Background
In my previous post - The Stormy Petrel of Eurozone Budget Deficits: Current Account Deficits - I examined this question:
Do pre-crisis current account balances, as opposed to pre-crisis fiscal balances, better predict crisis-period fiscal balances (quantitatively) for countries that are part of the European currency union (EMU)?
I showed that the answer was yes and discussed the implications in the context of the recently concluded EU summit. The analysis in the post focused on current account (CA) and fiscal balances in the 7 years preceding the start of the Great Recession (pre-crisis) and fiscal balances in the 2008-2010 (crisis) time period. Countries that joined the Euro currency union in 2007 or beyond were excluded from the analysis since they were not on the Euro long enough in advance of the financial crisis. The key finding was that the average CA balance/GDP by country for 2000-2007 is a much better quantitative predictor of the average fiscal balance/GDP for the same country in 2008-2010, than is the average fiscal balance/GDP of the country in 2000-2007. The EU "fiscal union" was largely a fiscal austerity pact, with inordinate emphasis on minimizing fiscal deficits and sovereign debt levels as a % of GDP, with little or no focus on the current account. However, as I discussed in the above post, pre-crisis fiscal balances are less effective in predicting future fiscal balances than are pre-crisis CA balances. Hence, I argued that a "fiscal union" that demands fiscal austerity as a way to ensure low fiscal deficits in the future, with no attention paid to CA balances, is unlikely to consistently prevent large future fiscal deficits.
In this post, I turn my attention to government debt - the other element of the EU pact. The question I ask here is:
Do pre-crisis current account balances/GDP, as opposed to pre-crisis government debt/GDP, better predict crisis-period debt/GDP or changes in debt/GDP (quantitatively) for countries that are part of the European currency union (EMU)?
The answer is mixed as I will explain below.
Summary
In this post, I summarize the results of a simple analysis of the average current account (CA) and government debt, as a percentage of GDP, for nearly a dozen European countries that use the Euro as their currency [Section 1]. With the usual caveat that correlation does not always imply causation, the key findings can be summarized as follows:
1. The average debt/GDP change from 2000-2007 to 2008-2010 was ~12.2 percentage points, with a standard deviation of ~11.9 percentage points. If Ireland and Luxembourg are excluded as outliers, the average is 10.0 and standard deviation is 9.8 [Section 1].
2. The % change in average debt/GDP between 2000-2007 and 2008-2010 was ~ 35.2%, with a standard deviation of 47.3%. If Ireland and Luxembourg are excluded as outliers, the average is 14.9% and the standard deviation is 13.9% [Section 1].
3. Pre-crisis average debt/GDP was a fairly good predictor of the absolute value of post-crisis average debt/GDP (slightly greater than 1X) but was poor in predicting the % change to pre-crisis average debt/GDP due to the crisis [Section 2]. The reason for this behavior is unclear.
4. Pre-crisis average current account balance/GDP is a low to moderate predictor of the absolute value of post-crisis average debt/GDP as well as the % change to pre-crisis average debt/GDP due to the crisis [Section 3].
Hence, if we were interested in knowing the likely % change to average debt/GDP [pre- to post-crisis], the pre-crisis average CA balance/GDP was more likely to give us some insight on this than the pre-crisis average debt/GDP. If, on the other hand, we were interested in knowing the likely absolute average debt/GDP [post-crisis], the pre-crisis average debt/GDP hits the mark much more effectively on this than than the pre-crisis average CA balance/GDP. In the context of the recently concluded EU "fiscal union" summit, a focus purely on nominal debt/GDP levels will likely provide less insight on the mechanisms that might allow for net debt reduction. A focus on CA balances/GDP is more likely to be useful in understanding how to shape the % change in debt/GDP in the desired direction.
1. Eurozone Countries - Assumptions & Data
The Euro country data set and assumptions for this analysis are summarized in my previous post. I picked 2000 as a staring point since the Euro was only introduced in 1999. The 2007 cut-off year was picked to align to the start of the Great Recession in the US and the attendant massive financial crisis. I also looked at the year 2010 by itself to sort of isolate the fiscal situation of the various countries in the first full year after the nominal end of the Great Recession in the U.S. in 2009. Finally, I excluded the following countries from the analysis because they were on the Euro currency for either zero years or a very minimal time period prior to the financial crisis:
- Estonia - since it only joined the EMU in 2011
- Slovakia - since it only joined the EMU in 2009
- Cyprus and Malta - since they only joined the EMU in 2008
- Slovenia - since it only joined the EMU in 2007
All of the debt, current account balance and GDP data were obtained from Eurostat and NPR. I calculated all averages. The data points that are the basis of the charts in this post are summarized in the table below. If you find any significant errors, please do alert me (via Twitter or by adding a comment to this post) so that I can make a correction if needed. The Appendix at the bottom of this post has a link to all of the charts that I put together.
