Macroeconomic effects of financial stress and the role of monetary policy: a VAR analysis for the euro area


This paper analyses an otherwise standard macro-financial VAR model for the euro area that includes - apart from conventional measures of output, inflation and monetary policy - a composite indicator of systemic financial stress, namely the CISS index, and total assets of the ECB balance sheet capturing the stance of unconventional monetary policy. I find that the CISS contributes significantly to the dynamics of the macroeconomy and exerts a strong influence on monetary policy when looking at both policy rates and the ECB balance sheet. The significance of the CISS appears robust to the inclusion of a broad set of real and financial control variables. Based on tests of direct versus indirect (Granger-)causality patterns proposed in Hsiao (1982), I also find that unlike unconventional policy as measured by ECB balance sheet growth, the policy rate does not seem to react directly to variations in financial stress but rather indirectly through the impact of financial stress on macroeconomic conditions. These different patterns of reaction are broadly consistent with the ECB’s “separation principle”. The estimated effects of the ECB’s standard and non-standard policy measures on inflation and economic growth are moderate, although an easier stance in both policy tools helps calm down financial stress.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6


  1. 1.

    The CISS project formed part of the Macro-prudential Research Network (MaRs) among researchers of the European System of Central Banks. The network aimed to develop core conceptual frameworks, models and tools to provide analytical support to macro-prudential supervision in the European Union (see

  2. 2.

    Each of the raw indicators is transformed using its empirical cumulative distribution function (ecdf), that is each observation is replaced by its ecdf value. This transformation is also called the probability integral transform (see, e.g., Spanos 1999). All transformed indicators are bounded by the interval (0,1] and approximately uniform distributed. See Hollo et al. (2012) for details.

  3. 3.

    A longer times series of the euro area CISS which starts in 1987 is shown in Hollo et al. (2012).

  4. 4.

    Estimation is implemented using the procedure DISAGGREGATE.SRC in WinRATS version 8.0.

  5. 5.

    I take the square root of the CISS to control for potential nonlinearities arising from the quadratic form of the formula with which the CISS is computed. The square root of the CISS is what has been called the “volatility-equivalent” CISS versus its standard “variance-equivalent” form as published by the ECB (see Hollo et al. 2012). However, all the basic messages of the emprical analysis presented in this paper do not alter when using the standard CISS instead.

  6. 6.

    Regular updates of the weekly CISS can be obtained via this link:

  7. 7.

    For an overview of different identifying assumptions see Christiano et al. (1999).

  8. 8.

    The structural shocks are thus distributed as u t i.i.d.N(0, D).

  9. 9.

    Given the low and insignificant correlation between their reduced-form residuals (see the estimate of coefficient a 21 in Eq. 3), this assumption is inconsequential.

  10. 10.

    The matrix P = A D 1/2 is known as the Cholesky factorisation of Ω. Like A, P is lower triangular, though whereas A has 1s along the principal diagonal, the Cholesky factor has the square roots of the elements of D, that is the standard deviations of the structural shocks u t , along the principal diagonal. The structural shocks from the Cholesky decomposition are obtained as v t = P −1 ε t = D −1/2 u t such that v t i.i.d.N(0, I n ). Thus, v j t is just u j t divided by its standard deviation \(\sqrt {d_{jj}}\). This decomposition is used to compute impulse response functions to one-standard-deviation shocks rather than one-unit shocks in u t .

  11. 11.

    However, Dufour and Tessier (1993) point out that the duality of non-causality restrictions on the coefficients of the autoregressive and the moving-average representation of a VAR does not hold in multivariate systems. Even if y 1 does not cause y 2 in the sense of Granger, the innovations of y 1 may account for a sizeable proportion of the variance of y 2. Conversely, even if the latter proportion is zero, it is quite possible that y 1 is found to Granger-cause y 2.

  12. 12.

    I prefer to report the exclusion F-tests based on the reduced-form model and thus to limit attention to the analysis of Granger causality rather than mingling it with assumptions about instantaneous causality between the variables as reported in Eq. 3, since the latter depends on the structural identification scheme. Exclusion F-tests based on the structural VAR form could be obtained by adding to each equation the contemporaneous values of those variables which precede the variable at hand in the order of the vector of endogeneous variables.

  13. 13.

    A block exogeneity test of the CISS with respect to the remaining variables delivers a p-value of 0.10. Interpreting this result as evidence against one-step ahead predictability would imply that the CISS is Granger causally priorto the other model variables, which in turn implies that the other model variables do not help predict the CISS even beyond the one-step-ahead forecast horizon (see Doan and Todd 2010 as well as Jarocinski and Mackowiak 2013).

