Training on SF, June 15–16

Principles of Calculating the Financial Stress Index for the Russian Federation

1 Definitions and scope of application

These Principles (hereinafter — Document) of the Analytical Credit Rating Agency (hereinafter — ACRA or the Agency) set forth the goals and methods for calculating the financial stress index for the Russian Federation (hereinafter – the Index), as well as the main approaches to its use.

These Principles, based on the concepts of systemic risk and financial stability1, imply that the financial system is interconnected and can contribute to more frequent defaults (whatever their causes) on contract payments of agents on some markets to agents on other markets. Large-scale episodes of this kind (financial crises) can lead to disruptions of economic activities in the real sector (initially due to the emergence of local liquidity crises), which makes them a focal point of consideration.

In order to evaluate how close the financial system is to a crisis, ACRA uses two approaches (see Figure 1):

  • Financial Stress Index (ACRA FSI) measuring financial instability indirectly based on external signs of a financial crisis: volatility in the key financial markets, interest rate spreads, etc. (this methodology describes that approach).
  • Structural Financial Stress Index (ACRA SFSI) that measures potential financial instability directly by aggregating information on the financial position of economic agents and by assessing their vulnerability to specific types of risk (see Principles of Calculating the Structural Financial Stress Index (ACRA SFSI)).

In some countries where liquidity in the financial markets is low even in normal times or market participants disregard information on financial position of counterparties and issuers or regard it as secondary due to some reasons, the latter approach may be the only one possible. As in such case we would not be able to obtain information of interest indirectly from the available market data regarding vulnerability of companies and households, and instead of analyzing external signs of stress, we have to analyze the structure of their financial balances directly.


1 Moiseev S.R., Lobanova M.A. The Concept of Macroprudential Policy. Dengi i Kredit. N 7, 2013; Schinasi G. J., Defining Financial Stability. IMF Working Paper. WP/04/187. October 2004. р. 13.

Figure 1. Difference in approaching construction of ACRA FSI and ACRA SFSI

The level of financial stress is the degree of materialization of systemic risk and financial instability, the financial system’s proximity to a financial crisis. Given the complexity of this phenomenon, within the framework of the ACRA FSI Index calculation, we do not aim at tracking down the statuses of all financial institutions and relationships between them; instead, we use the external manifestations of changes within them that are indirectly expressed in market price dynamics, interest rates (and their spreads), and currency exchange rates.

Based on international experience2, ACRA singles out five major external manifestations of financial stress, as follows:

  1. Uncertainty about fundamental values of financial assets or commodities: it is correlated with the volatility of financial instrument prices and forces market participants to respond more energetically and expeditiously to any new information (even when it is irrelevant).
  2. Lack of information about the behavior of other market participants: can lead to misinterpretations of price dynamics and episodes of sudden and violent expectation adjustments.
  3. Asymmetry of information about asset quality (Seller Knows More) or borrower quality (Borrower is Better Informed): occurs as a result of a growing spread in quality assessment and exacerbates problems of adverse selection and moral hazard, which in turn lead to eroded trust and more reluctant lending.
  4. Flight to quality: growing probability of losses forces investors to prefer less profitable and less risky investment mechanisms, which often triggers rapid changes of their relative prices.
  5. Flight to liquidity: fewer lending opportunities create additional incentives for preferring more liquid assets that are necessary to meet the demand for cash gap financing.


2 Hakkio, Craig S.; Keeton, William R., Financial Stress: What Is It, How Can It Be Measured, and Why Does It Matter? Economic Review, Federal Reserve Bank of Kansas City, 2nd Quarter 2009.

The Index dynamics is meant to do the following:

  • Draw a simple quantitative portrait of the financial market operating mode;
  • Indirectly warn about an increased or decreased likelihood of rapid changes in the creditworthiness of economic agents due to financial market malfunctions.

At the same time, changes in the level of financial stress do not reflect ACRA’s views on possible changes of credit ratings and/or outlooks.

The Index aggregates information on the dynamics of a number of factors selected in accordance with the principles outlined in Clause 2. The aggregation is performed through the use of the method outlined in Clause 3. Approaches to interpretation of the Index level and dynamics are outlined in Clause 4. The Index construction took into account the global experience of using similar indicators, as listed in Appendix 1.

2 Index factor selection

Financial stress manifestations (see Clause 1) are measured through the use of 12 quantitative factors. These factors (see Table 1) are selected in such a way as to guarantee the efficiency, continuity, and versatility of the Index mechanism (the factor selection does not change with time). Special attention in the factor selection and construction is given to the comparability of their values in various financial system operating modes (structural surplus vs. liquidity deficit, soft vs. tight monetary policy) and under a priori equal economic conditions (various average price levels, high / low real economic growth rates).

