Jul132011

The market value of Irvine home features by Global Decision and IHB

Over the last two posts on the Irvine Housing Market, we have looked at aggregate totals for Irvine and at individual neighborhoods. The nitty-gritty is in the details. What features in homes do buyers value in the market? Today we take a careful look at the devil in the details.

A presentation by Jaysen Gillespie of Global Decision

[email protected]

Global Decision is an analytics consulting firm.  While our methods are not industry-specific, our engagements are skewed towards specific industries in Southern California, such as real estate (along with online gaming and restaurant chains).  We specialize in applying both foundational and advanced analytics to better understand business and economic issues.

Today’s post is part three of our series on hedonic housing valuation in Irvine.  The goal of a hedonic housing valuation model is to use all information about a sale, including both the sale price and the characteristics of the home (number of beds, number of baths, square footage, etc.) to understand how the home’s value is derived from its constituent parts.  Wikipedia offers a good overview of hedonic regression or see the Global Decision tutorial on how to build your own hedonic regression model.

What factors are used in Hedonic Housing Models?

The choice of which factors to include in a hedonic housing model, or any regression model for that matter, are often driven by (1) availability of data and (2) general insight from the model builder or wisdom of the team.  In our case, a background review of other hedonic models along with generally accepted intuition drove the initial set of factors.  For some example results, please see http://www.hnb.hr/publikac/istrazivanja/w-019.pdf.  The .pdf link is a well-written and thoughtful hedonic model created by the National Bank of Croatia to better understand home price trends in that country.  On page 22 of this English-language document, the authors list the factors that fed into their model (which focused more on attached housing, the most prevalent dwelling type in European cities).  For the city of Zagreb, they have many of the same factors as we have for Irvine:  floor area, number of rooms, garage, neighborhood, and a few specific to them such as type of heating, presence of an elevator, and other items that generally add value in urban attached housing units.

What factors are using in the Irvine Hedonic Housing Model?

In Irvine, it turns out that most of the variation in home values is captured in a small number of factors.  For our model, we us the following list:

  • Bedrooms
  • Bathrooms
  • Lotsize
  • Square footage of house
  • Number of garage spots
  • Number of stories
  • Age of structure
  • Neighborhood
  • (Year-Quarter)

What about the quote that you can make an elephant’s trunk wiggle with 5 factors?

Overfitting of a model occurs when you use so many factors that you are losing the general meaning of each factor and instead creating a level of fit that is very specific to the data you are analyzing.  There is always a danger of overfitting when using small datasets with lots of factors.  Because our model has over 7,000 data points, using 8 factors plus time is not likely to produce overfitting.

One useful output of the Irvine Hedonic Housing Model is that we can understand how the factors associated with a property drive market value.  We looked at the “neighborhood” factor in detail in the previous post, and today we explore results from the other factors.

Good models often confirm the intuitive:  all else equal, we’d expect a one-story property to command a premium and a 3-car garage is a plus for most buyers.  The numbers reflect these facts.  A one-story property commands a premium of 4.3%, and a three-story property creates a reduction in value of 3.8%.  While the direction of these impacts is intuitive, the model helps us understand the best estimate for the magnitude of such impacts.

Add a bedroom:

The impact of just “adding a bedroom” is often near zero.  To understand this result, first recall that the result is really the impact of “adding a bedroom where all else is held constant.”  So a Woodbridge, 2-story, 1977, 3-bed/2-bath, 1800 sq ft home becomes a Woodbridge, 2-story, 1977, 4-bed/2-bath 1800 sq ft home.  Yes, you’ve added a bedroom – but you’ve really stolen the space from other rooms because the total square footage of the home is held constant.  For this reason, it’s not uncommon that hedonic housing models actually produce a slight negative impact from just the addition of a bedroom.  Most homes are constructed so that the addition of a bedroom is coupled with the addition of square footage – leading to an overall gain in value given both impacts.

Add a bathroom:

Irvine buyers love bathrooms.  Each one adds almost 4% to the value of the property.  So if that’s the case, why don’t we see 3-bed/6-bath properties?  The Irvine Hedonic Housing Model fits a relationship between property attributes and value based on the data fed into the model.  The vast majority of homes in our dataset have between 2.0 and 4.0 bathrooms.  Thus, the specific relationship indicated by the model is generally valid in that range.

Three Car Garage:

A three car garage adds 5.3% to the value of a property, relative to having the baseline 2-car garage.  One can debate the impact of a 3-car garage on the aesthetics of a SFR, but it’s beyond question that the market value of a home is increased with such an attribute.  As an aside, the standard error of that particular measure is just 0.3% — meaning that it’s very probable that the 5.3% estimate is close to the true incremental value of the 3-car garage.

Age of Structure:

All else equal, adding a year to the age of the structure creates a small reduction in value (0.6%).  Over many years, however, this creates a large premium for newer properties.  There are many possible reasons for this situation:  newer properties have lower repair and maintenance costs; they are built from superior materials (no lead/asbestos); they have more functionality (electrical outlets on every wall); and they are more energy efficient; and they meet modern design aesthetics.

Size (Square foot of home and lot):

To model the impact of a larger home or lot we need to take a slightly different approach.  Empirical evidence strongly suggests that each square foot of home size adds a fixed amount of dollars.  As an example, consider the following chart from last week’s post:

You can see the generally linear relationship between the square footage of a property and the sales price.