  14. 14.

    Hence, the null hypothesis states that the three endogenous variables as a block are exogenous with respect to the CISS.

  15. 15.

    The number of lags (p) for the endogenous and exogenous variables is set to four, the same number of lags as used for the benchmark VAR.

  16. 16.

    Sims (1980) suggests using a correction equal to the number of regressors in each unrestricted equation in the system. In the present case, the correction equals (n + m)p = 20.

  17. 17.

    The case of the VSTOXX may appear different, though, since the p-value of the block exogeneity test with respect to the CISS increases to 16 %. However, within the full benchmark VAR, the CISS clearly retains its predictive power when including the VSTOXX with a p-value of the block exogeneity test of 0.001. In addition, when testing for block exogeneity of the core variables with respect to the VSTOXX without including the CISS, the VSTOXX turns out to be statistically insignificant even at the 10 %-level. Its predictive power seems to depend on the presence of the CISS. In general, the fact that the block exogeneity zero restrictions cannot be rejected for both variables when the VSTOXX is included along with the CISS, may also point at problems of multicollinearity.

  18. 18.

    For an economic interpretation of such shocks see Christiano et al. (1999) who describe possible sources of exogenous variations in monetary policy, such as (i) exogenous shocks to the preferences of the monetary authority, for instance due to stochastic shifts in the relative weight given to certain data which, in turn, could reflect shocks to the preferences of the members of the decision-making body, or to the weights by which their views are aggregated; (ii) strategic considerations with respect to the policy expectations held by private economic agents; and (iii) technical factors, e.g. measurement errors in the preliminary data available to policy makers at the time they make their decisions, a point raised by Bernanke and Mihov (1998).

  19. 19.

    For the balance sheet instrument also the contemporaneous MRO rate is treated as predetermined.

  20. 20.

    The weekly MROs were based on a variable rate tender until October 2008, with a minimum bid rate as the interest rate below which the Eurosystem would not accept any bids. Thereafter, a fixed rate tender procedure was introduced and the minimum bid rate became the rate at which all bids were alloted (European Central Bank 2014).

  21. 21.

    In the present context “contemporaneous” information ignores the issue of publication lags in order to simplify the analysis.

  22. 22.

    The absence of direct causality may suggest that the CISS is likely not to emerge as a significant explanatory variable in estimated augmented (dynamic) Taylor rules if endogeneity issues are not properly dealt with.

  23. 23.

    See Baxa et al. (2013).

  24. 24.

    This interpretation assumes that the MRO rate predicted by the lagged endogenous variables approximates a short-term equilibrium rate. Positive or negative deviations of the actual MRO rate from that short-term neutral rate hence determine whether the policy stance is contractionary or expansionary, respectively. A long-run natural interest rate could be computed from the steady-state solution of the VAR model.

  25. 25.

    This interpretation is subject to several caveats. For instance, the regression format restricts the overall sum of shocks to be equal to zero. Hence, the procedure implicitly assumes that on average the policy stance is neutral. In addition, whether the cumulated policy shocks can represent the prevailing overall policy stance also depends on the appropriate choice of the starting date of the cumulation. Since the cumulated shock series start at a value of zero, the starting date should coincide with a period in which the policy stance can be considered as neutral. In the present case, I let the summation start at the earliest possible date.

  26. 26.

    In Hoffmann et al. (2015), we propose a composite indicator of financial integration in the euro area (FINTEC) that documents a strong price dispersion across euro area counties which took hold of basically all major market segments during the crisis.

  27. 27.

    This list of unconventional monetary policy measures by the ECB is not exhaustive. See various issues of the ECB Monthly Bulletin for complete references.

  28. 28.

    The correlation coefficient between the (square root of the) CISS and annual growth in ECB total assets is about 75 % when computed for the sample August 2007 to December 2013. The correlation over the entire sample period drops to 55 %.

  29. 29.

    The idea for this chart is borrowed from Boeckx et al. (2014) who furthermore offer a detailed account of events related to ECB non-standard measures.

  30. 30.

    Their VAR is estimated with Bayesian methods for a shorter sample that only covers the crisis and the post-crisis years. They use sign restrictions to identify the structural innovations.