Table 1. Rationale for including factors in the Index and their respective data sources

Factor

Characteristics

Data sources

Spread between money market interest rates zero-coupon short-term FLBs 
(3 months)

Reflects differences between perceptions of credit risk by members of the banking system and the Russian Federation, gives an assessment of the relative average credit quality of banks. Also, gives a measure to the ratio of interbank market liquidity to sovereign debt market. Insufficient interbank market liquidity may coincide with the formation of additional liquidity reserves by banks (or anticipation of a need for such a process).

The MIACR 2-7 days RUB rate and estimation of zero-coupon FLB rates are used for calculating this factor.

Money market interest rates: The Bank of Russia

FLB: Moscow
Exchange

Interest rate spread between large issues of liquid corporate bonds and zero-coupon FLB rate 
(five years)

Reflects differences between perceptions of credit risk by large corporate borrowers and the Russian Federation, gives an assessment of the relative credit quality of the large corporate sector. An increased rate spread between these two assets can also reflect the growth of demand for more liquid financial assets.

The IFX-Cbonds index and zero-coupon FLB rates assessment are used for calculating this factor.

Corporate bonds: http://cbonds.ru/indexes/

FLB: Moscow Exchange

Stock market volatility

High stock market volatility may signal the growth of uncertainty of the issuers fundamental characteristics and, consequently, a potential increase of information asymmetry between holders and issuers.

The MOEX Russia Index is adjusted (deflated) by the level of the consumer price index (weekly inflation linearly interpolated into a daily rate) in order to ensure comparability of the long-term fluctuation magnitudes.

Volatility is calculated as a standard deviation of the adjusted MOEX Russia Index within the last calendar week and additionally smoothed out by applying a moving average per calendar week.

Moscow
Exchange*

Financial sector stock price index

Stock prices of financial companies may decline against the backdrop of fundamental factor deterioration of the issuers: reduced solvency, increase in credit risk, observed or expected balance sheet deterioration. This index is one of the measuring tools with respect to the expected stability of the banking system as a whole (although its base includes a limited number of securities, not all of which were issued by banks).

For calculating this factor, ACRA uses the industry-based MICEX index for financial and banking sector stocks (MICEX FNL)3. The inflation adjustment necessary for the comparability of factor values year-on-year is made by way of weekly CPI (consumer price index) assessments. Concentration of this index on specific financial institutions is high (shares of four issuers account for around 85%4 of the index). We regard the high concentration as a potential problem for this and the below factors. However, the dynamics demonstrated by the shares of above issuers even in the current composition is very well interpreted explicitly alerting of stress at the time it occurs, rather than reacting to unique but not stressful news of issuers. High liquidity of these shares is of higher importance for us, rather than their classification as banking industry stocks. The calculation base may be expanded after performing additional research.

Index: Moscow Exchange*

CPI: Federal
Service of State Statistics*

http://www.cbr.ru/
content/document/file/
26398/kulikov_06_17.pdf

Divergence of financial institutions stock returns

A substantially opposite stock price movements of peer group banks can indicate an emerging information asymmetry about their characteristics (between holders and issuers, between holders and other holders) and correlate with the growing uncertainty of financial asset prices. Individual characteristics of financial institutions have more importance to stock holders at times of heightened financial stress.

For calculating this factor, ACRA uses stock prices of 4 to 10 financial institutions (by selecting the most liquid ones from the MOEX Russia Index calculation base according to Bloomberg data). Profitability is considered to be the stock price percentage increase over one calendar week. Spread is calculated as the cross-section variance of profitability indices.

Moscow Exchange*

Spread between interbank loan interest rate and 1-day liquidity interest rate offered by the Bank of Russia

Under the conditions of structural liquidity deficit, this factor shows to what extent, on average, an average-sized commercial bank participating in the interbank market, has access to operations with the Bank of Russia. The growth of this indicator may signal of an insufficient collateral for secured lending, of a significant difference in credit quality between the different banking system levels (e.g., between banks that borrow directly from the Bank of Russia and banks that borrow from them), and of Bank of Russia refinancing that is insufficient for short-term debt refinancing.

The National
Banking Association*

The Bank of Russia*

Difference between crude oil spot forward prices (one year)

Growth of price difference may indicate the presence of expectations (or at any rate, a prevailing uncertainty) of a change of average crude oil prices within the 12-month horizon. This may entail greater volatility of financial instrument prices, since in a crude oil price-dependent economy, correction of price expectations can directly create new expectations for currency exchange rates, interest rates, monetary and fiscal policy, budget policy, and economic growth. All of the above may affect the creditworthiness of borrowers and credit risk assessments by lenders, and hence their propensity to lend. This factor is designed in such a way as to make non-zero contribution to the Index only in the event of emerging expectations that oil prices will fall: Russia is an oil exporter, and ceteris paribus, such changes may make a stronger negative impact on its financial sector.

