Abstraction and time in economics

Economics Economics Questions
, , , ,
1

the following concerns a problem I’m working on for a course I am taking.

The basic question is whether the idea of non-precisive abstractions is still useful when we apply it to events over time in the form of counterfactual comparison.

The main reason that it may not be useful would be that it is impossible to isolate causal factors through time, to apply non-precisive abstractions to events. Below I explain this in some more detail

Two types of abstractions
Roderick Long argues for a distinction between two types of abstraction (Long 2006). The first type of abstraction specifies that certain things are absent in the theory or model. So for example, this kind of abstraction could specify that uncertainty is absent, that entrepreneurial error is absent, that time is absent, that choice is absent, and, more innocuously, that sellers have no hair color, have no history and so on. Such statements are unrealistic or just downright false.

The second type of abstraction does not specify that certain factors are absent but simply does not specify them. So if I for example think of a horse I can think of it as not having a color, or I consider it not as having a specific color. In the latter case I don’t say that the horse is colorless but I simply think of the horse without thinking about its color. The latter type of abstraction is called non-precisive abstraction and the former type is called precisive abstraction.

Counterfactual comparisons
A prediction is a comparison of a before-state and an after-state. Suppose unemployment in the Netherlands is currently at 5 percent. Suppose furthermore that now the minimum wage increases from 7 euro an hour to 10 euro an hour. This means that the costs for firms to employ people will stay the same (in case all their employees already earned 7 euro an hour or more) or increase (in case some employees earned less than 7 euro an hour) Our prediction says that unemployment will be greater in the new state than it was in the old state.

But it may very well be that at the time that the minimum wage was increased for example all sorts of licensing laws were abolished, making entry into certain professions easier. In this case it may very well be that unemployment will actually decrease despite the increase in the minimum wage. Does this make the economic theory that says that unemployment will stay the same or increase when the minimum wage is increased false or less realistic? Should we for example specify in more detail the conditions under which the prediction would be true?

One possible answer is to hold that economic theory is unaffected but when applied to the real world, that is infinitely more complex, predictions made with the help of theory may only point to tendencies rather than be absolute, precise predictions. This is the approach the marginalists and the evolutionary economists took: they speak of tendencies and they may try to refine the theory by building in more restrictive initial conditions or by becoming clearer about the causal mechanisms involved.

Another type of approach would take the form of for example ceteris paribus or ceteris absentibus clauses . The first type of clauses specifies that all else remains equal and the second clause specifies that all other forces are inoperative. At the risk of belaboring the point, both types of clauses then specify a condition, thereby making them into precisive abstractions. Since in reality both conditions cannot be fulfilled (things do not remain the same, and other things do have an influence) such precisive abstractions are false.

A non-precisive abstraction on the other hand would not compare the before-state with the after-state and specify unrealistic conditions pertaining to them and to the transition from the one into the other, but would instead look at the change only in terms of the influence of the increase in the minimum wage. So we would not say that everything stays the same or that other factors exert no influence. We are only concerned with this specific factor.

By using a non-precisive abstraction we make a different kind of comparison, a counterfactual one . We compare two wholly realistic states: one in which the minimum wage did increase and one in which it did not. We do not say that all the other factors stayed the same or that they did not exert any influence, but only that changes in them make no difference for the comparison of the actual scenario with that of the counterfactual scenario. The changes in or influence from the other factors is real but it is also the same in both scenario’s, so in terms of the comparison of the two scenario’s they are not relevant.

In other words, a non-precisive and thus counterfactual analysis would not say ‘An increase in the minimum wage results in equal or greater unemployment’ but ‘Unemployment will be equal to or greater than it would have been without the increase in the minimum wage’. So the comparison is not between a before-state and an after-state but between the actual and the counterfactual scenario. And the difference between the actual and the counterfactual scenario is only one with respect to the one element of an increase in the minimum wage or not.

Problem
The problem with this solution (and one Long also casually acknowledges) is the following:

It may very well be the case that the increase in the minimum wage for example prompts entrepreneurs to invest more in technology and that a new technological finding that emerges from this will exactly increase employment, or that unions because the minimum wage is increased now are a-okay with abolishing some laws that make it hard to fire people, etc.

