Skip to main content
Log in

Greece’s Three-Act Tragedy: A Simple Model of Grexit vs. Staying Afloat inside the Single Currency Area

  • Research Article
  • Published:
Open Economies Review Aims and scope Submit manuscript


Against the backdrop of the Greek three-act tragedy, we present a theoretical framework for studying Greece’s recent debt and currency crisis. The model is built on two essential blocks: first, erratic macroeconomic policymaking in Greece is described using a stochastic regime-switching model; second, the euro area governments’ responses to uncertain macroeconomic policies in Greece are considered. The model’s mechanism and assumptions allow either for a Grexit from the euro area or, conversely, the avoidance of Greece’s default against its creditors. The model also offers useful guidance to understand key drivers of the long-winded negotiations between the Greek government and the “institutions”.

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

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4

Similar content being viewed by others


  1. Blanchard (2014) has argued that macroeconomic policies should make avoiding dark corners a high priority.

  2. On the latest World Bank “Ease of Doing Business Index 2016” ( Greece ranks 60st, as the country with the very worst business environment in the EU, and far behind Iceland (12), Ireland (19), Portugal (23) and Spain (33).

  3. Rodrik (2014) has emphasized how difficult it may be to implement good governance and structural reforms in politically fragile countries.

  4. Refusal to strike a deal with the euro area would have terminated the ECB’s emergency lending to Greek banks, sending them into insolvency. Such a course would have led to the implosion of the Greek economy.

  5. The evolution of the ECB’s “Harmonized Competitiveness Index”, based on unit labour costs (, shows that Greece had indeed experienced one of the greatest losses in unit labour cost competitiveness prior to the start of the crisis, but from 2009Q4 to 2014Q4 unit labour cost competitiveness improved by 23 %. The considerable improvements in unit labour cost competitiveness were due to the massive drops in wages and salaries, which the “institutions” continued to insist upon with a view to improving the price competitiveness of Greek exports and import substitutes.

  6. Some observers have warned that Greece could turn into an Argentina, implying that the situation could get much worse (see, e.g., Reinhart 2015). Others claimed that following Argentina into default and into a strong depreciation could help Greece to start a recovery (see, e.g., Krugman 2015). Reinhart and Rogoff (2009) have documented that historically it has been quite common for sovereigns to default on their debts.

  7. Dynamic stochastic general equilibrium (DSGE) models are a conceivable modelling alternative. See, for example Gourinchas et al. (2016) The common practice is to solve and estimate linearized DSGE models with Gaussian shocks. These models have tangible micro foundations and are now widely used for empirical research in macroeconomics. Because these models are built on real business cycle foundations, political economy issues play a distinctly second fiddle role, if they play any role at all. Therefore, it remains challenging for policymakers to use them in the formulation of policies.

  8. The sustainability of the Greek debt was an important issue in the negotiations on how to resolve the Greek crisis. Correspondingly, cutting the effective interest burden on the Greek sovereign debt was an important part of the various assistance programmes.

  9. All uncertainty is associated with the fiscal policy process. For simplification and to sharpen the focus of the model, other stochastic shocks have been omitted. Furthermore, to keep the analysis transparent, the adequacy of the recommendations for action from the “institutions” are taken as given and are not scrutinized. For example, one might argue that austerity in a weak economy could be self-defeating as fiscal tightening curtails economic growth. One might also argue that Greece has a debt overhang. Finally, while structural reforms could possibly create a favourite environment for growth in the long term (IMF 2015c, pp. 104–107), an immediate payoff is doubtful given the largely unchanged governance structure. These topics of discussion indicate that ongoing membership in the euro area does not necessarily mean a “Grecovery”. For a thorough examination of the channels through which structural reforms may promote growth and the optimal sequencing of reform, see Christiansen et al. (2013) and Eggertsson et al. (2014).

  10. The volatility of the political system is clearly not an exogenous variable. Arguably, as regime transitions occur exogenously, they can be regarded as a measure of our ignorance rather than our understanding. Yet, the impact of political uncertainty is interesting, regardless of the actual causes. Our approach follows Zellner’s (1992) “KISS” (i.e. keep it sophisticatedly simple) principle. Ours is not the first paper to show that many economic variables can be modelled with the aid of Markov switching models. There is an extensive empirical literature modelling economies as following regime-switching processes. For background, see Hamilton (1989) and Kim and Nelson (1999) and the references contained therein. These findings motivate models that build regime-switching policy rules directly into theoretical frameworks.

