Predicting the impact of systemic risks

The cyclical Systemic Risk Indicator (cSRI) is the only indicator thus far which delivers the expected prediction for Malta.

When economic or banking systemic risks materialise, they can have large detrimental effects on the output and welfare of societies   ̶   something that was amply illustrated by the global financial crises between 2008 and 2012.

One particular study by Glenn Hoggarth of the Bank of England and other authors found that, on average, crisis periods result in cumulative output losses of 15-20% of annual GDP.  Marco Lo Duca of the European Central Bank and others estimated that output losses during systemic financial crises in EU countries amounted to 8% of GDP on average.

In times like these, it is imperative that policymakers in governments and central banks have at their disposal measures of systemic risk that provide sufficient lead time for them to act in a countercyclical manner, with a view to prevent systemic financial crises or mitigate their impact in the future.

Economic indicators are statistics and data that are used to analyse the health, size, and direction of a country’s overall economy or specific economic sectors. Economic Indicators provide information on key factors such as production, employment, spending, and prices that influence Gross Domestic Product (GDP), inflation, productivity, and economic growth. They help monitor current economic performance and also act as leading or lagging signals of future economic conditions.

By tracking these indicators over time, economists and policymakers gain insights into business cycles and make more informed decisions related to fiscal and monetary policy. There are different types of economic indicators that provide key snapshots of different spheres of economic activity. For one, production indicators offer a look at the output of goods and services from various industries on a monthly or quarterly basis. These include indices for industrial production, manufacturing, and housing starts, which respectively measure the volume of goods made in factories, the volume of output from the manufacturing sector, and the number of new residential buildings being constructed.

The total credit-to-GDP gap (the “Basel gap”) is a useful starting point for measuring the cyclical dimension of systemic risk because various studies have shown that it provides good aggregate early warning signals for systemic banking crises.  However, the Basel gap has some shortcomings when it comes to measuring cyclical systemic risk, being biased downward the longer credit booms last, sensitive to the length of the underlying time series of 10-15 years, and can be ambiguous where the credit-to-GDP ratio increases strongly, but at a slower pace than the trend component.

Academics and financial authorities all around the globe have been stepping up their efforts to improve the suit of tools and models in the field of systemic risk and macroprudential analysis, respectively.  If they can develop models that signal financial crisis vulnerabilities sufficiently in advance, mitigating macroprudential policy action can be taken earlier.

A composite indicator for cyclical risk for Malta

Recently Ms Sarah Vella, a senior research economist at the Central Bank of Malta, published a working paper on ‘Constructing a cyclical Systemic Risk Indicator (cSRI) for Malta’.  Although the CBM naturally monitors many indicators as part of its mission to conduct macroprudential oversight and analysis, to date it did not have an in-house country-specific composite indicator for cyclical risk for Malta.

Ms Vella addressed this gap by building a cyclical systemic risk indicator (cSRI). The cSRI is driven by the two-year growth rate in real bank credit, the one-year change in the debt service to income ratio, the house price to income ratio and the two-year growth rate in real total debt. These sub-indicators are believed to have early warning characteristics on financial distress. This indicator will complement other tools used by the CBM.

As Ms Vella points out, cyclical systemic risk depends on the phase of the financial cycle.  When economic output is expanding, risk builds up due to growing credit and surging financial and real estate asset prices   ̶   both of which increase private sector debt and collateral values.  However, once the general risk appetite decreases and doubts about the financial system’s sustainability emerge, a contraction in the demand for these assets leads the financial cycle to reach its peak.

If the financial market stress is large enough, the resulting severe financial market stress will spill over into the real economy.   Successive boom-busts have amply demonstrated how costly this can be, due to decreased output and the negative impact associated with the wellbeing of society.

However, the monitoring of cyclical systemic risk and the identification of financial cycle periods when implementing macroprudential policies is fraught with difficulties. One instrument which the policy-makers use to address cyclical systemic risk is the countercyclical capital buffer (CCyB)   ̶   a measure of the credit gap that aims to signal banking crises.  The CCyB is calibrated on the Basel gap mentioned earlier.

