Sunday, October 06, 2013

As Good As It Gets In Latvia?

For Maurice Pialat, champion of the marginal centre.
"This raises a final question, which, while not central to the issues of this paper, is nevertheless intriguing: How can a country with a low minimum wage, weak unions, limited unemployment insurance and employment protection, have such a high natural rate [of unemployment]?"

"To summarize, the actual unemployment rate is still probably higher than, but close to the natural rate of unemployment. Latvia may well want to take measures to reduce its natural rate, but the recovery from the slump is largely complete."
Boom, Bust, Recovery Forensics of the Latvia Crisis, Olivier Blanchard, Mark Griffiths and Bertrand Gruss

With these words three IMF economists (hereafter BGG) effectively signed off on their study of "what just happened on Latvia" and, they hoped, drew to a close a debate which has been going on now for some 6 years. In fact, far from closing the debate, what they may have done is effectively extend it into new terrain, since these apparently harmlesss words - "the recovery from the slump is largely complete" - have far reaching implications, as does the methodology they use for reaching it. These implications reach well beyond Latvia, and even far beyond the Baltics and the CEE in general, despite the conclusion that everyone seems to be reaching that Latvia was just a "one off". Possibly without intending to do so, they have drawn onto the clinical investigation table issues which have been mounting  up in the theoretical lumber rooms of neoclassical growth theory for some time now, issues which begin to assume a paramount practical importance in the context of our rapidly ageing societies. What, for example, do we understand by the term "convergence" these days? And if "steady state" growth can no longer be understood as implying a constant growth rate (trend growth in developed economies is now systematically falling) should we be considering the possibility that headline GDP growth will at some point turn negative, even if GDP per capita may continue to rise, due to the fact that populations are steadily starting to shrink. And if the answer to the former question is "yes", then what are the implications of this for the financial system, for the system of saving and borrowing, and for the sustainability of legacy debt? Not little questions these, but ones which will need to find answers and responses in countries like Latvia over the next couple of decades.

And again, returning to a question I raise about Ukraine (here), while Latvia's recovery may be complete and thoroughgoing, what satisfaction can we really take  from our knowledge of this when - according to the country's President Andris Berzins - the end state leaves the very survival of the country as an independent entity ten years from now as an open question? The problem - the country's population is falling, along with its workforce, and young educated Latvian's continue to leave looking for a brighter future elsewhere, even if they now do so at a slower rate than they did during the height of the crisis.

This is the first time I have written anything on Latvia in some time. In 2007 and 2008 I argued for Latvian devaluation, but refrained from continuing to do so in 2009 since the will of the Latvian people was so obviously against taking this path. I think policy has to work in the real world and not in the one we - like visitors to Andrei Tarkovsky's "room" - might wish we were in. But more than the going back over the debate  about whether or not it would have been better to devalue - we will never know the answer to this one, although although the viewpoint still seems to me a more defensible view than many imagine - what I would like to stress here are the reasons which lead me to arrive at the conclusion is was a good option, along with the factors which influenced me in getting there. These are set out in my June 2007 monster post: Is The Latvian Economy Running Out Of People?. The post is a long one, extraordinarily so as I say there, even by my standards. But going back over it, and with more than six years of hindsight to benefit from, I can't help feeling there is not a great deal I would change or even add. As I say in the introduction to that post:
"It is generally recognised by most external observers that this malaise has its origins in structural problems in the Latvian labour market, and it will be argued here that these structural problems have their roots in recent characteristics of Latvian demography (namely high out-migration and a sustained low birth rate). As such there is no easy solution. Even in the longer run the position will inevitably be difficult, since demography almost inevitably casts a long shadow. This does not mean, however, that we should be complacent. There are steps which can be taken to address the issues which Latvia faces in the short term, and it is important that such appropriate measures are enacted. These measures clearly include policies to reduce the dramatic overheating which is taking place, but they also should include policies to loosen the labour supply, not only by encouraging increased labour market participation and mobility, but also by actively encourage inward migration. Such policies may be seen as short term measures which are vital to move Latvia away from an unsustainable and towards a sustainable economic path."

Measuring Trend Growth

The facts of the crisis in Latvia are by now more or less well know. As BGG outline it the story runs as follows:
"The basic and striking facts to be explained are given in Figure 1 (see chart reproduced above - EH): An increase in GDP of almost 90 percent from 2000:1 to 2007:4, followed by a decrease of 25% from 2007:4 to 2009:3, and a recovery, as of 2013:1, of 18 percent. A mirror image in terms of unemployment, with a decrease in the unemployment rate from 14% in 2000:1 to 6% in 2007:4, followed by an increase to more than 21% in 2010:1, and a decrease since then, down to 11.4% in 2013:2."
For anyone seeking more background BGG gives an excellent and informative summary. What went on in Latvia was not a fiscal overspending issue (which is not to say the administration should not have been running a higher surplus during the latter part of the boom), but an accelerated credit-driven consumer demand and (housing) investment boom financed by external borrowing. This boom massively structurally distorted the economy, in the process taking output to levels well above those which were sustainable in the longer run. As BGG point out, "the ratio of private consumption to GDP (in constant prices) increased from 62% to 72%, [and] the ratio of investment to GDP (also in constant prices) from 22% to 36%." Now you don't have to be a mathematical genius to spot that 72 and 36 add up to 108, ie consumption and investment total more than 100% of GDP. How can that be, you may ask. The answer to the apparent inconsistency is that the difference is made up by imports (or the trade deficit), ie the Latvians were consuming all their own GDP and part of someone else's, with the difference being made up by external borrowing. BGG put it more elegantly:
"As a matter of arithmetic, the result of increasing consumption and investment ratios was a steady deterioration of the current account balance, with the ratio of the current account deficit to GDP increasing from 5% of GDP in 2000 to peak at a very large 25% in mid-2007."

So it is clear the Latvian economy was running above capacity, but how much above capacity? This is really what the present debate is about, since depending on the answer you give to that question the estimated current trend growth level of the country will be either higher or lower, as will the non-inflationary unemployment rate. Using various vintages of output gap estimates taken from real time EU Commission economic forecasts (12% positive  in 2007 as estimated in  2013) the authors derive a series of cyclically adjusted fiscal balances which show how, at least from the current vantage point, the size of the output gap, and hence the degree of laxity in the fiscal stance, was systematically underestimated. In 2007, for example, the EU Commission only thought the positive  gap (ie degree of overheating) was some 3%. Well its always easier to see things more clearly with hindsight might be the common sense response. Would that things were so simple!

An Interlude Concerning Production Function Metaphysics

What is involved here is a really important and hard to resolve methodological (and even, god help us, epistemological) issue (especially in countries which pass thorough a deep and protracted economic slump) - what is the special privilege of the present as a valid vantage point, when compared with the virtual infinity which time will eventually offer us?

After all, in the "present" which was 2007 things did look very, very different. Perhaps our current evaluation of our own "present" is just as conditioned as earlier perceptions of earlier "presents" were. The problem is we are using our present appreciation of the way things are to reach conclusions about the past which may look very different in some other, future, present. Yes, you're right, there is an element of circularity in the kind of argument that is used by BGG. As the people in the trade put it, potential output is an unobservable latent variable, you know, a bit like the Higgs particle, something you can't see or measure, but which you have to assume to exist for everything else in your theory to make sense.

