Most, if not all, of the deficit nations that make up the Bretton Woods II edifice and subsequent global imbalances have seen notable housing and construction bubbles as part of their path towards excess leverage. The demand for housing and thus in some sense construction on the aggregate macroeconomic level can be tied to demographics and specifically the idea of a life course [1]. So, could we say that this was one of the channels through which demographics have indirectly affected global imbalances?
I think so and below I will argue why, but first things first. In fact, blame it on that double shot latte that I enjoyed a couple of days ago early in the morning reading this paper by Pedro Gete from Georgetown University, but what follows will be terribly wonkish. Yet, intellectually and theoretically I think it represents an important piece of the puzzle so if you have the energy I welcome you to indulge me.
The Model,
Mathematical economics, rational expectations, and representative agent modeling have taken their share of the flak in the context of the macroeconomic shake-out following the financial crisis. For the most part I agree with the critics, but let the following be a counter example on the general blurred and fuzzy nature of economic modeling. In this way, the model developed and discussed in Gete (2010) is neat, concise, and easily allows for an intuitive expansion of perspective. In short, I like it; a lot! For good measure, here is the abstract;
This paper makes a theoretical and an empirical contribution to the debate on what caused the “global imbalances”. On the empirical side, I provide different types of evidence to support that housing demand shocks (shocks to the aggregate marginal rate of substitution between housing and tradables) help to explain the global imbalances. On the theory side, I show that shocks to the demand for housing generate trade de…cits without need for the standard ingredients used by others to model housing (wealth effects or trade in capital goods). I model housing as a durable and nontradable good. Countries import tradable goods during periods when more domestic labor is devoted to produce nontradables to smooth consumption between tradables and nontradables. Housing booms are larger if the country can run a trade de…cit because the de…ficit lowers the opportunity cost of building, which is the foregone consumption of tradable goods due to reallocation of labor to the construction sector. Concerning the empirical evidence, I first document that over the last decade there has been a strong cross-country correlation between housing variables and current account dynamics. Second, I show that using the cross-country dynamics of employment in construction as the explanatory variable, the model generates current account dynamics matching recent global imbalances. Finally, I use sign restrictions implied by the model to estimate a vector autoregression and identify the effects of housing demand shocks on the U.S. trade de…cit. The results suggest that housing shocks matter for current account dynamics.
In order to spare my readers the nitty gritty of the model I will not do the math here, but merely describe the model verbally. In this way, the nice aspect of the model and thus intuition in Gete (2010) is that it assumes that housing shocks are exogenous (which of course they aren’t), but by doing this he creates a simple laboratory through which to impose different assumptions on what might actually make shocks to the demand for housing endogenous (hint: this is where demographics come in, but first things first).
The economy in Gete (2010) is very simple. It consists of two sectors, a tradable and a non-tradable (housing), and only one input to production which can be shifted without adjustment costs between the two sectors (but not between countries). Now, in a model like this with exogenous preferences there is really not a lot we can do and this is especially the case if we assume a two economy world where relative preferences between tradables and non-tradables in both economies are the same. In such an economy, we simply solve the respective optimization problems and each economy consumes the same amount of tradables and non-tradables (housing). However, let us introduce a deus ex machina and assume that the demand for housing suddenly increases in one of our economies. What will happen?
Well, those of you with a fondness for core economic theory should start to feel that warm fuzzy feeling just about now while the rest of you … are you even still with me?
Moving back on track, and given that our only input to production is labour, the economy will have to shift labour from the tradables sector to the non-tradables sector and thus production of tradables will give way for production of non-tradables (housing) [2]. But does it also mean that we have to give up consumption of tradables? No, and this is the key mechanism by which Gete (2010) explains how an increase in the demand for housing/construction leads to a trade deficit and thus may explain global imbalances to the extent that the deficit countries are the ones who have seen housing booms. Specifically and in the jargon of the trade, the economy who sees a positive shock to housing demand may smooth consumption of tradables and non-tradables by running a current account deficit; effectively borrowing to consume tradables which it no longer produces itself because labour has been allocated to the construction of housing.
The empirical evidence presented in Gete (2010) is circumstantial although the simulation based on the model turn out quite well. One chart however which shows the proposed relationship quite well is the following which plots the change in the CA/GDP ratio (in % points) on the y-axis against the change in the labor share of construction on the x-axis. Moreover, I have made my own small replication with different data and a slight change methodology (and sample [3]) and the picture is the same.
