The research and modeling team at Freddie Mac Multifamily is always looking for new ways to provide insight and tools that support the multifamily rental market. In recent months, we’ve released lookup tools that highlight mission-focused areas and help renters identify properties that are affected by forbearance. Additionally we’ve created several reports that track the performance of our SBL and floating-rate loans.

Another important focus area of my team is modeling and data analysis. We’re currently partnering with our digital transformation group to create tools that streamline business practices and the user experience. In addition, I’ve recently put my modeling skills to use and co-authored a white paper titled, “Beyond the Cap Rate: Valuation of Multifamily Properties,” that closely looks at the methods used to evaluate and report multifamily property values. I am excited to share some of the key findings from this work.

Valuation Methodologies

Estimating the value of multifamily properties and other commercial real estate properties is an important but difficult task for real estate investors and researchers. Models that are commonly used, such as the capitalization rate (cap rate) model and trend-based mark-to-market approach, have limitations in accurately valuating properties in times of financial distress or market volatility.

This report introduces a new property valuation methodology that directly estimates the value of multifamily properties by accounting for both market trends and property operating performance. Our results show that the accuracy and valuation of multifamily properties was significantly and consistently improved when employing the model. These results were also true for other types of commercial real estate properties and was especially notable in valuating distressed properties.

Model Framework and Data

Modeling this new method was tricky, however we were successful in identifying the variables and process that helped to improve and better define concepts around cap rates. An important feature is that we developed advanced algorithms that take into consideration property-specific characteristics that affect property values. These include location, size, year built, metro level and neighborhood effects, overall market trends, and the financial stability of the property. From there, we created a model framework chart and formula. Using this framework, the property value can be derived given the predicted property appreciation rate and the initial property value. The chart below presents the relationship between individual property price growth and market trend.

property value growth

Estimation Results

Using a multi-step approach, we analyzed the empirical data by examining the individual importance of the market price trend and the net operating income (NOI) performance in the multifamily valuation by including one effect at a time in our estimation. The five effects are:

  1. Include only market price in estimation
  2. Include only relative NOI growth
  3. Assume the NOI effect to be different for properties that out/underperform the market
  4. Combine market price trend and relative NOI growth and allow for different NOI effects for out/underperforming
  5. Include property-specific characteristics and estimate their parameters
Estimation results chart image

Back-Testing Results

One of the unique advantages of the new model is the improved accuracy in the valuation of distressed properties. By comparison, the direct cap rate approach often undervalues these properties. To back-test our model, we used a sample of Freddie Mac Multifamily properties from the watchlist records and then assessed these using several values for the new method as well as the cap rate and mark-to-market models.

preiction accuracy chart image

Even in the extreme case, with the distressed indicator being 1 or 0, the new model standard deviation values were more accurate than the cap rate model, confirming the limitation of the cap rate approach on valuing distressed properties. These results further concluded that our model outperformed both of the other valuation models, with smaller mean squared prediction error and lower volatility. The volatility is also significantly lower than the other two models.

In summary, this new multifamily property valuation approach provides more accurate insight into credit risk evaluation and improves our multifamily business practices. If you would like to learn more about this new tool, our methodologies, or how we are applying analytical modeling in our research, please feel free to email me directly. You can download the full report here.