In the calculation, ACRA uses the spot Brent crude oil price and the evaluation of the forward price of oil of the same brand computed on the basis of forward yield curve.

Moscow
Exchange*

Crude oil price volatility

This factor is meant to complement data obtained from the previous factor, by uncertainty estimate of crude oil prices regardless of the direction of their change.

Volatility is calculated as a standard deviation of the logarithm of the Brent oil price within the last 30 days.

Bloomberg

Currency exchange rate volatility

The Index concurrently includes this factor and crude oil price volatility, despite the fact that they usually strongly correlate. The inclusion of both factors is necessary to ensure that the Index dynamics be equally relevant at times when the Bank of Russia enforces different currency exchange rate policies, makes FX market interventions not related to the exchange rate policy, in order to incorporate the direct effect from other dynamic factors of the currency exchange rate. This way, we consider, among other things, the availability of different channels for external stress transmission to the Russian financial system. In theory, an FX rate shock may be related not only to trading volume or trade terms and conditions, but also to rapid cross-border financial flows. And conversely, the crude oil market environment may theoretically be dampened in some manner or other (by virtue of using international reserves offsetting cross-border financial flows). Volatility is calculated as a standard deviation of the USDRUB logarithm for the last 30 days.

Bloomberg

RUB inflation

On average, higher inflation is less stable and predictable than low inflation. This leads, among other things, to greater uncertainty of expectations with regards to future interest rates.

ACRA uses the most up-to-date data available, i.e. the seasonally adjusted percentage increase in consumer prices for the last calendar week.

For calculation, ACRA uses the most up-to-date data it can obtain: percentage growth of consumer prices within the last calendar week.

Federal Service
of State Statistics*

«Velocity» of the simultaneous stock prices drops of financial institutions and sovereign debt

If average stock prices of financial and banking industry companies that are part of MICEX FNL index and prices of five-year FLBs drop simultaneously, this factor assumes positive values proportional to the velocity of the price drop. Its growth can meet the increasing demand for more liquid assets, which in turn is often a response to the increased uncertainty about interest rates and asset quality.

МICEX: Moscow Exchange*

FLB: Bloomberg

«Velocity» of divergence between stock prices of financial institutions and quality lender-issued bonds

In case of a simultaneous decline of the MOEX Russia Index and growth of the FLB index, this factor assumes a positive value proportional to the difference between relative indices growths. Growth of this factor values may indicate that investors prefer higher quality and less risky securities while sacrificing potential returns.

MICEX: Moscow Exchange*

FLB: Bloomberg

Source: ACRA
3 The sub-index calculation base was as follows as of December 21, 2018: VTB Bank, Sberbank, Moscow Exchange, Bank Saint-Petersburg, Credit Bank of Moscow, SAFMAR Financial investments, QIWI Plc.
4 The data given here is as of the publication of the article. See: Financial Stress Index for the Financial System of Russia; by D.M. Kulikov, V.M. Baranova; Dengi i Kredit, Vol.6 2017; p.39-48.

Table 2. Correlation between factors and key external financial stress manifestations

Factor

Uncertainty of fundamental values

Insufficient current status data

Information assymetry

Flight to quality

Flight to liquidity

Spread between money market interest rates and zero-coupon short-term FLBs (3 months)

 

Interest rate spread between large issues of liquid corporate bonds and zero-coupon FLB rate (five years)

 

 

 

Stock market volatility

 

 

Financial sector stock price index

 

 

Profitability spread of financial institutions stock

 

 

Spread between interbank loan interest rate and 1-day liquidity interest rate offered by the Bank of Russia

 

 

 

Difference between oil spot price and oil forward price (one year)

 

 

 

 

Oil price volatility

 

 

 

 

Currency exchange rate volatility

 

 

 

 

RUB inflation

ᴠ 

 

 

 

“Velocity” of the simultaneous stock prices drops of financial institutions and sovereign debt

 

 

 

 

“Velocity” of divergence between stock prices of financial institutions and quality lender-issued bonds

 

 

 

 

Source: ACRA

3 Dimension reduction

Values of individual factors can grow both when local expectations fluctuate and when market liquidity tools used for their calculations undergo changes. Therefore, it is reasonable to regard the level of financial stress as heightened only when several factor values increase simultaneously. The ACRA Index is a one-dimensional weighted sum of factor values. This type of construction ensures that the factor has the desired property described above (the level of financial stress is regarded as heightened if several factor values grow simultaneously).

3.1. Preliminary data processing

Prior to being weighed and summed up, the original factors are transformed in the way that makes their value increases correspond to financial stress level increase.

