For the last five years or so the Federal reserve has been working hard to keep interest rates low (by printing money and calling it “stimulus”).  Last week, they announced they would taper their stimulus, which should cause a spike in interest rates.  As a fellow real estate investor, I asked the obvious question: how will this influence the home prices?

Pundits love to get television and spout how increased interest rates will make it more expensive to borrow money, which in turn will deflate home prices.  Any talking head can come up with a theory, but you can’t place any value on it until there is evidence to support it.

So…what does the data say?

## Interest Rates and Home Prices

There are a number of interest rates to choose from, but for the sake of this post I’m going to use the fed funds rate (the interest rate at which banks lend money to one another).  You can make an argument for using a different rate, but the results are similar.  For home prices, I’m going to use the Case-Shiller Home Price Index (HPI).

In all of my examples I use annual data from January 1975 through January 2012.

Here is a scatter plot showing how interest rates and home prices correlate with one another.

Above I did a simple linear regression on the data.

If you’re not very statistically inclined, what you should care about is the R² (0.4305).  This number ranges from 0.0 to 1.0.  The higher the R², the more likely two things are correlated.

An R² of 0.4305 is low.

How low?  Well let’s look at another correlation.

## Population and Home Prices

At the end of the day, home prices should be driven by housing demand.  The most obvious way to calculate housing demand is by looking at the population.  So…let’s do that.

Here, the R² is 0.9376 (1.0 is the maximum).  That’s huge!  We can say with an incredible amount of certainty that home prices and the US population are correlated.  I could make an argument that home prices are set by population, but we don’t have enough data to say that for certain.

The difference between the R² of 0.9376 for the US population and 0.4305 for the Fed funds rate is enormous.

To drive that point home, I’m going to look at one other indicator.

## Oil and Home Prices

Let’s run our regression on oil prices versus HPI over the last 37 years.

That’s an R² of 0.543…which is higher than the R² for the Fed funds.

## Wrap it Up: How Much Does the Fed Taper Matter?

What’re the odds interest rates influence home prices?  Not much.

Indeed, oil has a higher correlation.  Now, I’m not saying oil prices drive the real estate market.  Rather, I’m using it as an example to show how insignificant interest rates are on historical home prices.

As a real estate investor you should be focusing on more fundamental drives of the market: like housing demand.

Photo: Paul Nicholson

During Kenny’s decade in finance he bought many single family rentals in rural areas, as a hobby. Along the way, he talked some brave souls into joining him as investors and recently retired from finance to take his hobby to the next level. Find more by and about Kenny on his personal blog and his recently created twitter account!

1. Is my understanding correct then in assuming what you were saying was the R2 is the impact on home prices a specific set of data represents as having an effect? The higher the value of R2 the more directly proportional the impact a set of data has as an influence on the HPI?

• It represents a correlation which is different from causation. This means that (as Kenny alluded to) population growth does not necessarily cause an increase in home prices however this data set does show a correlation. More data an imputs would be required to make the jump from correlation to causation.

• Yes, correlation cannot be translated into causation. Even if we see 100% correlation (or equal to 1) it does not mean that one correlating factor caused the other, it just means these factors have a correlation (they occur at the same time). I can find a correlation between “blue skies” and “cat’s purring,” it doesn’t mean that a cat caused the sky to be blue, or that the blue sky caused the cat to purr. There might be an owner that likes to pet the cat when the sky is blue, but we didn’t check that correlation. Or my husband’s drinking coffee in the morning does not cause me to drink my tea, or the other way around, we drink because we are thirsty. But if you correlate our morning drinking habits on a plot, they would correlate probably close to 100%.

• One additional observation is that a high correlation doesn’t imply causation, but a low correlation can increase the probability of no causation.

Granted, a complicated causation (like a long time to manifest), can still not demonstrate a high correlation, but as a general rule of thumb, I am more comfortable saying one thing doesn’t cause another when the correlation is low (like this example).

2. I love the data, although I do not assume that interests rates will rise just on the basis of the FED tapering. I believe the FED tapered because of growing strength of the economy and interest rates will also rise with increased economic health. It will be an interesting scenario.

• It will be an interesting scenario. Personally, I think it would be difficult for interest rates to not rise following the taper.

3. Have you run the numbers against unemployment?

How about at the regional level? Ad you drop down from 30k feet, it be fun to watch how the data sets are correletated.

• Intuitively, I would think that if interest rates (an aggregated number) aren’t correlated with total real estate prices, the regional data would be noisier and less correlated. I’ve not run the numbers though.

