The Impact of Sovereign Shocks

“The Impact of Sovereign Shocks” is a research work first developed in 2013 with the purpose to investigate the implications of sovereign systemic risk for the banking system.

At that time, most of the research on systemic risk was focusing primarily on banking systems, as too-big-to-fail or too-interconnected-to-fail banks were posing serious concerns about the stability of the financial system.

However, massive government interventions in banking systems, especially in Europe, made countries more fiscally constrained, that is, their ability to finance further bailouts was under question.

As a result, correlation between sovereign and banking default risk increased sharply and significantly.

In this work we shed light on the interrelationships between sovereign credit risk and macro-financial systemic risk by

  • Studying the propagation mechanism of sovereign and banking shocks;
  • Exploring potential mechanisms of shock transmission;
  • Quantifying the implications for the real economy.

In a preliminary analysis we provide empirical evidence of tail dependence in the distribution of sovereign and financial risk. In particular, we estimate quantile regressions on a panel of credit default swap (CDS) spreads of 163 European financial institutions and 28 European countries. The econometric model has the following form:

\Delta CDS^i_t=\alpha_q + X^{'}_t \times \gamma_q + \beta_q^{i\leftarrow j}\times\Delta CDS_t^j + \epsilon_t^i

where \Delta CDS^i_t and \Delta CDS^j_t are changes in credit default swap spreads of entities i, j \in \{sov, fin\}. X_t contains a set of macro variables and q=98\%.

The Figure below plots the Kernel densities of the impact coefficients \beta_q^{i\leftarrow j}. The notation Y|X identifies the distribution of Y’s impact given that X is in distress, that is, in the top 2 percentile of the distribution. In other words, the solid black line (Sov|Fin) plots the impact distribution of financial distress shocks on the sovereign sector, whereas the solid gray line (Fin|Sov) plots the impact distribution of sovereign distress shocks on the financial sector. The dashed lines represent the means of the distributions.


The impact distribution of sovereign risk presents a fatter tail than that of financial risk with more mass around large spillover rates ranging from 1.5 to 3.5. Moreover, the average sovereign spillover is 1.53 times as larger as that of financial risk (\beta_q^{fin\leftarrow sov}/\beta_q^{sov\leftarrow fin}=1.53), and they are significantly different. These results suggest that a distressed sovereign system may have systemic implications for financial institutions, and may lead to disastrous consequences for the real economy.

In order to understand the dynamics of shock propagation within and across the two systems, we present and address the two main challenges:

  • How do we measure sovereign and financial systemic risk?
  • What are and how do we identify sovereign and financial shocks?

We employ a portfolio approach to measure systemic risk in a way that is coherent with the structure of a macro-financial system. In particular, we quantify systemic risk with the DIP (Distress Insurance Premium – Huang et al. (2009)), that summarize in one number the size of the macro-system, the degree of interconnectedness, and the probability of occurrence of a catastrophic event (see Methodology). In other words, the DIP is similar to the spread on a super-senior CDO tranche, as it quantifies total losses exceeding a certain threshold.


Interestingly, even if sovereign and financial systemic risk are measured independently, they share similar magnitudes in both macro-systems and overreact to macroeconomic shocks, such as policy announcement and actions. A closer look at shocks shows how these can be grouped into sovereign and banking categories, as they hit directly that specific macro-system.

To investigate how shock impact and spill over macroeconomic systems, we employ a simple Vector Autoregressive approach (VAR), as follow:

y_t=\Phi \times z_t + u_t

where y_t=[\Delta DIP^{Sov}, \Delta DIP^{Bank}]^', z_t includes the constant and lagged values of y_t, and u_t=B \times \epsilon_t where B is the matrix of shock spillover, and \epsilon are i.i.d. innovations.

Our novel approach to study the complex interrelationship between sovereign and banking systems is to decompose shock into “shock of interest” and other shocks. Specifically,


We then instrument \epsilon_t^{sov} and \epsilon_t^{bank} as follows

(1)   \begin{equation*} u_{t}=\beta^{sov}\underbrace{\mathbf{\mathbbm{1}}_{sov}\xi_{sov}}_{\text{\ensuremath{\epsilon^{sov}}}}+\beta^{bank}\underbrace{\mathbf{\mathbbm{1}}_{bank}\xi_{bank}}_{\text{\ensuremath{\epsilon^{bank}}}}+v_{t} \end{equation*}

where \mathbf{\mathbbm{1}}_{i} with i=Sov,Bank are signed indicators that take value of +1,-1 or 0, if there is a positive, negative or no shock on a specific day, \beta^{sov}=\left[\beta^{sov\leftarrowsov},\beta^{bank\leftarrow sov}\right]^{'}and \beta^{bank}=\left[\beta^{sov\leftarrow bank},\beta^{bank\leftarrow bank}\right]^{'}with \beta^{Y\leftarrow X} being the impact of the X shock onto systemic risk of Y, and \xi measuring the average size of the exceptional event within networks to be estimated. We normalize the loading matrix such that sovereign (banking) shocks have a one-to-one impact on sovereign (banking) risk. Finally, the impact (or spillover) matrix becomes