The transformed factors are normalized to ensure that each of their historical dynamics has a zero sample mean and a single-unit standard deviation within a fixed timeframe. It is necessary for preventing the factor measuring units from influencing comparative factor importance for the Index dynamics when weights are determined.

3.2. Weight determination and Index calculation

Summation weights5 are calculated as the coordinates of the first principal component6 of normalized and transformed factors. The Financial Stress Index is considered to be the first principal component that is in turn normalized in order to have the maximum recorded value equaling 10 and the minimal one equaling 0 at the time of approving this Document (selected for convenience, see Figure 2). The Index is constructed in a way that does not cap it from below or above. Its value dropping below zero is theoretically possible in a situation when individual factors drop to the levels below the historical input data available at the time of approving this Document. Index growth above 10 will indicate financial stress more formidable than those observed in 2008-2009 and 2014-2015.


5 ACRA does not stipulate the weights of the Index factors in the text of this Document (neither has the Agency published them before) due to the requirements of the Agency’s internal documents, but is ready to discuss them in more detail, as well as input data and calculation details with all interested researchers. For further information, please, contact dmitry.kulikov@acra-ratings.ru and vasilisa.baranova@acra-ratings.ru.
6 I.T. Jolliffe, Principal Component Analysis, Springer Verlag, New York, 1986.

The construction of synthetic indices (including the ACRA FSI Index) is different in sense from the construction of forward-looking indicators, just like “unsupervised learning” classification methods are different from “supervised learning” ones. In the first case, the functional form and parameters of the Index are selected based on the auto-informativeness criterion, and in the second case based on external informativeness. ACRA tries not to repeat the dynamics of either indicator, but briefly describe a complex phenomenon.

This determines the choice of the principal component method in order to reduce dimension — it guarantees minimum loss of information about the multidimensional variance of the original factors among all of their possible linear combinations.

3.3. Special cases

Weights of factors included in the Index are checked for positivity. In order to ensure the continuity of the Index time series, weights do not change over time as the sample grows, but only when the Document is amended (then the historical series is recalculated to the maximum possible historical depth). In order to keep this Document up to date, it is reviewed annually. Changes can be due to:

  • Mismatch between the principles of Index calculation to the definition of financial stress, systematic over- or underevaluation of the stress level compared to its observed manifestations and consequences;
  • Emergence of new reliable supported sources of input data or financial market segments, allowing to expand the concept of financial stress or information base of the Index;
  • Changes in the principles of Index calculation or input data collection, their availability, and lower efficiency of information provision.

In case a new factor emerges that must be included in the Index, but the inclusion would lead to a sample imbalance (since full data is available only in a sub-sample), it is possible to apply an iterative EM-algorithm to simultaneously estimate a non-existent or non-observed factor values and new weights through the use of the Kalman filter7.


7 Based on the approach outlined in the article by S. Brave and R. A. Butters, titled Diagnosing the Financial System: Financial Conditions and Financial Stress, International Journal of Central Banking, June 2012.

Figure 2. Index Dynamics from 10.01.2006 through 04.11.2018

Source: ACRA

4 Interpretation and use

4.1. Financial crisis identification

Analysis of the Index transition matrices8 allows to interpret Index states that correspond to values of 2.5 p. and higher, as financial crises.

Starting from this level, the probability that the Index will change by 0.5 p., or more within the coming week (with the median weekly change of 0.14 p.) becomes steadily higher than the probability of it staying within the current range (see Figure 3). In other words, stable Index states occur only when its values are low; elevated Index factor volatility is associated with its high values.

Stable and unstable Index states correspond to different operating modes of the financial system. Sample distribution analysis of Index levels also speaks in favor of the chosen bracket. Values in excess of 2.5 p. sampled between January 2006 and November 2018 occurred in less than 7% of cases. Thus, the Index exceeding the 2.5 p. threshold corresponds to those rarely occurring states of the financial system that are marked by great uncertainty and high velocity of change with regards to rates, spreads, and prices.


8 The number at the intersection of line A and column B of the matrix shows the frequency of the Index moving from range A to range B within a week’s time throughout all its available history. The values under consideration vary by units of 0.5, e.g. 1.0–1.5.

Figure 3. Correlation between the probability of stable and unstable Index behavior at different levels of financial stress, and Index value distribution

When using the above-described approach, ACRA dates the two most recent financial crises as follows:  

  • 16.09.2008–11.04.2009;
  • 09.12.2014–24.04.2015.

4.2. Assessment of time prior to exiting the financial crisis mode

To evaluate the time before the Index exits the financial crisis mode, we use two estimates with different calculation principles, namely base and alternative ones. The base estimate is obtained in two steps.

Step One

By comparing the current Index value to the last 30-days moving average, the two-week average growth of the Index is used as a basis for classifying the current state as going through:

  1. a growth phase;
  2. a decline phase;
  3. a stable phase.

