A couple of days ago I looked at how one might best classify countries, as to whether or not they had experienced a “financial crisis” since 2007.  But this chart is one reason why I’ve become increasingly sceptical that “financial crises”, however one defines them, have large or enduring adverse real economic effects.  I think I first saw it in a sets of slides by Nobel laureate Robert Lucas, and every so often I would use it to try to stir up a bit of debate at the Reserve Bank.

It is a quite simple chart of real per capita GDP for the United States, back as far as 1870.  These are Angus Maddison’s estimates, the most widely used set of (estimated) historical data, and as Maddison died a few years ago they only come as far forward as 2008.  The simple observation is that a linear trend drawn through this series captures almost all of what is going on.  More than perhaps any other country for which there are reasonable estimates, the United States has managed pretty steady long-term average growth rates over almost 140 years.  And yet, this was a country that experienced numerous financial crises in the first half the period.  Lists differ a little, but a reasonable list for the US would show crises in 1873, 1884, 1893, 1896, 1901, 1907, perhaps 1914, and 1929-33.  There were far more crises than any other advanced countries experienced.

And yet, there is no sign that they permanently impaired growth, or income.  One never knows the counterfactual, but right through this period the US kept on towards establishing the dominant position it was to hold in the decades after World War Two.  Even the Great Depression, awful as it was (costly as it was to many people) does not look to have had permanent adverse effects.  Another source I’ve bored people with over the past few years is Alexander Field’s excellent relatively recent book on US economic growth, productivity, and the Great Depression, A Great Leap Forward.  Field reports the best estimates for TFP growth in the US over the last century, and growth was faster from 1929 to 1941 than in any of the other periods he presents.  One might quibble about when to start and end these sub-periods, but 1929 was before the downturn became well-established, and 1941 is around when GDP per capita got back to pre-crisis trend (before temporarily going well above it in World War Two).

The United States in the Great Depression had almost all the factors that are often cited to explain why financial crises might have permanent or very long-lasting adverse effects:

• Lots of bank failures, and in a system without nationwide banks, disrupting the intermediation process.
• Lots of corporate failures
• Big changes in the price level (steep deflation)
• Huge regulatory uncertainty (including around the robustness of the judicial system –  see, eg Roosevelt’s attempt to stack the Supreme Court)
• Significant fiscal costs (in this case, not bank bailouts, but the defaults by other Western countries on the huge US World War One loans).

And yet the underlying rate of innovation is estimated to have gone on just as strongly as before.

This is not, remotely, to trivialise the Great Depression.  But it still looks a lot more like an event that became as severe as it was because of inadequate demand, and was resolved when sufficient strong aggregate demand returned (in the US case not until World War 2).  Output lost in the interim is a real and substantial cost to the people involved, but the numbers get really big if something changes the long-term future path of growth.  And there is no sign of that having happened in the many financial crises the US experienced from the 1870s to the 1930s.

A few years ago, Andy Haldane of the Bank of England got a lot coverage for a speech in which he presented this table, suggesting that the cost of the 2008 financial crisis could be huge –  100 per cent or more of annual GDP.

If so, it could be argued that everything should be done, all resources of the state thrown at, avoiding such events, which –  as Haldane put it –  our children and probably our grandchildren might be paying for.  But if there is little or no permanent reduction in the future path of per capita income, as a result of the financial crisis itself, the real economic costs of crises are much much smaller.  And the benefits of any regulatory measures to reduce the risk of crises are commensurately smaller –  all the more so when we allow for how little any of us know about the long-term costs and benefits of such regulatory restrictions.  Even recessions occurring at the time of a financial crisis can’t all (or perhaps even mostly) automatically be ascribed to the crisis itself.

I’ll have a few more posts on related issues in the next few days or weeks.  But recall where I started on this, and where I will loop back to. Per capita income GDP growth in New Zealand and the United States since 2008 has been very similar, even though New Zealand had only a minor domestic financial crisis, while the US was at the epicentre of a major global liquidity event,  and many significant US institutions failed or came close to failing, and US lenders experienced very large losses.   Sadly, as earlier posts have illustrated the relative productivity performance in New Zealand (relative to the US)  has been even weaker.

I’ve been working my way through a series of posts on the post-2007 economic experience of a large number of advanced countries, with a particular focus on trying to make some sense of New Zealand’s (no better than middling) experience.

Today I wanted to have a quick initial look at the incidence of financial crises, since the shorthand for what has gone in recent years has often been “since the financial crisis” or “since the [so-called] GFC”.  Many authors, including some pretty serious and respected ones, have ascribed much of the poor economic performance to the “financial crisis”, adducing the experience as evidence supporting a case for much tighter regulatory restrictions all round.  And there are plenty of theoretical discussions as to how financial crises might have detrimental economic effects.