As one of the IMF authors, Bertrand Gruss, puts it in his paper on the topic, there are "many different methodologies" which can be used "each of them encompassing a different precise definition of potential output and entailing advantages and disadvantages". All of them have, however, one thing in common:  "potential output estimates are subject to substantial uncertainty." As he also notes, in the case of a country like Latvia, emerging from a substantial slump, the degree of uncertainty is especially large. So those who would use the arguments in BGG to argue something simplistic, be chastened, the room for error is large. But then "substantial uncertainty exists over the past and future" doesn't make for good headlines, and, perhaps more importantly, doesn't inspire confidence in the policymakers who admit this.

So does each historical moment have its own special "truth" as far as potential output goes? This point - present moment bias - is described by Paul Krugman like this: "These methods automatically interpret any sustained decline in actual output as a decline in potential, and they cause that re-estimate to propagate backward through time." This approach could be described as "present moment reductionism" in the sense that events in the past are viewed and evaluated from the standpoint of the present, in a way which makes them explicable and comprehensible only in terms of the present they give rise to. The German philosopher Liebniz once put it this way,  we live in "the best of all possible worlds", if not in the best of all imaginable ones (back to Tarkovsky's room).

Basically, it is difficult to avoid the bad performance generated during the slump  "contaminating" the data. What we really need is information on Latvia's future performance, then we could situate the present. We need a time series from the future, then we could see much more clearly what is happening now. Unfortunately for us we can't have access to one. The "set up" (or world) we live in has this characteristic.On some views this is precisely what makes it interesting.

At the same time recognising this reality doesn't make the problem simply go away. As macroeconomists we are constantly forced to make what come near to being ad hoc judgements, and we need to do so time and time again, as we go forward and on the fly. As the Spanish poet Antonio Machado put it, "el camino se hace andando" - we make the path we walk along as we walk. The difficulty is that we are in a bit of a "garden of forking paths" here, since the decisions taken in 2008 and 2009 are the reason we have reached reach the endpoint we are at now, and it is this (momentary) endpoint which conditions our judgement about the initial conditions we set out from. And this is the case even though, had we taken another path  at the outset we would surely have arrived at another "now" from whence the starting point would have been seen differently.

That master of neo-classical growth theory Robert Solow put it thus in his Nobel acceptance speech:  
Growth theory was invented to provide a systematic way to talk about and to compare equilibrium paths for the economy. In that task it succeeded reasonably well. In doing so, however, it failed to come to grips adequately with an equally important and interesting problem: the right way to deal with deviations from equilibrium growth........if one looks at substantial more-than-quarterly departures from equilibrium growth........... it is impossible to believe that the equilibrium growth path itself is unaffected by the short- to medium-run experience.......So a simultaneous analysis of trend and fluctuations really does involve an integration of long-run and short-run, or equilibrium and disequilibrium. 
As he says, it is impossible to believe that the longer term path of the economy is unaffected by the trajectory taken during the deviations from trend - whether upwards or downwards.

(Incidentally, I used the comparison with Liebniz above because it seemed appropriate, because it seemed to me that Liebniz's "rationalisation of the real" was exactly what is going on here. This attitude was famously satirised by Voltaire in his Candide. Curiously when I went back to the Solow speech to dig the above extract out what else did I find - a reference to Candide. Happy to be in good company).

Now in fairness our IMF authors are well aware of this issue, although I'm sure they'd like to put it all very differently. Indeed, while they cite the EU Commission output gap estimates, they also carry out their own calculations (at least one of them Bertrand Gruss (as mentioned above) does, with the results being published in the 2013 edition of IMF Latvia selected issues). As his says in his commentary on the study findings:
"Many different methodologies have been used to estimate potential output, each of them encompassing a different precise definition of potential output and entailing advantages and disadvantages. No specific approach can be taken to be “the” correct one and potential output estimates are subject to substantial uncertainty. This uncertainty is probably even larger for countries like Latvia, a transition economy still going through substantial structural changes and coming out of a severe crisis that has likely rendered obsolete a significant part of the economy’s productive capacity."

For technical reasons which we don't need to go into here, BGG decide to use a production function methodology broadly similar to the one in the diagram above (click on image for better viewing), which is in fact the one they use over at the European Commission (Roeger, 2006) where they got the 12% 2007 output gap result.  In fact the IMF variant isn't identical. Their result (at least as of last January when the study was reported):  "Output was probably about 5–10 percent above potential before the crisis, although the extent of overheating at the pre-crisis boom is particularly uncertain."

[For the wonks, the benchmark PF model they used suggested the output gap peaked at around 5 percent of potential output before the crisis - well below the 12% level suggested by the EU Commission. Then, since they were worried about possible cyclical contamination of the TFP input, they used an alternative potential TFP series (cleaned up by applying an HP filter) and this gave them a gap estimate of about 9.5% much nearer to the EU Commission figure, which ain't that surprising since it is Roeger's preferred technique (see right hand path in diagram).]

Just to give us a feel for the kind of range of certainty involved here, Bertrand Gruss concludes his results by stating the following, "While acknowledging the uncertainty of estimates, staff believes output was significantly above potential before the crisis, but probably in the 5–10 percent range rather than in the 15–20 percent range".

More important than the actual result in my opinion is how they achieved it. A quick inspection of the left hand path in the diagram will reveal that a very significant part of the calculation revolves around labour inputs which ultimately depend on demographic dynamics. Indeed Gruss justified his preference for the production function approach precisely for this reason: "The emphasis on a production function approach reflects both staff view that it represents an adequate framework for Latvia (where, for instance, population dynamics and structural unemployment play an important role in potential labor and potential output estimates)....".

Put simply the only real positive impetus to trend growth we can expect in the future from Latvia will be on the TFP side, since the labour input component will turn negative at some point (if it hasn't already done so). Bertrand Gruss in fact puts it quite bluntly: "Labor is not expected to contribute to potential growth in the coming years."

Demographic Destiny?

Now, quite coincidentally, the IMF is finally getting round to thinking about the demographic side of the European periphery problem (not sure why it took them so long since they've been using the kind of production function methodology described above  for years). Well, at least in the Latvian context it is. I say "finally" because for whatever reason there seems to be some sort of resistance among fund economists to thinking about demographic issues (including migration flows) as part of the core macro picture, yet as can easily be seen above it really is, and Robert Solow wouldn't doubt it for a moment.

Anyway, their current thoughts on the Latvian demographic outlook can be found in the form of an appendix to their 2012 Latvia Article IV consultation report. Coincidentally this report was published at the same time as the second part of their program monitoring reflections, ie the signal being given would seen to be that while demography is important, it is an "issue pending" which can be safely passed over to the post program environment. This is in complete contrast with the methodology being advocated here which is that the program should in part have been designed with this central issue in mind. I have been advocating this since 2007 and I will continue to do so.

Be that as it may, as they inform us in their appendix, Latvia’s population is shrinking rapidly.  