So in these cases the increase in the minimum wage sets in motion another process that would not have occurred without the increase but that exactly results in a decrease in unemployment. In such cases even the counterfactual analysis that holds that unemployment would have been equal to or smaller in the scenario in which the increase in the minimum wage had not occurred seems to just be false.

This also points to a problem concerning the role of time in such counterfactual analyses: for what amount of time do these counterfactual laws hold? Given the strong interdependence of factors sooner or later a process may be set in motion by a certain change that countervails the effects of the change itself.

It would be interesting to investigate this topic in more detail, to see if this aspect can somehow be incorporated into counterfactual analysis.

Relevant literature:
Hülsmann, Jörg Guido. 2003. “Facts and Counterfactuals in Economic Law”, Journal of Libertarian Studies. Volume 17, no. 1 (Winter 2003), pp. 57–102.

Hülsmann, Jörg Guido. 2000. “A Realist Approach to Equilibrium Analysis.”. The Quarterly Journal of Austrian Economics, Vol. 3, No. 4 (Winter 2000), pp. 3-51

Long, Roderick T. 2006. “Realism and Abstraction in Economics” The Quarterly Journal of Austrian Economics, Vol. 9, No. 3 (Fall 2006): 3–23

2

k, so below is (first a quick summary of and then) my very preliminary response to the questions I raised in the first post in this thread. it’s quite unclear still (and possibly will remain so), but there may be something to it. At this point it’s mostly just a hunch:

So far we have seen that in our abstractions we can simply not specify the presence or absence of other factors and in that way isolate the factors we want to study. Through counterfactual analysis we can then see the effects that the factors that we do include in our abstraction have in isolation. But what happens when one of these factors A exactly sets in motion another factor B whose effects counteract the effects of factor A?

For example, it may very well be the case that an increase in the minimum wage leads unions to agree with a decrease in regulation which exactly would make labor cheaper for employers. Or it may be the case that if a central bank increases the money supply and private banks lend this new money for a price lower than that on the market to companies who finance investments that increase the productivity of labor. In both cases the effects of factor A have set in motion a factor B that counteracts the effects of factor A: less regulation means lower costs for employers which makes it possible to hire more workers; increased productivity means more or higher quality goods which exactly increases the purchasing power of the money.

In such cases even the counterfactual analysis becomes problematic for we can no longer say that for example unemployment would be equal to or higher than it would have been in the absence of an increase in the minimum wage or that the purchasing power of money will be the same or lower than it would have been without the increase in the money supply. We can no longer say this because these changes set in motion counteracting forces that would not have occurred without them. It then seems that we would have to include such possible consequences in our abstractions.

But this raises at least two questions: 1. How can we know beforehand what forces may be set in motion by a certain change that exactly counteract the presumed effects of this change? This seems like an empirical matter that we cannot solve through a priori reasoning. If this is the case then we cannot know beforehand what we should abstract from and what not. 2. For what amount of time does the counterfactual analysis hold? It seems that the more time passes the greater the chances of forces being (indirectly) set in motion through the initial change that counteract the presumed effects of the initial change.

One possible reply to these concerns is that the concerns stem from a confusion between theory and application, and a confusion between understanding and prediction, as I will briefly try to explain below.

Suppose a counterfactual economic law says that an increase in the minimum wage leads to equal or higher labor costs for employers and hence to equal or higher unemployment than there would have been in the absence of such an increase. Now we look at the real world and see that there has been an increase in the minimum wage and so we can state that the labor costs are now higher than they would have been without this increase. But we also see that unions now thanks to the increase in the minimum wage agree on a decrease in regulation which results in lower labor costs. If we were in the game of predicting consequences we would be in a pickle because it is not clear to what extent the decrease in regulation will counteract the increase in the minimum wage.

But if instead we stick to understanding and thus to applying economic laws to reality in order to make sense of the latter then we can say that we can apply the economic law until the moment that counteracting processes are set in motion due to the increase in the minimum wage. We then do not say anything about the empirical consequences at time t, nor do we make predictions about when exactly this moment will occur, but we just state that we can apply the law to make sense of the period before counteracting forces are set in motion. Over time, as I just stated, chances will increase that counteracting forces are set in motion due to the initial increase and so the more time passes the more limited the possible application of the economic law to reality may be.