  11. As shown in Appendix 1, fiscal policy has been conducted in a stop-and-go fashion over the past decade and thus the Markov switching process is able to capture the recurrent fiscal balance regime changes in Greece. The two regimes correspond roughly to periods in which most observers believe that fiscal policy actually differed. In any case, the data belie the fragility of the fiscal stance and provide useful clues in terms of where to look for potential model components.

  12. The following assumptions stem from the currency crisis literature. The models in this literature start with the seminal continuous-time, perfect foresight model developed by Krugman (1979). Masson (2007, pp. 3–60) provides a thorough review of the currency crisis literature. Note that the paper isn’t really a currency crisis paper. Contrary to the first generation crisis approach no speculative attack occurs which forces a transition to a floating rate prior to the depletion of the reserves. What the current paper does have in common with first generation approaches is the use of a monetary model and the modeling of money growth in the post-crisis regime as a function of debt monetization. For recent contributions to the “sudden stop” literature, see Montiel (2013, 2014) and Sula (2010).

  13. In the course of the debate, a temporary Grexit has also been suggested. If the ECB allowed the Greek authorities to introduce an emergency parallel “currency”, Greece might in effect suspend its euro area membership without technically leaving. This could then be reversed if Greece struck a deal with its creditors at a later date. By continuing as part of the euro area, the Bank of Greece might retain credibility, which it would otherwise lack. The positive impact would be that the Greek economy might not slump as far as it would otherwise and the drachma might depreciate less than otherwise. Technically, this can also be modelled by the Markov switching approach.

  14. The average maturity of the Greek sovereign debt is approximately 16 years (De Grauwe 2015). This is considerably longer than the maturities of the government debt of the other euro area countries.

  15. The numerical results should be viewed as largely illustrative. Applying the modelling framework in practice yields an intuitive interpretation of the model without requiring a background in stochastic calculus to understand the arguments in the text.

  16. We use the term “creditors’ decision rule” as it better captures the economic concept than the more technical counterpart b τ , i.e. our specific measure. This also has the virtue of easing exposition and avoiding ambiguity.

  17. In 2016, the institutions will resume their talks with Greece concerning another debt restructuring, with the main objective of ensuring that Greece does not face unstable dynamics. Note that default on sovereign debt is rarely full and absolute. Generally, payments are suspended and restructuring takes place. This process typically involves both a reduction in total commitments and a rescheduling of payments. Reinhart and Trebesch (2015) have shown that since its independence in 1829, Greece has defaulted four times on its external creditors. In other words, bailouts are a recurrent theme in Greek history.

  18. There is no cast iron rule for what debt ratio is too high, but there is no question that where Greece is heading with large (small) p 01 (p 10) will not be sustainable in the medium to long term.

  19. For a recent DSGE analysis of the costs and benefits of delaying austerity in Greece, see House and Tesar (2015). In this analysis, the authors assume the full credibility of the Greek government with regard to future reforms. In other words, policy uncertainty has been supressed in the analysis. Cacciatore et al. (2015) have recently explored the consequences of labour and product market reforms within monetary union in a DSGE setup.

  20. After exit, Greece would have to negotiate continued EU participation. The EU treaties have a provision for leaving the union, but not just the eurozone. That negotiation would be all the more difficult were new authorities in Greece to default on debt to the European Financial Stability Facility, the ECB and the IMF. Furthermore, Greek firms would face legal and financial disaster. Some contracts governed by Greek law would be converted into a new drachma, while other foreign law contracts would remain in euros. Many contracts could end up in legal disputes over whether they should be converted or not.

  21. In Mohsen and Economidou (2005, Table 3) interest rate semi-elasticities for Greece are presented. The way to interpret the coefficients in the log-linear specification is as the percentage change in (m t  − p t ) that we obtain from one unit change in r t . This is in contrast to the log-log specification in Eq. (15), in which α 1 gives the percentage change in (m t  − p t ) which we obtain from a percentage change in r t . Therefore, one has to convert the semi-elasticity to full elasticity.

  22. During exchange rate turmoil, financial markets may revise the perceived probability of realignment upwards and may even anticipate an overshooting of the exchange rate after a collapse.

  23. A similar robustness analysis can be conducted for other model parameters but leads to no substantial additional insights.

  24. The potential gains of structural reforms depend critically on the interaction between several policy areas, such as competition in product markets, streamlined labour law, the efficiency of public administration and the ease of doing business, to name but a few. In other words, jointly implemented reforms tailored to national conditions are needed to obtain sizeable growth effects.


