated in the current breakeven methodology contribute to differences between the current and Q2 breakeven numbers, we think these are mostly driven by variance in metro level home and rental appreciation rates.

The metros with the largest quarter-over-quarter differences in breakeven horizons are San Jose, CA with a current breakeven of 3.7 years compared to 8.2 years in Q2 (difference of 4.5 years) and New Haven, CT current breakeven is 7.0 years compared to 3.0 years in Q2 (difference of 4 years). The home appreciation rates in San Jose, CA ranges from 4.2% to 6% and in New Haven, CT ranges from -3.4% to 0.7% over the span of first five years of owning a home. Since homes in New Haven, CT historically appreciate at a mere 0.7% as opposed to 6% in San Jose, CA; it takes much longer to breakeven in New Haven than it does in San Jose.

In the previous breakeven methodology, the breakeven numbers were primarily determined by home values and rental values. In the current methodology home and rental values as well as home and rental appreciation rates influence breakeven numbers which is definitely a more realistic scenario. This is due to the fact that even though higher priced homes increase the net costs of owning a home; they can also be a profitable investment depending on the rate the possibly affluent neighborhood it belongs to appreciates over the years. Also, the historical rate of rental appreciation in New Haven, CT is 2.5% and hence it is profitable for one to rent if one is not planning to stay there for at least 7 years. The historical rate of rental appreciation for San Jose, CA, on the other hand, is much higher at 4.7%; so it makes sense to buy a home if one is planning to stay in San Jose for at least 4 years.

At the city level, we find considerable variance in the breakeven horizons within a metro. This is shown when we select a particular metro from the list of metros and observe the breakeven numbers for the cities within that metro as listed in the bar chart below. For example; the breakeven horizon for the New York metro is 4.9 years; however the city level breakdown of the New York metro reveals that the breakeven numbers range from 1.5 years in the city of Rossmoor to 16.5 years in the city of Sleepy Hollow. The possible reasons for a wide spread in the breakeven horizon within a metro are due to large city level differences in home values, rental values and property taxes which are reflected in the breakeven horizon computation since these numbers are computed at the household level and then aggregated to the city and metro level.

Zillow will use the previous breakeven methodology in conjunction with the updates listed above to publish the average and median breakeven numbers at the metro and city levels, and update these numbers periodically.