I have not run the numbers on unemployment. I considered it, but that’s a really hard number to come by. They regularly change the numbers to suit whoever the power that be are, so it’s a comparing apples and oranges.

4. I wish I were a little more statistically advanced and real-estate savvy, but it sounds like a convincing argument. Two things I am a bit incredulous about: the idea that the rates won’t rise due to the tapering, and that the population will continue to grow. I guess the latter is mostly dependent on immigration being worked out well at the Congressional level, and our country continuing to be a go-to place. The first is totally dubious; the second, less so. Also, some metro areas have more immigration than others, so it’s a bit local. I think my area is net-positive. People are leaving Ohio and coming to my state, for example.

I read a book called “Aftershock” by Weidemer (I also read the other one, by Reich) and it predicted in no uncertain terms the impending surge of interest rates, the popping bubbles, etc. For example, if housing prices are higher in ratio as compared to other causal factors such as median pay increasing over time (it hasn’t) and connected to interest rates which are in turn connected to many other things, then the environment is dangerous to invest in. They literally tell you to sell real estate and wait until all the connected bubbles have popped (the last, biggest bubble in the chain being the federal debt, which is completely unmanageable, and based a lot on the opinion that domestic and foreign bond buyers think of the U.S.). So, I’m not out of real estate investing, but I am going to hedge it by diversifying in 2014.
Incidentally, the Reich book is I think better regarded, and is based on the idea of income inequality, the squeezed middle class, and how those things are putting a lot of pressure on the economy since we are 2/3 consumer-based. That of course is bad news all around, for many reasons. But since we aren’t about to try to deal with income inequality head-on due to the political situation, it’s at least positive that there have been some good numbers on the economy lately. Still, mostly not attacking the core issue of the fact that folks can’t afford stuff like they could in the decades after World War II. The rich who have most of the money don’t spend it on goods and services like the other classes would, and they pay like 15% tax after loopholes. Well, that’s pretty much a tangent so I will stop.

I would be more sanguine about your ostensibly great blog if I knew more about economics and the housing market. At the lightest level of analysis, though, it is a good and a heartening post.

• Hey Jason,

Thanks for the comment.

Personally, I think interest rates will rise, and I have no opinion on population growth (all I said in this article is that in the past population growth has a high correlation with real estate prices).

You might be interested in another blog that I wrote about real estate vs stocks. One of the charts that I like is the one showing rental demand vs housing prices…both are a a measure of housing demand. It’s interesting that following the ’07/’08 bubble burst, they fell right back in line with their historical ratios: http://goo.gl/NLCJzA

Cheers,

Kenny

5. This is how I have typically focused on is demand vs supply. Here is something that was taught to me as an indication of how a certain market will pan out as far as housing prices are concerned.

1. Go down to your city planning or permits department
2. ask them about the number of commercial permits have been pulled in the last 12 to 24 months.

3. Commercial Permits means job openings or hirings are going to happen
4. If you study this in many areas you will get an indication of the number of jobs coming into a local area

5. If this happens, housing demand will begin to increase

If you have property in an area where alot of commercial is being built up, you should anticipate an increase of housing prices.

Love the article!

• Thanks for the comment Gerald. Commercial development and jobs definitely increase housing demand.

Cheers,

Kenny

6. Kenny,

Thanks for the post, it got me thinking about what in fact cause prices to fluctuate. I think that your methodology is off though. What you’ve done is take the value of the index and tracked it with rates, which ostensibly seems like it should yield the results your looking for, but because home prices tend to increase over time, what you’ve becomes less meaningful. For example, a value of the Case Shiller today at 180, vs. 1995 when the index was 80ish, does not take into account the steady drift of prices due to inflation. See below link.

http://research.stlouisfed.org/fred2/graph/?s%5B1%5D%5Bid%5D=SPCS10RSA

There are three possible ways, statistically, to correct for this that I can see:
1) Change in home prices annually vs. interest rate.
2) Inflation adjusted home values vs. interest rate.
3) Change in home prices vs. change in interest rate.

I tend to think the 1st or 3rd may yield more accurate results, because what you are really trying to examine is “Do rising interest rates have a negative impact (cause them to decline) on home prices. This makes me think #3 would yield the best results. Might make an interesting study…

Happy to discuss.

Riley

• Thank you for posting this – it’s exactly what I was going to say. Population goes up over time (at least in the US) as do prices due to inflation – so of COURSE they will be correlated.

