The declining sensitivity of asset prices to events in Greece

Menno Middeldorp.

The risk of Greece exiting the euro area (Grexit) has unsettled financial markets regularly over recent years. A New Year poll suggested that most Greeks feel that 2016 will see the threat of Grexit return. However, even if the probability of Grexit rises again, that does not necessarily mean that financial markets will respond with similar volatility. Indeed, this post shows that, based on the sensitivity of international asset prices to those in Greece itself, each successive episode of Greek stress has in turn caused less stress abroad.

To measure the sensitivity of global financial markets to Grexit risk I regress euro area, UK and US asset prices on a composite of Greek asset prices. I do this for three different episodes when Greek financial markets exhibited signs of stress and there was also a high volume of news articles on Bloomberg that referred to Grexit risk. For most euro area, UK and US asset prices, their sensitivity to Greek stress declined in each successive episode.

Identifying stress in Greek financial markets

The related risks of Greek sovereign default, banking system instability and Grexit are likely to manifest themselves most clearly in Greek financial markets themselves. Specifically, if Grexit risks increase one would expect rising Greek government bond yields while Greek equity prices, particularly bank shares, decline. One good way to capture overall developments in Greek asset prices is to use a statistical technique called principal components, which transforms the common movement in a set of time series into a new synthetic variable. The main practical problem with this approach is that during the height of Grexit worries in 2015, Greek equity markets were closed, so I use daily changes in the following data to construct the principal component (PC) measure of Greek asset prices:#

*FTSE/Athex Large cap index: This equity index is tracked by an exchange traded fund (ETF) that is listed in the US, so I use its price (adjusted for exchange rate changes) to estimate the equity index on days the market in Athens is closed.

*Euro-equivalent price of American Depository Receipts of the National Bank of Greece: Bank equities are likely to be closely related to Grexit risk, but with the market in Athens closed the best alternative is to use the US traded shares of the National Bank of Greece. Given that this is the largest Greek bank, it is likely to be representative of the banking sector as a whole.

*Greek 10-year sovereign bond yields: Greek sovereign bonds have continued to trade so can be included in the dataset. I have excluded the 12 March 2012 observation from the time series when yields dropped from 46% to 18.6% as private sector debt was written down.

Volatility is a common way to measure stress in asset prices, so in Chart 1 I show a centred rolling 60 weekday standard deviation of the PC (pink line).

Over the last five years volatility in Greek financial markets has mostly been related to risks emanating from Greece itself.  However, there are also periods when global events apparently moved Greek markets, for example the 2013 “taper tantrum” and more recent volatility. Because I am interested in identifying the effect of Grexit risk on international financial markets, these periods when the causality went the other way around are excluded from the analysis.

One way to distinguish between moves in the PC due to Grexit risks from other drivers is to make sure that they were accompanied by news about Grexit. The blue line in Chart 1 shows a 60 weekday centred moving average of the percent of news articles on Bloomberg that contain keywords related to Grexit.

There are three episodes when the volatility of the Greek principal component and the Grexit news index measure rise and fall more or less simultaneously. The regression coefficients discussed below are estimated on these three periods. The exact start and end dates for the regression samples (represented by the shaded areas in Chart 1) correspond to notable dates in the Greek crisis timeline that are chosen so that the Grexit and Greek PC lines are roughly at similar levels in each cut-off, e.g. the pink line in (c) is at a similar level to the pink line at (e).

Chart 1: Greek financial market principal component and Bloomberg Grexit news


(a) 23/12/09 String of downgrades by major rating agencies culminating in a rating cut by Moody’s.
(b) 07/07/10 Greek parliament passes key pension reform as part of EU/IMF support package.
(c) 27/07/11 Another string of rating downgrades ending with one by S&P.
(d) 26/07/12 ECB President Mario Draghi gives his “Whatever it takes” speech.
(e) 08/12/14 Parliament starts a failed attempt to elect a new president, later resulting in new elections.
(f) 14/08/15 Greek parliament passes package of laws for third support program
“Grexit news” represents the share of Bloomberg articles that meet the following search criteria
Sources: Datastream, Bloomberg, Wikipedia, author’s calculations.

