Protectionist backlash or falling consumer demand: Is the world in danger of deglobalization?

What if unemployment in Ireland reaches 25% next year? What if GDP falls a quarter between 2007 and 2012? The spectre of the Great Depression looms over us large at the moment and there has been much commentary of late – see for example Robert Samuelson’s recent blog post – on whether and how our current global recession/depression compares with the last one of similar scale, that of the 1930s.

Is it pointless spooking of the public or is it a relevant comparison worth exploring further? Recently, Kevin O’Rourke and Barry Eichengreen make the case that the comparison is at least worthy of further investigation in an analysis of some key global indicators, including output, stock markets and interest rates, comparing ‘now’ and ‘then’. Their conclusion was that “world industrial production, trade, and stock markets are diving faster now than during 1929-30” – something that previous US-centric comparisons hadn’t concluded. One thing that worried me as a student of the history of globalization was the inclusion of trade in that set of statistics. Here is O’Rourke and Eichengreen’s Figure 3, Trade Then and Now, which worried me so much:

World trade then and now - Source O'Rourke/Eichengreen 2009

World trade then and now - Source: O'Rourke/Eichengreen 2009

Are things that bad? Is the world going to “deglobalize”? Is trade going to collapse and bring us – in a trade-dependent Ireland and a trade-dependent world – unemployment, poverty and misery along similar lines as the world saw in the 1930s?

The first thing to know is whether or not the world is more globalized in trade terms now than it was in the 1920s and 1930s. This may seem like a dumb question at first: just as the world is more urbanized and more industrialized on a totally different scale now compared to a century ago, surely it’s more globalized too, right? But those who research globalization have shown that it’s rowed back and forth over the decades and centuries, whether one looks at trade, migration or capital markets.

For example, as Kevin O’Rourke writes elsewhere with two co-authors, the globalization of international investment was greater in 1914 than it was at any point later until the early 1970s. “Deglobalization” of capital markets meant that while foreign assets accounted for nearly 20% of world GDP during 1900-14, the 1930-1960 figure was just 5-8%, similar to levels in 1870.

A table in the same paper outlines trade and the integration of goods markets in the pre-1914 period. The best estimate for Europe is that the trade-output ratio – how much of what was produced was traded – increased from 30% in 1870 to 37% in 1914. What happened next? In particular, what happened in the Great Depression and how does that compare with now? The graph below shows two lines, the orange line being the world trade-output ratio from 1991 to 2013, as estimated by the IMF’s World Economic Outlook. The blue line is my own estimate, based on Mitchell’s Historical Statistics and research I did while in Trinity, of the global trade-GDP ratio in the 1920s and 1930s, using 25 prominent economies (not dissimilar to a proto-OECD).* The shaded part shows the future, for the orange line – i.e. 2009 on.

Global trade-output ratios across the two Depressions

Global trade-output ratios across the two Depressions

Three things strike me:

  • The first thing to notice is that in 1991, one third of what was produced globally was traded. The world was about as globalized in 1991 as the OECD was in the mid-1920s and as Europe was in 1900. So we certainly not talking exponentially different levels of trade intensity now compared to a century ago – probably just greater geographical spread.
  • Protectionism, deglobalization and the destruction of trade kicked in in the early 1930s. The change was a steady four-year shift to a new lower level of trade intensity. For all intents and purposes, the Great Depression led to a halving of how integrated global trade markets were.
  • The world’s global trade intensity since 1991 has been marked – it has essentially doubled, meaning that almost two thirds of what is produced is now traded. Not only that, unlike in the 1930s when trade intensity almost halved, global integration of trade markets is likely to increase over the coming years of global recession and recovery, if the IMF’s latest statistics are to be believed.

Does this make any sense? How can trade in 2009 be falling faster than it did in 1929 – at the start of a period of dangerous protectionsim – and yet the world is still globalizing? Mathematically, the answer has to be that trade is contracting, but slower than output. Economically, the answer – I think – is that trade is much more integrated into daily life now than then. Or put another way, trade in the 1920s and 1930s was more easily substitutable than now. Globally integrated supply chains and consumer networks mean that when output falls now, trade falls – and vice-versa, as countries are trade-dependent. Just look at Japan’s exports – that to me tells a story of global consumers cutting back on buying new cars, not British or German consumers deciding to buy local rather than buy Japanese cars.

So, having looked at the stats, I’m a little less worried than before. Firstly, politicians seem much more acutely aware of the dangers of protectionism now (… although perhaps a historian can correct me on the political economy of the early 1930s). Secondly, while 1920 and 1990 were not dissimilar starting points, in terms of the level of trade intensity, we have entered our recession at a different level of trade intensity than our forebears 80 years ago. While the 1920s were a stuttering decade for global trade, the nineties and naughties have seen solid expansion of trade networks. The 20-year build-up before recession set in, coupled with the technologically-enabled disaggregation of value chains, has created global trade networks of a much more integrated nature than those of the 1930s. It would be much harder now – even if we all wanted to – to destroy our trading networks, as we’d be trimming our own consumption possibilities far more than consumers had to back in the 1930s. Hopefully, my optimism is not misplaced.