During 2000–11, the population declined by about 14 percent (340 thousand people). Emigration was responsible for about two thirds of this decline while natural change due to low fertility accounted for the remainder: 

Emigration: an estimated 200–215 thousand people, mainly young people—roughly 9 percent of the population—have left Latvia during 2000-11 (Hazans, 20111; and Central Statistics Bureau); and 

Low fertility: the decline of the population for natural reasons was about 125–140 thousand people (5 percent of the population). The number of births has halved since the early 1990s—from around 40,000 annual births to around 20,000—falling below replacement levels.

In fact saying that fertility has fallen below replacement levels is putting it mildly, since the Latvian fertility rate is currently around 1.3 (one of the lowest in the EU) and has been effectively below replacement since the country came into existence. The number of births has been falling more rapidly since the onset of the crisis due in part to the harsh economic conditions but also aided and abetted by the fact that the majority of the women emigrating are of childbearing age.

So Latvia is facing a massive challenge. A combination of low fertility and emigration mean that the population is shrinking rapidly and at the same time ageing. The proportion of over 65s is set to surge between now and 2030 as it is all over Europe. Naturally with the hole in the pyramid left by the "missing births" and the working-age-population migration-loss the country is bound to be an example of one of the worst case scenarios, far worse than Japan, since Japan has only been resisting immigration, it has not lost population through emigration. Fortunately, the country has a possible solution - it belongs to the EU, is about to join the Euro, and the possibility exists that the Euro Area will become a transfer union over the next decade. At least that's the theory, I don't doubt the reality could well be different. But really the creation of this transfer union is Latvia's only hope now, and obviously it would be a substantial net beneficiary, since otherwise it is hard to see how the country will be able to offer its elderly population modern minimum standard welfare services like health and non-poverty-inducing pensions.  

Emigration and the IMF Program

Actually BGG do try and address some of these issues. They do so since, as they say, "an important part of the adjustment has taken the form of emigration". As they also point out Latvian emigration long predates the crisis. The average net emigration rate was 0.5% from 2000-2007. It increased to an average 1.3% from 2008 to 2011, but by 2012, was roughly back to its pre-crisis average. So emigration isn't a product of the crisis, it was simply made worse by it, but still, and going back to Solow  ( it is impossible to believe that the equilibrium growth path itself is unaffected by the short- to medium-run experience.) how far was Latvia's longer term future being put at risk by the form in which the adjustment occurred.

[Just as a side issue it is worth noting that exactly the same question arises in the context of the Greek adjustment. Had the IMF forced the EU to accept debt restructuring and an EFF rather than the initial SBA, the pace of the fiscal adjustment could have been slower, and the loss in output lower. Mein Gott, we might not now be talking about a current estimate of a Greek output gap of plus 10% in 2007 (if you follow the logic of the argument advanced earlier). See my "Second Battle of Thermopylae" post].

In fact BGG do attempt to address this issue:
"The question however is whether this emigration is, in some sense, a failure of the adjustment program. In the United States, migration rather than unemployment is the major margin of adjustment to state specific shocks ..... These adjustments are typically seen as good, indeed as the main reason why the United States functions well as a common currency area: If there are jobs in other states, and if moving costs are low, it is better for workers to move to those jobs than to remain unemployed."
This is an argument that it commonly advanced in the context of Euro Area issues (let's leave aside for the moment the fact that Latvia wasn't in the Euro) - in an optimal common currency area this sort of labour mobility is a good thing. In addition let's leave aside the question that Europe isn't the United States, that it is a continent made up of nations, and that these nations form part of our identity as Europeans in a way which is hard to quantify economically and in a way which can't simply be wished away by waving a magic wand (or paying another visit to Tarkovsky's room), the fact of the matter is that the Euro Area isn't an optimal common currency one. At least institutionally it isn't. To become one of those it would need to have a common treasury and a common unemployment benefit and pension system, etc.

Unfortunately, this is an issue which BGG, like so many before them, simply slide silently past - "the largely permanent departure of the younger and more educated workers may indeed be costly for those who stay" -  like a ship in the night looking for open water while at the same time carefully evading the enemy  minefield.
"Is the answer [to the above question:EH] different for a small country than for a US state? Some economic aspects are different: Some of the costs of running a country are fixed costs, and thus may not be easy to support with a smaller population. In the United States, many of those costs are picked up by the Federal government (although, as we have seen for Detroit, the remaining fixed costs per capita may become too large for a state or a city to function). This is not the case for a country, which must for example finance its defense budget alone."
The reference to Detroit is of course salutory (this is exactly the problem), although it is curious that the example they take for the fixed costs of having a separate state is defence, an area where Latvia obviously benefits from the existence of external institutions like NATO and the EU. Again, the extent would be hard to calculate, but one of the factors which must have influenced Latvian's in their decision not to offend their EU partners by devaluing the Lat must have been a consideration of just this issue.

Still, the question remains, from a demographic point of view could things have been done differently? It's very hard to give a conclusive answer. My argument in favor of devaluation was always based on the potential demographic dynamics  which it might induce. Obviously there would have been a large drop in output, but Latvia had one of those in any event. Would less people have migrated out? That is very doubtful, and indeed, as BGG point out, people were emigrating even at the height of the boom. But then again, would the post crisis potential growth rate have been higher? Would the country still have had to face a non inflationary unemployment rate of 10%, or would the additional international competitiveness achieved have meant it was much lower? Would immigrants be arriving to do some of the lower skilled work?

The thing about this last point is, more than just ending the emigration what Latvia really needs (like Japan, like South Korea) is immigration to shore up the population pyramid, to make the welfare system sustainable in the longer run, especially since although the country's future currently depends on the creation of an EU transfer union there is no guarantee there is actually going to be one.

It is unlikely that the emigration hemorrhage would have been avoided even with devaluation - large numbers of Argentinians, for example, arrived irregularly in Spain in 2002 and 2003 and the two countries weren't in any kind of bilateral Schengen arrangement. But would the natural rate of unemployment have been different following the adjustment? We will now never know.

However,  an argument from two of my Economonitor colleagues - Andris Strazds and Thomas Grennes - should give us some food for thought. According to these authors, when it comes to emigration dynamics "Unemployment Matters Much Less Than Relative Income Levels". Now despite the fact that one might have some reservations about the actual methodology they use (they seem, for example, to confound the migration component in population dynamics and the birthrate one where in fact these are quite distinct channels) they are certainly digging in the right area, as the following chart which comes from a pre crisis IMF report makes clear.

The problem, of course, isn't only relevant to Latvia. Despite the fact that Spain's unemployment rate is currently around 27% immigrants continue to arrive in the country (often risking their lives to do so), a fact which puzzled the Financial Times demography correspondent Norma Cohen when we spoke about this article. "Why on earth," she asked me "would people want to come to Spain with such a high rate of unemployment?" Because salaries are better than in their home countries would be the simple answer, and because they are willing to do work which many Spaniards are reluctant to do, at least at the salaries which are on offer. So economic migrants continue to arrive, an estimated 300,000 of them last year, even though the net migrant flow reversed since more left (both native Spaniards and immigrants) with Spain's population falling for the first time in modern history as a result.

