The Greek crisis: An autopsy

Pierre-Olivier Gourinchas, Thomas Philippon, Dimitri Vayanos

05 August 2016



For its sheer intensity and duration, the Greek crisis has been quite unprecedented. One measure says it all – real income per capita declined every single year between 2007 and 2013, a cumulated drop of 26%. Since then, it has barely risen.

To put the Greek crisis in perspective, we compare it in Figure 1 with the sample of all ‘Trifecta crises’ since 1980 – the combination of a sudden stop with output collapse, a sovereign debt crisis, and a lending boom/bust. This is the who’s who of financial crises, a distinguished group that includes Argentina and Turkey in 2001, Ecuador in 1999, Indonesia and Russia in 1998, Chile and Uruguay in 1983, and Mexico in 1982. Greece's drop in output was significantly more severe and protracted than any of these episodes.

Figure 1 Greece versus the universe of all `Trifecta crises’ since 1980

Source: Gourinchas et al (2016)

Much has been written about the drivers of the Greek crisis and its severity (Blanchard 2015). Some say that it was the inevitable consequence of the boom that preceded the crisis, a boom fuelled by unsustainable fiscal profligacy –and fraud – and cheap foreign capital. Others point to the collapse of the Greek banking system saddled with non-performing loans, and the uncertainty about euro exit, depressing investment and aggregate demand. Some blame the nasty regimen of fiscal austerity imposed by the country’s creditors, while yet others emphasise the strictures of the common currency, compounded by a lack of wage and price flexibility that make it difficult to restore competitiveness.

Quantifying the role of each of these factors is a difficult yet essential task for understanding the crisis and for formulating appropriate policies going forward. This is what we attempt to do in a recent paper (Gourinchas et al. 2016). We do this by building a tractable but realistic DSGE model of the Greek economy, which we use to replicate the evolution of key macroeconomic variables from 1999 onwards. The model is designed to capture a number of important features of the Greek crisis – foreign capital can suddenly dry up, the government can default and so can private firms and households, the banking sector faces capital constraints, and price and wages adjust slowly.

Incorporating debt, default, and price stickiness yields rich interactions between the creditworthiness of the government and that of banks, firms, and households. Consider a shock that raises the prospect of a sovereign default, increasing the government’s funding costs. The government responds by raising taxes and cutting expenditure. This causes the economy to contract, increasing the probability of default on corporate loans and household debt. To protect themselves against the higher probability of default, banks raise the rates at which they lend to firms and households. This causes private investment to decline and aggregate demand to contract further. Conversely, consider a shock that drives up the funding costs for banks – such as a sudden stop. Banks pass the higher rates onto firms and households, which causes investment to decline and output to contract. The subsequent decline in government revenues negatively affects the fiscal outlook, increasing the prospect of a sovereign debt crisis and raising the funding costs of the government.1

We put the model to two main uses. First, we decompose the movements in output, investment, and other key macroeconomic variables into the contribution of each type of shock. This helps us determine which shocks were the most important in driving the crisis dynamics. Second, we perform a number of ‘counterfactual exercises’ to identify the role played by different aspects of the institutional environment.

Figure 2 Decomposition of output

Note: Output relative to trend and the cumulated effect of each shock on output in each year.

Figure 2 shows the decomposition of output. It contains many important lessons. First, during the early stages of the boom (1999-2002), credit demand in the private sector (brown bars), low private sector default risk (light green) and moderate price inflation (dark blue) were all important drivers of output growth. Starting in 2003 and increasingly until 2009, however, an overly stimulative and ultimately unsustainable fiscal policy (red) became the main driver of output. Second, the proximate cause of the turnaround in output was the ‘sudden stop’ of private capital that caused funding costs in the private sector to increase (dark green). The drop in output accelerated in 2010 and 2012 mainly because of the decline in government spending (red), high funding costs for the government (light blue), and increasingly high funding costs for the private sector (dark green). Third, the unavoidable fiscal consolidation (red and orange) accounts for close to 50% of the output drop from peak to trough (2007 to 2013). Much of the remainder is accounted for by the sudden stop affecting the government (light blue) and the private sector (dark green).

Figure 2 also suggests that the recovery of the Greek economy is increasingly hindered by two domestic factors – elevated levels of non-performing loans in the private sector (turquoise), and excessive price rigidities in product markets despite the significant decline in wages (dark blue).

Our model allows us to perform a variety of counterfactual exercises. These exercises help shed light on why some of the shocks had large effects.2 A first exercise examines the role of leverage. Debt levels in Greece were high at the onset of the crisis. In particular, Greece’s government debt and external debt were larger (as fractions of output) than in Ireland, Italy, Portugal, and Spain. And all of government, private, and external debt was more than twice as large compared to the average of the emerging-market economies at the onset of a sudden stop episode.