It's worth noting that:
- The average debt/GDP change from 2000-2007 to 2008-2010 was ~12.2 percentage points, with a standard deviation of ~11.9 percentage points. If Ireland and Luxembourg are excluded as outliers, the average is 10.0 and standard deviation is 9.8.
- The % change in average debt/GDP between 2000-2007 and 2008-2010 was ~ 35.2%, with a standard deviation of 47.3%. If Ireland and Luxembourg are excluded as outliers, the average is 14.9% and the standard deviation is 13.9%
Let's dive into the data using a series of charts.
2A. Eurozone Government - Pre-Crisis Debt/GDP v. Crisis Period Debt/GDP
The recent EU summit resulted in what one might objectively characterize as a fiscal austerity pact rather than a true "fiscal union" - as discussed before. The pact focused primarily on fiscal deficits (or budget deficits) and sovereign debt of the member countries and sought to set limits on those entities as a percentage of GDP. The strong emphasis on debt naturally raises the question - is past performance a good predictor of the future? To start with, I therefore compared the average debt/GDP for each country during the 2000-2007 time period to the average debt/GDP for the same countries during the 2008-2010 period, as well as to the 2010 debt/GDP for the same countries.
The nominal positive correlation is strong for the data set below (R-squared is ~0.85). A similar chart showing just the 2010 debt/GDP in the vertical axis is available for reference here. As I will show later, Ireland and Luxembourg are unusual and in some ways outliers (not necessarily in this basic debt progression chart below) and the same charts excluding them are available here and here.
2B. Eurozone Government - Pre-Crisis Debt/GDP v. Crisis Period % Change in Debt/GDP
A related, interesting question is whether pre-crisis average debt/GDP can be used to effectively project the % change in average debt/GDP that one might expect as a result of the crisis. Here is a chart that plots these two parameters. The R-squared on this plot is moderately high but quite misleading because the fit is overall quite poor and is skewed primarily by Ireland and Luxembourg. Without these two countries, as I'll show below, the fit is poor.
I've explained before why Ireland is clearly a massive outlier. As this study shows, Ireland' asset price bubble was much higher than the ridiculously large bubble in the U.S. and the recklessness of some of Ireland's banks is well known. In the early 2000s, Ireland was actually running CA surpluses or small CA deficits - its CA deficits only exploded in the latter years. A key difference between Ireland and some other countries currently suffering from the threat of financial collapse is that sovereign debt was quite modest to small in Ireland - the problem was the enormous private debt of Irish banks, which the Irish government decided to address by going deep into debt at the expense of taxpayers, in order to bail out the banks. This is a key reason why Ireland is an outlier (small debt prior to crisis and massive increase in debt during the crisis). Of course, other Eurozone countries also accumulated high private debt, but Ireland was #1 by far. Hence, Ireland's surging sovereign debt was largely the consequence of dealing with large private debt. This bailout-induced single data point significantly distorts the entire data set. I was therefore curious to see what the charts would look like without Ireland in the mix.
Luxembourg is the other major outlier. It had extraordinarily low public debt/GDP prior to the crisis - in the low single digits! Hence, with even a high single digit percentage point increase in actual debt/GDP, the % change in the debt/GDP looks massive for Luxembourg. No other Eurozone country had such small debt/GDP prior to the crisis - hence, I would argue that Luxembourg didn't quite fall into the pattern of more highly indebted countries. As a result, I've created a version of the above chart that excludes Ireland and Luxembourg - as shown below. Without those countries, any superficial correlation vanishes. In other words, for most Eurozone countries, the absolute level of average public debt/GDP in the pre-crisis period is a poor predictor of the % change in average debt/GDP during the crisis. This is true even when looking at 2010 data alone (here and here).
The next question then, in keeping with the discussion in the previous post, is whether current account balances might provide more insight.