In this sense the relationship seems solid enough to vindicate the overall claim that raising the weight of housing/construction leads to excess investment over the capacity of domestic savings and thus an external deficit. Note especially that my sample, data, and methodology are quite different from those in Gete (2010) Whether Mr. Gete is right in that it relates to tradables vs. non-tradables is a theoretical discussion in its own right, but I am willing to go with this issue for now although, from a life cycle/permanent income perspective, I would not discount the wealth effect argument entirely.
Endogenising Housing Shocks, introducing Life Course Theory
Stepping outside the realms of Neo-Classical representative agent modelling and real business cycle simulation, how might we operationalize this argument in a realistic theoretical context [4].
Before moving on, remember the key premise set up by Gete (2010). Preferences are exogenous and thus what this paper really sets out to show is how the distinction between housing as a non-tradable and "other goods" as tradables may lead to capital flows between economies. Specifically, this is put up as an alternative explanation to how an increase in housing demand may generate a trade deficit relative to the traditional account of a strong wealth effect from housing which translates into a demand for consumption today relative to savings tomorrow (or through the fact that it increases permanent income). For the purpose of what follows, this particular distinction is not so important. What is important however is to find a way or mechanism through which to endogenize the preference for housing and thus a way to model this purposed shock.
Enter life course theory.
Life course theory essentially deals with the timing and significance of key events in an individual’s life from birth to death. As it is presented in presented e.g in (ed. Mayer (2006)) it primarily operates on core sociological parameters such as age of leaving school, age of labour market entry, age of retirement, age of marriage (fertility decisions). This makes it a broader framework than the life cycle framework, but in fact the two are tightly joined at the hip. In this sense, one could also imagine more strict economic parameters such as age of first acquisition of house, car, or other “durable” goods. Specifically, one would probably tie this together to the capacity and willingness to take on debt to finance large durable purchases which are, by definition, very rarely financed through non leveraged lump sum transfers.
Slowly but surely we are moving towards a working model here. One way to conceptualize the way demographics may serve to generate international capital flows would then be to take a life course/life cycle perspective of the demographic transition in which an economy harbours the capacity sustain and develop a construction/housing boom which, through the mechanism described in Gete (2010), may serve to facilitate an ongoing trade deficit. Specifically, we could say that an economy must have a certain and relative amount of workers in their most productive age (say 30 to 45) to generate this dynamic which, by far, is not automatic since evidently; for an economy run a large current account deficit there need to be a corresponding pool of foreign savings.
As shown, Gete (2010) provides tentative evidence to suggest how the correlation between a large share of construction as percentage of GDP is closely associated with a trade deficit.
In general however, does my theory square off with reality?
Well, consider the fact that no economy with a median age over 40 have seen a sustainable housing boom which has led into an external deficit. In fact, based on the assumptions laid out above in the small laboratory set up by Gete (2010) and my endogenous imposition through demographics, we could say that as long as there is a balanced amount of economies with relatively stable population pyramids contrary to a group of rapidly ageing economies the system may work. Apart from mercantilist Asia and the petro exporters, I would hold this to be an important part of the source of the underlying stability of Bretton Woods II. The problem is that we are all ageing beyond the threshold where we can reasonably expect the economy to shoulder a large and ongoing deficit based on a rapid increase in construction and housing. This then becomes just another, and very specific, perspective on the Gordian knot which is the global economic system with so many would be savers (exporters) and too few economies willing and able to suck up the excess liquidity [5].
Consequently, and if we can say that if the characteristics of a classic external deficit economy based on demographic fundamentals include a large share of construction/housing as a percentage of GDP, the recent economic crisis has drastically limited the peloton of such economies while simultaneously increasing the number of economies more than willing to finance whatever housing bubble there might be left somewhere.
Is this a new proposition then?
Not quite and if we look at the argument as a two step theoretical construct, the first step in terms of linking the demand for housing/construction to demographics is not new. Mankiw and Weil (1989) who examine the effect of the boomer generations move through the population pyramid is a seminal contribution with Green and Hendershott (1996) and related study. Far more interesting however are studies who try to combine both steps and thus tying together construction, demographics, and external account dynamics. Here especially Malmberg and Lindh (1999a and 1999b) is interesting as it shows how the disaggregation of investment reveals significant effects in a classic empirical context à la Higgins (1998) and Lürhmann (2003). Note especially this from the abstract;
Disaggregating investment we find that young cohorts have a positive correlation with housing investment while older but still active cohorts have a positive correlation with business investment. The differences in saving and investment effects are, nevertheless, sufficient to generate persistent and sizeable age effects on the current account. Our results suggest that policies concerning current account balance should take into consideration age distributions and the degree of development.
