Sunday, January 29, 2012

More on the taxation lie

Last year, I posted a piece about what I called the "taxation lie that won't go away." Simply put, the idea--the lie--is that by increasing a tax rate, there will be a directly proportional increase in tax revenues for the following  years. And the corollary of the lie is that decreasing a tax rate will lead to a corresponding decrease in revenues for the following years. This is, of course, a principle argument for those wishing to see a repeal of the Bush era tax cuts (for the upper income earners, anyway) and/or the creation of a "millionaire tax."

In this simplistic and tragically flawed model, the economy is like a computer program: change an input and the output changes, according to some mathematical formula. Thus, by applying this formula in reverse, liberal politicians, pundits, and economists are able to proclaim that the Bush tax cuts have cost the government some amount of dollars. Here's an article from October of last year that does just that. The number is the total cost to the government since 2001 and stands at a staggering $1 trillion dollars. But again, it's just plain nonsense.

To understand why that is, why the economy does not work like a computer program, why revenues cannot be predicted with simple mathematical formulas, it is important to understand what the economy really is and what it really is not.

It is fair to call the economy a system, though it is not a system that was simply set down in pre-ordained fashion. Rather it is a system that has developed over time. This is true for both the world economy as a whole and the U.S. economy in particular, but for now we are primarily talking about the latter.

Now, there are many different kinds of systems, from eco-systems, to air conditioning systems, to computer systems. But all of these systems can be classified in a variety of ways. One important--nay, critical--classification is whether a system is closed or open. A closed system is basically self contained; nothing is exchanged beyond the bounds of the system, changes are wholly internal, and therefore discoverable and predictable. An open system is one that is constantly interacting with things (other systems) outside of itself, one in which unpredictable external change can impact the system at any time. Additionally, for closed systems parameters are limited, as a matter of definition. For open systems, they are unlimited.

Classical--or traditional--economic theory treats the economy as a closed system. It has to do this, in order to theorize and graph things like market equilibrium. But the economy is not a closed system, at all. As we know, outside influences are ever-present, from other markets and economies at a policy level, to basic changes like capital investment, tourism, and--yes--immigration. In modeling the economy--to predict it's future--these outside influences must be accounted for. But can they be, really?

Because not only is the economy open, it is also complex. Non-complex systems are made up of interconnected parts and exhibit identifiable overall properties, based on the properties of the various parts (and therefore predictable properties). In contrast, complex systems exhibit properties that cannot be anticipated from the various parts of the system. This is, of course, a huge problem for people that would like to model the economy. In a way, it is related to the idea of unintended consequences. Changing something in a system--like say, a tax rate--can (and does, really) have consequences outside of the essential components that the change addresses(taxes paid by the individual and taxes collected by the government).

It is not necessary to accept some sort of trickle-down economic theory to understand the validity of this point. The complexity of the system means that variables assumed to be constant in a traditional economic paradigm really are not. Thus, any model that makes such assumptions is fundamentally flawed from the get-go. How flawed is it, though? Well, the simple rule of thumb here is that the larger the system is--if it is complex--the less trustworthy such assumptions can be.

So, what we have--in the economy--is an open, complex system (much like the global eco-system). And such a system simply cannot be modeled with any degree of accuracy by using basic mathematical formulas and concepts. In fact, the system is not functional in nature, at all (meaning direct input/output equations are next to useless). Instead, the system is algorithmic; every change or input has a range of possible outcomes, based on a map of some number of steps with a variety of choices.

Eric Beinhocker addressed many of these ideas in his 2006 book, Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics. In it, he notes that the economy is much more akin to an evolutionary system than it is to anything else. And he has the right of that, I think.

But the point here is that people who think they can directly measure something like the cost of a tax cut are simply mistaken. They've deluded themselves into believing the economy is essentially a simple thing, entirely predictable, with basic near-constant features. And this delusion--now centuries old--has created the ideas of economic planning and economic policies, the idea that the government can control and direct the economy with precision, with minimal error.

If the past three years tell us nothing else, they tell us that this is pure fantasy.

Cheers, all.


  1. Nicely done. Another factor that economic analysis fails to take into account is Bastiat's "that which is seen and that which is not seen." Even if you accurately predict the impact on the economy of collecting X in taxes for purpose "Y," there's no way to predict what won't happen because those tax dollars have been diverted from use "A" to use "B". That impact is unknowable.

  2. I don't know. While I agree that predicting things exactly is problematic, some empirical observations are possible (and, of course, basic laws of mathematics can't be violated). For example, take the Houser law. I think it is fair to say that we have enough data to state that as long as changes in the tax code are around the edges (not a fundamental revamp), the revenue will remain within certanin bounds (at least for the most part). You can also draw some predictions within those bounds based on observed correlations -- i.e. gdp growth rates (or possibly first derivative of this rate) drive unemployment and revenue (again within certain bounds). Such observations can dictate policy directions etc.

  3. But Dm, there's no way to say--with any certainty--where those edges are; you can't test for them and no formula can provide them. It's like the Laffer curve. It may be a reality, but no one's actually constructing the real one, anytime soon.

    That said, looking at such things is certainly useful. But I would argue that all either really do is confirm that incentives are all that matter.

  4. I disagree. Since the 40th the revenue remained around 18% of GDP. It went above 20% 2 or 3 times in all this time, and went below 16 in cases of severe recessions. Granted, this still leaves a lot of wiggle room, but _some_ predictions are possible. As to your last sentense, I said nothing about causality and such. Empirical observations are what they are -- observations.

  5. Fair enough. But I'm still opposed to constructing policy based on those observations, as if outcomes can be positively predicted in that regard.

  6. Well, it depends what you mean by "policy" :) Given the above data, it is clearly unwise to plan your spending at 25% of GDP. Also, given some correlation between GDP growth rates and revenue, a general approach would be to spur growth if you want to increase revenue (this would be true even if revenue remained completely flat as a percentage of GDP). Obviously, these are just broad strokes how to approach things. Such attempts, by the way, at least try to take into account the dynamism of the system as opposed to static calculations you made fun of in the op :)