3A. Eurozone Government - Pre-Crisis CA Balance/GDP v. Crisis Period Debt/GDP
The companion charts to the ones shown in Section 2A have average current account balances (sometimes loosely referred to as trade deficits or surpluses) in the X-axis. The chart showing the average debt/GDP for 2008-2010 in the Y-axis is below. As you can see in the chart below, past CA balances only modestly correlate (negatively) to the absolute magnitude of future debt/GDP - the correlation improves when comparing 2010 data alone here. However, removing Ireland and Luxembourg from the data-set makes the correlation worse (see here and here).
3B. Eurozone Government - Pre-Crisis CA Balance/GDP v. Crisis Period % Change in Debt/GDP
Current account balances for a particular time period might not necessarily correlate strongly to the debt/GDP in that same time period (see here and here), in part because the baseline debt/GDP at the time was probably significantly influenced by past policy. But it is worth asking whether they are more likely to influence the change to the debt/GDP value from where it started, than the actual value itself. In the next chart, we plot the average CA balance/GDP for the 2000-2007 time period in the X-axis and compare it to the % change in debt/GDP in the crisis years (2008-2010). The correlation is not only poor but also counter-intuitive - but this is largely because of two outliers - Ireland and Luxembourg.
When we remove Ireland and Luxembourg from the data set, here is what we get - a modest negative correlation roughly similar to what we observed in Sec 3A. Notice that, in comparison to the relevant chart in Sec. 2B (R-squared of 0.01), the R-squared is higher here (0.36). The relationship is not conclusive by any means, but the pre-crisis average CA balance/GDP appears to do a better job of predicting the changes in debt/GDP in the crisis years, than does the pre-crisis average debt/GDP itself. The correlation for the former improves if we focus just on 2010.
The data presented here suggest the need for deeper analysis as well as for a focus on both debt/GDP and current account balances in looking for long-term solutions to the financial crisis.
Appendix: List of All Charts Available
1. 12 EMU Countries: Average CA Balance/GDP (2000-2007) v. Average Government Debt/GDP (2000-2007)
2. 12 EMU Countries: Average CA Balance/GDP (2000-2007) v. Average Government Debt/GDP (2008-2010)
3. 12 EMU Countries: Average CA Balance/GDP (2000-2007) v. Average Government Debt/GDP (2010)
4. 12 EMU Countries: Average Government Debt/GDP (2000-2007) v. Average Government Debt/GDP (2008-2010)
5. 12 EMU Countries: Average Government Debt/GDP (2000-2007) v. Average Government Debt/GDP (2010)
6. 12 EMU Countries: Average CA Balance/GDP (2000-2007) v. % Change in Average Government Debt/GDP (2008-2010 v. 2000-2007)
7. 12 EMU Countries: Average CA Balance/GDP (2000-2007) v. % Change in Government Debt/GDP (2010 v. 2000-2007)
8. 12 EMU Countries: Average Government Debt/GDP (2000-2007) v. % Change in Average Government Debt/GDP (2008-2010 v. 2000-2007)
9. 12 EMU Countries: Average Government Debt/GDP (2000-2007) v. % Change in Government Debt/GDP (2010 v. 2000-2007)
---
10. 10 EMU Countries excluding Ireland and Luxembourg: Average CA Balance/GDP (2000-2007) v. Average Government Debt/GDP (2000-2007)
11. 10 EMU Countries excluding Ireland and Luxembourg: Average CA Balance/GDP (2000-2007) v. Average Government Debt/GDP (2008-2010)
12. 10 EMU Countries excluding Ireland and Luxembourg: Average CA Balance/GDP (2000-2007) v. Average Government Debt/GDP (2010)
13. 10 EMU Countries excluding Ireland and Luxembourg: Average Government Debt/GDP (2000-2007) v. Average Government Debt/GDP (2008-2010)
14. 10 EMU Countries excluding Ireland and Luxembourg: Average Government Debt/GDP (2000-2007) v. Average Government Debt/GDP (2010)
15. 10 EMU Countries excluding Ireland and Luxembourg: Average CA Balance/GDP (2000-2007) v. % Change in Average Government Debt/GDP (2008-2010 v. 2000-2007)
16. 10 EMU Countries excluding Ireland and Luxembourg: Average CA Balance/GDP (2000-2007) v. % Change in Government Debt/GDP (2010 v. 2000-2007)
17. 10 EMU Countries excluding Ireland and Luxembourg: Average Government Debt/GDP (2000-2007) v. % Change in Average Government Debt/GDP (2008-2010 v. 2000-2007)
18. 10 EMU Countries excluding Ireland and Luxembourg: Average Government Debt/GDP (2000-2007) v. % Change in Government Debt/GDP (2010 v. 2000-2007)
Comments