The idea of "centre and periphery" seems like a useful analogy here, since more than simple emigration or immigration what we seem to have is a steady displacement of population with migrants of lower skill entering one side of a country while higher skilled natives exit across the other. In this sense one can truly speak about "population flows". Naturally the net human capital loss involved  is substantial. Italy had some three million immigrants during the first decade of this century, but the overall annual rate of growth was not much above zero.

Beyond implementing the maximalist programme of a completely federal Europe with population moving in one direction and transfers moving in the other  it is hard to see what the solution is here.


"Do these lessons extend beyond Latvia? The evidence from adjustment in Euro periphery countries suggests great caution." - BGG

"I’m not sure I believe this [BGG] story but if you do, what lessons does Latvia hold for other countries, and the euro in general? And the answer, in brief, is none. Latvia’s story as I’ve just told it looks nothing like anything we’ve seen in the past, and probably not like anything we’re likely to see in the future – including, by the way, Latvia’s future." Paul Krugman, Latvian Adventures

The general consensus seems to be that Latvia is an interesting case study, but one where the lessons learned have little application beyond the country's frontiers. I'm not sure I buy this. Let's start at the beginning.

We all know what happened in Latvia - the country's economy massively overheated - but are we so sure why it happened? The answer isn't as obvious as it seems. The quick synthesis explanation offered by BGG runs as follows:
In short, the anticipation of a large scope for catch up growth, together with cheap external financing, led to an initially healthy boom. As time passed, the boom turned unhealthy, with overheating leading to appreciation and large current account deficits, with lower credit quality, and with balance sheet risks associated with FX borrowing.
Yep, but why was there so much external financing available, and why did it continue even after it was obvious to all bar the Latvian government that the accumulating imbalances were putting the country at risk of disaster? Paul Krugman puts my question in a little more elegant fashion:
 First of all, on a conceptual level, how does an economy get to operate far above capacity? We understand operating below capacity: producers may fail to produce as much as they want to if there isn’t enough demand for their products. But how does excess demand induce producers to produce more than they want to?
I think part of the answer here is that we all generally thought that in an epoch of large scale globalisation with extensive migrant and fund flows "open" really did mean open, in the sense that to erect a well functioning economy all you needed was a large strip of land (of which Latvia has plenty), cheap tax rates and flexible labour laws, then the entrepreneurs, the capital and the labour would all flow in. The problem in Latvia's case was they didn't. The capital was there, so were the entrepreneurs, but one of the other factors was in short supply, and indeed instead of flowing in it was flowing out. Then bang!

That's over-simplifying a bit, but it is the bare bones of the situation, a situation which surely has lessons to be learnt for other CEE countries (or far flung places with similar underlying demographics like Vietnam). In particular the word "Ukraine" comes into my head.

But beyond this, why was all that capital flooding in to finance something which to the careful eye was evidently not working? My reply would be, and taking us back to the literature of the time, the operation of the Global Financial Accelerator, a term coined by the Danish economist Carsten Valgreen to describe what was happening in Ireland and Latvia before the crisis actually hit. Essentially, in an environment of ample global liquidity being generated by central banks in countries which don't have the capacity to absorb all the liquidity phenomena like Latvia and Iceland simply happen, as we have been seeing in recent months as the Fed tapering debate lead to a sudden stop in one Emerging Market after another. Fortunately on this occasion the liquidity was being withdrawn before the kind of massive imbalances we saw in both Latvia and Iceland had time to occur. I for one, at least, think it's worth considering what happened in Latvia, and what can be learned, in the context of the current EM debate.

Another issue worthy of note, as I say in the introduction to this post, concerns the issue of convergence. Historically it has been assumed that per capita incomes in countries forming part of the EU would tend to grow at faster rates than those in richer economies with the result that all member state economies should eventually converge to some common living standards band in terms of per capita income. This now seems unlikely to happen, especially given the demographic and growth outlook on the periphery, Latvia included. The economy is growing well right now, but as we can see it is labouring under severe structural problems (the unemployment rate) and the demographic outlook suggests that growth will now steadily weaken. What we have is as good as it gets.

Ironically GDP per capita has been performing well in relative terms since the bust, and in ways the textbooks never envisaged - through a drop in the population numbers. Despite the fact that output is still well below the pre crisis level, as BGG note, Eurostat estimates PPP GDP per capita to now be at 9% above its 2008 peak.

Finally there is the point about how the adjustment took place. As BGG explain, the majority of the internal devaluation took place not through wage and price reductions, but through productivity - the mysterious factor X. But is it that mysterious? What happened was that there was massive labour shedding, as unemployment shot up to 22%. Then, as growth resumed, employment didn't follow (mirroring a pattern which arguably we are seeing in a milder form elsewhere, in other countries which are recovering from sharp housing busts). So while output recovered employment didn't which simple arithmetic tells you results in a strong productivity boost. As BGG explain, there was a strong underlying improvement in TFP taking place due to the "catch up" effect, and this undoubtedly helped Latvia in ways we don't yet fully understand. More study would be useful, since again I do think there are things to be learnt.

As a last word I would say that if you are reading these lines you have probably struggled your way all through this inexcusable indulgence in  verbiage. In which case thank you. You may also have noticed I haven't referred to the issue of fiscal austerity once. Not even a teensy weensy bit. There is a simple explanation for this, the Latvia debate was all about whether or not to devalue, it never was a for or against austerity one. As Paul Krugman puts it: "if we were really looking at an economy with a double-digit inflationary output gap, even the most ultra-Keynesian Keynesian would call for fiscal austerity". For reasons I have outlined above, I don't fully grant the whole inflationary output gap estimate, but still I think the point holds, this was never about for or against fiscal austerity, since among other reasons it was never about public sector debt.


The paper published by Blanchard, Griffiths and Gruss relies heavily on the work of the Latvian demographer Mihail Hazans whose groundbreaking studies effectively forced the Latvian authorities to amend their population and migration estimates. I had the pleasure of meeting Mihail when I shared a platform with him in a colloquium organised in 2012 by the American Chamber of Commerce in Riga. The title of the gathering was, not surprisingly, Latvia's Demographic Future (you can find my presentation here).

Basically every country on the EU periphery needs its Mihail Hazans, since we have no accurate or systematic system for measuring these important migrant flows.

In response to what I perceive to be a major lack of knowledge and information I have established a dedicated Facebook page in a vain attempt to campaign for the EU to take the issue of  emigration from countries on Europe's periphery more seriously, in particular by trying to insist member states measure the problem more adequately and having Eurostat incorporate population migrations as an indicator in the Macroeconomic Imbalance Procedure Scoreboard in just the same way current account balances are.

If we don't have the necessary information then how can we hope to formulate the adequate policy responses. If you are willing to agree with me that this is a significant problem that needs to be given more importance then please take the time to click "like" on the page. I realize it is a tiny initiative in the face of what could become a huge problem, but sometimes great things from little seeds to grow.

Friday, August 16, 2013

Can demography explain Portugal's slump before the crash? Is the Eurozone suffering from a “Shortage of Japanese”?