In order to model this in our model (which predicts the % change in values) we have to change our thinking.  Instead of looking at adding an extra 100 sq. ft. of size, we need to think exponentially.  Our model looks at adding x% of size to the house.  By doing so, we end up with a value flow-through ratio.  We define this flow-through ratio as the ratio of increase in size to the increase in property value.  In our case, we have a flow-through ratio of 44% for square footage and 11% for lotsize.  If we double the size of the square footage, we generate a 44% increase in property value.  Similarly, for lot size, if we double the size of the lot, we generate an 11% increase in property value.

Because “doubling” of either lotsize or property square footage is not often a reasonable design choice, we prefer to couch our metrics in terms of “adding 10% to” the relevant factor.  An increase of 44% for doubling of size works out to an increase of 3.7% for each 10% increase in size.

Have Irvine buyers changed preferences over time?

An interesting discussion regarding the Irvine housing market often involves much speculation about the buyers themselves.  Between the years of 2000 and 2011, much has varied about how buyers have financed their homes, the types of homes purchased (newer vs. older), and even the size of the homes.  We can also look at how the results of the Irvine Hedonic Housing Model vary over time by constructing separate models for various time ranges.  To investigate this issue, we’ve created four models that break the 2000-2011 timeframe down into 2000-2003, 2004-2006, 2007-2008, and 2009-2011.  We’ve determined the “value add” for the factors listed above for each timeframe the show the results below:

One conclusion from the above chart is that Irvine buyers have consistently assigned similar values to the underlying factors of the properties purchased.  With the exception of an extreme spike in the value-add for a 2007-2008 “one story” property, the other value-adds are relatively constant over time.  This is a surprising result:  we know that property financing was dramatically different in pre-bubble and post-bubble periods.

There is also much talk about the “foreign cash buyers” (FCBs) and other buyer prototypes that impact the Irvine housing market.  We can now rest assured that these shadowy figures assign the same general valuation to the features of the property as John and Jane Q Public who are out shopping for a home.  Another possibility is that the ratio of buyer prototypes active in the market has remained constant over the 11 year period of study.  Perhaps the FCB has consistently purchased the same percentage of Irvine housing stock over time.  Further study of loan data by property is needed to confirm or reject theories regarding composition of the buyer pool.

The one variable that appears to be trending down in impact is age:  the fact that homes are older appears to be of less concern to 2009-2011 buyers (vs. 2000-2003 buyers).  This is, in part, explained by the fact that Irvine experienced a number of new neighborhoods from 2000 to 2011.  Back in 2000, there was little “new” housing in Irvine – almost all sales were between 15 and 32 years old.  In 2010, there was a sizable cluster of homes between 4 and 15 years old and a cluster of homes between 22-26 years old, and then homes 30-40 years old.  This distribution reflects the boom-and-bust history of Irvine home values.  When values are dropping (see 1991-1996), demand is low and few homes are built.  As the vintage Irvine neighborhoods get older, the average degradation in value (per year) is reduced.

IrvineRenter’s Commentary

Having worked for homebuilders most of my career, I was intrigued by the data on home features (below).

Builders have long known there is a penalty for only having one bathroom. Originally, one bathroom homes were created for cost: bathrooms are expensive. However, over time, it became apparent that buyers were willing to put dollars toward bathrooms, so builders seldom fall below the standard of two bathrooms per home. For a time, builders experimented with larger and larger bathrooms and more of them. There seemed to be no limit to the buyer’s appetite for bathrooms. Have you noticed how master bathrooms are almost as large as the bedroom itself? This is no mistake. Buyers are demanding this space.

Builders have also wrestled with the provision of a third story to add square footage to houses with very small ground floor footprints. Buyers don’t like them. The third floor is generally limited to 500SF for code reasons, so when builders add this space, they can only add 500SF. Since they also diminish the value of the final product by 3.8%, there is almost no additional value created in third story construction. The Irvine Company has experimented with this design in several communities, and it continues to prove unpopular.

The addition of a third garage stall has also been a difficult builder decision. If these three garage stalls are side-by-side, it makes the house much wider, or consumes the entire front elevation which makes the house unattractive. There is unquestionably an increase in value when adding the third car, so builders and architects have experimented with tandem garage arrangements where three cars can fit into the width of two. The three-car with tandem configuration is popular because most people use the tandem space to store their stuff. Rarely do people actually park three cars in their three car garages. As long as the third garage stall — which is very inexpensive square footage — adds 5.3% to the value, expect builders to provide this space. They would be foolish not to.

I was not surprised to see additional lot size added little or no value. The Irvine Company figured this out years ago. Lots in Irvine are 10% to 20% smaller than lots in surrounding communities, yet Irvine actually commands a premium. On a per-square-foot of lot basis, the Irvine premium is remarkable. And since the Irvine Company is in the business of selling land, obtaining a huge land premium is the focus of all their efforts. I expect to see lot sizes shrink even further as the Irvine Company increases density to further increase their land value. This may create opportunity for competitors, but the Irvine Company will likely ignore what competitors are doing and focus on increasing their land value.

I was not surprised to see the preference for new or the flattening of values as properties age. A 40-year old property is not much different than a 25-year old property if both have received the same owner care. I rented a 40-year old property in University Park for over three years that was renovated prior to my moving in. The property felt new to us, and I wouldn’t have valued it any more or less than a comparable property in Woodbridge that may have been 15 years newer. Newness wears off after about 10 years, and properties older than that are valued based on how recently the interior was updated.

I want to thank Jaysen again for this great analysis. I hope you have found it as valuable and as enlightening as I have.