  • Blanchard O. (2014) “Where danger lurks”. Finance & Development, International Monetary Fund, September 2014, vol. 51, No. 3

  • Cacciatore M, Duval R, Fiori G, Ghironi F (2015) “Short-term pain for long-term gain: market deregulation and monetary policy in small open economies”. NBER Working Paper No. 21784, Cambridge (Mass.)

  • Christiansen L, Schindler M, Tressel T (2013) Growth and structural reforms: a new assessment. J Int Econ 89:347–356

    Article  Google Scholar 

  • Consiglio A, Zenios S (2015) “Greek debt sustainability: the devil is in the tails”., 12 August

  • De Grauwe P (2015) “Greece is solvent but illiquid: policy implications”., 3 July

  • Eggertsson G, Ferrero A, Raffo A (2014) Can structural reforms help Europe? J Monet Econ 61(C):2–22

    Article  Google Scholar 

  • Gourinchas P-O, Philippon T, Vayanos D (2016) The analytics of the Greek crisis. NBER Working Paper No. 22370, Cambridge (Mass.)

  • Hamilton JD (1989) A new approach to the economic analysis of Nonstationary time series and the business cycle. Econometrica 57:357–384

    Article  Google Scholar 

  • House CL, Tesar LL (2015) “Greek budget realities: no easy options”. NBER Working Paper No. 21688, Washington

  • IMF (2012) “Greece: request for extended arrangement under the extended fund facility – staff report; staff supplement; press release on the executive board discussion; and statement by the executive Director for Greece”. IMF Country Report No. 12/57, Washington

  • IMF (2015a) “Greece: preliminary draft debt sustainability analysis”. Country Report No. 15/165, Washington

  • IMF (2015b) “Greece: an update of IMF staff’s preliminary public debt sustainability analysis”. IMF Country Report No. 15/186, Washington

  • IMF (2015c) World economic outlook. April, Washington

  • Kim C-J, Nelson CR (1999) State-space models with regime switching. MIT Press, Cambridge

  • Krugman P (1979) A model of balance of payments crises. J Money Credit Bank 11:311–325

    Article  Google Scholar 

  • Krugman P (2015) “Argentine lessons for Greece”. The New York Times, 9. July 2015

  • Masson PR (2007) Lectures in international finance. World Scientific Publisher, New Jersey

  • McQuinn K, Whelan K (2015) “Europe’s long-term growth prospects: with and without structural reforms”. University College Dublin, UCD Centre for Economic Research Working Paper No. WP 15/08, Dublin

  • Mohsen B-O, Economidou C (2005) How stable is the demand for money in Greece? Int Econ J 19:461–472

    Article  Google Scholar 

  • Montiel PJ (2013) The simple analytics of sudden stops. Open Econ Rev 24:267–281

    Article  Google Scholar 

  • Montiel PJ (2014) Capital flows: issues and policies. Open Econ Rev 25:595–633

    Article  Google Scholar 

  • Reinhart CM (2015) “What Greece can expect”. Bloomberg Views, 9 July 2015

  • Reinhart CM, Rogoff KS (2009) This time is different: eight centuries of financial folly. Princeton University Press, Princeton

  • Reinhart CM, Trebesch C (2015) “The pitfalls of external dependence: Greece, 1829–2015”. NBER Working Paper No. 21664, Cambridge (Mass.)

  • Rodrik D (2014) When ideas trump interests: preferences, worldviews, and policy innovations. J Econ Perspect 28:189–208

    Article  Google Scholar 

  • Sula O (2010) Surges and sudden stops of capital flows to emerging markets. Open Econ Rev 21:589–605

    Article  Google Scholar 

  • Zellner A (1992) Statistics, science and public-policy. J Am Stat Assoc 87:1–6

    Article  Google Scholar 

Download references


We would like to thank the editor and two anonymous referees for helpful comments on an earlier draft. The usual disclaimer applies.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Yu-Fu Chen.


Appendix 1: Regime Switching in Greece’s Budget Surplus/Deficit to GDP Ratio

A two-regime Markov switching model, allowing for regime switching in coefficients and variances, is confronted with quarterly Greek data for the government budget surplus/deficit-to-GDP ratio δ t  from 2006. The model takes the following form:

$$ -{\delta}_t={\alpha}_0+{\upalpha}_1{s}_t+{u}_t $$
$$ {u}_t={\phi}_1{u}_{t-1}+{\phi}_2{u}_{t-2}+\cdots +{\phi}_r{u}_{t-r}+{\varepsilon}_t $$
$$ {\varepsilon}_t\sim N\left(0,{\sigma}_0^2+{\sigma}_1^2{s}_t\right) $$

where s t  = {0, 1}, α 0, and \( {\sigma}_0^2 \) are the mean and variance in state s t  = 0 and the parameters α 0 + α1 s t  and \( {\sigma}_0^2+{\sigma}_1^2{s}_t \) are the mean and variance in state s t  = 1, respectively. The lags in Eq. (26) are estimated ex post, i.e. r has no effect on the Markov switching estimates.