* For those interested in the details, the 25 counties were weighted by their non-agricultural labour force, to strike a balance between GDP and population weightings.

(PS. I still think a comparison of the real economy effects of the financial crisis of the early 1870s is worth a go… Here’s hoping I’ll find the time!)

Intergenerational outsourcing and the consequences of building 10% too much: A look at Ireland’s property market in 2013

With Davy Stockbrokers predicting a 70% fall in Irish construction activity from its peak over the coming ‘medium term’ (2009-2011 or so), I though it might be timely to review some headline statistics for Ireland’s property overhang.

Recently, I’ve been peddling the idea that between 2004 and 2007, we were building twice as many homes as we needed and building twice as many for 3/4 years implies building half as many as you need for 6/8 years to return to equilibrium. Does that stack up? Or, put another way, if we start in 2002 with Census statistics on the stock of housing, use Dept of Environment statistics for the period 2002-2008 and turn Davy’s figures into ballpark estimates for 2009-2013, how bleak will things look in five years time?

The answer, much to the chagrin of those who loathe two-armed economists, seems to be that it depends – in this instance on what part of the country you’re talking about, but also about what you think is the appropriate long-term need for new houses in this country. If we take 2001 figures (technically March 2002 figures) as our ‘departure from normality’ point, how far off course are we? Between 2002 and 2008, we churned out over half a million properties, off an existing base of just 1.3 million households. Back-of-the-envelope estimates, based on an overview of economists’ figures on this topic, suggests that we should have been building perhaps 300,000 households in that same period. (That’s using an equilibrium figure of 40,000 properties a year, rising temporarily after the accession of new EU member states.) So, enough with all the stats, what’s all this for, you wonder. Well, I was hoping to use all this to answer two key questions:

  • Where suffered worst from Ireland’s properties building bonanza? Where is housing inventory lying around most?
  • How long will we have to sit around building hardly anything until we’re back to some semblance of normality in the property market?

Where did we build our extra properties? By the end of 2008, we were about 5 years ahead of schedule – i.e. we’d built 12 years supply in just 7 years. To give a regional flavour, based on insights gleaned from the property overhang per county figures I calculated in December, I split Ireland into three regions – Dublin, Connacht/Ulster and the rest of the country. (The data allow for a full county-by-county analysis, however time constraints and poor formatting in the various external sources has prevented me from threatening another heatmap!) Over the period in question (2002-2008), more houses were built in Connacht/Ulster than there were in Dublin, which has almost twice the population! As a result, in terms of years of “pre-production”, if you will, while Dublin had under 2 years excess supply by end-2008, Connacht/Ulster had almost 8 years. Once more emphasis: builders managed to produce 15 years output in Connacht/Ulster in just 7 years.

How long will we have to sit around building nothing? It’s all very well for someone to come along after the fact and say “You shouldn’t have done that”. What’s more interesting is to shed some light on where the adjustment will come first and where it will be hardest. One option would be just to close up our construction sector for a few years until inventory shifts sufficiently and prices start to rise. Practically, of course output doesn’t and shouldn’t collapse to zero and, as per Davy’s figures, will be in the range of 10,000 to 25,000 over the coming 5 years.

Therefore, I’ve assumed output of 20k in 2009 (still slowing down), 10k in 2010 (bottom of the market) and then a simplistic 5k increase in output every year after that, rising to 25k in 2013. Let’s call this the ‘post-Section 23′ scenario. This is contrasted with a ’20:20 foresight’ scenario where steady-state output in construction remains 40,000, apart from a minor blip of 35,000 in 2009 due to global economic circumstances. In both scenarios, new houses are allocated according to a region based on its Census weight – crucially, and we can relax this later, even in our post-Section 23 world, output resumes in Connacht/Ulster, not at the distorted rates we saw but in proportion to its size. The result of all this is the chart below. The figures show the excess of properties as a percentage of the total property stock in each of the three regions.

Ireland's excess properties, % of total properties, by region, 2003-2013f

Ireland's excess properties, % of total properties, by region, 2003-2013f

The results are pretty clear:

  • Even with some major internal restructuring of the construction industry (i.e. rebalancing output of houses according to a region’s weight in the economy), Connacht and Ulster will still have a significant property overhang, more than 10% by 2013 – and that itself based on a drastic 70% contraction in building activity from peak levels.
  • For most of the country – and indeed the country on average – the overhang will have halved by 2013 but will still be in the region of 5/6%.
  • In Dublin, shortages in housing may emerge as quickly as 2012.