If instead the decrease in regulation happens at the same time as the increase in the minimum wage (through an agreement between the government, employers and labor unions for example) then we can simply not apply our economic law: we can simply not say that unemployment will be equal to or higher than it would have been without the increase. But this too is a problem of applying the economic theory to reality, not of the economic theory itself. All it means then is that we cannot apply this law to this situation, it does not mean that the economic law becomes problematic or inexact.

In principle this problem is no different from the general empirical problem of seeing when and where economic laws can help us understand reality. Economic reality is complex and the trick with any economic theory is to ‘hook’ the features of the abstract theory onto features of reality in order to shed light on the latter. The problem of seeing where, to what extent and for how long certain economic laws can help us understand certain aspects of reality is a general one. The objections discussed in this section are specific instances of this general problem because the fact that the antecedent in the law sets in motion a process that counteracts its implied effects, decreases the scope for the application of such laws. But just as the general problem of application of economic laws to reality does not invalidate these laws or counterfactual analysis in general, these specific problems do not either. If we accept this perspective then we can say, pace Mill and Hausman, that economic laws can in fact be exact and wholly realistic.

3

I think the confusion stems from confusing praxeology and thymology. Praxeology = economic theory; thymology = application of theory to reality. As you point out: “But this too is a problem of applying the economic theory to reality, not of the economic theory itself.”

In his Introduction to Rothbard’s “A History of Money and Banking”, Salerno gives an example similar to yours (p. 22-23):

“The Fed increases the money supply by 5 percent in response to a 20-percent plunge in the Dow Jones Industrial Average—or, perhaps now, the Nasdaq—that ignites fears of a recession and a concomitant increase in the demand for liquidity on the part of households and firms. At the same time, OPEC announces a 10-percent increase in its members’ quotas and the U.S. Congress increases the minimum wage by 10 percent. In order to answer the question of what the overall impact of these events will be on the purchasing power of money six months hence, specific understanding of individuals’ preferences and expectations is required in order to weight and time the influence of each of these events on the relationship between the supply of and the demand for money. The ceteris-paribus laws of economic theory are strictly qualitative and only indicate the direction of the effect each of these events has on the purchasing power of money and that the change occurs during a sequential adjustment process so that some time must elapse before the full effect emerges. Thus the entrepreneur or economist must always supplement economic theory with an act of historical judgment or understanding when attempting to forecast any economic quantity. The economic historian, too, exercises understanding when making judgments of relevance about the factors responsible for the observed movements of the value of money during historical episodes of inflation or deflation.”

Salerno makes an important point here that economic laws are “qualitative and only indicate the direction of the effect” of different factors. So with your example of a minimum wage and deregulation, both factors produce qualitative effects in opposite directions, and we cannot say what the end result will be. Just like when supply rises and demand falls, we cannot say anything about the effect on quantity; the effect is indeterminate.

In this light, I think the temporal aspect is irrelevant. It doesn’t really matter if the deregulation comes after the minimum wage, or at the same time, or before. The important point is that the policies have opposing qualitative effects, and so produce an indeterminate outcome.

Returning to counterfactuals and abstraction: The economic theory of the minimum wage abstracts (non-precisively) from counteracting policies, e.g. decreased regulations. It just says that a minimum wage causes more unemployment than otherwise.

If the theory did include the effects of counteracting policies, then it wouldn’t be able to say what the overall effects would be. I guess we need to abstract from counteracting policies in order to have a useful theory, i.e. a theory that gives us practical knowledge.

In the end, I don’t think this is a counterfactual problem at all, but just a praxeological and thymological problem: On the praxeological level, we must abstract from counteracting policies to create useful theories. If we include counteracting policies in the theory, then our analysis will be problematic, regardless of whether it is counterfactual or not. So this isn’t about counterfactual theories, but about theories in general (counterfactual or not).

On the thymological level, the problem is applying the theory to reality. If there are two counteracting policies, e.g. minimum wage and decreased regulations, then because theory tells us the effect on unemployment is indeterminate, we have to employ subjective understanding of the situation to apply weights to the different factors (See Salerno’s introduction, p.20-23).