Measuring the sensitivity of international asset prices to Grexit stress

The sensitivity of international financial markets to Grexit risks are measured by regressing daily changes in the asset prices listed below to the daily change in the Greek PC.

  • UK, US, German and euro area periphery (Italian, Spanish, Portuguese) 10-year state bond yields.
  • FTSE All-share, S&P 500 and Eurostoxx equity indices.
  • Spreads of the Bank of America Merrill Lynch sterling, US dollar and euro-denominated investment grade indices for:
    • financial corporate bonds
    • industrial (i.e. non-financial non-utility) corporate bonds
  • Bilateral exchange rates of the euro against the pound, dollar and yen.

The time-series have been normalized, so the coefficients from these regressions represent the sensitivity, in standard deviations, of the relevant asset prices to a one standard deviation change in the Greek asset price PC. Standard deviations are calculated using daily data since January 2007, rather than just the regression windows, which means that the coefficients can be compared across each episode.

Chart 2 shows the coefficients for each episode as a black bar. A confidence interval of 2 standard errors above and below the coefficient is shown as a grey line. Financial/industrial spreads and periphery sovereign yields have positive coefficients. Other sovereign yields, equity markets and euro exchange rates have negative coefficients, but to make the results easier to compare the sign of these coefficients are flipped.

Chart 2: Coefficients for each episode ±2 standard errors
Chart2_a Chart2_b Chart2_c Chart2_d Chart2_e Chart2_f

*/**/***             The coefficients for episode 1 and 3 are statistically different, i.e. Wald test rejects equivalance at the *10%, ** 5% or *** 1% probability level
Sources: Datastream, Wikipedia, author’s calculations.

Markets are less and less afraid of Grexit . . .

Clearly a number of global financial markets have responded to Greek asset prices during episodes of Grexit stress. Even in the most recent episode, most of the confidence intervals do not cross the zero line, indicating the coefficients are statistically significantly different from zero. However, each subsequent episode has seen declining sensitivity of international asset prices to the stress in Greece, as the coefficients have generally declined. In most cases the declines vs. the 2010 episode are statistically significant (see asterisks on the Chart 2 asset price labels). The sharp drop in responsiveness of financial bond spreads (Chart 2e) to Grexit stress is particularly notable and welcome, because higher bank funding costs were an important channel through which earlier episodes threatened euro area financial stability (the IMF quantified such sovereign spillovers to banks in 2011).

. . . does that mean there is nothing to fear?

The sensitivity of markets to Greek stress should reflect both the likelihood of Grexit and the degree of contagion should it occur. During the most recent episode, a Reuters poll of 57 economists on July 7th 2015 gave a 55 percent chance of Greece leaving the euro area, which was “the first time the median probability has shown Greece is more likely than not to leave the euro in many years of Reuters polls asking the same question” (this probability declined sharply to 30% in the next poll a week later, after Greece reached agreement with its creditors) Although the forecasts of economists may be different to market participants, under the assumption that markets were also pricing in a higher probability of Grexit than in the past, their more muted response would suggest a less dire view of the consequences of Greece leaving. That could reflect changes in the euro area such as stronger capital ratios of banks, the reduction of fiscal deficits and new policy tools for containing contagion. Of course, it does not follow that market participants were actually correct. Even if they were, the fact that asset prices did respond to the latest episode suggests that market participants still thought of Grexit as a disruptive event, just less disruptive than a few years ago.

Menno Middeldorp works in the Bank’s Macrofinancial Analysis Division.

If you want to get in touch, please email us at You are also welcome to leave a comment below. Comments are moderated and will not appear until they have been approved.

Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.