Step Two

  1. If the current state is classified as a decline phase, ACRA uses the average rate of decline estimated within all the selected historical decline phases. The sought-for time is calculated as the number of days required to reach down to the borderline level at the estimated constant rate.
  2. If the current state is classified as a growth phase9, the sought-for time is calculated as the number of days required to reach the level of the historical maximum of the Index estimate, and time required to return to the stable level.

Average growth and decline rates are estimated at the periods of respective phases based on the entire available Index history.

Alternative assessment is obtained by making simplifying assumptions about the lack of autocorrelation of Index increments. In this case, ACRA applies the calculated result of the expected time of the Markov chain reaching the absorbing state (values<2.5 p.)10 to the Index transition matrix. In reality, the Index growth increments are positively autocorrelated, therefore this estimate may be excessively optimistic in the growth phase and excessively pessimistic in the decline phase. It may be more preferable with regards to the base estimate, being error-proof in terms of phase determination and not requiring explicit assumptions about the highest level the Index can reach at times of financial crises.


9 Or as a stable phase, which is unlikely due to reasons outlined in Sub-Clause 4.1.
10 Kemeny, John G.; Snell, J. Laurie (July 1976) [1960]. "Ch. 3: Absorbing Markov Chains". In Gehring, F. W.; Halmos, P. R. Finite Markov Chains (Second ed.). New York Berlin Heidelberg Tokyo: Springer-Verlag.

Figure 4. Evaluation of time required to overcome a financial crisis using the 2008–2009 example

Source: ACRA

4.3. Assessment of growth “significance”

Comparison of the observed Index dynamics and typical volatility allows to draw conclusions about whether to interpret growth over a certain period as a reflection of the real financial stress dynamics or simply as "noise." For this purpose, ACRA evaluates the Index growth module distribution11 (over a time interval equal to the period of interest) in the crisis-free mode (values<2.5 p.). If the magnitude of the observed growth is comparable with the rarer values on the right side of the distribution, it can be interpreted as "significant." The level of this significance (frequency below which the observed growth is considered to be relatively rare) can be the standard statistical tests significance levels of 5%, 10%, 15%, and 25%. The growth frequency of the same level or bigger for one day, one week and two weeks is marked in Figure 5.


11 This approach is considered justifiable, since in the crisis-free mode, growth increments can be seen as symmetrically distributed around the zero mean.

Figure 5. Frequency of index growth of the same or larger magnitude (under crisis-free conditions)

Source: ACRA

Table 3 shows examples of interpreting the Index dynamics in the wake of events that are potentially capable of affecting the level of financial stress in Russia. It is important to clarify that the Index growth after an event(s) does not necessarily mean it was precisely that event (those events) that had made a difference necessary to change the state of the financial system.

Table 3. Example of calculating an Index growth “significance”12

Date

Event

Index growth after the event

Frequency of Index growth of the same or larger magnitude in crisis-free periods

   

1 day

week

weeks

1 day

week

weeks

August 7, 2007

BNP Paribas freezes its deposits with three investment funds amid the U.S. mortgage sector crisis

0.0113

0.1121

0.2882

0.84

0.53

0.24

August 22, 2008

In the course of 10 days, the Federal Reserve Service (FRS) slashes its rate by 1.75 pps

-0.0194

0.19

-0.0696

0.75

0.3

0.75

September 15, 2008

Lehman Brothers declares bankruptcy

0.7994

2.3223

1.6492

0.00

0.00

0.00

April 23, 2010

Greece officially turns to the EU for support

-0.1105

-0.1138

0.4482

0.18

0.53

0.10

May 7, 2010

Spreads between government bond rates in Europe widen sharply (Wave 1)

0.0530

0.1532

-0.0368

0.43

0.39

0.87

August 5, 2010

Spreads between government bond rates in Europe widen sharply (Wave 2)

-0.1928

-0.203

-0.2946

0.06

0.27

0.23

October 28, 2010

Government bond rates in Europe surge abruptly

-0.1750

-0.4567

-0.5737

0.07

0.05

0.05

May 4, 2011

Oil prices plunge abruptly (by 12 pps in 4 days)

0.4617

0.5207

0.4056

0.01

0.04

0.13

August 22, 2012

Russia joins the WTO

-0.0072

-0.1234

-0.1064

0.90

0.49

0.63

April 7, 2013

Oil prices plunge sharply (by 9 pps)