Table 1 Maximum Likelihood Parameter Estimates and Standard Errors

Delving into the regression output reveals the presence of two regimes, one characterized by low and the other by high the government budget surplus/deficit. Smoothed probabilities are estimated using the entire sample.

Fig. 5
figure 5

Model Estimates

Appendix 2: Derivation of the Solutions for v 0 and v 1

Subtracting Eqs. (8) from (7) gives us:

$$ \left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)\left({v}^1-{v}^0\right)={\delta}^0-{\delta}^1+\frac{\partial \left({v}^1-{v}^0\right)}{\partial t}. $$

It can clearly be seen that the solution to v 1 − v 0 is:

$$ \left({v}^1-{v}^0\right)=\frac{\delta^0-{\delta}^1}{r_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\left(1-{e}^{-\left({r}_b-\pi -{\eta}_y+p+{p}_{10}\right)\left(\tau -t\right)}\right). $$

Substituting (29) into Eqs. (7) and (8) in the main text yields:

$$ \left({r}_b-\pi -{\eta}_y\right){v}^0=-{\delta}^0+\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{r_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\left(1-{e}^{-\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)\left(\tau -t\right)}\right)+\frac{\partial {v}^0}{\partial t}, $$
$$ \left({r}_b-\pi -{\eta}_y\right){v}^1=-{\delta}^1-\frac{p_{10}\left({\delta}^0-{\delta}^1\right)}{r_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\left(1-{e}^{-\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)\left(\tau -t\right)}\right)+\frac{\partial {v}^1}{\partial t}. $$

As Eqs. (30) and (31) are no longer coupled, they can be solved separately. To keep the analysis simple, and without loss of generality, we can assume that v 0 has the following solution:

$$ {v}^0=\alpha \left(1-{e}^{-\left({r}_b-\pi -{\eta}_y\right)\left(\tau -t\right)}\right)+\beta \left(1-{e}^{-\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)\left(\tau -t\right)}\right). $$

Substituting Eq. (32) back into Eq. (30) and rearranging leaves us with:

$$ \begin{array}{l}\ \left[\left({r}_b-\pi -{\eta}_y\right)\left(\alpha +\beta \right)+{\delta}^0-\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{r_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\right]\\ {}-\left[\left({r}_b-\pi -{\eta}_y\right)\alpha -\left({r}_b-\pi -{\eta}_y\right)\alpha \right]{e}^{-\left({r}_b-\pi -{\eta}_y\right)\left(\tau -t\right)}\\ {}-\left[\left({r}_b-\pi -{\eta}_y\right)\beta -\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{r_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}}-\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)\beta \right]\\ {}{e}^{-\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)\left(\tau -t\right)}=0.\end{array} $$

Equation (33) holds if all items in parentheses are equal to zero. The second set of parentheses related to \( {e}^{-\left({r}_b-\pi -{\eta}_y\right)\left(\tau -t\right)} \) is already equal to zero. From the first and third sets of parentheses, we have:

$$ \upbeta =-\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{\ \left({p}_{01}+{p}_{10}\right)\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)}. $$

and the first set gives us:

$$ \alpha +\beta =-\frac{\delta^0}{\left({r}_b-\pi -{\eta}_y\right)}+\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{\left({r}_b-\pi -{\eta}_y\right)\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)}. $$

Substituting Eq. (34) into Eq. (35) yields:

$$ \begin{array}{l}\alpha =-\frac{\delta^0}{\left({r}_b-\pi -{\eta}_y\right)}+\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)}\left[\frac{1}{\left({r}_b-\pi -{\eta}_y\right)}+\frac{1}{\ \left({p}_{01}+{p}_{10}\right)}\right]\\ {}=-\frac{\delta^0}{\left({r}_b-\pi -{\eta}_y\right)}+\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{\left({r}_b-\pi -{\eta}_y\right)\left({p}_{01}+{p}_{10}\right)}.\end{array} $$