I think you should use, as Riley suggests, inflation adjusted home prices v. average mortgage rates. Not sure how close avg mortgage rates correspond to the fed rate, but the fed rate is worthless in this comparison unless it is very very closely correlated to mortgage rates.

• Thanks for the comments folks! I see what you’re saying about there being a bias in my methodology. I’ll have a think about how to fix it. (This was more of an exploration I was doing turned into a quick and dirty blog post.)

Unfortunately, none of Riley’s methodologies are without problem as well:

1) The change in home prices is still going to have the positive skew you rightly pointed out the inflation will cause. Expected value of that series wouldn’t be 0.
2) Comparing inflation adjusted numbers with interest rates is…non trivial (I’ve not found a way to adjust for inflation that I like as they regularly change how these numbers are calculated, again for political reasons).
3) Looking at the home price delta has the same skew problems as 1.

Would love to work through this with you folks if you want to drop me a message on twitter (I don’t check my BP inbox often).

Cheers,

Kenny

• Hi Kenny,

Per your comment toward #1, it’s not necessary for the series to have an expected value of 0, be a perfectly normal distribution, or have no skew. Often a data series will have a positive expected value but can still vary dramatically with the independent variable (think S&P 500 returns and earnings growth).

I’ve put a bit more thought into this today, and I think that the #3, change in home prices vs. change in interest rate, is the right methodology, with one caveat, namely that the change in home price be 1 year lagged. Here is my logic. While the change in home price data series doesn’t have an expected value of 0, it does normalize the data, that is reset the data at each concurrent period to 0 and allow it to vary in proportion to the previous period’s level. The data will necessarily have a positive drift because inflation is (usually) always positive, and inflation and interest rate changes have a high correlation to one another as well (conjecture). You would use the 1 year lagged housing change because it takes time of the effects of an interest rate rise to show itself in real estate, much the same way that private equity returns are lagged because of infrequent strikes.

Would love to hear anyone’s critiques of the above, particularly Kenny’s.

7. Kenny, I understand the point that you illustrate. I would point out that very few borrowers (like none) can finance a home purchase at the Fed Funds rate.

I’m sure the (limited) correlation would also appear if you used the 30-year mortgage rate.

• I concur.

Cheers,

Kenny

8. I believe what the evidence supports is that long term, gradually increasing or decreasing rates have a stabilizing effect on prices of real estate (or any asset for that matter). Rapidly increasing or decreasing rates have the opposite effect. Combine “non-gradual” increases or decreases in interest rates with other targeted government action and you’ll see bubbles or busts. (which by the way are perfect opportunities for gov to step in and save us from ourselves)

In simple terms; let interest rates jump up 4, 5, 6+ percent in the next 6-18 months and tell me interest rates don’t effect real estate values. If the same rates come to be over a longer period, and as you point out, are done with demand in mind, the effects on value won’t be as evident. But rates do effect value!

I believe there’s no denying- significant decreases in interest rates over a relatively short period combined with other government policies which moved towards “easy money” in the same period led to the real estate bust of 2008- charts be damned!

• Thanks for the comment Payne. You talk a bit about regulatory impacts on real estate prices, and I concur they effect is significant. I don’t have a good way to quantify that though.

On the other point: significant changes in interest rates influences home prices, I’ve not seen any data supporting that stance. Granted, I’ve not looked very hard as the Federal reserve is good at making sure there is no massive spike/drop in interest rates (at least on paper).

Cheers,

Kenny

9. This sort of dialogue here is exactly why I joined BP. As a new person to real estate (in the private sector side), I hear lots of dread from my colleagues over the “monitory easing” thing, but maybe not so much…according the R factor comparison here.

I worked in the PLANNING DEPT for Kansas City, MO when for the first time ever recorded in the City’s history, NO BUILDING PERMITS were issue for several months during the last economic crash, which backs up Gerald Harris’s (above).

• Welcome to real estate and and thanks for the comment Lon!

Cheers,

Kenny

10. I strongly believe interest rates are tied to home prices. In my opinion, wages are the most important factor, which is affordability and affordability is tied to monthly payments and monthly payments are tied to interest rates. Insurance and property taxes are just as important since it factors into monthly payment. What about health insurance rates? Do we think that has an effect on home prices? Heck yeah! A lot of your data does show trends, but we are in a completely different world with a global economy in the last 15 years. I also think our economy will grow very slowly for many years to come because of insane governmental financial issues, everything is just unsustainable. I would not be surprised at all by some kind of massive downturn or correction because of artificially low interest rates.