Homeownership is a key housing and economic measure of social conditions. Due to the housing downturn, the national homeownership rate declined precipitously, and has failed to recover to its historical norm. In fact, the homeownership rate reached a 50 year low in the second quarter of 2016 and has only edged up slightly since then, rising to 63.5% at the end of 2016.

The following analysis uses county-level data to explore how changes in key variables (age, marital status, income, and home prices) affect local homeownership rates. The results confirm that among other factors, lower housing costs are positively correlated with homeownership.

**Background **

According to the latest data from the American Community Survey the mean homeownership rate is 71.4% among all U.S. counties (i.e. the average rate for all counties). **Figure 1 **shows the county-level variation in homeownership rates for the continental U.S., and **Figures 1A** and **1B** show the rate variation in Alaska and Hawaii, respectively (Scale in **Figure 1** is the same for **Figures 1A** and **1B**). There are notable areas in the U.S. that are below this typical rate. For example, a number of counties in California have homeownership rates that are lower than the mean, as do clusters of counties throughout the Southeast region. The same applies to some Northeast counties that have large cities.

**Regression Model**

Using data from the 2015 American Community Survey (5 year data), a simple regression model was constructed that takes a deeper look at factors affecting homeownership on a county level (linear-log OLS model). The independent variables included in the model are median age, the share of households with married couples, median home value, and median household income. The dependent variable in the model is the homeownership rate, or the share of owner-occupied households. Each variable included in the model is statistically significant. The independent variables are jointly statistically significant (F-statistic=1266.81) and the R-squared is 0.617, which means that 61.7% of variation in the homeownership rate can be explained by the model.

**Homeownership Model Scenarios**

With the model, a series of scenarios were created to show the estimated impacts on homeownership due to changes to the independent variables.[1] First, a “mean case” or typical county was created. The “mean case” county has a population with an average of age of 41, 51% of households consist of married couples, the average home value is approximately $134,000, and the average household income is about $47,000. With these assumptions, the model predicts that the average homeownership rate of the “mean case” county is 71.5%, as seen in **Figure 2**.

**Figure 2: County Homeownership Rate Model: Variable Simulations**

**Figure 2** also displays homeownership rates for each scenario in which each variable is changed approximately by its standard deviation among counties. The first scenario shows when the average age in the county is increased and decreased by five years. When the average age increases to 46, the county homeownership rate increases from 71.5% to 74.5%. On the other hand, if the county average age drops by five years to 36, then the homeownership rate slips to 68.1%. This scenario shows that age has a positive impact on the homeownership rate: as the average age increases, the homeownership rate increases.

Another variable with a positive effect on county homeownership rates is the share of “married” households. In the “mean case” scenario, the share of married households is 51%. When the share of married households increases by 10 percentage points to 61%, the county homeownership rate jumps from 71.5% to 76.6%. In contrast, when the share of married households decreases by 10 percentage points to 41%, the homeownership rate drops to 65.2% from 71.5%.

Another variable included in the regression model is home value. In the regression model, home value has a negative effect on county homeownership rates. In the “mean case” scenario, the average home value across counties is about $134,000. When the average county home value increases by $75,000, the county homeownership rate dips slightly, going from 71.5% to 69.2%. If the average home value decreases by $75,000, it results in a higher homeownership rate of 75.6%. The results indicate that homeownership is more attainable in areas with improved housing affordability and low costs of new home construction.

The final variable included in the regression model is county household income. In the “mean case” scenario, household income is approximately $47,000. When this is increased by $10,000, the county homeownership rate increases slightly to 72.1% from 71.5%. Decreasing income by $10,000 has a small, but negative impact on the homeownership rate: it drops from 71.5% to 70.7%.

Overall, this model provides a first-pass analysis of the impacts that age, marriage, home values, and income may have on county level homeownership rates. Some of the explanatory variables are relatively fixed going forward, such as age. However, the results emphasize the importance of housing affordability on local homeownership rates.

__ [1] ____Data for all variables were taken from the 2015 U.S. Census American Community Survey (2011-2015).____ __