4

I would just say this. A minimum wage above the market wage necessarily means there will be unemployment. At the very moment it is instituted, that is, a wage above the market wage is manifested, there must necessarily be unemployment. To say otherwise would mean that there was no deviation from the market wage, thus leading to a contradiction. Of course, this is a timeless law.

Once you introduce time it becomes a bit more fuzzy. It’s realistic relevance is partially wiped out if, for example, you could say that a minimum wage necessarily sets into motion the process of a rising nominal market wage (if the min. wage is stated in nominal terms). But still, for that period of time in which the min. wage deviated from the market wage there will have been unemployment which would not have existed had it not been put in place. So you would then have to say that it (min. wage) would somehow cause even less unemployment beyond the moment in time in which the market rate had surpassed it than would have attained in its absence. Thus, in total there will be less unemployment with the min. wage than without it.

But if you give it some thought, the whole concept of unemployment doesn’t really make any sense in the absence of coercion. A free business will not propose a wage which it does not wish to pay. An individual will only forgo a given wage if he prefers being “employed” in whatever else he is doing.

Unemployment and a wage being forcefully held above its natural level are mutually exclusive. Thus, in the absence of minimum wage, or any form or regulation which holds wages above their natural level, there can be no such thing as unemployment. Unemployment is a concept which only makes sense in the context of coercion being exacted on wages.

The only instance in which I can think of where something like a min. wage or any type of regulation could be argued for, is if you were to necessarily show that its implementation would eventually lead to less overall coercion. I don’t think this can be done though. You run into the trouble of already assuming the existence of regulation, meaning that things become really hairy.

5

thanks for your reply and apologies for the delay

Okay, I think I see what you mean: because economic laws are only qualitative anyway, the fact that there may be countervailing forces in operation means that we cannot say anything about the outcome anyway because it would require quantification to see what force would be stronger than the other and hence what the outcome is. The temporal aspect then is irrelevant: my problem concerned the situation in which a cause sets in motion anther, countervailing, cause that would not have occured without that first cause. This qua problem of quantification is similar to the problem of countervailing forces in general.

Is this an accurate summary of your point?

If so, then I think there still is a problem in terms of realism. We (or I in any case) want economic propositions to be true of something, of a realistic situation in a realistic world (so not ceteris paribus or ceteris absentibus world). Counterfactual analysis is an attempt to preserve such truths in the sense that it uses counterfactual (and actual) realistic worlds and uses a comparison between the two to give meaning and truth to the economic propositions (unemployment is equal to or higher than it would have been in the absence of an increase in the minimum wage).

My problem is not so much not being able to predict an outcome (let alone do so with just praxeology and without thymology) for which the quantification problem indeed seems insurmountable, but to give a realist interpretation of economic propositions, to make them true of a realistic situation, and whereas predictions do so in the form of a before-and after-state, counterfactual analysis does so in the form of a comparison between the actual and the counterfactual world. And here a situation of (for lack of a better word) ‘simultaneous determination’ (such as in the union negotations) or a situation in which a cause ets in motion a countervailing cause that would not have occured without it, poses a challenge because even with counterfactual analysis it is not possible to isolate the factors over time. My solution then would say that this problem is in fact a special instance of the general problem of the application of theory to reality, of seeing the circumstance in reality that conforms to the elements of the theory.

I could be totally missing your point about how temporality is not relevant to my problem though.

or just a realistic interpretation. Again, I’m not so much concerned with being able to predict anything, but with being able to make economic propositions true of realistic worlds. Abstracting from counteracting policies that were set in motion by the original cause (either at the same time (e.g. union negotiations) or afterwards) seems impossible over time, so then I would say that this simply means that a counterfactual interpretation of the economic proposition can only be applied up until the point where the countervailing force set in motion by the original force starts. After that (and with ‘simultaneous determination’ this is instantly) After that we can no longer say that the proper interpretation of the economic proposition that ‘an increase in the minimum wage leads to equal or higher unemployment’ is true in a counterfactual sense.

But this too leaves a problem because we would still want to say that there is a force operative (as described by the economic proposition) that is relevant to understand or make sense of the situation (even though we don’t pretend that we will be able to predict an outcome). So how do we give a realistic interpretation of that?