In Portugal, unlike other countries now experiencing economic difficulties in the Eurozone - Spain, Greece and Ireland-, anemic economic growth was all too evident before the financial crisis of 2007-08.  Accession to the Euro club brought a significant reduction in interest rates in those countries that had maintained historically high interest rates, which triggered real estate bubbles - Ireland and Spain, and public overspending - Greece, allowing these countries to maintain rates of economic growth above the European average. In Portugal, who suffered not only a mild housing bubble but also from government overspending,  growth was much weaker than the group mentioned above and more in line with countries that  have had lacklustre performances in the European context: Germany; Italy. (See chart below)

What’s the reason for the low growth relatively to other countries that are now in the same situation? To put another way, what’s the cause that differentiated Portugal from the other countries before the crisis of 2007-08, and how it faded since? Some recent papers and articles attempt to explain the causes of this relative weak performance, but although some of the causes advanced can in part explain the weak growth in Portugal, they do not fully explain the low growth relatively to the group mentioned above because many of the issues were also present in the other countries before the crisis.

Long term economic growth can be split into labor force growth and productivity growth.  Invariably, the main cause cited for the Portuguese low growth has been productivity growth, or the lack of it, but although this certainly has been a problem in the last decade, not so much before, it hardly appears enough to explain the relative week performance. The differences in productivity gains don't seem big enough to justify the differences in GDP growth, even in the cases of Ireland and Greece, countries where productivity growth was higher than Portugal; but certainly this is true for Spain as its productivity growth was smaller than Portugal, as can be seen in the chart below. In the last decade, productivity growth in Portugal was threefold that of Spain but its growth was less than half.

As such, we can safely conclude, that productivity on its own cannot explain the differences in economic growth between Portugal and the other group of countries. On the other hand, the stagnation and decline of the working-age population can not only explain the weak economic growth that afflicted Portugal, but also the GDP growth differences between several countries in the Eurozone before the crises.

As explained in the previous post, between 2003 and 2008, working-age population growth in Portugal was negligible and as such the “workforce effect” - contribution of labor force growth to GDP growth - was non-existent, as can be seen in the chart above.  Starting in 2008, working-age population growth became negative and thus the “workforce effect” began to act as drag on the economy. To maintain a healthy economic growth Portugal had to gradually increase its productivity growth, and/or alternatively increase its labor participation rates - to compensate for the declining workforce -, but this is not easily attainable, as such, trend growth will surely steadily fall. Therefore, when the labor force starts to stagnate or decline, economic growth stalls. This explains the weak economic growth in several European countries in the last decade. In particular, it explains why Portugal and Spain had very different economic performances before the 2007-08 financial crises and how this began to converge after this. 

Population change is comprised of natural growth, the difference between births and deaths, and net migration, the difference between immigration and emigration. Natural growth, both for Portugal and Spain, had been barely edging positive in the turn of the century since in both countries the total fertility rate fell below replacement level in the early 80’s. In Portugal, natural growth turned negative in 2007, the year that for the first time there were more deaths than births. On the contrary, Spain experienced a slight recover of its natural growth in recent years, as a result of an increase in its total fertility rate, as can be seen in the chart below, although this is due almost exclusively to foreigners, who have a higher fertility rate than the native-born.

As a result, both in Portugal and Spain, population growth in the last decades depended almost exclusively on having a positive net migration, and this resulted in a large influx of immigrants. But while in Portugal this growth began to slow down since 2002, in Spain immigration exploded until 2007, as can be seen in the chart below. "No modern country on Earth experienced such a massive increase in its immigrant population as Spain. In 1990, one in 50 people in Spain was an immigrant. Today, it's one in seven."

Portugal not only received fewer emigrants from 2000 onwards, but also witnesses a massive exodus of its nationals.  As Edward Hugh pointed out, the entry in the European Union was accompanied by steady emigration flows, which clearly sets Portugal apart from the other countries, Spain in particular (see map on page 11), and resembles more the path that would be later trodden by Eastern Europeans countries when of their accession to the European Union.  Consequently, and according to the Instituto Nacional de Estatística (Statistics Portugal), during the inter-census period, the resident population of Portugal increased by only 1.9% while in Spain, the increase was 12.9%. (Chart below)

With the accession to the European Monetary System and later the Euro, interest rates declined significantly for both Portugal and Spain and as a result the two increased their debt levels. According to the McKinsey report, Debt and deleveraging (see page 14), in the second quarter of 2011 Portugal and Spain had total debt of 356 and 363 (as % of GDP), respectively. The consequence of cheap and easy credit was to create a housing bubble, both in Portugal and in Spain, but while the Portuguese began to deflate in 2002, the Spanish continued to inflate until 2008. This outcome was the result of the substantial increase in Spain’s population as a result of immigration, many of them Portuguese, while the increase in immigration in Portugal was just enough to replace the ones who were leaving. This population growth allowed the housing bubble to continue for much longer in Spain, while in Portugal there were no longer enough people to buy the excess homes being built, and so prices didn't skyrocket; but the housing units were built regardless. As such, rather than a classical bubble with inflated house prices, in Portugal, it was more a case of oversupply, given that 800,000 homes were built in the last decade while the population only grew by 200,000. On the contrary, in Spain, in addition to the excess construction, prices went through the roof, with emigration given a great contribution to both. It is estimated that the immigration inflow increased house prices by about 52% and was responsible for 37% of the total construction of new housing units between 1998 and 2008From 2002 until the 2007-08 financial crises, the growth of the Portuguese economy began to be more in line with the growth of economies where the labor force was stagnant or declining, namely Italy and Germany, as can be seen in the chart below.

The chart above is easier to understand if we group countries into two groups: countries with weak economic growth – Portugal, Germany and Italy-, and countries with more healthy growth - Spain, Ireland and Greece.  With the exception of Greece, countries with sound economic growth before the recession were also the countries with higher labor force growth in the same period, as shown in the graph below.  By contrast, in countries where economic growth was weaker, the labor force growth was also more moderate or even negative, as in Germany. (Chart below)

However, the dynamics changed completely with the onset of the 2007-08 recession. In Spain, where working-age population growth depended exclusively on immigration, the rates of labor growth collapsed, only matched by the plunge of its GDP growth, and its workforce has actually started to shrink. In Ireland, despite a more abrupt fall, growth nonetheless remained positive, this was due to the fact that its population growth did not depend only on net migration but had an important natural component as well.  In fact, Ireland has the highest fertility rate amongst European countries and therefore, unless immigration returns to numbers only seen in previous centuries, growth of its working-age population should stabilize in positive territory, although at a level well below the pre-crisis.

In Portugal and Greece, even before the 2007-08 recession, the labor force growth already showed clear signs of a slowdown, as growth came to a standstill in 2005, and despite a slight recovery after, more pronounced in Greece than Portugal, working-age population went into decline with the onset of the recession.  Italy, which had reversed the decline of its working-age population initiated in the 90’s, appears once again to slide back into negative territory. On the other hand, in Germany, the workforce began to grow for the first time since 1998 due to an increase in immigration, many of them Portuguese, Spanish, Italian and Greek.

As explained by Daniel Gros, when comparing economic growth performance between countries with very different rates of population growth, the best indicator is undoubtedly GDP per Working Age Person (GDP/WAP).