But what do we mean by a “financial crisis”?  Many authors who try to classify events do so by looking at the gross or net fiscal costs of a crisis (eg the costs of bailouts, recapitalisations, guarantee schemes and so on).  I’ve long thought that was a flawed basis for classification, for a variety of reasons:

• A country that allowed its banks to fail, with no bail-outs, might have no direct fiscal costs at all, yet on any plain reading could have experienced a very substantial crisis.
• Most (though not all) of any fiscal support for troubled financial institutions tends to benefit citizens of the country concerned, and so redistributes wealth rather than destroying it.  There may be all sorts of adverse incentive effects, and deadweight losses from the taxes required to cover the fiscal costs, but the level of fiscal support is not a very meaningful indicator of the cost to society.  In rare cases (eg Ireland) the fiscal costs themselves can become quite directly problematic, in terms of on-going market access, but that is not the general experience.
• A country’s banking system might be very highly capitalised, such that no banks actually failed even under severe stress, and yet on most reckonings the country concerned would have experienced a financial crisis.
• In some cases, banks will have experienced the bulk of their losses on offshore operations, while the intermediation business in the home economy might have been fine.  The home government might choose to bailout the bank concerned, but there are few obvious reasons to think that those offshore losses (and the choice to bail) will have much effect on the home economy.  A good example, in the most episode, was the losses sustained by German banks.  On many classifications, Germany shows up as having had a financial crisis, and yet almost all of the increased loan losses resulted from offshore exposures (particularly in the United States housing finance market).  If we are trying to understand the economic implications, it probably makes more sense to think of those losses as a US event than a German event.

So instead, I proposed a classification based on non-performing loan data, which the World Bank collects and reports by country.   The proposition here is that, if there are sustained economic consequences from something we can label a “financial crisis”, they are likely to arise primarily from the initial misallocation of resources that led to the loan losses in the first place.  Gross over-investment in a particular sector (say, commercial property in New Zealand in the late 1980s, or in Ireland in the last decade) eventually leads to losses.  The projects don’t live up to expectations, and real resources devoted to those projects can’t easily or quickly be reoriented to other uses. Buildings lie empty, or even half-completed. If there are problems, they arise whether or not any bank ever fails, or is bailed-out by the state.  And banks have to reassess their entire models for generating income, and are likely to become more risk averse as a result of the losses their shareholders faced.  There might be additional costs if a large number of major financial institutions actually close their doors permanently after a crisis (that argument is part of the case for the OBR tool in New Zealand), but relatively few major institutions actually closed their doors in the period since 2007.

The table below classifies countries based on data on banks’ non-performing loans (NPL) from the World Bank’s World Development Indicators. The 20 countries that had a substantial increase in the stock of domestic NPLs after 2007 (the final two columns), are treated as having experienced a domestic financial crisis.

Non-performing loans since 2007

Source: World Bank.

Note:

Very low and stable: NPLs:                     less than 2.5 percent of loans throughout, remaining at a stable percent of loans

Moderate and stable: NPLs:                  2.5 to 5.2 percent of loans throughout, remaining at a stable percent of loans

Moderate and small increase:               : NPLs greater than 2.5 percent of loans throughout, increasing by less than 2.5 percentage points

Mostly, the classification looks intuitive, and much as expected. All advanced countries that had a support programme with the IMF during that period are in the financial crisis category. At the other end of the spectrum, the commodity-exporting advanced countries (including New Zealand) all avoided a domestic financial crisis on this, as well as other, measures.

Two points may be worth noting. First, although the extreme liquidity crisis in the United States in 2008/09 attracted headlines – and had global ramifications – loan losses in the United States have been overshadowed by those in many European economies.  One element of this is that more loans in the United States were not on the books of banks (and this is bank data). Second, although some banks in countries such as Germany and Switzerland got into material difficulties, in most cases those were the result of losses incurred in the United States.  As discussed earlier, these losses probably had different implications for the performance of the home (German or Swiss) economy than losses arising out of domestic lending business.

If this classification looks broadly sensible, it is somewhat ad hoc, and might not easily generalise to other crises and other times, even if broadly comparable data were available for earlier periods.

One can’t just jump from this classification to the chart of GDP per capita performance to see whether countries that had financial crises, on this measure, did worse than those that did not.  Apart from the many other influences on any country’s performance, it is also important to recognise that any causation can run in both directions.  The argument that a quite-unexpected period of very weak economic performance will have generated large loan losses (many projects will have been based on assumptions that, however apparently reasonable, did not play out as expected) is at least as strong (I’d argue stronger) than the proposition that “financial crises” themselves cause sustained economic underperformance.  Loan losses did not cause Greece’s problems, but were an integral of an overall process of economic mismanagement and misallocation of resources that led to Greece’s disastrous underperformance in recent years.

And what of New Zealand?  We have had pretty low banking system NPLs throughout.  Finance company losses mattered a lot to investors in those companies, but were still quite modest relative to the overall stock of loans in New Zealand.  There were fiscal costs, through the deposit guarantee scheme, but without the international rush to guarantee schemes there would have been much the same loan losses and probably no direct fiscal costs.

As no major institutions failed, and (economywide) there was little evidence of sustained over-investment in any particular class of asset, one would not generally think of New Zealand having experienced a domestic financial crisis.  Funding was disrupted for a time in late 2008/09, and that was among the factors inducing a greater degree of caution among lenders, but 6-7 years on it seems unlikely that any domestic financial stresses have materially affected New Zealand’s overall performance.  For some sub-sectors, the picture might have been a little different, as finance companies had been major financiers for property developments, but for the economy as a whole the effect seems likely to have been quite peripheral.  And yet, our economic performance has been similar to that of the United States, the epicentre of the initial crisis, and where the increase in actual loan losses was substantial.