0.1436

0.5114

0.4436

0.11

0.04

0.10

March 6, 2014

First anti-Russian political and financial sanctions are introduced

0.2218

-0.0647

0.2607

0.04

0.72

0.28

December 16, 2014

Currency market goes into a “panic” mode

0.5278

3.9284

0.5836

0.00

0.00

0.05

April 30, 2015

The Bank of Russia reduces its key rate by 1.5 pps

-0.1687

-0.2586

-0.3671

0.08

0.20

0.16

June 23, 2016

Brexit referendum results are made public

0.3642

0.2521

0.124

0.01

0.21

0.58

November 30, 2016

OPEC meeting

-0.0891

0.0303

-0.1440

0.24

0.86

0.53

March 8, 2017

Higher crude oil stocks reported in the US

0.2603

0.3174

0.0880

0.03

0.13

0.68

May 3, 2017

AFK Sistema shares plunge on the news of a lawsuit filed by Rosneft

0.1267

0.7342

0.0526

0.13

0.01

0.81

August 29, 2017

Financial rehabilitation procedure announced with respect to Bank “FC Otkritie”

0.0676

0.2999

0.2432

0.33

0.14

0.3

January 30, 2018

US Treasury releases first part of report on Russia sanctions

-0.0107

0.0188

0.2004

0.88

0.93

0.39

March 14, 2018

UK Prime Minister Theresa May announces new sanctions on Russia

-0.0108

-0.0672

0.0023

0.88

0.72

0.99

March 19, 2018

Results of Russian presidential elections

0.0242

0.1115

0.0823

0.72

0.55

0.70

April 6, 2018

US sanctions on private Russian companies

-0.0533

0.7721

0.5107

0.49

0.01

0.07

August 9, 2018

US announces new sanctions

-0.0132

0.1839

-0.0848

0.85

0.34

0.70


12 Color-marked are “significant” Index growth instances that occur less frequently than in 25% cases in crisis-free periods.

4.4. Use as forward-looking indicator for beginning of a recession   

Although initially ACRA had no plans to build a forward-looking indicator for economic activity, the Agency believes that the fact that the Index value is sustainably above the threshold value can be used as a signal of an increased probability of a recession. Both economic crises in the sample were preceded by the growth in financial stress and the Index. At the same time, the increased financial stress without the subsequent economic downturn did not occur in the sample. It is noteworthy that this observation may not characterize the forward-looking features of the Index, but could rather say something about the type and consequences of the recessions (in the sample) in Russia. A recession is theoretically possible if there is no episode of a systemic financial crisis. Currently, the accumulated information is not enough to obtain general conclusions about the leading features of the Index, although the available data does not refute their presence.

5 Sustaining quality and information disclosure

The current and historical values of the Index are regularly made public on the official website www.acra-ratings.com, with frequency of publications being subject to ACRA internal regulations.

In order to keep this Document up-to-date, ACRA may review or amend it based on the following rationale:

  • Inconsistency of these Index calculating Principles with the definition of financial stress, systematic overestimation or underestimation of the stress level compared to its observed effects and consequences;
  • Emergence of new reliable and sustainable original data sources or financial market segments that would allow for expanding the definition of financial stress or the Index knowledge base;
  • Amendments to the methodology used for calculating or collecting the original data, changes in its availability, or a decline in promptness of data reporting.

Not later than one calendar year after the latest revision date of this Document, ACRA shall conduct a review thereof in accordance with its internal regulations. As a result, the Document may be amended or remain unchanged.

ACRA shall disclose information about amendments to this Document, its up-to-date version, and historical Index values calculated using the current Document, on its official website www.acra-ratings.com.

Appendix 1. Global practices of calculating similar indices and comparison with the ACRA Index

Financial stress and financial condition indices are widely used around the around the world allowing regulators to oversee financial systems and enforce economic stabilization policies. Investors rely on these indices when assessing the overall risk of investing in financial instruments of a country or region. Researchers apply them when analyzing phenomena that depend on the financial system operating mode.

Comparison of Index dynamics with indices of other countries provides an opportunity to gauge the potential propagation of financial stress from the global financial system into Russia (see Subclause A).

While maintaining the Index, it is essential to keep track of the knowledge base contents of peer indices outside of Russia (see Subclause B). Global practices allow for an early detection of potential financial stress symptoms that have not been observed in the Russian market economy due to its comparatively “young age.” Indicators that can track these symptoms are included in the Index when quality criteria data is available.

A. Dynamics comparison

Figure 6. Financial stress indices of ACRA and St. Louis Fed

Source: ACRA

The Asian crisis of 1997–1998, the "dot-com" bubble (2000), the mortgage crisis in the U.S. (2007), the debt crisis in Europe which broke out in 2010, and other stress episodes show that local crises may or may not have spill into other countries. The potential for such spillage is determined by both the amplitude of shock and the openness level of the trigger financial market, on the one hand, and the network properties of the global financial market, on the other hand. A comparison of index dynamics for various economies may give a general idea of the potential influence of external stress on domestic financial systems within the assumption of maintaining openness to external shocks at the same level as when the previous stress episodes occurred. Accounting for openness dynamics and shock proliferation requires a more detailed analysis.