Hence, if we compare the per working-age person GDP growth between the various countries in the last decade, as such taking working-age population growth out of the equation, it can be said that neither the growth of Spain and Ireland was so spectacular, nor the growth of Portugal, Germany and Italy fared so badly in comparison. In fact, only Greece seems to have had a spectacular growth, which would be in line with its productivity growth before the recession, but Greece’s population statistics might be underestimated, as Greece has not only a large population of illegal immigrants but some weakness in data collection have also been highlighted. Anyway, its growth was probably due to other unrepeatable factors, such as the Olympic Games. It also should be highlighted the economic growth achieved by Germany despite the decline of its workforce, proving that growth is possible with a declining population.

Despite some regional variations as a result of internal migrations, the reality is that working-age population in the Eurozone as a whole has initiate a long downward trend that will have major repercussions in terms of its economic growth, as explained in the previous post, and therefore, we can also conclude that Europe suffers from a "shortage of Japanese" as shown in the graph below.

As such we cannot fully comprehend the situation that Portugal, Spain and Greece face at the moment without looking into their adverse demographics.  This already exerted an disproportional role in the last decade, namely in Spain,  whose working-age population growth  goes a long way explaining its outstanding economic growth, while for Portugal the contrary it’s true, as the lack of population growth made its economy  lose steam  as it joined the Euro. More worryingly tough, is that working-age population in Europe as a whole has started a long, perhaps irreversible, path of decline that will act as a drag on its economic growth, making the economic recovery for these countries even more difficult.

Friday, June 14, 2013

On the problems of David Goodhart on immigration in the United Kingdom

I was alerted by a post by Yorkshire Ranter Alex Harrowell about Jonathan Portes' review of David Goodhart's new book about immigration in the United Kingdom, The British Dream. The title? "An Exercise in Scapegoating".

The problem with Goodhart's book, simply put, is that it doesn't appear to deal with facts.

There are two major problems with contemporary British society, according to David Goodhart in The British Dream, and both are primarily caused by immigration. The first problem is economic: the plight of the white working class, especially the young, and the decline of social mobility. Goodhart argues that low-skilled immigrants have taken jobs from unskilled natives, leaving them languishing on benefits, while high-skilled immigration reduces both the incentives and opportunities for ambitious and talented natives to move up the ladder. Many find this thesis convincing, and it has been accepted as fact by much of the political elite. There is, however, almost no evidence to support it. The second problem is social: the decline of a shared sense of community, local and national, which Goodhart relates to the failure of at least some immigrants to integrate, either ‘physically’ (where they live, who their kids go to school with, what language they speak and so on) or ‘mentally’ (in terms of the degree to which they identify with Britain, or share a common set of values). Some may think this argument has more force, but again, his conclusions far outrun the facts.

I’ll start with the economics. Goodhart gives a fair summary of the current consensus about the effects of immigration on the labour market. It comes in two parts. First, in the medium to long term the ‘lump of labour’ fallacy is just that: it isn’t true that the number of jobs in the economy is fixed, and more jobs for immigrants doesn’t mean fewer for natives. Second, the evidence suggests that in the UK immigration has little or no impact on employment even in the short term; it may drive down wages for the low-skilled, but the effect is small compared to that of other factors (technological change, the national minimum wage and so on). These things are now well established, and Goodhart appears to accept them. So it comes as something of a surprise that he should cite with approval Fraser Nelson’s observation that mass immigration ‘broke the link between more jobs and less dole’[.]

[. . .]

Goodhart cannot escape from his instinctive view that the political economy of immigration is a zero-sum game, even as he accepts that both economic theory and the evidence say no such thing. ‘A disproportionate number of new jobs,’ he writes, ‘seem to have been going to recent immigrants.’ But this is an expression of the ‘lump of labour’ fallacy he elsewhere dismisses. The belief that if immigrants get ‘more’ of something (jobs, education, opportunities, political power), natives (or whites) must get less. This guides his discussion of the local economic impact of immigration in Merton, South-West London: ‘Poor whites [are] doing the worst of the lot.’ Such people have ‘mainly opted out: they seldom vote, and a lot of the younger people are “Neets” – not in employment, education or training.’ It isn’t entirely obvious from Goodhart’s description of Merton why immigration is responsible for this. Do the immigrants displace natives from jobs, schools and polling booths, or do they somehow drag them down? Either way, such facts as there are in this sentence appear to be wrong. Do fewer working-class whites vote in Merton than elsewhere? I’m not aware of the existence of any statistics on local voting by ethnicity or class, but both the Merton constituencies (Mitcham and Morden, and Wimbledon) had a higher than average turnout at the last election, and more than two-thirds of Merton voters are white. As for Neets, the Merton Council report cited by Goodhart found 87 white Neets aged 16-18. Is that a lot? Not really: there are about 2300 whites in the relevant age group. Whether the council’s data are directly comparable with official statistics isn’t certain, but if they are then the chances of a white teenager being Neet are considerably lower in Merton than they are nationally. (The official statistics suggest that Merton’s Neet rates are pretty low.)

More, Goodhart's problems with facts seem to have unsettling inclinations.

Goodhart is of course right that young Brits, especially those not from middle-class backgrounds, are having a pretty hard time. But the question is why. Think about it this way. Suppose you’re poor, young and white: where in the UK don’t you want to be? That’s a subjective question. But from an economic point of view, one might want to consider such criteria as the proportion of young people who don’t get decent GCSEs and the number who are out of work. By those yardsticks, the answers are reasonably clear. Nationally, just under 60 per cent of kids whose first language is English get five good GCSEs including maths and English. There are eight local government districts where that figure dips below 50 per cent: Hartlepool, Middlesbrough, Knowsley, Blackpool, Barnsley, Hull, Nottingham and the Isle of Wight. Job prospects for young white school-leavers in these areas are, not surprisingly, poor: the proportion on the dole ranges from 12 to 18 per cent, compared to the national average of about 8 per cent. Yet of these areas, only Nottingham has a substantial non-white or immigrant population. In London, children whose first language is English are somewhat more likely (about 62 per cent) than the national average to get five good GCSEs, despite considerably higher poverty rates. Young whites in London don’t often end up on the dole: only about 5.5 per cent, close to the lowest rates in the country. Things are a little worse in Merton, where GCSE results are about average, and the proportion of young whites claiming benefits is about 6.5 per cent, but that’s still well under the national average.

So, to put it bluntly, if you’re going to be white, British and poor, all the statistical evidence suggests you’d be better off being born in Merton – or anywhere else in London, surrounded by immigrants – than in the mostly white areas where educational outcomes, in particular, are worse. We need to be careful here. Correlation doesn’t imply causation. There are lots of possible explanations for the figures I have given; in particular, the remarkable improvement in London’s schools, especially for more disadvantaged children, over the last decade. And there’s little doubt that the depression of local economies in some Northern cities is responsible for both the high unemployment and relatively low immigration in those places. Econometric analysis suggests that there’s little or no association between high immigration and employment or unemployment rates.


