27/04/2011

So the below points towards an imprecise set of microfoundations for the older generation of simultaneous equation models not derived from an intertemporal consumption/saving optimisation problem.  Consumption and investment decisions are separated, essentially, allowing for a return to the Kalecki/Goodwin worker/capitalist models.  Obviously, if savings are only allowed to take the form of equity then this distinction effectively holds anyway.  The problem, of course, is how to reconcile the consumption and saving decision (in a form as close to the neoclassical method as possible) with the now independent investment decision.  There is an enormous amount of literature for inspiration, fortunately.

26/04/2011

I => S etc

Thoughts whilst reading Chiang's Fundamental Methods of Mathematical Economics chapter on dynamic analysis.  Really need to work through this whole book next year (as I'm rubbish at maths):

In neoclassical dynamic equilibrium models, I = S is decided by an optimal intertemporal consumption decision give accessible rates of return on S => I.  With the existence of firms, if owners can communicate their chosen level of investment, nothings changes.  Alternatively, if this communication is impossible for whatever reason, shareholders can communicate these decisions through efficient capital markets (by Modigliani Miller).

However, if both production and financial intermediation exist, there is no mechanism by which the system can ensure S => I from max u(C) problem (see argument contra MM).  In this world, clearly investment may be induced to a degree dependent on firms' behaviour; ie we get multiplier-accelerator dynamics and I <=> S (see Chiang pp540 for Samuelson M-A model).

Alternatively, the Lucas surprise supply model is essentially an induced investment model based on noise in firms' demand (see Fisher Black's paper, Noise)!!  'Noise' models with irreversible processes should, incidentally, provide lots of material for simulated disequilibrium macro-models. . .

Stephen Hawking . . .

Re-reading A Brief History of Time for the nth time (probably the best book ever).  Ch.9:
A computer memory is basically a device consisting of elements that can exist in either of two states.  Before an item is recorded in a computer's memory, the memory is in a disordered state, with equal probabilities for the two possible states.  After the memory interacts with the system to be remembered, it will definitely be in one state or the other, according to the state of the system.  So the memory has passed from a disordered state to an ordered state.  However, in order to make sure that the memory is in the right state, it is necesssary to use a certain amount of energy.  This energy is dissipated as heat, and increases the amout of disorder in the universe.  One can show that this increase in disorder is always greater than the increase in the order of the memory itself.  Thus the heat expelled by the computer's cooling fan means that when a computer records an item in memory, the total amout of disorder in the universe still goes up.  The direction of time in which a computer remembers the past is the same as that in which disorder increases. 
Our subjective sense of the direction of time, the psychological arrow of time, is therefore determined within our brain by the thermodynamic arrow of time.  Just as a computer, we must remember things in the order in which entropy increases.  This makes the second law of thermodynamics almost trivial.  Disorder increases with time because we measure time in the direction in which disorder increases.  You can't have a safer bet than that!

20/04/2011

Milton Friedman, FT interview 2003

 "I've been wrong so far, so I don't have too much confidence in my view. But I think within the next 10 to 15 years the eurozone will split apart. The British government, on balance, should stay out of it."

The Heart of the Matter

Also, finished The Heart of the Matter over the weekend.  Quite possibly the most depressing book I've ever read - not existentially depressing in the continental manner, more heavy and oppressive in a more emotional sense.  Good though - now I just want to read more GG. .

'Market Completing'

Part of an interesting thread on market completeness on Beyond Microfoundations.  I asked:
@David/Keshav,

Is there any proof in general equilibrium theory that, given incomplete markets (say S states of the world, N = S-2 AD securities) a 'non-completing' increase in markets (say to N = S-1 AD securities) invariably results in a Pareto improvement?
Keshav responded:
Rob,

No, in fact an increase in markets (even a `completing' increase) need not be Pareto improving. Here is a counterexample:

There is one commodity (corn), 2 periods and 2 states which are realized between period 0 and 1. There are 3 agents: `savers' who have only a period 0 endowment and value consumption in period 1, state 2 more highly than period 0 consumption; `safe borrowers' who have an endowment in period 1 in either state, and only value period 0 consumption; and `risky borrowers' who only have a period 1, state 2 endowment and only value period 0 consumption. Suppose first that the only asset traded is a riskless bond which pays out one unit of corn in period 1, in either state. The only agent who can issue such a bond is the `safe borrower', since only he can meet his obligation to pay out corn if state 1 occurs.