B. Possible sets of factors

In contrast to stress indices that focus more on quick assessments of the current state of a financial system, conditions indices13 are usually constructed for the purpose of being used as forward-looking recession indicators. Therefore, they more often rely on economic data and aggregated financial balances, as their specifics allow for the use of data that are less frequent and timely. The ACRA Index belongs to stress indices group, but it may also include factors typical for conditions indices. The values of less frequently updated factors remain constant between updates.

A brief description of the main regularly published indices is given in Table 4. Special attention in the course of composition analysis of other indices is given to methodological errors that lead to a calculation termination or a methodology review. For example, the Cleveland Fed’s financial stress index has not been calculated for at least six months since May 2016, as its methodology had been recognized to be systematically overestimating the level of stress on the property and asset securitization market. The Bank of Canada’s financial conditions index has been on hold since December 2015, after its dynamics was recognized to have potential for being incorrectly interpreted. Conclusions about the causes of such episodes (more often, the disparity in the factor values in a priori equal conditions) are included in the factor selection requirements for the Index.

The Russian experience of similar developments was analyzed on the basis of the five earlier proposed methodologies for calculating indices and forward-looking indicators.

Most of the previously developed indices for Russia are not published or published irregularly since they have been developed for internal use, or as part of separate studies14.

 

13 The traditionally used abbreviations are FSI (financial stress index) and FCI (financial conditions index).
14 Apart from the indices outlined in Table 4, information about expertise of the Centre for Macroeconomic Research of Sberbank and the development of an Index analogue as part of research on the monetary policy rule [E.A. Fedorova, A.S. Mukhin, S.Ye. Dovzhenko Modeling Rules of Monetary Policy of the Central Bank of the Russian Federation with the Financial Stress Index // Journal of the New Economic Association. 2016. № 1. P. 84–105] was available when writing this Document. As far as we know, the Bank of Russia has a non-public analogue of the Index.

Table 4. Some regularly published financial stress or financial conditions indices

Index

Country/ Region

Frequency15

Description

Chicago Fed FCI

USA

W

One of the most detailed, developed indices for the USA. Its uniqueness is that is uses a large number of factors (100) of different frequencies. The raw data for most of the factors used are only available for recent years, but the algorithms for estimating unobservable values have allowed analysts to calculate a continuous series of the index since 1973 (when only 25 factors were available). A wide range of factors ensures good sensitivity to any possible changes in the money, stock, and banking markets.

Kansas City Fed FSI

USA

М

The methodology provides the conceptual framework for a complex definition of financial stress based on five external manifestations

· Uncertainty in fundamental prices of financial assets or exchange commodities;

· Lack of information on the current state of the economy or financial market;

· Asymmetrical information on the quality of the asset or the borrower;

· “Flight to quality;”

· “Flight to liquidity.”

This index is the first principle component of 11 source factors.

St. Louis Fed FSI

USA

W

The main manifestations of stress are divided into three classes:

· Yield spreads;

· Interest rate levels;

· Inflation and exchange rate risks.

This index is the first principle component of 18 source factors.

Bloomberg FCI

USA, European Union, Asia (not including Japan)

D

This index is calculated based on 10 financial indicators, which are grouped by money, stock, and bond markets. The market sub-indices are Z-scores that show the number of standard deviations of the weighted sum of the initial factors from its average value for the period. The contributions of the sub-indices to the final index are the same. The set of indicators (mainly spreads of market instruments) provides high updating efficiency.

Goldman Sachs FCI

USA, Europe, and Japan

Q

Four indicators with constant weights are used to calculate this index:

· Long-term corporate bond yields;

· Short-term bond yields;

· Exchange rates;

· Stock price.

Bank of England FSI for UK

UK

М

This index dates bank to 1971 and is one of the longest financial stress indices for the UK16. The calculation uses 13 market indicators, which include stock market, government and corporate bond, money market and real estate market indicators. As in the methodology for the Kansas City Fed FSI, manifestations of financial stress are highlighted. Indicators for the sub-indices are first divided, then portfolio theory is used to aggregate these indicators. 

ECB Composite Indicator of Systemic Stress

European Union

W

The methodology uses the concept of systemic risk to justify the need for a portfolio approach for the primary aggregation of indicators (directly taking into account their paired correlations). Source factors (mainly financial indicators) are combined into sub-indices. Weights of sub-indices are determined taking into account their potential impact on the dynamics of industrial production. The impact assessment is based on vector auto-regression models, therefore it can change over time and depends on the state of the economy.