Figures 3 to 5 explore what would have happened to output, investment and the current account-to-output ratio, if Greece had half the levels of government, private, and external debt in steady state, i.e. levels of debt more typical of emerging-market economies.

Figure 3 Output counterfactual

Note: The figure reports (log) real output, in deviation from steady state. Low leverage: divide steady state government, private and external debts by two. Low stickiness: divide price and wage stickiness parameters by two.

Figure 4 Investment counterfactual

Note: The figure reports (log) real investment, in deviation from steady state. Low leverage: divide steady state government, private and external debts by two. Low stickiness: divide price and wage stickiness parameters by two.

Figure 5 Current account to output ratio counterfactual

Note: Low leverage: divide steady state government, private and external debts by two. Low stickiness: divide price and wage stickiness parameters by two.

Under this ‘low-leverage’ scenario, output growth during the boom would have been smaller as the government would have been unable to raise expenditure to overly high levels. The output gap in 2007 would have been 9.2% rather than the 14.1% estimated by the model. The fiscal contraction and sudden stops would also have had a substantially more muted impact on the economy. The peak-to-trough decline in output would have been smaller by one third, and that in investment by one sixth. Investment would have declined substantially less, and the required turnaround in the current account would have been much milder.

Figures 3 to 5 suggest one important reason for the depth of Greece’s recession. Greece experienced a sudden stop episode typical of emerging market economies, but with the debt levels of an advanced economy. Here, an analogy might help. Catching the measles as a child is painful, but often relatively short-lived. Catching the measles as an adult can be much more serious and is more likely to lead to complications. Similarly, experiencing a sudden stop as an emerging market economy with moderate levels of debt can be painful but short lived. Experiencing a sudden stop as an advanced economy with much more elevated debt levels can be much more serious and more likely to lead to complications. Had debt levels been more in line with emerging-market economies, Greece would have experienced a more typical emerging market ‘trifecta’ crisis.

A second counterfactual exercise examines the role of price and wage stickiness. More flexible wages and prices may have cushioned the effects of lower domestic demand on output – and indeed the output decomposition in Figure 2 suggests that the lack of price adjustment is a factor that hindered the recovery in 2014 and 2015. What would have happened from 1999 onwards if prices and wages had been twice as flexible?

Figures 3 to 5 suggest that with more flexible prices and wages, Greece would have avoided a significant share of the boom-bust cycle. The peak-to-trough decline in output would have been smaller by 40%, and that in investment by 30%. The recovery in output would also have been sharper, with output 8% above the baseline estimates in 2015.

This exercise is different from simulating the effects of Eurozone exit. Indeed, ‘Grexit’ would have manifested itself not only via more flexible prices, but also via additional effects (e.g. on the solvency of banks and firms), which we do not model.

Summarising, our model makes it possible to quantify the role that different factors played in driving macroeconomic dynamics during the Greek crisis. We find that, while an unavoidable fiscal consolidation was the most important factor driving the drop in output, it accounted for only for half of that drop. Much of the remainder can be explained by the higher funding costs of the government and the private sector due to the sudden stop. Lower leverage would have cushioned the Greek economy somewhat from the sudden stop. The peak-to-trough decline in output would have been smaller by about a third if Greece’s levels of debt were half of their pre-crisis values. More flexible prices and wages would also have softened the effects of the drop in domestic demand – the peak-to-trough decline in output would have been smaller by about 40% if prices and wages could adjust twice as fast.


Blanchard, O (2015), “Greece: Past critiques and the path forward”,, July

Gourinchas, P-O, T Philippon, and D Vayanos (2016), “The Analytics of the Greek Crisis”, NBER Macroeconomics Annual, forthcoming (also CEPR DP No. 11334).


[1] The model features eight stochastic shocks – government spending and revenue shocks, a sovereign risk shock, private default shocks, a sudden stop shock, a credit demand shocks, and price and wage inflation shocks. We estimate the model using Bayesian methods and annual data on government revenue and spending, household debt, non-performing loans in the private sector, borrowing rates for the government and private firms, as well as price and wage inflation. Importantly, while we do not feed the model any data on aggregate output, employment, investment or the trade balance, it is able to reproduce closely the evolution of these variables.

[2] Our model could also be used as a tool to evaluate policies going forward, although we do not perform such exercises in our paper.



Topics:  Europe's nations and regions Macroeconomic policy

Tags:  eurozone, Greece, sudden stop, Grexit, debt, EU, Eurozone crisis, Fiscal crisis, sovereign debt

Professor of Economics at the University of California, Berkeley and CEPR Research Fellow

Professor of Finance, Stern School of Business, NYU and CEPR Research Affiliate

Professor of Finance and Director, Paul Woolley Centre for the Study of Capital Market Dysfunctionality, LSE