Thursday, February 28, 2008

Civil disobedience

Arnold Kling writes:

The idea of civil disobedience is that it allows a minority to confront a situation that the majority of people are passively tolerating. It allows the minority to show intensity of feelings, and it forces others in society to look at the issue and choose sides... Another way to think of it is that there are multiple equilibria in politics, and in social norms in general... And maybe the only way to [change the] equilibrium is through civil disobedience.

The way I see it, the issue behind civil disobedience is one of the transaction costs needed to coordinate the actions of a group who already agrees with an idea (or perhaps is willing to agree with it if exposed to it). It has little to do with changing the opinion of the majority (in fact, that group could already be the majority itself). Good ol' Coase, good ol' collective action.

Let me explain: let's say that

  1. Change requires a simple majority (we will lift this assumption later).
  2. There is a minority with a true "better idea:" a blueprint for an alternative set of institutions that leads to a Kaldor-Hicks improvement.
  3. This minority is failing to convince enough of the electorate to demand a new course through the existing mechanisms.

The problem could be that: (a) the majority is not interested in paying the cost to become informed, but would agree if informed; or (b) is informed and agrees, but fails to act because of the usual issues of collective action augmented by ignorance with respect to the size of the group (even if we all acted, would it be enough to generate change?); or (c) would disagree if informed or is informed and disagrees.

If the problem is (c), then no amount of civil disobedience would lead to much. Perhaps the majority is truly irrational (a la Caplan's "Myth of the Irrational Voter") or perhaps they are rational but mistrust the new institutions to (re)distribute the Kaldor-Hicks welfare improvement in a way that benefits them (could that be called "a constrained Pareto improvement"?). This is precisely the case in which the majority approves of state repression of those violating the rules. If their minds can be changed, it's not likely to be through civil disobedience. End of story. (We do it too: nobody thinks that every thief is a Robin Hood.)

But if the problem is either (a) or (b), it is then equivalent to one in which a majority already exists, albeit one suffering from transaction costs that are too high either at the information stage or, being informed, in overcoming the usual barriers to collective action/public good provision. I would further argue that (a) is really caused by (b): why would an agent choose to remain ignorant if not because he or she cannot see a positive return to acquiring the information? So I will concentrate on (b).

And in this situation, civil disobedience is one of the many alternatives used to overcome collective action issues, much like talk radio or grassroots movements.

Say a small group with a low activation threshold protests and gets arrested; if their case makes it to the evening news and is seen by a larger group with a higher activation threshold, this might be enough to signal to them the existence of others and to inform them of coordination mechanisms (like a march on Saturday to show support for those arrested). From then on, the meme can spread through coverage of each successive (and larger scale) act, and so activate other agents with even higher thresholds as two things happen: information about the size of the group is revealed (as long as it is revealed to be large enough to show that a cost-benefit analysis of taking action can be passed in the positive); and coordinating action becomes less costly through the spread of high-profile information about what, where, and when to do.

Here, we can lift the assumption of a simple majority to enact change: large enough minorities can exert sufficient pressure on legislators when they are better coordinated than the majority, as any lobbyist worth his/her salt knows.

This poor-researcher's model might be enough to suggest when civil disobedience is more likely to be effective as a mechanism for social change. There needs to be a group large enough to enact change, but ignorant of its own size and facing informational costs that make coordination mechanisms too expensive. There can either be heterogeneity within this group with respect to the activation thresholds, all the way down to some individuals willing to pay a really high cost to see change enacted; or some random shock on an agent's environment that leads this one to act first. Finally, there must be a medium which allows information to spread cheaply once someone gets the ball rolling.

Two things are missing from this model: first, details about the characteristics of the medium and the message. Let me go out on a stereotyping limb for a second. Blogs are cheap to consume and can spread quickly, but they transmit information with a (relative) low emotional quotient as it must be read and thus relies on a (relatively) higher appeal to reason compared to emotion. The evening TV news are also cheap to consume and can also spread quickly, but they are transmitted through moving images of the real world. This allows them to appeal more directly to emotion and rely less on reason. And on top of all, there are the selection issues involved in who relies more on blogs and who more on evening TV news. Through which mechanism is civil disobedience more likely spread at an early stage, at an intermediate stage, at a late stage? I would place my bet on blogs triggering the pioneers into action and then TV kicks in to spread the call to action massively.

The other missing element is, of course, something to set the whole thing in motion. Again, I'm totally guessing, but I venture two triggers. In one scenario, there are low-activation threshold agents acting all the time, but the technology is not there to spread the news about it to inform the troops it's time to rally. Along comes a technological shock such the printing press, radio, TV, or the Internet that makes this cost-effective. (The exogenous shock could be something subtler, like a change in the news market that makes it more likely that this type of stories will be transmitted; for example, a change in ownership rules that make it possible to transmit more local news.)

