Tuesday, 5 February 2013



On Marginal Revolution and its post on Dr. Cochrane, I came across another commenter 'Claudia', aka Claudia R Sahm, an economist working for the Fed. She was discussing the determinants of individual consumption and hinting at very different modeling ideas than the ones traditionally seen behind the DSGE model so beloved by Central Banks.

For non-economists, DSGE stands for Dynamic Stochastic General Equilibrium. You're meant to be impressed, by the way. Those models, especially in their New Keynesian variants, are a fusion of supposedly Keynesian macro-economics ideas with Neo-classic micro-economics (for more details, if you're really interested: http://en.wikipedia.org/wiki/Dynamic_stochastic_general_equilibrium).

Here, Claudia R. Sahm was clearly interested in re-writing macro-models with 'permanent income' and 'expectations' playing key roles.

In her own words, "I think that models which rely on significant interest rate sensitivity by households are going to have a hard time explaining actual consumption (...) I think the permanent income explanation is more plausible (...) The resources available to households took a big hit in the recession, lower net worth, little income gains, high unemployment AND households appear to have projected those conditions forward. The income expectations series in the Michigan survey fell like a rock in the recession"


Her thoughts on the subject rekindled my interest in trying to formalise a bit what I thought about Macro. You might remember that I posted a first, not very good, post about this. The take-away was that, despite some validity of supply-side concerns, the key to growth was 'solvent demand'.

I now would like to extend that thinking a bit.

First, let's define what we're trying to achieve. It's not original but I am interested in the same variables that interests most macro-economics i.e. Total output and/or output growth (GDP as the best and most standard approximation), (un)employment, business cycles, inflation and, ultimately, the policies governments should adopt in order to maximize material wealth for their people.

However, not only do I reject those DSGE models most macro-economists now use as barbaric monstrosities but I even disagree with things like IS/LM because, in my opinion, interest rates are not fundamental to the real world macro-economic situation and its evolution. It's not that they do not count but they are variables of lesser importance.

So what's left? Well, I quite like the old

Y = C + I + G + (X-M) 

i.e. GDP is equal to the sum of Private Consumption, Private Investment, Public spending and the difference between exports and imports (net exports).

What we're trying to do is maximize Y, smoothly, over time.

Let's take a look at 'Consumption', the subject of Dr Cochrane's article and Claudia R. Sham's studies. Traditionally, it is represented as follow:

 C = a + b(Y-T)

where b is the marginal propensity to consume. Keynes seem to have considered it a function only of present real income. Milton Friedman added a permanent income hypothesis i.e. transitory changes on income have little effect on consumption and Franco Modigliani formulated a Life Cycle Hypothesis i.e. people vary their consumption according to their stages in life.

Both Friedman and Modigliani obviously have a point. However, from Friedman intuition about permanent income, it seems that little useful was done: the formalization seems to give a determining role interest rates

What would I personally add or change to this body of work?

I would say that Consumption is a factor of income - both current and permanent as well as 'windfalls'.  Depending on the form that windfall takes, reactions vary a lot i.e. people can save it all or spend it all in a fairly observable fashion but without much distinguishable rational. An interest in 'windfalls' may sound strange, until you realise that a lot of stimulus, a classical Keynesian tool, can be classified as 'windfall' i.e. a temporary boost to present income, unlikely to be repeated. This is not trivial.

Furthermore, I would add that a key element determining consumption is uncertainty. The more uncertain people are about their incomes projections, the less they're likely to consume. Hence, high unemployment or job insecurity are big factors limitating consumption, above and beyond the destruction of solvable demand unemployment represents.

Finally, another element to add to the mix is debt i.e. the ability and willingness of consumers to increase consumption/investment using debt. Clearly, here, the existing level of debt, the ability to service it and expectations about future net worth as well as the confidence one feels in such expectations all matter.

Importantly, I think interest rates are NOT crucial, both in the taking of debt and in the arbitraging between consumption and saving - Except when it comes to housing as mortgage loads are sensitive to reasonably small changes in interest rates and, particularly, finance. A lot of financial products are quite sensitive to even minute changes in interest rates so banking activities and banking performance are indeed well linked to interest rates. But, in the real economy, housing asides, given the large degree of uncertainty about the future (even in better times than 2013), it takes big changes in interest rates or in interest rates expectations for people to move one way or the other.

Life cycle does matter but, if your demography is stable, it's not a big deal at the aggregated level. However, in the West, with an aging population, it will be a factor unless people can be reasonably certain of the permanency of social safety nets - retirement, health care programs etc.

I don't have much of a clue as to how to model these hunches mathematically but this is definitely the kind of stuff I'd look at to explore the formation of consumption decisions.