This would seem to fit not only with the picture laid out in Gete (2010), but also crucially with a strong life course effect and thus an argument to support a strong eye to life course/life cycle dynamics when modelling current account dynamics.
With respect to the global imbalances this perspective also offers a valuable lesson.
In a nutshell, if demographics drive housing and construction activity, and the latter drive current global imbalances it then follows naturally that demographics have something to do with current account imbalances. It may not be so easy however. Consequently, it is dubious to claim that the reason why excess global liquidity was channeled into housing in the first place falls exclusively on the effect from demographic change. However, this would also be the wrong way to present the main argument. In this way, we can say that the extent to which a given economy was able (willing) to respond to excess global liquidity by developing a housing/construction bubble was a function of a specific demographic structure. More to the point even; the inability to harbour or develop a housing/construction bubble and thus a matching current account deficit for all this liquidity splashing around is marked by the absense of a certain demographic structure namely that of a median age below 40 (to put it really strict).
This final point is perhaps the most important point to take home from my considerations above.
Summary
Well, I told you that this would be wonkish, but even if you have not read all those academic references above I hope that it is still possible to dissect the main message here. Demographics matter, but much more than this a strong methodological foundation in a life course/life cycle framework enables us to see how the demographic transition casts a long and significant shadow over the economic profile of individual economies as well as the aggregate global economy. Of course, this is my all time favorite hobby horse, but I do think that the facts are there to back me up.
As for Gete (2010), I would warmly recommend you to read it especially if you have lost faith in classic economic modelling where, I think, it provides a good example of an intuitive and easily comprehendible model of the world.
For future reference the topic above is naturally one that I will be pursuing vigorously as I move forward, so this is not the last time that I have treated this topic since it has much, much more to offer.
————–
This was a post by Claus Vistesen, who also blogs at his own site Alpha.Sources.
List of References
Gete, Pedro (2010) – Housing Markets and Current Account Dynamics, Working Paper (Georgetown University)
Green R. and P.H. Hendershott (1996) – Age, housing demand, and real house prices. Regional Science and Urban Economics, 26(5):465–480,
Higgins, Matthew (1998) – Demography, National Savings, and International Capital Flows, International
Economic Review, Volume 39 (1998) Issue (Month): 2 (May) pp 343-69
Malmberg, Bo and Lindh, Thomas (1999a) – Age Distributions and the Current Account A Changing
Relation? Working Paper Series 1999:21, Uppsala University, Department of Economic
Malmberg, Bo and Lindh, Thomas (1999b) – Demography and housing demand—what can we learn from residential construction data. Workshop on Age Effects on the Economy, Stockholm, pages 2–3, 1999
Mankiw, N. Gregory and Weil, David N (1989) – The baby boom, the baby bust, and the housing market, Regional Science and Urban Economics Volume 19, Issue 2, May 1989, Pages 235-258
Mayer, Karl Ulrich (2006) – Handbook of the Life Course (review) Social Forces – Volume 84, Number 4, June 2006, pp. 2363-2365
Lürhmann, Melanie (2003) – Demographic Change, Foresight and International Capital Flows, MEA
discussion paper series 03038, Mannheim Research Institute for the Economics of Aging (MEA),
University of Mannheim
—
[1] – No, not life cycle which is also important, but indeed life course.
[2] – Think undergrad microeconomics here with allocation of labour in a two goods economy with a corresponding production possibility frontier and indifference curve.
[3] – I have the following countries; Australia Austria Belgium Czech Republic Denmark Finland France Germany Greece Hungary Ireland Italy Japan Korea Luxembourg Mexico Netherlands Poland Portugal Slovak Republic Spain Sweden Switzerland United Kingdom United States Estonia Slovenia.
[4] – And yes, here I AM implying that RBC simulations and neo-classical economic modelling are not very realistic even if they may hold some intuitive appeal.
[5] – … which again is created by the fact that OECD central banks are in QE mode in an attempt to spur growth in their domestic economies, but where the liquidity simply ends up moving through the back door ending up wherever there is yield to be found. And round and round we go …