In the other scenario, ideas and rhetoric become important again: the low-activation threshold agents are those who have lower costs in acquiring information (or in being swayed by it). Along come the intellectuals or the marketers as an exogenous shock that sets the whole thing in motion.

It is in this sense only that civil disobedience can disguise itself as a way to change the social consensus; but I would still argue that it does no such thing as it needs an audience receptive to the message, just waiting for enough of a spark (from the low-action threshold agents) to show to the its own self that it is out there and that it is large enough to merit action.

Hmmm... maybe I should write a paper... But no, I can't: I must go and file my taxes like a responsible, law-abiding patsy.

CLICK to go on reading "Civil disobedience"

Wednesday, February 27, 2008

More on the tired Aggregate Supply curve

As a total n00b in the blogosphere, it's nice when one finds that the big shots and little-oneself are aligned. A little under a week ago, I typed my thoughts on the slowing of the AS curve, as opposed to AD-based slowdown or even stagflation, and the policy implications of seeing things this way. And today... Samuelson himself writes on The Specter of Stagflation! OMG, I feel like a groupie watching the concert from the backstage.

p.s. Thanks to Greg Mankiw for the pointer!

p.s.2. But seriously, the one point where I'd disagree with Samuelson, at least in emphasis: it seems to me that the current situation is one of an AS curve expanding more slowly than the AD curve. Inflation is not, right now, not yet, strictly speaking a monetary phenomenon, but one driven by rising input costs (directly and via imported goods). We may, in fact, still be in the arena of price shocks rather than inflation itself. It's might seem like picking at straws, but there it is: not all periods of increasing prices are really inflation, but when rising prices get entrenched in expectations, it's inflation alright. So it's from here on when the Fed's policy (specially when combined with an expansionary fiscal policy) can drive these rising prices into inflationary expectations or not. And then it would be a monetary issue, and be stagflationary, and be another case of "the more something changes..."

p.s.3. How do y'all think Bernanke is evolving on this issue? Answer's here. Some interesting (and scary) back-of-the-envelope results by Larry White.

UPDATE (a few hours later): As usual, James Hamilton illuminates the issue... and did so six days ago (he's the one who makes me wish I had studied more time-series metrics). Need more time to keep up with'em blogs!

CLICK to go on reading "More on the tired Aggregate Supply curve"

Monday, February 25, 2008

Decoupled or Decoupling?

The global impact of the U.S. slowdown indicates that the U.S. economy still exerts influence over the rest of the world, despite predictions that other countries are "decoupling," Johnson said. [Simon Johnson, economic counselor and director of the research department at the IMF.]

All the talk in the recent weeks about decoupling along the lines of "now that decoupling has been proven dead..." is making me think I've missed something. The argument seems to be: markets in the rest of the world are reacting to negative news in the US, so the rest of the world is not immune to US jitters, so there's no decoupling.

Wait: if decoupling meant literally complete detachment, well, then, yes, it's a concept in line with other embarrassments of recent history like "real estate prices can never go down" and "the dot come revolution has changed the laws of (asset pricing) nature." Only true autarky from the US economy would achieve this.

So I was under the impression (misunderstanding?) that this was not a binary issue, rather a question of degree: "decoupling" is a process, "decoupled" is a destination, and we're "decoupling," not "decoupled." On a politically-loaded parallel, it's the difference between "drowning" and "drowned," but let's not go there. It means lower, not zero (nor negative!) correlation.

That's why decoupling can coherently co-exist with globalization (and in fact be spawned by it). Decoupling makes a duh-type of sense (as in "is at least plausible") if taken to mean that the relative importance of the US as a source of net demand for other economies' outputs (or as a financial market accessible to them) has fallen, on average, for countries around the world, both because of the emergence of new regional powerhouses AND because of the expansion of their own domestic markets. (And that's an "if": this statement might, of course, be proven wrong by the data by looking at trade and capital flows at different levels of geographic aggregation.)

In fact, the US could be the most important partner for every other country and decoupling still hold: the US just needs to be less important (on average) than before.

I see it as "Globalization, part II": the US' insatiable appetite for consumption (and debt) allowed other economies to grow; as access to the larger US market allowed them to expand and thus exploit competitive advantages and economies of scale, these economies began to link with each other, to produce for their own and their neighbors' rapidly expanding markets; and now that the US is ailing, they will most certainly slow down their expansion since one (still very important) source of demand for their output is weakened, but not so much as they would have slowed if the domestic and non-US foreign drivers of expenditure hadn't increased in importance.