Therefore, we have the solution for v 0 as follows:

$$ \begin{array}{l}\ {v}^0=\left(-\frac{\delta^0}{\left({r}_b-\pi -{\eta}_y\right)}+\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{\left({r}_b-\pi -{\eta}_y\right)\left({p}_{01}+{p}_{10}\right)}\right)\left(1-{e}^{-\left({r}_b-\pi -{\eta}_y\right)\left(\tau -t\right)}\right)\\ {}-\frac{p_{01}\left({\delta}^0-{\delta}^1\right)}{\ \left({p}_{01}+{p}_{10}\right)\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)}\left(1-{e}^{-\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)\left(\tau -t\right)}\right).\end{array} $$

The next step is to obtain the corresponding solution for v 1. Following the same algebra steps as before, yields:

$$ \begin{array}{l}{v}^1=\left(-\frac{\delta^1}{\left({r}_b-\pi -{\eta}_y\right)}-\frac{p_{10}\left({\delta}^0-{\delta}^1\right)}{\left({r}_b-\pi -{\eta}_y\right)\left({p}_{01}+{p}_{10}\right)}\right)\left(1-{e}^{-\left({r}_b-\pi -{\eta}_y\right)\left(\tau -t\right)}\right)\\ {}+\frac{p_{10}\left({\delta}^0-{\delta}^1\right)}{\ \left({p}_{01}+{p}_{10}\right)\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)}\left(1-{e}^{-\left({r}_b-\pi -{\eta}_y+{p}_{01}+{p}_{10}\right)\left(\tau -t\right)}\right).\end{array} $$

For the original guess to be correct, we need to cross-check whether prs (37) and (38) satisfy Eq. (29). Substituting (37) and (38) back into Eq. (29) shows that Eq. (29) does indeed hold. Equations (37) and (38) resemble Eqs. (9) and (10) in the main text.

Appendix 3: Derivation Equation (24)

We normalize P * to be 1. Equations (19), (22) and (23) then give us

$$ {r}_s={r}_s^{*}+\frac{\beta_1}{\beta_0+{\beta}_1{B}_s}\left[{y}_s{P}_s\left(-{\delta}^1-\frac{p_{10}\left({\delta}^0-{\delta}^1\right)}{\ {r}_s-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\right)+{r}_s{B}_s\right] $$


$$ {\alpha}_1 \ln {r}_s={\alpha}_1 \ln \left({r}_s^{*}+\frac{\beta_1}{\beta_0+{\beta}_1{B}_s}\left[{y}_s{P}_s\left(-{\delta}^1-\frac{p_{10}\left({\delta}^0-{\delta}^1\right)}{\ {r}_s-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\right)+{r}_s{B}_s\right]\right) $$

We assume that the central bank counters the effects on the money supply via sterilisation. Notice that prior to the sudden stop, we have

$$ m-{p}^{*}- \ln (1)=m-{p}^{*}={\alpha}_0-{\alpha}_1 \ln {r}_s^{*} $$

From Eq. (40) we know that after the attack the solution is

$$ m-{p}^{*}-s={\alpha}_0-{\alpha}_1 \ln \left({r}_s^{*}+\frac{\beta_1}{\beta_0+{\beta}_1{B}_s}\left[{y}_s{P}_s\left(-{\delta}^1-\frac{p_{10}\left({\delta}^0-{\delta}^1\right)}{\ {r}_s-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\right)+{r}_s{B}_s\right]\right) $$

and thus

$$ s={\alpha}_1 \ln \left({r}_s^{*}+\frac{\beta_1}{\beta_0+{\beta}_1{B}_s}\left[{y}_s{P}_s\left(-{\delta}^1-\frac{p_{10}\left({\delta}^0-{\delta}^1\right)}{\ {r}_s-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\right)+{r}_s{B}_s\right]\right)- \ln {r}_s^{*} $$

Therefore, we have the solution for the exchange rate as follows:

$$ \ln S={\alpha}_1 \ln \left(1+\frac{\frac{\beta_1}{\beta_0+{\beta}_1{B}_s}\left[{y}_s{P}_s\left(-{\delta}^1-\frac{p_{10}\left({\delta}^0-{\delta}^1\right)}{\ {r}_s-\pi -{\eta}_y+{p}_{01}+{p}_{10}}\right)+{r}_s{B}_s\right]\ }{r_s^{*}}\right) $$

This completes the proofs.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Chen, YF., Funke, M. Greece’s Three-Act Tragedy: A Simple Model of Grexit vs. Staying Afloat inside the Single Currency Area. Open Econ Rev 28, 297–318 (2017).

Download citation

  • Published:

  • Issue Date:

  • DOI:


JEL Classification