Lesson 5 of 7
The closest mainstream cousin
The BIS measures a financial cycle from credit and property prices, about sixteen years long, with peaks close to crises. How it differs from this account.
By Javier Sanz
Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.
Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)
In one minute
- The BIS measure is deliberately parsimonious: credit and property prices together, because those two co-vary and carry most of the information about crises.
- It runs at a much lower frequency than the business cycle. The average length in a sample of seven industrialised countries since the 1960s is about sixteen years.
- It is a measured indicator with no accepted theory behind it, and the institution that publishes it says so in print.
Of everything in mainstream economics, this is the work that sits closest to what this track has been describing. It is also the clearest case of a claim being turned into a published series with a stated method, which is what makes it worth a lesson.
Five things the BIS says about the financial cycle
- 1Credit and property pricesThe most parsimonious description; equity prices co-vary far less
- 2Lower frequencyAbout sixteen years on average, against one to eight for the business cycle
- 3Peaks near crisesThe peaks of the cycle sit close to episodes of financial distress
- 4Usable in real timeIt carries information about distress risk ahead of the event
- 5Regime dependentLength and amplitude vary with the policy and financial regime
What is being measured
Two series, combined: credit, and the price of property. The survey's argument for that choice is empirical. The two move together closely at low frequencies, which reflects the role of credit in financing construction and property purchase, and the variance in both is dominated by slow-moving components. Equity prices are excluded on the grounds that they co-vary much less and move mostly at higher frequencies.
The cycle so defined is long. Using frequency filters that target eight to thirty years, and separately using turning-point dating, the average length in a sample of seven industrialised countries since the 1960s comes out at about sixteen years, against the one to eight years conventionally used for the business cycle.
The derived indicator most used by policymakers is the credit-to-GDP gap: the deviation of the credit-to-GDP ratio from a statistical trend. Basel III suggests using it in countercyclical capital buffer frameworks.
What the BIS says against its own measure
This is the part a reader should not skip, because it is where the institution's own honesty is visible.
A 2018 BIS working paper states the criticisms of the gap plainly: that the statistical filter used produces spurious dynamics and has end-point problems, that its typical implementation is at odds with its own statistical foundations, and that normalising by GDP correlates the gap negatively with output. The authors write that they agree with these criticisms.
Their defence is narrow and is the interesting part. In the absence of clear theoretical foundations, they say, all such gaps are indicators rather than measurements of anything, so which to use is an empirical question. They run the alternatives against each other over 42 economies with quarterly data from 1970 to 2017 and report that none outperformed the baseline credit-to-GDP gap as an early warning indicator.
Where it differs from the account in this track
Three differences, and none is small.
The measure is credit and property prices, not the time structure of production. Nothing in it says which stages of production expanded, which is the distinctive Austrian claim.
The framework is one of self-reinforcing interactions between perceptions of value and risk, attitudes to risk, and financing constraints. That is closer to Minsky than to Mises, and the survey names Minsky and Kindleberger as the writers who kept the subject alive.
And it is offered as an indicator without an agreed theory, which is the opposite of a tradition whose claims come from a theory and whose evidence is thin. One camp has a mechanism and few measurements. The other has measurements and an unsettled mechanism.
Check yourself
Which two series does the BIS use to describe the financial cycle?
- Equity prices and interest rates
- Credit and property prices
- Money supply and inflation
Show the answer
Answer: B.
The 2012 survey argues these two are the most parsimonious description, and excludes equity prices because they co-vary far less and move at higher frequencies.How long is the financial cycle on the BIS measure?
- About four years
- About sixteen years on average, in seven industrialised countries since the 1960s
- About thirty years
Show the answer
Answer: B.
That figure comes from frequency-based filtering targeting eight to thirty years, corroborated by turning-point dating.What do the BIS authors concede about the credit-to-GDP gap?
- That it has never predicted a crisis
- That the filter has real statistical problems and that all such gaps are indicators rather than measurements
- That it should be abandoned
Show the answer
Answer: B.
Their 2018 paper agrees with the criticisms and defends the measure only on the empirical ground that nothing else performed better.Sources
- Borio, 'The Financial Cycle and Macroeconomics: What Have We Learnt?', BIS Working Paper 395, December 2012, section one: the definition in terms of self-reinforcing interactions between perceptions of value and risk and financing constraints, the five stylised features, the parsimony argument for credit and property prices, the average length of about sixteen years in seven industrialised countries, and the naming of Minsky and Kindleberger.
- Drehmann and Yetman, 'Why You Should Use the Hodrick-Prescott Filter, at Least to Generate Credit Gaps', BIS Working Paper 744, September 2018, abstract and introduction: the definition of the credit-to-GDP gap, its use in Basel III, the criticisms the authors accept, the argument that all gaps are indicators without theoretical foundations, and the horse race over 42 economies from 1970 to 2017.
What this leaves out
- This is a heterodox tradition. The mainstream account of the same phenomenon differs, and nothing here is a forecast or a view on any security.
- The BIS series and the papers cited are not an Austrian account and are not presented as one. They are the closest mainstream work on the same phenomena, and the differences are stated in the lesson.
- Nothing here is a statement about current credit conditions anywhere. This lesson is about how a measure is built and what its publishers say it can bear.