Goodhart [makes] an absurd, some would say offensive, analogy with the US. Talking about Waltham Forest in North-East London, he writes: ‘In America, this is called Sundown Segregation; people mixing during the day but going home to quite separate neighbourhoods.’ ‘Sundown towns’, as the sociologist James Loewen has documented, were places in the US where, before the enactment of civil rights legislation, black people (sometimes in earlier years also Chinese, or Mexicans) were not permitted to remain after dark. I’m not sure which is worse: the irrelevance of this concept to modern Britain; the failure to do the elementary research required to establish the origin of the term; or the ugliness of comparing Waltham Forest to American towns that once displayed signs reading ‘Don’t let the sun set on you, nigger.’

Even worse is Goodhart’s discussion of specific communities where there are very real economic and social problems. I know little about Bradford, so have no firm basis on which to judge his plausible-sounding claims about the Pakistani immigrant experience there: much higher segregation than in London, the dominance of clan politics, the impacts of chain marriage migration etc. A friend in Bradford – a professional, leftish woman of Pakistani origin, working in education – told me that what he has to say is ‘unscientific’, based on ‘cheap anecdotes and quite frankly baloney’. No reason you should take her word for it, so let’s provisionally accept Goodhart’s suggestion that there is a prima facie case that first-cousin marriage is responsible for a higher than average proportion of birth defects among Bradford Pakistanis. He cites some relevant research, but follows it up with assertions which are unsubstantiated, inaccurate and alarmist: ‘Bradford has just opened two more schools for children with Special Educational Needs,’ he writes. ‘On some measures nearly half of all children in the area qualify for special help.’ No source. However, the Department for Education publishes the statistics, and it turns out that in Bradford, the proportion of children who ‘qualify for special help’ is about 21 per cent. Well, 21 per cent is not ‘nearly half’. It is, in fact, only slightly higher than the national average of 20 per cent. And it is significantly lower than in some other, mostly white places. Nationally, the highest figure is 27 per cent, in north-east Lincolnshire. What’s going on? Maybe the locals marry their cousins there too. But I think I’ll restrict myself to saying that this is a complex phenomenon on which I’m no expert. It’s a pity Goodhart didn’t do the same.

David Edgar's Guardian review touches on this.

Many elements of this narrative are questionable. Set against comparable countries, Britain's current immigration level is average: both Germany and France have higher numerical foreign-born populations, and 11 EU countries have immigrant populations which are proportionately higher than ours. It's true that less than half of current immigration comes from the EU, but emigration by non-EU citizens is higher too. As Goodhart acknowledges, more than 70% of current immigrants stay less than five years, because so many of them are students (only half of the headline four million have settled here). Studies of the 2011 census indicate that large cities such as Birmingham and Bradford have seen a decrease in segregation for most ethnic groups; in London, the decrease is particularly notable among Bangladeshis.

The argument that racism is greatly exaggerated doesn't really stand up either. The five-fold increase in the number of racist attacks since the early 90s may be partly due to changing definitions, but the absolute 2011-2012 figure of 47,678 racist incidents in England and Wales, of which 35,816 were recorded by the police as race-hate crimes, is a dramatic figure in itself. Goodhart's argument that press demonisation of immigrants contributes positively to race relations by providing "a psychological safety valve" is clearly self-serving. On pay, a 2008 report to the Equality and Human Rights Commission found that the earnings of Pakistani and Bangladeshi men at the low and middle levels of education were only two thirds of those of similarly qualified white men. On employment, male Chinese graduates are over three-quarters less likely to be employed than their white peers. And while Goodhart acknowledges the results of the so-called "CV tests", revealing that employers are still less likely to employ applicants with ethnic minority names, he justifies the discrepancy on the grounds that such people might prove "a source of tension and embarrassment" in the workplace, as, after all, "people will generally give preference to, and feel more comfortable being around, people they are familiar with". Acting on such attitudes as an employer has, of course, been illegal since the passage of the 1968 Race Relations Act.

[. . .]

Of course Goodhart acknowledges the cultural and (in a selective kind of way) the intellectual contribution of Britain's postwar immigrant communities. But recognising the changes that were brought about not in restaurants or at concerts or universities, but in workplaces and on the streets, through campaigns that were initially resisted by large sections of the host community, challenges the notion of a unified, linear national story. Goodhart's insistence that integration cannot be a "two-way street" and that immigrants "must carry the burden of any adaptation that is necessary" raises the question of what is being adapted to.

Harrowell has written at length about Goodhart's radicalization, here, noting (after a blog post by Portes) Goodhart's own claims that the United Kingdom is being run by people who care more for the benefits of people in Burundi than of people in Birmingham.

Well, we've been warned.

Thursday, June 13, 2013

"Hans Rosling: Religions and babies"

I'd like to thank Will Baird for pointing me towards this 2012 TED talk by the Gapminder Foundation's Hans Rosling, examining the question of the relationship between religion and fertility.

Sunday, June 09, 2013

Three links on cities as enormously productive, and predictable, social organizations

Cities, as forms of human settlement that are becoming increasingly dominant across the world, have been on my mind recently. I wanted to share with our readers three papers of interest regarding the structure and importance of cities, economically and demographically.

The first paper I came across via via blogger James Nicoll. The 2007 article in the Proceedings of the National Academy of the Sciences, "Growth, innovation, scaling, and the pace of life in cities", by Luís M. A. Bettencourt, José Lobo, Dirk Helbing, Christian Kühnert, and Geoffrey B. West, argues for the existence of structured urban hierarchies.

Humanity has just crossed a major landmark in its history with the majority of people now living in cities . The present worldwide trend toward urbanization is intimately related to economic development and to profound changes in social organization, land use, and patterns of human behavior. The demographic scale of these changes is unprecedented and will lead to important but as of yet poorly understood impacts on the global environment. In 2000, >70% of the population in developed countries lived in cities compared with ≈40% in developing countries. Cities occupied a mere 0.3% of the total land area but ≈3% of arable land. By 2030, the urban population of developing countries is expected to more than double to ≈4 billion, with an estimated 3-fold increase in occupancy of land area, whereas in developed countries it may still increase by as much as 20%. Paralleling this global urban expansion, there is the necessity for a sustainability transition toward a stable total human population, together with a rise in living standards and the establishment of long-term balances between human development needs and the planet's environmental limits. Thus, a major challenge worldwide is to understand and predict how changes in social organization and dynamics resulting from urbanization will impact the interactions between nature and society.

The increasing concentration of people in cities presents both opportunities and challenges toward future scenarios of sustainable development. On the one hand, cities make possible economies of scale in infrastructure and facilitate the optimized delivery of social services, such as education, health care, and efficient governance. Other impacts, however, arise because of human adaptation to urban living. They can be direct, resulting from obvious changes in land use [e.g., urban heat island effects and increased green house gas emissions ] or indirect, following from changes in consumption and human behavior, already emphasized in classical work by Simmel and Wirth in urban sociology and by Milgram in psychology. An important result of urbanization is also an increased division of labor and the growth of occupations geared toward innovation and wealth creation. The features common to this set of impacts are that they are open-ended and involve permanent adaptation, whereas their environmental implications are ambivalent, aggravating stresses on natural environments in some cases and creating the conditions for sustainable solutions in others.