Yet, sometimes different indicators can give different signals, raising doubts as to which measures are appropriate to respond to the crises. It would be extremely helpful if as much information as possible from the vast range of financial cycle indicators in existence is summarised into one index.  This consolidation permits the sub-indicators’ dynamics to be more easily monitored whilst conveying relevant information about the build-up of systemic risk in the economy.

As Ms Vella says, when such a composite indicator has effective signalling properties, the negative impact caused by the systemic financial crisis can be mitigated pre-emptively.   The eurozone’s domestic cyclical systemic risk indicator (d-SRI), combined with its exposure-based systemic risk indicator (e-SRI), give early warning features that can predict vulnerable periods prior to a systemic crisis.  But these are not enough when one needs to examine the risks for each particular member of the currency zone. Country-specific macroprudential policies, including a country-specific SRIs, then come into play.

This is where Ms Vella’s paper comes in.  She constructs a country-specific indicator for Malta that can be used to complement other cyclical systemic risk measures and to identify periods of high cyclical systemic risk, leading to timely macroprudential policy response. This country-specific indicator is constructed as a subset of early warning sub-indicators which include the two-year real bank credit growth, the one-year change in debt service to income ratio, the house price to income ratio, and the two-year real total debt growth.

This is easier said than done, since the outcome depends on the availability of suitable sub-indicators and their weights.  Ironically, these were rather flimsy because of the fact that the history of financial crises in Malta is rather limited.  Thank God for that, but it presents a problem to the model-builder.  So the CBM researcher decided to use what is known as a Principal Component Analysis (PCA) to capture important co-movement among the set of variables.

I won’t go into the statistical analysis and gymnastics which the author had to do to compile the data necessary for her indicator.  This was by no stretch of the imagination a hit-and-miss approach, but required a good understanding of certain basic relationships between variables and the reliability of the data concerned. In certain circumstances, she had to derive substitute variables for those where data did not exist.  In this, she used to good effect EU and UN guidelines.

The cSRI derived for the period between 2006Q1 to 2022Q4, together with its contributions, is shown in the figure. Where positive values build up, this reflects the net accumulation of systemic risk when above zero, while when the indicator trends down (even if above zero) this signals a very low cyclical risk.

This qualitative time-series analysis of the cSRI shows that the indicator developed by the researcher delivers a credible narrative about the accumulation of cyclical systemic risk in Malta over time. The cSRI’s macro-financial variables are closely aligned with the evolution of cyclical systemic risks in Malta, yielding valuable insights on Malta’s recent financial and macroeconomic environment.

The paper mentions sensitivity analysis conducted by Ms Vella, for example by adding the unemployment rate as a variable.  The evidence shows that this macroeconomic variable does not provide additional contribution to the cSRI.  Another test revealed that the CBM advertised house price index was more reliable than the National Statistics Office (NSO) transacted house price index.

The most important question of all was whether the cSRI contained early warning properties about future economic outcomes.  Vella deployed so-called Impulse response functions (IRFs) to local projections to formally test the predictive power of the indicator on future real annual GDP growth. Òscar Jordà of the Economics Department of the University of California had already used local projections to compute impulse response functions in the application of monetary policy shocks in the USA. The author had found a delayed yet significant impact at three to four years ahead, equivalent to a decline in real GDP growth of approximately 4 p.p. on average.

Vella developed an equation for a model to shock future real annual GDP growth by the cSRI, using the sample period from 2006Q1 to 2022Q4. The impulse responses for Malta are based on a horizon period of up to 20 quarters.  The impulse response function was found to be very sensitive and wide confidence intervals implied that the estimate was highly imprecise over such a long period. 

Where the model comes into its own is when confidence intervals narrow.  There is a greater degree of precision in predicting a decline in future real GDP growth, three to four years ahead, as the chart shows. Hence, the cSRI conveys sensible early warning properties. The signalling properties of the cSRI are also superior to the signalling properties of the Basel gap for Malta which still provides a counter-intuitive and insensible prediction on real GDP growth.

The cSRI is the only indicator thus far which delivers the expected prediction for Malta.  A note of warning is due. It is not the case that the cSRI can be used mechanically.  It can only be one of an array of tools, since no policy-maker or central bank will want to rely on a single indicator.

Illustration: Dilok Klaisataporn/iStock/Getty Images Plus

0 0 votes
Article Rating
Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments

Menu