The radical transformation of the ECB

Students of my generation will remember what we learned about how the European Central Bank conducts monetary policy: The ECB makes one-week loans to commercial banks against top-rated collateral. This was called “Main Refinancing Operations”. The interest rate charged on these loans was known as the Main Refinancing Rate and was considered the key policy rate of the ECB, like the Fed Funds Rate in the United States. Then we learned something about Marginal Lending Facilities and Long-Term Refinancing Operations, but were told they were relatively unimportant.

This was indeed how the ECB operated – before 2009. Since then the ECB has changed its operations. It seems to me that the radical nature of this change has not been recognized enough by economists – let alone the general public.

Look at the chart below. It shows the assets held by the ECB system for purposes of monetary policy operations. The Main Refinancing Operations (the yellow area) have disappeared. In 2019 they constituted a mere 0.25% of the total monetary-policy related assets! The Long-Term Refinancing Operations (blue area) have replaced them and make up about 20% of the total now.

But the elephant in the room is the grey area that first appears in 2009 and then explodes after 2014. The ECB labels it “Securities held for monetary policy purposes”. What are those securities? Government bonds and a couple of corporate bonds as well. The ECB started to buy them with the “Securities Market Program” in 2009 and hugely expanded the purchases with the “Public Sector Purchase Program” in 2015. Last year, the ECB system held 2.6 trillion of assets in relationship with those programs. That is more than 80% of their total policy-related assets.

This means that the ECB of our old textbooks, the ECB that was envisioned by the founders of the euro, has ceased to exist. It has been replaced by an altogether different beast. The primary way in which the ECB conducts monetary policy these days consists in buying Eurozone government bonds in the open market.

This has made the ECB the single biggest lender to Eurozone governments. As I showed in my last post, 91% of all new government debt issued after 2010 is now being held by the ECB. It resembles a 3.2 trillion euro hedge fund, financed by short-term commercial bank deposits (aka “reserves”), holding a diversified portfolio of Eurozone government bonds. The equity owners of this fund are the Eurozone government themselves: they “own” the ECB, they are responsible for replenishing its equity if and when it is deemed necessary.

One implication of this radical transformation should be immediately obvious: Eurozone governments have in effect mutualized 91% of their post-2009 debt. Whenever a Eurozone government defaults on the bonds held by the ECB, the losses would be absorbed, eventually, by the other Eurozone governments.

I’m not saying that’s a bad thing or a good thing. I’m not saying it is illegal or legal. But nobody should delude themselves or others that this is not what has been happing.

Italy’s public debt, decomposed

Why is Italy’s debt so high?

Is it because the Italian government was fiscally irresponsible, spending too much and taxing too litte? Or is it because investors demand such high interest rates on Italian government bonds? Or is it a consequence of Italy’s dismal economic performance in recent years?

To answer this question, we can take a simple decomposition of the debt-to-GDP ratio. First, remember the government budget constraint:  \displaystyle dB = G - T + rB. 

where B is public debt, G is spending, T is revenue and r is the interest rate. Second, take the time derivative of the debt-to-GDP ratio  \displaystyle d\left(\frac{B}{Y}\right) = \frac{dB}{Y} - \frac{B}{Y}\frac{dY}{Y}. 

Combine the two equations and denote the GDP growth rate dY/Y by g:  \displaystyle d\left(\frac{B}{Y}\right) = \frac{G-T}{Y} + (r-g)\frac{B}{Y}. 

This equation allows us to decompose the total change in the debt-to-GDP ratio into a primary deficit component, an interest component and a growth component. The graph below shows this composition for Italy during the pre-crisis period (2000-2008) and the post-crisis period (2009-now).


In the years between the introduction of the euro and the financial crisis, Italy’s debt ratio decreased slightly by about 2 percent of GDP. During the years after the crisis, it increased by almost 30 percent of GDP.

What changed? As you can see by looking at the yellow and blue areas in the graph, it wasn’t interest payments or the primary surplus. Interest payments were around 5 percent of GDP both before and after the crisis and the Italian actually ran a primary surplus in both periods. What changed was the green area: the recent rise in the debt ratio is almost entirely due to Italy’s shrinking economy.