  1. Aboura S, van Roye B (2013) Financial stress and economic dynamics: an application to France. Kiel Institute for the World Economy Working Paper No. 1834, March

  2. Adrian T, Liang N (2014) Monetary policy, financial conditions, and financial stability. Federal Reserve Bank of New York Staff Report No. 690

  3. Adrian T, Moench E, Shin HS (2010) Macro risk premium and intermediary balance sheet quantities. Federal Reserve Bank of New York Staff Report No. 428

  4. Altavilla C, Giannone D, Lenza M (2014) The financial and macroecnomic effects of OMT announcements. ECB Working Paper No. 1707, August

  5. Baxa J, Horvath R, Vasicek B (2013) Time-varying monetary-policy rules and financial stress: does financial instability matter for monetary policy?. J Financ Stability 9:117–138

    Article  Google Scholar 

  6. Beirne J, Dalitz L, Ejsing J, Grothe M, Manganelli S, Monar F, Sahel B, Sušec M, Tapking J, Vong T (2011) The impact of the Eurosystem’s Covered Bond Purchase Programme on the primary and secondary Markets. ECB Occasional Paper No. 122, January

  7. Bekaert G, Hoerova M (2014) The VIX, the variance premium and stock market volatility. J Econ 183(2):181–192

    Article  Google Scholar 

  8. Bernanke BS, Mihov I (1998) Measuring monetary policy. Q J Econ 113 (3):869–902

    Article  Google Scholar 

  9. Bloom N (2009) The impact of uncertainty shocks. Econometrica 77(3):623–685

    Article  Google Scholar 

  10. Boissay F, Collard F, Smets F (2015) Booms and banking crisis. Journal of Political Economy, forthcoming

  11. Bordes C, Clerc L (2012) The ECB’s separation principle: does it ‘rule OK’? From policy rule to stop-and-go. Oxf Econ Pap 65:i66–i91

    Article  Google Scholar 

  12. Boeckx J, Dossche M, Peersman G (2014) Effectiveness and transmission of the ECB’s balance sheet policies. CESifo Working Paper No. 4907

  13. Cardarelli R, Elekdagb S, Lall S (2011) Financial stress and economic contractions. J Financ Stability 7:78–97

    Article  Google Scholar 

  14. Christiano LJ, Eichenbaum M, Evans C (1999) Monetary policy shocks: what have we learned and to what end?. In: Taylor, JB, Woodford, M (eds)Handbook of Macroeconomics, edn 1, vol. 1, chap 2. Elsevier, Amsterdam, pp 65–148

    Google Scholar 

  15. Ciccarelli M, Maddaloni A, Peydro J-L (2013) Heterogeneous transmission mechanism: monetary policy and financial fragility in the Eurozone. Economic Policy, July, pp 459–512

  16. Constâncio V (2011) Challenges to monetary policy in 2012. Speech at the 26th International Conference on Interest Rates, Frankfurt am Main, 8 December (

  17. Cœuré B (2013) Non-standard monetary policy measures: where do we stand? Speech at the International Monetary Seminar “Sovereign Risk, Bank Risk and Central Banking” organised by the Banque de France, Paris, 10 July 2013 (

  18. Curdia V, Woodford M (2010) Credit spreads and monetary policy. J Money Credit Bank Supplement 42(6):3–35

    Article  Google Scholar 

  19. Davig T, Hakkio C (2010) What is the effect of financial stress on economic activity? Federal Reserve Bank of Kansas City Economic Review, 2nd Quarter

  20. De Bandt O, Hartmann P (2000) Systemic risk: a survey. ECB Working Paper No. 35, November

  21. Doan T, Todd R (2010) Causal orderings for multivariate linear systems. Mimeo, March

  22. Dovern J, van Roye B (2014) International transmission and business-cycle effects of financial stress. J Financ Stability 13:1–17

    Article  Google Scholar 

  23. Draghi M (2012) Speech at the Global Investment Conference in London, 26 July 2012 (

  24. Dufour J-M, Tessier D (1993) On the relationship between impulse response analysis, innovation accounting and Granger causality. Econ Lett 42:327–333

    Article  Google Scholar 

  25. European Central Bank (2014) Recent developments in excess liquidity and money market rates. Monthly Bulletin, January, pp 69–82

  26. Eser F, Schwaab B (2015) Evaluating the impact of unconventional monetary policy measures: empirical evidence from the ECB’s Securities Markets Programme. Journal of Financial Economics, forthcoming

  27. Fahr S, Motto R, Rostagno M, Smets F, Tristani O (2013) A monetary policy strategy in good and bad times: lessons from the recent past. Economic Policy, April, pp 243–288

  28. Gambacorta L, Hofmann B, Peersman G (2014) The effectiveness of unconventional monetary policy at the zero lower bound: a cross-country analysis. J Money Credit Bank 46(4):615–642