OECD FCI

OECD countries (The Organization for Economic Co-operation and Development)

Q

Factor weights of are determined taking into account their potential impact on the dynamics of GDP within a four to six-quarter horizon. The impact assessment is based on the vector auto-regression model which includes indicators like population income and real exchange rate. The weight of each indicator is equal to the relative change in GDP for four to six quarters after the indicator increases by one. The small number of factors used (six), their broad definition, and low data frequency allow for cross-country comparisons.

IMF Advanced Economies FSI

17 developed countries

М

Having a fairly standard set of indicators (seven) and fixed equal weights makes cross-country comparisons as easy as possible.

Bank of America Merrill Lynch Global FSI

Global

D

This index characterizes the state of the global financial system, not the market of each individual country. It includes 41 factors with equal weights, each of which describes the relative value, liquidity, risk perception, and special distribution properties of the profitability of financial instruments traded on the world's largest exchanges. Factors can be considered in the context of three semantic subgroups:

· Credit risk and liquidity risk;

· The cost of insurance against major losses;

· Risk appetite.

OFR Financial Stress Index

Global

D

For calculations, this index uses 33 indicators; for example, spreads of bond yields of different credit quality, interest rates in the United States and other developed and developing countries and regions. Symptoms of stress are highlighted in the same way as in the methodology for the Kansas City Fed FSI. The principal component method is used to determine weights.

Summary of leading indicators -CMASF

Russia

М

CMASF calculates seven separate leading indicators of various stress events or conditions, four of which directly describe the state of the banking system:

· Banking crisis indicator;

· Banking crisis continuation indicator ;

· Systemic credit risk indicator;

· Systemic liquidity risk indicator.

These indices are designed on a signal approach basis (counting the number of factors that crossed the threshold) to get ahead on the quarter and more quantitatively identified events, such as the outflow of deposits of at least 1% per month, etc. In October 2016, the methodology for indicators was changed to respond more adequately to the increased average volatility of the ruble exchange rate after the change in the currency regime, as well as to the flow of private deposits to larger banks and the reduction in the number of banks as a whole.

Financial stability index of IEP (Institute for Economic Policy)

Russia

М

This index uses a signal approach. The thresholds at which factors are considered to signal coming instability are calibrated based the periods of financial instability identified by experts. These factors are growth rates of market rates, stock indices, the ruble exchange rate, monetary aggregates, and inflation.

15 D — daily, W — weekly, M — monthly, Q — quarterly. 
16 Chatterjee R., Chiu C.-W., Duprey T., Hoke S.H. (2017). A financial stress index for the United Kingdom. Bank of England Working Paper № 697.

Appendix 2. Index factor dynamics based on the 2014–2015 example

One of the most notable episodes of high financial stress in Russia began in 2014 with a gradual decline in commodity prices, which by the year-end amounted to almost 50% (oil prices started sliding in July 2014). The expectations of their quick recovery that were prevalent in the beginning were phased out by an understanding that one of the key causes of the meltdown was supply shock. That meant that the market was adjusting to the new environment and not simply being highly volatile. Lack of information about factors influencing the medium-term equilibrium oil price had created an uncertainty regarding the future economic conditions and fundamental values of most assets.

Figure 7. Selected Index factor dynamics during the 2014-2015 crisis

 

Source: ACRA

In particular, exchange rate expectations had shifted dramatically. With the full transition to the unrestricted exchange rate floating regime (November 10), the RUB exchange rate volatility increased significantly. The Bank of Russia continued smoothing it out through the use of the currency repo mechanism, but the persisting expectations of a weakening RUB and lack of information about reasons motivating other investors and lenders created incentives to save in foreign currencies and reduce RUB deposits. This was accompanied by significant increases of both inflationary expectations and actual inflation.

The worsening economic prospects brought down expectations about corporate earnings and pushed up corporate credit risk estimates. The partially coordinated price decrease on the stock and debt markets, which is being observed since October 2014, can be explained by the flight to quality.

The dropping prices on most securities pledged under repo agreements curbed the banks’ ability to obtain funding (for a brief period, the collateral utilization ratio reached almost 80%17). The increased corporate risk and limited abilities for refinancing heightened the bank credit risk perception and widened the money market spreads (after December 16). AS a result of the uneven distribution of available collaterals and a significant growth of information asymmetry about borrower quality, the average number of counteragents dropped by 15% and concentration of flows on key players became more pronounced.


17 See Bank of Russia’s Money Market Review, N 1, 4Q2014.

Log in

Forgot password

Sign up

Reset password

Reset password

Termsofuse

Полное использование материалов сайта разрешается только с письменного согласия правообладателя, АКРА (АО). Частичное использование материалов сайта (не более 30% текста статьи) разрешается только при условии указания гиперссылки на непосредственный адрес материала на сайте www.acra-ratings.ru . Гиперссылка должна быть размещена в подзаголовке или в первом абзаце материала. Размер шрифта гиперссылки не должен быть меньше шрифта текста используемого материала.