A perfect prediction (self-congratulation)

Three years ago, Christoph Zwick and I wrote a paper about the sustainability of Austria’s public debt. Under our preferred model, we forecasted the debt-to-GDP ratio, which at the time stood at slightly over 80 percent, to recede towards 60 percent within the next decade. We concluded that the long-run probability distribution of Austrian public debt given current data indicated no cause for alarm. How good was our projection?

Well, the new Austrian minister of finance just held his budget speech, in which he announced a zero budget deficit in the coming years. Assuming the government follows through on this plans, this would indeed bring down the debt-to-GDP ratio to 62 percent, exactly as our main projection predicted.

Here is the finance ministry’s proposed budget path:

Bildschirmfoto 2018-03-21 um 09.00.09

And here our main projection from the paper (note that the initial debt level is slightly lower due to different definitions of public debt; the dark and light blue areas indicate the 75 and 95 percent probability bands around the median, which is in black):

Bildschirmfoto 2018-03-21 um 08.54.12

Ladies and Gentlemen, this is what a perfection prediction looks like.


Debt crises in a monetary union: the case of Indiana

As Europeans we tend to think of America as new, young, and modern, whereas in Europe everything is old and traditional. At least that’s what I thought, until I noticed this while driving around in the American Midwest:


The license plate celebrates the 200th birthday of the State of Indiana in 2016. 200 years! This means, in a sense, Indiana is older than many of the states of the European Union. In 1816, Germany was still a patchwork of small territories, loosely connected through the German Confederation – of which Austria was a part. Italy was merely a geographical description – the process of Italian unification had not even begun. Greece was just a province of the Ottoman empire. Belgium, until 1815 known as „Austrian Netherlands“, was part of the United Kingdom of the Netherlands. France did exist as a nation then, however, while the people of Indiana lived for 200 years under the same political system and only once made marginal changes to their constitution (more about this below!), the French during the same time went from the post-Napoleonic Bourbon monarchy to the Second Republic to the Second Empire, then back to the Third Republic, then to totalitarian rule under the Nazis, and finally back to Fourth and now the Fifth Republic. As far as I am aware, there is no European country which has had the same constitution for the past 200 years without interruptions or major changes – the single exception I can think of is the United Kingdom which always had the same constitution: none.

There is another striking fact about the history of Indiana. Indiana has been in a monetary union with the rest of the United States for as long as it existed. And during its early history, it has had its own debt crisis which bears a striking resemblance to the recent history of the much younger European monetary union.

When Indiana became a State in 1816, it was mostly a wilderness at the margin of civilization. The only major road in the country was the Buffalo Trace – literally a trace created by migrating bison herds. Population was only 65,000 initially, but growing fast. The government of the young state decided to take the country’s infrastructure into the 19th century. And 19th century infrastructure, they figured, was going to be canals. So, they launched a giant public investment program, called the Mammoth Internal Improvement Act, spending 10 million dollars (equivalent to 260 million current dollars, roughly 100% of GDP at the time) on canals and toll roads. The heart of the project was the Wabash & Erie Canal connecting the Great Lakes with the Ohio River. „Crossroads of America“  was the official state motto of Indiana.

To finance these projects, the governor of Indiana, a certain Noah Noble, had a plan: some money was to be raised by selling public lands, some by raising taxes, and some by borrowing from the Bank of Indiana, which was partly state-owned. The Bank of Indiana refinanced itself by issuing bonds, backed by the state, at the London exchange.

Initially, the plan looked like a big success. The construction works employed many thousands of people and provided a stimulus for the economy. Borrowing costs were low and spirits were high. But soon, problems started to appear. It turned out that the government had greatly underestimated the costs of building the canals, mostly because they failed to take into account the damage done by muskrats who burrowed through the walls of the dams. Critical voices in the State Congress regarded the canals as a total waste of money. Railroads, they argued, were the future! Nobody seemed to listen.