Instead of thinking in terms of the old saying "if the US catches a cold, the rest of the world catches pneumonia," we should switch to "if the US catches a cold, the rest of the world sneezes and feels rather chilly, but still manages to go to work."

UPDATE (Feb 27th): Here's John Authers, Investment Editor at the FT, pointing out, towards the end of this vlog, that decoupling is alive and well.

CLICK to go on reading "Decoupled or Decoupling?"

The Oscars: A disappointing result (oh, really?)

No, I don't have a thing against No Country for Old Men or against the Cohen brothers. On the contrary. But Best Animated Feature to Ratatouille when Satrapi's Persepolis was in competition? Don't get me wrong: Ratatouille was fun. Cute. Entertaining. And then out of mind. Persepolis is a work that, at the very least, asks you to ask questions. And the animation style is a delight. One can, of course, feel some smug superiority over the self-serving, $-eyed Academy, and feel one knew better. I know I do, but shhhh, keep it secret.

If you haven't experienced Persepolis: (a) go get the graphic novel (and, if possible, read it too); and (b) watch the movie. Persepolis is really two novels with different angles (although you can now get them in one volume). While both books are autobiographical and, thus, about emotions embedded in historical circumstances, the first book, the more historical/political of the two, focuses on the fall of the Shah and the rise of the Islamic regime; the second book is a heart-breaking piece on alienation (and, if you've spent a prolonged puberty crossing cultural lines while trying to get a degree or two, it is very likely going to resonate).

CLICK to go on reading "The Oscars: A disappointing result (oh, really?)"

Friday, February 22, 2008

AD vs AS

I'm having an issue with the pro-fiscal stimulus talk: it seems to be based on the assumption that there is insufficient aggregate demand. Here's one example from a source I deeply respect:

In particular, when the economy is particularly weak, the key constraint on short-term economic growth is demand for the goods and services that firms could produce with existing resources. (source)

But insufficient demand cannot be the (main) cause for sluggish growth when combined with rising inflation.

Granted that a (dwindling) minority might want to say "if combined with rising inflation", but the non-inflationary scenario is weakening. Here is a view in favor from the Fed via FT; view in favor from the financial markets (a little old by now, relatively speaking); and yet another view in favor, this from the Economist making the argument that commodity prices are not going to be headed down anytime soon. Then we have the jump in wheat prices, with all the linkages that has (geez, thanks misguided subsidies for corn-based fuels). And I will leave it with the falling US$ and rising costs in China, meaning more expensive imports in the US.

On the other hand, here is a good view against.

I am obviously on the side of the inflationary scenario. Let's accept inflation is a problem. Now, to the basics: in an aggregate supply (AS) and demand (AD) model, a drop in AD lowers both price level and output; a drop in the AS lowers output, but increases the price level. Here's the classic graphical representation of the model. In the long-run equilibrium, the AD and the (short run) AS intersect together and with the Long Run AS (LRAS), which is insensitive to the price level (and assumption to be challenged later). We alternatively contract the AD and the AS and see that it's the second case that mimics what we're observing in the economy:

Of course, it's best to rethink these graphs in dynamic terms and, with a lot of hand waving, substitute inflation for the price level and output growth for the output level. In this case, the curves' positions measure their expansion (the "AD dot" and "AS dot"). I know, there's more to it, but bear with me: the point is that, while both curves may be slowing down, the aggregate supply has to be slowing down more if inflation is rising. Rising input and import prices and the cost of credit will do the trick. Also, rising inflationary expectations would increase expenditure today and reduce supply today, specially if nominal input prices remain constant, because of, say, contracts (buy at a lower relative price today, sell at higher relative price tomorrow).

(Of course, the AD might be contracting too because of fears on unemployment, credit constraints, etc. The point is that the AS must be contracting more.)

The issue is not moot. If the AD is the limiting factor, expand it until it can sustain itself (leaving aside the serious issues of timing and overshooting, and of fiscal deficits). But if it's the AS...

An AD-expanding package like the fiscal stimulus will give a temporary boost to the rate of expansion and then take that boost away. Inflation and growth will be higher for a bit than without it. But what happens next will be a return to long term equilibrium growth in output and, very likely, higher equilibrium inflation. How much higher will depend on what happens with the causes behind the slowing expansion of the AS curve and with inflationary expectations.