    Article  Google Scholar 

  29. Giannone D, Lenza M, Pill H, Reichlin L (2012) The ECB and the interbank market. Econ J 122:F467–F486

    Article  Google Scholar 

  30. Gilchrist S, Zakrajsek E (2011) Monetary policy and credit supply shocks. IMF Econ Rev 102(4):195–232

    Article  Google Scholar 

  31. Gilchrist S, Zakrajsek E (2012) Credit spreads and business cycle fluctuations. Am Econ Rev 102(4):1692–1720

    Article  Google Scholar 

  32. Granger CWJ (1969) Investigating causal relations by econometric models and cross-spectral methods. Econometrica 37:424–438

    Article  Google Scholar 

  33. Hamilton J (1994) Time series analysis. Princeton University Press, Princeton

    Google Scholar 

  34. Hartmann P, Hubrich K, Kremer M, Tetlow RJ (2015) Melting down: systemic financial instability and the macroeconomy. Mimeo

  35. Hoffmann P, Kremer M, Zaharia S (2015) Financial integration in Europe through the lens of composite indicators. Mimeo

  36. Hollo D, Kremer M, Lo Duca M (2012) CISS - a Composite Indicator of Systemic Stress in the financial system. ECB Working Paper No. 1426, March

  37. Hsiao C (1982) Autoregressive modeling and causal ordering of economic variables. J Econ Dyn Control 4:243–259

    Article  Google Scholar 

  38. Hubrich K, Tetlow RJ (2015) Financial stress and economic dynamics: the transmission of crises. J Monet Econ 70:100–115

    Article  Google Scholar 

  39. Illing M, Liu Y (2006) Measuring financial stress in a developed country: an application to Canada. J Financ Stability 2(4):243–265

    Article  Google Scholar 

  40. Jarocinski M, Mackowiak B (2013) Granger-causal-priority and choice of variables in vector autoregressions. ECB Working Paper No. 1600, October

  41. Kapetanios G, Mumtaz H, Stevens I, Theodoridis K (2012) Assessing the economy-wide effects of quantitative easing. Econ J 122:F316–F347

    Article  Google Scholar 

  42. Kliesen KL, Owyang MT, Vermann EK (2012) Disentangling diverse measures: a survey of financial stress indexes. Federal Reserve Bank of St. Louis Review, September/October, pp 369–398

  43. Lenza M, Pill H, Reichlin L (2010) Monetary policy in exceptional times. Economic Policy, April, pp 295–339

  44. Li F, St-Amant P (2010) Financial stress, monetary policy, and economic activity. Bank of Canada Review, Autumn, pp 9–18

  45. Litterman D (1983) A random walk, Markov model for the distribution of time series. J Bus Econ Stat 1:169–173

    Google Scholar 

  46. Mallick SK, Sousa RM (2013) The real effects of financial stress in the eurozone. Int Rev Financial Analysis 30:1–17

    Article  Google Scholar 

  47. Miles D, Schanz J (2014) The relevance or otherwise of the central bank’s balance sheet. J Int Econ 92:S103–S116

    Article  Google Scholar 

  48. Mittnik S, Semmler W (2013) The real consequences of financial stress. J Econ Dyn Control 37:1479–1499

    Article  Google Scholar 

  49. Peersman G (2011) Macroeconomic effects of unconventional monetary policy in the euro area. ECB Working Paper No. 1397, November

  50. Sims CA (1980) Macroeconomics and reality. Econometrica 48:1–49

    Article  Google Scholar 

  51. Spanos A (1999) Probability theory and statistical inference. econometric modeling with observational data. Cambridge University Press, Cambridge

    Google Scholar 

  52. Taylor JB (1999) A historical analysis of monetary policy rules. In: Taylor JB (ed) Monetary Policy Rules, NBER Studies in Business Cycles, vol 31. Chicago University Press, Chicago and London, pp 319–347

    Google Scholar 

  53. van Roye B (2014) Financial stress and economic activity in Germany. Empirica 41:101–126

    Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Manfred Kremer.

Additional information

I thank my discussant, Cillian Ryan, and seminar participants at the joint bdvb Research Institute/University of Wuppertal International Conference in Düsseldorf, the 35th International Symposium on Forecasting in Riverside, and the 2nd Annual Conference of the International Association for Applied Econometrics (IAAE) in Thessaloniki for fruitful discussions and comments. The views expressed in this paper are mine and do not necessarily reflect those of the European Central Bank.


Appendix A: Description of control variables

This appendix describes the control variables used in the block exogeneity tests reported in Table 3.