Now suppose we replace the riskless bond with the two Arrow securities. The security paying off in state 1 will have no value since no-one values state 1 consumption. Now, both types of borrower can issue securities paying out in state 2. Since the supply of these securities is greater than under incomplete markets, their price will be lower. This fall in price reduces the safe borrower's income, so he consumes less in period 0. So completing markets is not Pareto improving.

16/04/2011

Irreversible Changes

"Entropy measures the irreversible changes in a system"

Equilibrium = maximum entropy; in an economic system in equilbrium, all possible irreversible processes have been completed.  SELF ORGANIZATION is order in non-equilibrium states - where entropy inside a
system is lowered by increasing entropy increase outside the system (thus net entropy increases) . . . this cannot be examined using equilibrium mechanics.

See Sethi (2006) Statistical Mechanics: Entropy, Order Parameters, and Complexity.  Thanks to a lecture handout on graphical models here.

I will now stop talking about things I don't understand for a while and put down sensible posts.

14/04/2011

Idea

For PhD as my proposals stand irt solving search models with microscopic simulation:
  • Define a lattice and relevant objects (agents)
  • Allow searching agents to randomly change location every period; this will give an ex ante objective probability distribution to finding other agents (trade opportunities, jobs) based on relation of searching agents to non-searching agents to lattice size etc
  • Use the above method to estimate an analytically tractable search model (eg basic job-search in Pissarides (2000))
This probably won't work perfectly, as the rate at which worker find jobs/jobs find workers will become endogenous in a presumably complicated manner, which it isn't in the basic job-search model.  However this is only an aesthetic problem . . . the method should be easier (certainly more intuitive) than searching for fixed points in the parameter space, and should also improve upon these methods by allowing the balance-sheets of each individual agent to be kept track of.  This will be useful in search-theoretic models of financial intermediation (my actual topic). 

Things needed prior to completing the above:
  1. Learn search-theory
  2. Learn object-oriented programming
  3. Take classes in combinatorics 

13/04/2011

Larry Summers & Martin Wolf, INET Conference

Two brilliant passages from Larry Summers, being interviewed by Martin Wolf at the 2011 INET conference at Bretton Woods (thanks to beyondmicrofoundations for pointing out the video):

"We have a broad social problem that covers everything from finance to deep sea drilling to nuclear, that in all kinds of areas that are technical and hugely important to society, there is roughly nobody who knows about them who does not also have some sort of deep interest in them, and that creates all kinds of questions of legitimacy and knowledge, in that we don’t want regulation by the co-opted, but nor do we want regulation by the ignorant . . .  there is hardly anybody who is both knowledgeable and uncoopted."
"I find the idea of ‘expansionary fiscal contraction’ in the context of the world in which we now live, to be every bit as oxymoronic as it sounds, and I think the consequences are likely to be very serious for the countries involved . . . but I’ve always been a believer in being an empiricist about my convictions, so I would be happy to say that if Britain enjoys a boom over the next two years coming from increased confidence I would be required to radically re-think my view as to how the macroeconomy operates and to be quite contrite about the seriousness of my misjudgement.  You can make a judgement, those of you who know me, how likely I would be to put myself in a position of great contrition, and you might therefore conclude that I am fairly confident this experiment is unlikely to work out well!"

Normal Distribution

Yesterday evening I found out that the normal distribution was discovered/invented by Gauss in order to justify the OLS method.  It was then shown to be the maximum entropy distribution for any distribution with a given mean and variance, the central limit theorem was proved, &c, and then found its way into theoretical physics (as any gas in equilibrium is in its maximum entropy state, the particles' velocities must be normally distributed as this the m.e.d.).  The history of science is full of wierd stories like this!  I think Feyerabend would have enjoyed it . . . I'm personally still quite attached to a 'logic of scientific discovery', however . . .