I. Nominal real estate asset and debt levels are unique and not comparable to other assets.
First I see a problem in the way household debt is frequently described as a sign of middle class stagnation and of a housing "bubble". I have discussed this before. Very short version - most middle class debt is mortgage related. In other words, most middle class debt is related to savings - deferred consumption, not debt-fueled consumption.
The manifestation of many recent market trends is interest rates. With regard to housing, the most important factor is very low long term real rates. Emerging market savers and developed market baby boomers (and their pension managers) have a tremendous demand for long term safe cash flows, so that long term real interest rates have been bid down to very low levels. (Real rates were very low in the 1970's also.) Changing interest rates have different effects on different securities, depending on what is held constant.
1) Treasuries. As treasuries mature and are re-issued, face values are reset. So, coupon payment levels change over time, but the changing asset value of, say, a treasury ETF reverts to a stable mean over time, even if interest rates don't reverse. And, government debt is the product of various political factors that are not related to interest rate levels. So, the quantity of government debt and the market value of that debt do not have a systematic relationship to interest rates. In futures markets that use a stationary bond maturity as the basis, bonds trade based on a premium or a discount to a stationary interest rate, but this is not a factor in the public image of bond values or in the total nominal quantity of bonds over time.
2) Corporate bonds. Corporate bonds have many of the same characteristics as treasuries. It is usually assumed that corporations would sell more bonds when rates are low. I have not found that to be the case. But, even if it were the case, a change of 1% or 2% would only be expected to change corporate debt levels marginally. Corporations don't borrow to target a set interest expense level. They borrow to fund a set nominal investment level.
3) Corporate equities. It is frequently asserted that the stock market is fueled by Fed-induced low rates. Risk premiums for corporate equities tend to move counter to risk free interest rates, so equities also don't react systematically to interest rate changes. Most capital gains in equities come from growth in economic activity and stabilizing aggregate demand during recoveries.
4) Real estate. Real estate is different than all these other categories of assets. The "coupon" on a piece of real estate is the rent. Rent tends to track income, more or less. And, there is no reset on real estate face values - real estate is like a perpetual bond with no maturity reset. So, real estate in the real world, and in our collective consciousness, acts like those bond futures contracts. And since real estate has a very long life, it's nominal value is very sensitive to changes in real interest rates, especially when they are very low. This doesn't only increase the nominal value of real estate assets. Since rents tend to change very slowly, relative to interest rates, this has a similar effect on real estate debt. In a low real long term interest rate context, it can be reasonable to purchase leveraged real estate with very high levels of nominal debt, because the cash flows will compare favorably to renting.
So, when real long term interest rates are very low, like they have been this century, the only nominal asset value they really affect is real estate. I don't have a precise suggestion for adjusting for this fact. But, given that it is a fact, any analysis that uses the changing levels of mortgage debt and real estate values as the signal for some broad social issue is simply baseless. They aren't measuring what they think they are measuring.
But, it's a problem for all of us. I am saying that, if real long term interest rates change over time, the nominal value of assets at a given point in time is not a reliable or useful piece of information. And I don't have a suggested replacement.
This is a difficult condition to accept. But, if you accept the most common story in defiance of my position, then you have to believe that American middle class households have been stumbling under the weight of stagnating incomes and taking on debt to mask the effect on their lifestyles. And this problem has been coincident with an unprecedented and relentless bidding war on owner-occupied middle class housing. That's simply unbelievable, notwithstanding the widespread belief in it.
II. It was not unreasonable to model MBS's based on historical experience.
I come here today to ("gulp", straightens tie nervously) defend David X. Li and the widely derided risk models that were applied to MBS's during the housing boom.
There are many legitimate arguments to be made about assuming normal distributions, continuation of historical trends, etc. These arguments can be made about MBS's as well as many other types of investments. And, to be honest, I am not enough of a statistician to get too far into the weeds on the topic.
But, the point I would like to make is that there is nothing exceptional about MBS's that make the models in use in the 2000's especially bad. The breakdown in the models basically arose from the fact that correlations all rose toward unity. There were admittedly securitizations that, in hindsight, seem to have been made up of especially flimsy mortgages. And, around the margins, we could second-guess some of the ratings that were applied to those securities.
But, in the end, the universal collapse in asset values was a product of Fed policy. My point is that, if you have a self-inflicted black swan - if the Fed is bound and determined to suck liquidity out of the economy - then how can you model that? The only way to be prepared for that is to....I don't know, bury a bunch of gold coins under your porch?
I mean, how'd your house do in 2008? How about your stocks? Even inflation protected bonds dipped in the chaos of late 2008. All these assets fell right along with MBS that had been designed with those models. Nobody needed an actuarial model to lose a substantial sum in 2008. It was a pretty easy thing to do. There were speculators who had the right side of some trades when the bottom fell out. You might have done quite well if you were long volatility. But is there someone out there who had a better risk model, who sailed right through 2008?
Homes lost 30% of their value - in the aggregate - in 2 years. Home prices were stable as interest rates rose. The losses began approximately 1 year after the yield curve inverted, when short and long term rates started to collapse. So, we shouldn't pin responsibility for this on monetary policy? We should blame the MBS models because they weren't robust in the face of 30% aggregate losses and a massive liquidity crisis?
So, can you get caught with your pants down if you own leveraged assets? Of course. Could we have an economy that was based on more robust forms of investment? You bet. Could the Fed give us more stability? Certainly. Our answers to all of those questions matter a lot more than whether or not there were some simplifying assumptions in some asset construction.
I suspect that MBS risk models that are based on historical correlations of defaults - even in MBS with relatively risky mortgages - will perform relatively well in the future. There will be times where they don't perform well. But in those times, it won't be the model that killed you. It will be a war or a pandemic, or a Fed that hasn't accounted for the topic of the first half of this post.