Commodity price index::

Annual change of the log HWWI commodity price index; Hamburg Institute of International Economics (HWWI) index for the euro area based on prices in euros; weights for individual commodities are based on their share in total euro area raw material imports between 1999 and 2001 (in 2000 prices); monthly data. Source: Haver Analytics.

Consensus inflation forecast::

Mean forecast of the one-year ahead percentage change of the euro area HICP, computed as the pro rata average of the mean forecast of the percentage change of the index for the current year and the subsequent year on the respective previous calendar year; in percent per annum; monthly data. Sources: Own calculations and Consensus Forecasts by Consensus Economics.

Unemployment rate::

Average euro area harmonised unemployment rate, seasonally adjusted; in percent; monthly data. Source: Haver Analytics.

Consensus real GDP forecast::

Mean forecast of the one-year ahead percentage change of the euro area real GDP, computed as the pro rata average of the mean forecast of the percentage change of the index for the current year and the subsequent year on the respective previous calendar year; in percent per annum; monthly data. Sources: Own calculations and Consensus Forecasts by Consensus Economics.

Business climate index::

European Commission business climate indicator for the euro area in standard deviation points, seasonally adjusted; monthly data. Source: Haver Analytics.

Policy uncertainty index::

The News-based Policy Uncertainty Index quantifies newspaper coverage of policy-related economic uncertainty. The index is based on news articles from 2 papers from each of the largest 5 European economies (Germany, the United Kingdom, France, Italy, and Spain). The papers include El Pais, El Mundo, Corriere della Sera, La Repubblica, Le Monde, Le Figaro, the Financial Times, The Times of London, Handelsblatt and FAZ. The primary measure for this index is the number of news articles containing the terms uncertain or uncertainty, economic or economy, as well as policy relevant terms (scaled by the smoothed number of articles containing “today”). Policy relevant terms include: policy, tax, spending, regulation, central bank, budget, and deficit. All news searches are done in the native language of the paper in question. Each paper-specific series is normalized to standard deviation 1 prior to 2011 and then summed. The series is normalized to mean 100 prior to 2011; monthly data. Source: http://

Bank loans::

Annual change of the log of euro area MFI loans to the private non-financial sector; monthly data. Source: ECB.

Effective euro exchange rate::

The nominal euro effective exchange rate is defined as a geometric weighted average of the bilateral exchange rates of the euro against the currencies of the EER-12 group of partner countries which includes Australia, Canada, Denmark, Hong Kong, Japan, Norway, Singapore, South Korea, Sweden, Switzerland, the United Kingdom and the United States. The bilateral exchange rates used in the calculation are the official ECB daily reference rates. Weights are based on trade in manufactured goods with the trading partners in the period 1999-2001 and are calculated to account for third-market effects; monthly data. Source: Haver Analytics.

10-year government bond yield::

Average yield to maturity of government bonds with maturity of ten-years (or the closest available maturity) of euro area member states, weighted by the relative amounts of relevant bonds outstanding; in percent per annum; monthly average of daily data. Source: ECB.

Term spread::

Difference between the euro area average 10-year government bond yield and the three-month Euribor; in percent per annum; monthly average of daily data. Source: ECB.

BBB corporate bond spread::

Yield spread between BBB-rated bonds of non-financial corporations and AAA-rated government bonds with five to seven years of maturity based on Bank of America Merrill Lynch bond indices for the euro area; in percent per annum; monthly average of daily data. Source: Datastream.

High yield corporate bond spread::

Yield spread between non-investment grade bonds of non-financial corporations of all maturities and AAA-rated government bonds with three to five years of maturity based on Bank of America Merrill Lynch bond indices for the euro area; in percent per annum; monthly average of daily data. Source: Datastream.

Option-implied stock volatility::

Measured by the main EURO STOXX 50 Volatility Index (VSTOXX). The VSTOXX does not measure implied volatilities of at-the-money EURO STOXX 50 options, but the square root of the implied variance across all options of a given time to expiry. The main index is designed as a rolling index at a fixed 30 days to expiry that is achieved through linear interpolation of the two nearest available sub-indices; in percent per annum; monthly average of daily data. Source: Datastream.

Appendix B: Additional table

Table 4 Forecast error variance decomposition for the benchmark VAR

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Kremer, M. Macroeconomic effects of financial stress and the role of monetary policy: a VAR analysis for the euro area. Int Econ Econ Policy 13, 105–138 (2016).

Download citation


  • Financial stress index
  • Financial stability
  • Non-standard monetary policy
  • Central bank balance sheet
  • Monetary policy shocks
  • VAR models