(2)   \begin{equation*} \mbox{\ensuremath{\mathbf{B}}}=\left[\begin{array}{cc} 1, & B^{sov\leftarrow bank}\\ B^{bank\leftarrow sov}, & 1 \end{array}\right] \end{equation*}

where B^{sov\leftarrow bank}=\beta^{sov\leftarrow bank}/\xi_{bank} and B^{bank\leftarrow sov}=\beta^{bank\leftarrow sov}/\xi_{sov} measure the spillover rates, that is, B^{sov\leftarrow bank} quantifies how much of the average banking shock \xi_{bank} is propagated to the sovereign system, and viceversa.

In order to identify the instruments \mathbf{\mathbbm{1}}_{sov} and \mathbf{\mathbbm{1}}_{bank}, we employ a methodology similar to the one proposed by Collin-Dufresne et al. (2010). It proceeds as follows:

  • Step 1. Collect news articles that are economically relevant
    for the researcher, using the crisis timeline provided by several sources such as the Financial Times, the Wall Street Journal, BBC, Reuters, Bloomberg, rating agency websites, the ECB, Brugel, the Saint Louis Fed, and Stratfor.
  • Step 2. Read news articles to understand if the content refers to countries or financials. In the next section we will provide more details about our choice.
  • Step 3. Use changes in systemic risk measures (DIPs) to compute the shock size as \Delta DIP_{t}^{i}\times\mathbf{1}_{t} with i=sov,bank and where \mathbf{1}_{t} is a dummy that takes value of 1 if at time t there is an event, and zero otherwise. Keep only those sovereign (banking) shocks that cause at least a 5 bps change in the sovereign (banking) DIP.
  • Step 4. Build \mathbf{\mathbbm{1}}_{sov} and \mathbf{\mathbbm{1}}_{bank} by setting the sign so that positive (negative) shocks will decrease (increase) systemic risk.

The following table provides some examples (for the complete list see Internet Appendix)


The first set of results pertains to the two macro-systems on aggregate. The following table reports the estimated coefficients

The average exceptional shocks, \xi_{sov} and \xi_{bank}, are negative and similar in magnitude, implying that a negative exceptional sovereign (banking) shock will increase sovereign (banking) risk by 7.72 (7.03) basis points a day, on average. Interestingly, the transmission rate of sovereign shocks to banking risk is 76 percent, twice as large as that of banking shocks to sovereign risk (\nicefrac{B^{bank\leftarrow sov}}{B^{sov\leftarrow bank}}\approx2.1). To look at these magnitudes from a different perspective, we see that \xi_{sov} and \xi_{bank} account for almost the 11 percent of their sample averages of changes in sovereign and banking DIPs. Moreover, unlike banking shocks, sovereign shocks are also significantly persistent. These results suggest that shocks to sovereign risk have important implications for the stability of the system, especially for banking risk.

The second set of results pertains to the impact of shocks on sub-systems. In particular, we investigate the sources of fragility and shock spillover by testing the shock impact on:

  • countries sorted on their degree of fiscal fragility (debt-to-GDP ratio and fiscal space);
  • banks sorted on their exposure to these fiscally constrained countries;
  • local versus foreign banks highly exposed to these fiscally fragile countries.

The main picture that emerges shows that sovereign shocks impact significantly and persistently highly-indebted countries, or those with small room for fiscal maneuver. Moreover, the impact of sovereign shocks on banks highly-exposed to these countries are as important as banking shocks. Finally, local banks are largely impacted by these shocks as highly indebted countries are not able to provide further implicit guarantees (liability-side effect), and as banks suffer higher volatility on their balance sheets (asset-side effect).

In the last part of the paper, we quantify the implications on the real economy. In a VAR framework, we test the response of aggregate industrial production and unemployment to systemic shocks. The following picture shows that banking shocks are responsible of lower production and higher unemployment. Therefore, sovereign shocks have an indirect impacts on the real economy through the banking system.

Understanding the complex interaction between systems of governments and financial institutions is of crucial importance for policymakers. We provide evidence that monitoring banks alone does not offer a complete picture of the sources of systemic shocks. Fiscally fragile governments make the financial system more vulnerable to shocks as their primary balance is not enough to support extraordinary expenses related, for example, to banking bailouts.