These unfolding complex demographic and social trends make it clear that the quantitative understanding of human social organization and dynamics in cities is a major piece of the puzzle toward navigating successfully a transition to sustainability. However, despite much historical evidence that cities are the principal engines of innovation and economic growth, a quantitative, predictive theory for understanding their dynamics and organization and estimating their future trajectory and stability remains elusive. Significant obstacles toward this goal are the immense diversity of human activity and organization and an enormous range of geographic factors. Nevertheless, there is strong evidence of quantitative regularities in the increases in economic opportunities, rates of innovation, and pace of life observed between smaller towns and larger cities.

In this work, we show that the social organization and dynamics relating urbanization to economic development and knowledge creation, among other social activities, are very general and appear as nontrivial quantitative regularities common to all cities, across urban systems. We present an extensive body of empirical evidence showing that important demographic, socioeconomic, and behavioral urban indicators are, on average, scaling functions of city size that are quantitatively consistent across different nations and times [note that the much studied “Zipf's law” for the rank–size distribution of urban populations is just one example of the many scaling relationships presented in this work]. The most thorough evidence at present is for the U.S., where extensive reliable data across a wide variety of indicators span many decades. In addition, we show that other nations, including China and European countries, display particular scaling relationships consistent with those in the U.S.

Via The Atlantic Cities' Emily Badger, meanwhile, I came across "Urban characteristics attributable to density-driven tie formation". Authored by Wei Pan, Gourab Ghoshal, Coco Krumme, Manuel Cebrian, and Alex Pentland, and originally presented in June 2012, the paper makes some interesting claims.

In this paper we propose social tie density (the density of active social ties between city residents) as a key determinant behind the global social structure and flow of information between individuals. Based on this we have described an empirically grounded generative model of social tie density to account for the observed scaling behavior of city indicators as a function of population density.

The model predicts that social tie density scales super-linearly with population density, while naturally accounting for the narrow band of scaling exponents empirically observed across multiple features and different geographies. We note that this is achieved without the need to recourse to parameter tuning or assumptions about modularity, social hierarchies, specialization, or similar social constructs. We therefore suggest that population density, rather than population size per se, is at the root of the extraordinary nature of urban centers. As a single example, metropolitan Tokyo has roughly the same population as Siberia while showing remarkable variance in criminal profile, energy usage, and economic productivity.

We provide empirical evidence based on studies of indicators in European and American cities (both categories representing comparable economic development), demonstrating that density is a superior metric than population size in explaining various urban indicators.

Our argument suggests that the reasons for creating cities are not that different from creating work environments like research institutions. While current technology makes remote communication and collaboration extremely easy and convenient, the importance of packing people physically close within each other is still widely emphasized. We argue that cities are operating under the same principle|as a consequence of proximity and easy face-to-face access between individuals, communication and ultimately productivity is greatly enhanced.

We of course note certain caveats and limitations of our study. The density of social ties is intrinsically a function of the ease of access between residents living in the same city. Consider the example of Beijing in China, which has a very high population density. Due to its traffic jams, Beijing currently is de-facto divided into many smaller cities with limited transportation capacities between them and consequently may not demonstrate a higher social tie density than other cities with a much lower population density. Thus a direct comparison of the model predictions with a similarly dense area such as Manhattan is not feasible.

[. . . ]

A number of theories of urban growth suggest the importance of specialist service industries, or high-value-add workers, as generative models of city development. While our model does not disprove these theories, it provides a plausible and empirically-grounded model that does not require the presence of these special social structures. The other theories must therefore appeal to different sorts of data in order to support their claims. Cities are one of most exceptional and enduring of human inventions. Most great cities are exceptions in their own right: a New Yorker feels out of place in Los Angeles, Paris, or Shanghai. However, this exceptionalism may be more due to our attention to human-scale details than tothe underlying structures. In this paper we have presented a generative theory that accounts for observed scaling in urban growth as a function of social tie density and the diffusion of information across those ties. It is our hope that this provides both a foundation for the commonalities across all cities and a beginning point for which divergence between specifc cities can be explored.

In Badger's interview, Pan suggests that the enormous benefits of urban life start to flatten out once urban areas pass the 40 million mark.

What might this mean? Well, via io9, I came across the April 2012 McKinsey Global Institute report"Urban America: US cities in the global economy", by authors James Manyika, Jaana Remes, Richard Dobbs, Javier Orellana and Fabian Schaer. It makes the claim that the United States is uniquely highly urbanized, and hence uniquely prosperous.

Today, large US cities have more weight in the US economy than do large cities in any other major region. In 2010, 259 large US cities generated almost 85 percent of US GDP. During the same period, large cities in Western Europe accounted for less than 65 percent of the region’s GDP. Among emerging regions, metropolitan China accounted for 78 percent of China’s GDP and the large cities of Latin America contributed 76 percent to regional GDP.

Large US cities have such relative economic weight for two reasons. First, they are home to 80 percent of the population compared with less than 60 percent in Western Europe. Second, they have a relatively high per capita GDP premium. The average per capita GDP of large US cities is almost 35 percent higher than in smaller cities and rural areas; in Western Europe, this premium is about 30 percent.

The relative weight of different regions in the world economy changes when we home in on the economic clout of their large cities. Even though Western Europe’s GDP exceeded that of the United States by nearly 10 percent in 2010, the combined GDP of large US cities exceeds that of large Western European cities by more than 20 percent.

It is America’s cities that explain why the United States continues to enjoy higher per capita GDP than Europe. The higher share of US urbanites—and the fact that they command a larger per capita GDP premium over US smaller towns and rural areas than do their European counterparts—explains three-quarters of the per capita GDP gap between the two economies.

The nation’s largest and well-known megacities of New York, New York, and Los Angeles, California, will continue to prosper. New York is on course to remain the second-largest city by GDP in the world in 2025, and Los Angeles to rise from sixth place today to become the fourth-largest city. But the weight of these megacities in the US economy is not decisive to the overall importance of cities in the United States. London and Paris have a smaller share of the overall Western European population—6 percent, compared with the combined population of the US megacities of 10 percent of the total US population—but they enjoy a significantly higher per capita GDP premium than their US counterparts. Paris and London contribute 9 percent to Western Europe’s overall GDP, compared with the 13 percent contributed by New York and Los Angeles.

Instead, the true vigor of America’s urban economy comes from a broad base of dynamic middleweights and the relatively high per capita GDP they achieve. There are just over 255 middleweight cities in the United States, compared with just over 180 in Europe. And they generate more than 70 percent of US GDP today, compared with just over 50 percent in Western Europe. In fact, the top 28 US middleweights alone contribute more than 35 percent of US GDP. The dynamism of middleweights in the United States is a characteristic of today’s global urban expansion, making them an interesting group to understand for both US and global growth prospects.


I do wonder whether comparing the United States to western Europe was the best thing to do, not least since despite western Europe's integration within the European Union the region's nation-states and their frontiers are still quite relevant. The argument doesn't seem immediately implausible, though, and has obvious implications on regional population balances within the European Union and its various member-states (and between the member-states). Should Europeans, for instance, try to promote increased urbanization and the growth of large cities to compensate for shrinking workforces? What of other world regions?