And then, in 1837, a financial crisis broke out. The crisis was triggered by the Bank of England which, in an attempt to curb the outflow of gold and silver reserves, raised interest rates. This had a direct impact on Indiana whose borrowing costs skyrocketed. It also had an indirect effect: since the United States was on a gold and silver standard, American banks were forced to follow the Bank of England in raising interest rates, which led to a credit crunch and a nation-wide recession. (A classic example of a monetary policy spillover effect!)

The combination of stagnant tax revenues, exploding construction costs and rising interest rates meant that State of Indiana was effectively bankrupt at the end of 1841. So they sent the head of the Bank of Indiana to London to negotiate a restructuring of the debt. The creditors agreed to a haircut of 50% of the debt. In exchange, Indiana handed over control of most of the canals and roads, many of them still unfinished. The Wabash and Erie Canal was held in trust to pay off the remaining debt. It operated until the 1870s yielding a low profit, but was soon made obsolete by – the railroads which turned out to be the key infrastructure of the 19th century.

The conclusion Indiana drew from this was that the long-run costs of government borrowing far exceed the short-run benefits. Which is why in 1851, they adopted an amendment to their constitution, forbidding the State government to get into debt (except in cases of emergency).

I’d say there is a thing or two our modern European states can learn from this story.

Sind Staatsschulden eine Belastung für unsere Kinder?

Dass Staatschulden eine Belastung für zukünftige Generationen sind, gehört zu den Dingen, die nun wirklich jedes Kind über die Volkswirtschaft weiß – neben der Tatsache, dass unbegrenztes Wirtschaftswachstum unmöglich ist, und dass eine negative Leistungsbilanz schlecht für ein Land ist.

Leider ist das, was jedes Kind über die Volkswirtschaft weiß, falsch. Die Frage ist nur wie falsch. Völlig falsch? Unter Umständen richtig, aber in der Regel falsch? Oder unter Umständen falsch aber in der Regel richtig?

Sagen wir unsere Regierung beschließt jedem österreichischen Haushalt 100 Euro zu schenken. Finanziert wird das über eine neue Anleihe mit einer Laufzeit von 30 Jahren (einer Generation). Belastet diese Staatsschuld die zukünftige Generation?

Ja sicher: Die zukünftige Generation wird höhere Steuern zahlen müssen um die Anleihe zu bedienen. Angenommen der Zinssatz beträgt r und das Bevölkerungswachstum n, dann muss jeder zukünftige österreichische Haushalt 100*(1+r)/(1+n) Euro an zusätzlichen Steuern schultern.

Unsinn: Die zukünftige Generation erbt ja auch die Anleihen, die ihre Eltern gekauft haben! Die Zusatzsteuern, die sie zahlen müssen, fließen ihnen selbst zu, weil sie die Anleihen halten. Ihre Eltern haben ihr „Geschenk“ vom Staat selbst finanziert indem sie die Anleihen gezeichnet haben. Für die zukünftige Generation entsteht überhaupt keine Belastung.

Moment mal: Da haben wir aber ein paar implizite Annahme gemacht. Wer sagt denn, dass die gesamte Anleihe an die nächste Generation weitervererbt wird? In der Realität erbt nicht jeder Haushalt was, und nicht alle Haushalte haben Nachkommen, denen sie was vererben könnten. Was, wenn die erste Generation die Staatsschulden an die nächste Generation verkauft anstatt vererbt?

Sagen wir die erste Generation hat die Anleihe als Altersvorsorge gekauft (entweder direkt oder über eine Pensionsversicherung) in der Erwartung sie im Alter wieder verkaufen zu können. Zunächst bekommt jeder Haushalt der ersten Generation 100 Euro vom Staat und zeichnet Staatsanleihen im gleichen Wert. Wenn die erste Generation in den wohlverdienten Ruhestand übergeht, verkauft jeder Althaushalt die Anleihe an einen jungen Haushalt der nächsten Generation um 100*(1+r) Euro. Nach 30 Jahren – die erste Generation ist mittlerweile tot – hebt der Staat von jedem Haushalt der zweiten Generation 100*(1+r)/(1+n) Euro an zusätzlichen Steuern ein und begleicht damit seine Schuld gegenüber dem Haushalt in derselben Höhe. Somit hat sich die erste Generation um 100 Euro bereichert – zu Lasten ihrer Kinder!

