HomeManaging RiskSignaling systemic risk

Signaling systemic risk


Systemic financial crises arise when vulnerable financial systems meet adverse shocks. A systemic risk indicator tracks the vulnerability rather than the shocks (which are the subject of ‘stress indicators’). A systemic risk indicator is by nature slow-moving and should signal elevated probability of financial system crises long before they manifest. A recent ECB paper proposed a practical approach to building domestic systemic risk indicators across countries. For each relevant categories of financial vulnerability, one representative measure is chosen on the basis of its early warning qualities. The measures are then normalized and aggregated linearly. In the past, aggregate systemic risk indicators would have shown vulnerability years ahead of crises. They would also have indicated the depth of ensuing economic downturns.

Lang, Jan Hannes, Cosimo Izzo, Stephan Fahr and Josef Ruzicka (2019), “Anticipating the bust: a new cyclical systemic risk indicator to assess the likelihood and severity of financial crises”, ECB Occasional Paper Series, No 219 / February 2019

The post ties in with SRSV’s summary lecture on systemic risk management.
The below are quotes from the paper. Emphasis and cursive text have been added. Acronyms have been replaced by conventional wording.

A very brief construction manual

“The domestic systemic risk indicator [called d-SRI] is constructed as the optimal weighted average of six early warning indicators, after normalising the individual indicators…The domestic systemic risk indicator is constructed by first selecting indicators based on their univariate signalling performance, and then aggregating the indicators based on a linear regression approach…the domestic systemic risk indicator strikes a balance between being a broad-based cyclical systemic risk indicator with appealing empirical properties and being simple, transparent and consistent across countries.”

“The four [stages of calculation] are: (i) selection of a set of relevant indicator categories for the risk of interest; (ii) selection of the optimal early warning indicator for each of the indicator categories; (iii) normalisation of each optimal early warning indicator based on the pooled median and standard deviation across countries and time; and (iv) linear aggregation of the normalised early warning indicators into a composite systemic risk indicator based on optimal indicator weights.”

“The domestic systemic risk indicator is therefore designed to cover five of the indicator categories for cyclical systemic risks that are recommended by the ESRB: (i) measures of potential overvaluation of property prices; (ii) measures of credit developments; (iii) measures of external imbalances; (iv) measures of private sector debt burden; (v) measures of potential mispricing of risk.
As a rule, the domestic systemic risk indicator includes one indicator per recommended category. The exception to this rule is credit: two indicators, one based on bank credit and the other based on total credit, are included in the domestic systemic risk indicator. This choice is guided by the fact that credit has historically played a prominent role in driving financial crisis vulnerabilities.”

“For each indicator category, the best univariate early warning indicator is selected as the relevant domestic systemic risk indicator sub-indicator… The specific “optimal” univariate early warning indicators that were selected for the domestic systemic risk indicator based on the comprehensive early warning exercise are: (i) the two-year change in the bank credit-to-GDP ratio; (ii) the two-year growth rate of real total credit; (iii) the two-year change in the debt-service-ratio; (iv) the three-year change in the residential real estate price-to-income ratio; (v) the three-year growth rate of real equity prices; (vi) the current account-to-GDP ratio.”

“Indicator normalisation is done by subtracting the median and dividing by the standard deviation of the pooled indicator distribution across countries. Optimal indicator weights are chosen to maximise the early warning properties…The optimal weighting procedure for the domestic systemic risk indicator assigns the largest weight to the bank credit-to-GDP change (36%), followed by the current account balance (20%), the residential real estate price-to-income ratio change (17%), real equity price growth (17%), the debt service ratio change (5%), and real total credit growth (5%)… Pooled indicator normalisation and constant weights across countries and time implicitly assumes that there are common indicator patterns across the crises experienced by individual countries at different points in time.”

“Expert judgment enters the domestic systemic risk indicator in two steps. First, via dictating variable selection along pre-defined risk categories (identified by experts); second, by imposing a minimum weight of 5% assigned to each of the domestic systemic risk indicator indicators.”

The empirical properties of the ECB’s systemic risk indicator

“The domestic systemic risk indicator contains useful information about both the likelihood and the severity of financial crises with a lead time of several years…The domestic systemic risk indicator displays long cycles and starts to increase above normal levels around four to five years ahead of systemic financial crises… In contrast to stress indicators, a key desirable property of the early warning indicators and the domestic systemic risk indicator is that signals flare up well before crisis events hit the economy. Historical regularities indicate that the seeds of a crisis are sown well in advance, as imbalances build up over time and leave the financial system vulnerable to shocks.”

“Both the in-sample and the out-of-sample early warning properties of the domestic systemic risk indicator are superior to those of the credit-to-GDP gap and other well-performing univariate early warning indicators. The early warning properties and dynamics of the domestic systemic risk indicator are robust to real-time indicator normalisation and real-time estimation of optimal weights for combining the sub-indicators.”

“Indeed, the level of the domestic systemic risk indicator around the start of financial crises is highly correlated with measures of crises severity, such as real GDP declines. [The chart below] shows that there is a high negative correlation (-0.67) between the maximum value of the domestic systemic risk indicator before the start of a systemic financial crisis and the maximum drop in real GDP that materialised during the ensuing crisis.”

Other empirical observations

Simple bank credit and household credit transformations tend to have the best early warning properties and outperform indicators that are based on total credit… For credit variables, changes in credit-to-GDP ratios tend to have superior early warning properties than real growth rates or gap measures based on a recursive Hodrick-Prescott filter… Changes in the bank credit-to-GDP ratio tend to accelerate around five years before the start of systemic financial crises, usually peaking around one to two years before a crisis materialises.”

“Changes in the debt service ratio and real M3 growth also have good early warning properties and perform slightly better than total credit indicators.”

“Medium-term changes in the price-to-income ratio are the best residential real estate indicators.”

On average, [financial] crisis periods result in cumulative output losses of 15-20% of annual GDP…During past banking crises across a large sample of countries worldwide, output losses amounted on average to 23% of GDP….Output losses during systemic financial crises in EU countries amounted to 8% of GDP on average.

Comparisons to the ‘Basel gap’

The total credit-to-GDP gap of the Basel III framework (the ‘Basel gap’) is a useful starting point for measuring the cyclical dimension of systemic risk. This is because various studies have shown that it provides good aggregate early warning signals for systemic banking crises…The ‘Basel gap’ refers to the total credit-to-GDP gap, which is calculated as the cyclical component of a recursive Hodrick-Prescott filter… However, the ‘Basel gap’…can be biased downward the longer credit booms last, because credit excesses enter the trend estimate as time progresses…[and] is sensitive to the length of the underlying time series.”

“The main result of the evaluation exercise is that simple transformations of credit and asset price variables can have similar or even better early warning properties than the Basel gap.”

Data on financial crises

“The new ECB/ESRB EU crises database provides precise chronological definitions of financial crisis periods in EU countries between 1970 and 2016… One important innovation of the dataset is that it contains information about whether a financial stress event was systemic or not and whether the event was of purely domestic origin, purely foreign origin or due to a combination of domestic and foreign factors…The baseline financial crisis definition for the exercise encompasses all systemic events from the ECB/ESRB crises database that were not purely due to foreign factors… The sample considered for the early warning exercise comprises all euro area countries plus Denmark, Sweden and the UK for the period Q1 1970 – Q4 2016. Out of the 22 countries in the sample, 18 experienced at least one systemic financial crisis of macroprudential relevance.”

Ralph Sueppel is managing director for research and trading strategies at Macrosynergy. He has worked in economics and finance since the early 1990s for investment banks, the European Central Bank, and leading hedge funds.