One More Advance in Creating a Better Price-to-Rent Ratio

In an earlier post, we described a better price-to-rent ratio and stressed the importance of comparing similar homes when looking at rental rates and home prices. In this post we aim to explore yet another approach that we believe is the most comprehensive and accurate measure of the price-to-rent ratio.

With the unique data that we have at Zillow we have four possible ways of calculating the price-to-rent ratio:

  1. We can use the age-old method of comparing the prices of current for-sale listings or recent sales to the prices of current rental listings. As we demonstrated here, this approach usually leads one to compare very different sets of homes because the set of homes for sale can differ quite a bit from the set of homes for rent. Note, these differences are typically more substantial than can be controlled for with just simple filtering (e.g., limiting the analysis to just two bedroom homes) and often a key difference between the two sets of homes is the geographic location of homes for rent versus homes for sale (e.g., rentals may often be located closer to the urban core or in areas with less expensive home values). In short, with this approach one is usually comparing apples and oranges.
  2. We can use current rental listings and pair the rent price with the estimated home value (Zestimate) for each rental listing. This is exactly the approach we analyzed here and found to be superior to the first approach which is comparing dissimilar sets of homes.
  3. We can use current for-sale listings (or recent sales) and pair the listing (or sale) price with the estimated rent price (Rent Zestimate) for each listing. This is the converse of Approach #2.
  4. We can use all homes that exist in an area and pair the Zestimate with the Rent Zestimate for each home. This approach is the most inclusive and informative one. It covers the largest number of homes since we are not dependent on only using homes that are listed either for sale or for rent. Therefore we are able to calculate price-to-rent ratios for many more regions, including regions that are much smaller and usually would have too few observations. Moreover, this approach makes the computation less vulnerable to biases that may be introduced when either the set of for-sale or for-rent listings are not representative of the housing stock or when the listing mix changes over time.


Using Approach #4, we calculate the price-to-rent ratio by dividing a home’s estimated home value (Zestimate) by twelve times the estimated rental price (Rent Zestimate). Intuitively, this ratio is interpreted as the number of years of annual rent that would be required to purchase the home. We repeat this procedure for the almost 100 million homes in our database having both a Zestimate and Rent Zestimate. This approach ensures that we are comparing rents and home values for the same set of homes and that this set remains consistent over time.

Table 1 shows the price-to-rent ratios for over 200 metropolitan areas. The “best” price-to-rent ratio is calculated using approach #4 described above, using the Zestimate and the Rent Zestimate. The third column of the table shows the “bad” price-to-rent ratio, which is calculated using the standard approach of comparing median listing prices to median rental prices (i.e., comparing two different sets of homes) as described in approach #1 above.

Notice that most of the “bad” price-to-rent ratios are higher than the “best” ratios, which can also be seen in Figure 1, where most of the dots lie above a 45 degree line (if the two approaches were identical, the points would all be located along this line). One possible explanation for this pattern is that rental homes may, in the aggregate, have a lower home value than for-sale homes, a pattern that would lead the “bad” ratio to be positively biased. Table 1 further presents data that suggests that this pattern is, in fact, exactly what we find. The last two columns are the estimated home values of current for-sale homes and current for-rent homes (i.e., the set of homes used to compute the “bad” ratio). Notice that in almost all cases the median home value of homes listed for sale is higher than the median home value of homes listed for rent, demonstrating again that these homes are not of the same type and will therefore produce higher price-to-rent ratios than is actually the case in the given metro region. In some rare cases the homes listed for sale are of lower value than the homes listed for rent and then, as we would expect, the price-to-rent ratio is actually lower than the “best” ratio.

Figure 2 shows the “Best” Price-to-rent Ratio over time. Again, this better approach has the advantage of not being subject to a changing mix of homes over time, since we cover all homes on Zillow at all times.

In closing, it matters a lot that one compares apples to apples when computing metrics that purport to provide insight into the relative advantages of owning versus renting an asset. Metrics that examine rent versus buy but do so using substantially different types of assets should be avoided. We think the approach outlined herein provides the most straightforward, stable and comprehensive metric comparing price to rents.