In the model above, picture the AS curve moving rightwards (say due to a decrease in nominal input prices, with the price level constant), or the AD curve moving also right (an increase in expenditures at all price levels), or both, until the three curves meet again. Along the way, the price level and output (or inflation and growth) adjust accordingly. In the case of the AS, movements along it up and right are due to a fall in the real prince of inputs (their nominal price remains constant, perhaps due to contracts, and the price level as a whole goes up).

How do we return to equilibrium? If the causes of the contraction of the AS go away before higher inflationary expectations settle in (or if other AS-expanding trends, like productivity growth or an increase in the stocks of inputs, speed up and counteract the contraction), then we are in the happy scenario: we could return to essentially the same long term equilibrium we would have had if the AS curve would have adjusted through unemployment's effect on input prices. So start from the long term equilibrium, AS contraction, now the stimulus (AD expansion, say up to an intersection on the LRAS, and now a well synchronized contraction of the AD as the AS expands so that we slide down the LRAS to the original point. If it can be achieved, we could stay at full employment with only a temporary spike in inflation which would settle down to its original level. Tough, but commendable, the job of economic titans and believers in the efficacy of governments.

But what if what caused the AS slowdown does not go away and higher inflationary expectations settle in? Then the AS does not expand and the AD does not contract back. We end up in a higher inflation equilibrium with the same long term growth as before. And before someone jumps up and says "so what if we have higher inflation, as long as full employment is back?": beyond the inefficiencies inflation brings to the price system (more on that later), keep in mind that movements along the AS curve occur because the real price of inputs is falling as inflation rises faster than wages. That is, in the end, workers are in the same position, real wage-wise, as if the return to long-run equilibrium came about because the expansion of the AS curve due to falling nominal input prices, holding the price level constant. And I say "workers" because, with all others input prices rising, it will be real wages falling disproportionately.

I'll hold on a second to say something in favor of the higher inflation scenario: the return to full employment (via lower real wages, one way or another) is usually faster when it's due to rising price inflation (over wage inflation) than when we wait for nominal wages to fall (or, more realistically, for wages to remain constant given a strictly positive rate of inflation).

But unfortunately, inflation does have a long-run cost. In the long run, perhaps a year from now in this case, workers are likely to end up worse off in the higher inflation scenario. As has been convincingly argued and experienced in high inflation and hyperinflationary environments, the long term AS curve could have a negative slope because inflation reduces the efficiency of the price system. Then we'd end up in a long term equilibrium with higher inflation and somewhat lower growth.

Now, for the full picture: it might be the case that the long term aggregate supply is shifting left (say, due to the end of cheap imported inputs). In that case, efforts to go back to the old growth rate would just be inflationary and unsustainable. Do we need to bite that bullet? I, for one, a natural born pessimist, see us moving into the higher input cost / higher inflationary expectations scenario and, possibly, lower long term growth.

Not a bad scenario to tell the truth, there's still robust growth and lowish inflation... just not as good relative to the last decade or two, in which price-less risk (who, what, how, argh!), cheap imports and inputs, and unlimited demand for US$-assets complemented productivity increases. Even with the same productivity increases, risk-pricing should approach something reasonable now (but I'm prepared to be surprised once again by the capacity of capital markets to live in denial), import/inputs will not help keep inflation down as much, and lower demand for US$-denominated assets might lead to higher interest rates.

A most dismal science.

CLICK to go on reading "AD vs AS"

Thursday, February 21, 2008


It seems appropriate to start with a declaration of intent. First, for full disclosure: I wanted the blog to be called "The Null Space," but that name (with and without "the") was taken. Now that that piece of fascinating trivia is behind us: a nullspace of a matrix A is the subspace composed of all vectors x such that Ax=0.

What? You don't get it? While nullspaces are packed with fascinating characteristics, here's the deal: they contain all the solutions to certain types of differential and difference equations, including sexy beasts like Markov processes. Oh, yeah!

Still don't get it? Nullspaces hold the solutions to (some broad types of) dynamic problems. These little beasts (the set of eigenvectors, together with their inseparable lovers, the eigenvalues) provide information about the path, stability, and final resting place, if any, of systems represented in those dynamic problems.

And what is a blog about if not about trying to make sense of that dynamic mess out there, aim for a little more stability in your path, find your own steady state?

Hmmm. OK, that was horrible. The truth is, I really liked "the nullspace" since I heard the word in math camp of the Econ PhD program and decided I had to open a bar named that one day. I haven't opened a bar; I opened a blog.


p.s. since you were just dying to know, "kernel" means the same in linear algebra... but not in all branches of mathematics. See how nuanced and insightful this blog is from the get go?

CLICK to go on reading "Hazing"