Okay, aber das ist wohl auch kein realistisches Szenario. Die Wahrheit liegt wie immer irgendwo dazwischen: Sagen wir ein Anteil s der Staatsschulden wird vererbt und der Rest verkauft. Dann beträgt die Belastung der zukünftigen Generation nur 100*(1-s)*(1+r)/(1+n) pro Haushalt.

Moment, es geht noch weiter: Wer sagt denn, dass die Anleihen nur von österreichischen Haushalten gehalten werden? Was, wenn die Staatsschulden in den Händen ausländischer Haushalte sind? Sagen wir der Anteil der von Inländern gehaltenen Staatschulden beträgt d. Dann muss jeder Haushalt der zukünftigen Generation 100*(1-d)*(1+r)/(1+n) Euro an ausländische Gläubiger zahlen. Die Last der Staatsschulden verteilt sich dann wie folgt: die erste Generation trägt 100*d*s*(1+r) Euro pro Haushalt, die zweite Generation 100*(1-d*s)*(1+r)/(1+n).

Zum Beispiel: d=1/3, s=1/2, r=10%, n=5%. Dann beträgt die Last der Staatsschulden für die erste Generation 17,5 Euro und für die zweite 87,3 Euro pro Haushalt. Gemessen in Gegenwartswerten wird somit die erste Generation zu Lasten der zweiten Generation um 83,3 Euro bereichert.

Das heißt, die populäre Doktrin, dass Staatsschulden unsere Kinder belasten, stimmt in aller Regel. Nur dann, wenn die gesamte Staatsschuld im Inland gehalten und zur Gänze an die nächste Generation vererbt statt verkauft wird, ist sie falsch.

Some More on the Crisis Vulnerability of Countries

A couple of weeks back we had a great post by Christoph Zwick on current account rebalancing in the Eurozone. In the comment section I mentioned I had my doubts on what the variables are that we should look at in order to determine which countries are most vulnerable to experiencing “sudden stops” of capital inflows, thus triggering a financial crisis, which is in essence what happened in the southern Eurozone periphery. In particular I found the argument unconvincing that debt ratios are of much use in this discussion, my thoughts on which I attempted to summarize in a follow-up post I did back then.

I don’t mean to pick on this, but I feel it is important since it is a fairly widespread view that excessive debt ratios (both public and private) are a large part of the reason why some countries fared so badly in the financial crisis while others did much better. The subsequent calls for austerity, often based on this flawed analysis, are also what is destroying the future of an entire generation of young people in the affected economies.

I just happened to stumble on some research by Frankel and Saravelos (.pdf) that seems to address specifically this issue, and I thought it would be a useful addition to the conversation. In essence it is a literature overview that tries to summarize the findings regarding which variables proved to be statistically significant as “early warning indicators” for financial crises. The top 5 indicators they found are as follows (Table 1): foreign exchange reserves, the real exchange rate, GDP growth, growth in credit and the current account. Clearly not all of these are applicable to countries in a monetary union, such as the Eurozone, which essentially by definition have no foreign exchange reserves of their own, for instance. But what struck me the most is the indicator at the very bottom of that list: External Debt. In fact, of the 83 studies reviewed, only 3 found external debt to be a statistically significant indicator to predict a financial crisis. “Budget Balance”, the meaning of which is unfortunately not really defined in the paper, was found to be significant in only 9 of the 83 studies surveyed. Both government as well as external debt simply do not seem to be what we should be looking at here.

P.S.: As a side-note, some economists at Well’s Fargo (.pdf) took these indicators to estimate which of the 30 biggest developing economies in the world are more likely to be subject to crises in the near future. Might be interesting to keep an eye open.

Edit: The survey I quoted in the post relates to pre-2008 crises. The authors then conduct some analysis of their own for 2008-2009 episodes, and for some reason do find external debt to be a significant indicator. I’m not sure how that works, but it does of course change the story I tried to tell.