A tale of two US lending curves
I was a little bit intrigued by my previous chart and decided to take a second look at it. Here it is, using a log scale (as this is a chart covering 47 years, please keep in mind that what looks like small temporary fluctuations actually represent large ones…):
The two dotted lines represent the 1950-1980 trend for each curve. Why did I pick 1980? As it takes many years to develop new international Basel regulations, numerous drafts and proposals are circulated over a few years. This raises expectations of what future regulatory requirements will look like and banks progressively adapt the shape of their balance sheet to be in compliance once the rules effectively kick in. For instance, current Basel 3 regulations consultations started in 2009, with almost final proposals published between 2010 and 2011, and planned to be enforced by 2019, although most banks already comply with many of its features.
A draft of Basel 1 rules was published in 1987, following years of consultations (I am unsure exactly when those started, hence the choice of 1980), followed by a final agreement in 1988 and an official enforcement from 1992 onwards. In the US, Basel 1 was in place until well into the 2000s, and only a few banks had started applying Basel 2 before the crisis.
The first shocking feature of this chart is the differential between trend and actual business lending volume. Business lending completely collapsed relative to trend from the end of the 1980s, coinciding pretty much exactly with the implementation of Basel 1, and despite real interest rates falling (see chart below). Can any other (macro or micro) economic event explain this very sudden change? Did we overnight move into a low-growth/secular stagnation/economic ‘abundance’ paradigm? This looks highly unlikely.
The second remarkable feature is that real estate lending volume did not offset the fall in business lending. For a long period, real estate lending seemed to be above pre-1980 trend, before getting back to trend level and then departing from it again from the early 2000s. Never real estate lending fell below trend following the introduction of Basel 1, except during the Great Recession.
Now, this chart is very hard to interpret by itself, and it will take much more analysis to come to any meaningful conclusion. Pre-1980 lending trend perhaps was too rapid (remember post-WW2 boom and stagflation)? This would have two possible consequences: 1. real estate lending growth would therefore have been way too fast post-1980 and 2. business lending was perhaps in line with the actual trend?
My guess is that the truth lies in between: pre-1980 trend was perhaps too rapid, but post-1980 business lending growth was too low. Remember that household debt to income kept increasing over the period, signalling that real estate lending growth was surely too fast.
There is another possible explanation: that my chart does not cover all possible lending instruments. We know that Basel rules strongly encouraged the use of securitised products (which benefited from better capital treatment), in particular securitised mortgage instruments such as CMBS and RMBS. Banks that originate those mortgages originally reported them on their balance sheet, before taking them off-balance sheet. Does the FRED real estate lending data include those products (off-balance sheet or purchased by other institutional investors)? I have no idea and will have to do a lot more research on the FRED database before coming up with a satisfying description of the chain of events.