- This topic has 43 replies, 11 voices, and was last updated 1 year, 9 months ago by
.
-
I’ve been thinking about your results in the context of my monthly payment chart:
My interpretation of this chart has always been that it shows that rates haven’t had a ton of influence on home valuations. If they had, then the red line would mean revert around a pretty flat line across the whole chart. In other words, monthly payments would be the constraint, and home prices/valuations would move up or down to accommodate a pretty constant monthly payment (as a % of income).
But that’s definitely not happening in this chart. In the 70s and 80s, valuations didn’t plummet to accommodate high rates. Instead, prices hung in there, and people paid a much higher monthly payment (as a % of incomes or rents).
Thus my conclusion that you can’t really value housing based off rates.
However, if you take a closer look at my graph, there are really 2 regimes there. In the 70s/80s, home prices weren’t influenced so much by rates. However, starting in the 90s, we DO see that more flat trend line where (with the exception of the bubble), monthly payments tended to stick to a somewhat constant level.
And that fits in perfectly with your conclusions — prices really have followed rates pretty closely. But if we were to extend your study back in time, we’d find that this wasn’t nearly so much the case.
What do we make of this? Well, it happens that this is actually a fairly controversial topic in finance generally.
One camp claims that lower rates justify higher stock valuations, pointing back to 30-40 years of a clear correlation there. The other camp points out that prior to that period, there was no such correlation! (Here is an article exploring some of the long term data). There are other arguments as well, such as: if low rates are such a benefit, why are stock valuations in Europe and Japan much lower than the US, while their rates are ALSO lower? IE, it seems that this relationship has only taken place in the US, which undermines the idea of a causal relationship.So I just wanted to put that out as a note of caution here. xbox’s analysis happened to cover the period where there was an unusually strong relationship between rates and the prices of RE and risk assets generally. But, this relationship is not written in stone.
Now, let’s say the relationship does remain intact. xbox’s results jibe with my general view that even through prices look high, when you consider interest rates, they aren’t nearly so bad.
However, based on these charts I’ve refined my view a bit. Looking at my monthly payment charts, I was thinking, we’re still below the historical (post-1977) median. But that includes the prior regime. If we only look at the 90s+ regime, with that tight relationship between rates and prices, then we ARE in fact overvalued even considering how low rates are. xbox’s charts do a good job of showing that. But, it’s a reasonable level of valuation, part of the normal ebb and flow, and nothing I’d consider remotely bubble-like. (Of course I mean overvalued based on this model… there are of course things going on outside the model, eg long term supply and demand changes, which could impact fair value.)
The other thing to consider is that if the rate/price relationship remains intact, this doesn’t tell you much about future prices anyway, UNLESS you want to posit that you know what rates are going to be. It’s my view that rates (and inflation, which heavily influence nominal rates) are poorly understood, and that the world’s most preeminent economists and asset managers disagree on what the course of rates will be. IE, I don’t think anyone can really make a high confidence forecast on rates. So IF the rate/price relationship remains intact, it is a fact that any home price forecast has an embedded interest rate forecast.
I’ve got some new data from Rich that I want to share, and I’ve been working with the charts and trying to develop some insight into what I’m seeing.
Here’s the same chart as I used above but with data from Jan. 1977 through April 2021.
The orange is the CaseShiller price and the blue is the price I would anticipate based on interest rate and inflation. (From here on out I’m going to refer to the anticipated price based on interest rate and inflation as the XValue, just because that’s easier to type.) There are two things to notice before jumping to any conclusions about this chart.
One of the first insights I had when looking at this chart is that the vast majority of the time house prices were overvalued in the last 45 years. Then I realized that this chart assumes that the Case Shiller and XValue are equally valued in January 1977. But the chart has very few times that is true. Simply because my data starts on January 1977 doesn’t mean the Case Shiller and the XValue would be equal at that time. So, let’s take a look at the chart with the Case Shiller undervaluing the XValue by 30% at January 1977.
Now we see about an equal amount of overvalue and undervalue.
This also lead me to realize that without some anchor to tell me that houses were fairly priced on a specific date, it becomes very difficult to use these charts to tell if currently (or at any specific time) housing is overpriced or underpriced.
Previously I claimed that over the long term the vast majority of house price rise could be explained by interest rate changes and inflation. (Short term that isn’t so true, but long term it is) Now that I have more data I think that claim still holds up very well. To help see that, I can fit a second order polynomial to each data series to find the slope.
Here we see a purple line which is the fit for the Case Shiller, and the yellow line which is a fit for the XValue.
The next issue is that as prices go up, price increase look bigger than they are when measured in percent. Imagine this scenario: You buy two houses, one for $500k and the other for $200k. Twenty years later the first house is worth $1.5m and the second house is worth only $600k. So the first house has appreciated $1m and the second house has appreciated $400k. Which one has a better return? The answer is neither, they both tripled in price. But the higher the starting price the bigger the increase should be numerically. So, when we look at the above chart, the most recent years look like a lot more appreciation but as a percent not so much. To compensate for this it’s important to work with log scale on the charts if you want to talk about growth rates. So, here’s the same chart but with a log scale on the Y axis and no slope lines.
And now with the slope lines.
- You must be logged in to reply to this topic.