Objections to the above might include one along the following lines: construction will not only contract 70% but also no-one will be building in Connacht/Ulster for years to come so even the rebalancing of output described above is not an accurate forecast. In that case, the overhang will just take the full 8 years from 2008. Section 23 and the property boom will have taken construction jobs from 2009-2015 and left them in 2002-2008 – a sort of integenerational outsourcing.

Another objection is that the optimistic (if 2012 is optimistic) scenario painted for Dublin hinges on that long-term need of 40,000 units a year (which translates into about 12,000 new units in Dublin annually, based on its Census weight). Significant and persistent net outward migration from Dublin from 2009 on – which incidentally is why I believe that Dublin Bus, so clearly an ‘inferior good’ in the economist’s sense of the word, is losing money when incomes fall – might mean that the demand for housing in the period 2009-2013 may fall to 20,000. Replacing 40,000 with 20,000, from 2009 on suggests that the average percentage overhang for the country stays stuck at 10% and Dublin – while still much lower – remains stuck at 3-4%.

In sum, we are where we are. We’ve more than enough houses everywhere in the country and plenty of houses in places where we won’t need them for another 10 years or so. Therefore, it would be wise for the Government to take this crisi-tunity, as Homer Simpson would say, to harness both supply and demand sides of the market.

  • On supply, it should focus the efforts of the much-trimmed residential construction industry, when that sector starts to medium-term plan in 2010/2011, on Dublin and other areas around the country most likely to show a shortage of property this side of 2015.
  • On demand, the Government should attempt to deliver balanced regional development, taking property overhang as an opportunity for affordable housing to create new centres of employment. Taking this to its most logical conclusion, firms outsource because they want to free up resources to specialize on what they’re good at. Therefore, we must adopt a mentality along the following lines: “Let’s take this opportunity to treat our property boom as intergenerational outsourcing, which has freed us up to focus on what we’re good at.” (Just don’t say all we’re good at is construction!)

The Pelosi scare-graph revisited: Private sector job losses and camel recessions

As some of you may know, Nancy Pelosi has been scaring people with graphs, recently. By her metric, namely the absolute numbers of jobs lost, the current recession is more severe – and faster – than the last couple of regressions.

Naturally, something that high profile gained a lot of attention and ultimately modification. The one people seemed to settle on was one by William Polley, who made some slight modifications and improvements. In particular, he made a greater number of comparisons all the way back to World War II, and also changing it from absolute number of jobs lost to percentage jobs lost, to give some idea of the scale of the recession.

Of course, I could hardly let a good graph pass me by. Below is a slightly different version, with three more tweaks, to tackle some remaining issues:

  • Firstly, I’ve averaged the four recessions between 1948 and 1961, calling them the “1950s recessions” in the absence of a more appropriate nickname. The reason for this is that the four recessions in that period were all remarkable similar. In particular, they all started in Autumn and ended almost two years later in Summer. They had similar length and similar severity, and there was a similar government response each time – increase the number of jobs in the public sector by about 6%.
  • Secondly, I’ve tried to further reduce the ‘clutter effect’ – or at least make the graph more intuitive – by using the colour scheme to indicate passage of time. The darker the colour, the more recent the recession.
  • Lastly, and most substantively, I’ve focussed just on private sector employment. Government employment tends to be acyclical and rising steadily (apart from the early 1980s – perhaps Reagan swinging the axe?), so the focus should be on total private sector (non-farm) employment.
% fall in US private sector employment during recessions, 1945-2009

% fall in US private sector employment during recessions, 1945-2009

It’s hopefully pretty self-explanatory – our recession looks like it’s going to be dead hard! Might as well get the discussion started with a few initial observations on the graph:

  • By this metric – arguably more down-to-earth than fuzzy GDP metrics – recessions have been getting longer, not shorter, since the war. 1950s recessions lasted two years, the last one (2000-2003) stretched to more than four years!
  • The worst point of the recession has been been getting milder and milder – at least until this recession. A good old-fashioned 1950s recession would wipe 6% or more off private sector jobs. The early 2000s recession never even reached 5%. Although it did come close twice, speaking of which…
  • Recessions are like camels. Just as camels can have one hump or more than one, it seems recessions can be Dromedary or Bactrian also and in fact are drifting towards the multi-humped latter species.

Unfortunately, we don’t have the monthly data to do the same for the 1930s recession – or indeed to collate data across a wide spectrum of countries. But, if the USA is a good bellwether for the downturn, whatever about it perhaps being more flexible in the upswing, then it seems that modern (i.e. post-1980) recessions hit a smaller number of workers for longer than their pre-1980 counterparts.

It looks like our current recession is going to mix the worst of old-time recessions, i.e. the 6%+ fall in private sector employment, with the worst of modern recessions, i.e. it’s going to last three years or more and probably come back with a vengeance a couple of times just when we think it’s getting better.