Do Individuals Measure Value? Mises vs. David Friedman
You guys may remember that a a week or so ago, I asked for help in sourcing a David Friedman quote, that Bryan Caplan had put at the top of his blog post. That led to…
My latest blog post at the Independent Institute. An excerpt:
As the quotation above indicates, Mises agrees with Friedman that people place all satisfactions on a single scale of value. However, I still maintain that Mises would not agree that these individuals aremeasuring value.
Now at this point, we can imagine Friedman pushing back by asking, “What’s the problem? If we evaluate the potential satisfaction from a work of art, along with the potential satisfaction from a pizza, using the same scale of value, then doesn’t that mean we are implicitly measuring value in pizza-units as well as in art-units? Hence my claim, that economists ‘think everything can be measured in anything’? Use an analogy with length: If you agree that we can place a baseball bat and a pencil on the same scale of length, then can’t I measure baseball bats in terms of pencils?”
But to repeat my original claim, the mere fact that we can place disparate satisfactions on the same value scale is not the same thing as saying we can measure the subjective value of the different satisfactions.
I must confess, it took me a minute to isolate exactly what was wrong (from a Misesian perspective) with Friedman’s discussion, since at first glance it seems like Friedman is just making an obvious “economistic” point. Yet he was smuggling in the concept of measurement, which does not follow from the fact that people put, say, lifespan and chocolate cake on the same scale. (And I grant you, even that term “scale” is probably dangerous–it would be better to use more ordinal language, like “preference ranking.”)
Incidentally, David Friedman and I have had this debate over cardinal vs. ordinal utility before. I am sure he would push back against Mises/me and say modern economics does too rely on cardinal utility.
If you can find a transaction where a party exchanged 100 pizzas for one painting then we could measure that painting in pizza-units and simultaneously measure the pizza in art-units.
Two problems come out of that: the next transaction might be different, and you still cannot use multiplication and division (no reason to presume that half a painting would exchange for 50 pizzas). You could say the value of the art is a floating thing, and changes all the time, but that seems a bit dissatisfying unless you feel good about everything as a retrospective tautology (it is what it is).
If you have a busy marketplace and a lot of art is changing hands around the pizza oven, then you can start with time series, statistical models and stuff like that. Those type of metrics do start to approximate cardinal numbers, with a better approximation the more liquid the marketplace. You can start making some future estimates on that, “Hey I’m opening a new pizza joint downtown, it’s gonna need three paintings in the dining room, so at current market rate how many pizzas do I need to produce?”
No, you would only have the market price for paintings in terms of pizzas. You do not know if the party that gave 100 pizzas were willing to give 150 or more, for example, or if the other party was willing to accept less than 100.
If your measurement is not based on real transactions, then what actually are you going to measure?
What does it mean, from an epistemology standpoint, to “measure” something that never happens?
I would argue that in a market with reasonable turnover, a supplier can map out the local slope of the demand curve, by injecting some small noise into the price they offer and then seeing what happens to volume of sales. This is again a statistical measurement based on real transactions, and you need a statistically significant number of transactions to achieve that. IMHO, measurement of potential parts of the demand curve that are far from current market price is impossible, and if you want to stick within a strict empirical framework you have to say those things simply don’t exist.
Let me give an example, so you have a driver who is trying to get from point A to point B in a city and the driver has a good map of the city. Of course there could be many thousands of possible routes, but the majority of those routes are highly inefficient and slow, driving around back streets, etc. The driver ignores most of that junk and quickly narrows it down to three different routes that are all plausible in terms of time and distance. At the same time Big Brother has a satellite GPS tracking system monitoring this vehicle and what data is collected here is exactly ONE route which is the final decision that the driver goes ahead with.
So did the two alternative routes not taken ever really exist? From the point of view of measurement, no they did not. They never happened. They might have happened, but they didn’t. Indeed, the thousands of intricate circuitous routes that the driver never even thought seriously about might have happened. I mean those were remote possibilities, but they still never did happen, so none of that can be measured either.
“If you can find a transaction where a party exchanged 100 pizzas for one painting then we could measure that painting in pizza-units and simultaneously measure the pizza in art-units.”
But you are not measuring either. Such measurement requires a universal standard of reference.
An exchange of 100 pizzas for 1 painting is not measuring the value of either the pizzas or the painting. Each individual has a different set of valuations (what you are calling “measurement”) of the same goods.
Exchanges are not set against a single standard of value or measurement. They are set against TWO such standards, namely, the two parties in the exchange.
You have to stop looking at the goods for a final answer and start looking at the people instead. Valuation is a human action. It is not a description or predicate or characteristic of material goods.
Good post.
Speaking of saying things in a careful and nuanced way to avoid reader misunderstanding, personally I would type this statement:
“But to repeat my original claim, the mere fact that we can place disparate satisfactions on the same value scale is not the same thing as saying we can measure the subjective value of the different satisfactions.”
To read:
“But to repeat my original claim, the mere fact that we can ground disparate satisfactions on the same individual actor is not the same thing as saying we can measure the subjective value of the different satisfactions.”
This I believe emphasizes the fact that just because you can link various satisfactions to some common ground, it doesn’t mean there is any objective magnitudes separating them.
I picture a large perfect sphere (individual actor) surrounded by many smaller spheres (thoughts/satisfactions) orbiting it, with each smaller sphere having no fixed location in 3 dimensions.
Didn’t Bryan claim in his “why I am not an austrian” that modern econ did not rely on a cardinal concept of utility contra Rothbard? Might be worthwhile to contrast this.
I think his point was that because one can parse out ordinality from the mainstream “util” mythology, that this somehow means it is not dependent on any presumption regarding the existence of cardinaloty or utils.
Personally I don’t buy it, because the way it is taught, is presuming cardinality and utils.
Sometimes economists blur the line and treat utility as cardinal, but as far as I know in their original formalization Arrow and Debreu were careful to emphasize that preferences were ordinal. The exact utility function used doesn’t matter as long as the ordinal preferences are preserved. In fact, economists often use transformations that preserve the order of preferences in order to simplify analysis. They wouldn’t be able to do this if they treated utility as cardinal.
Ok, imagine Billy (in the friendship analogy) has “friendship units” with which he measures his friendship with Bob, Tom and Sally. Imagine Bob has 100 FU, Tom has 70 FU and Sally 60 FU. Now we know for sure that Billy prefers to be with Tom and Sally together (ok, we must make many other assumptions, but you get it) than to be with Bob, which is arguably his best friend. However, if Bob has 150 FU, this will not be true anymore.
Even in the real world, where friendship units don’t exist, these two situations are possible — Billy may have a scale of preference that resembles:
(A)
1. Bob and Tom
2. Bob and Sally
3. Tom and Sally
4. Bob
5. Tom
6. Sally
Or it could be
(B)
1. Bob and Tom
2. Bob and Sally
3. Bob
4. Tom and Sally
5. Tom
6. Sally
Now comes my question: is it possible to know that, from
1. Bob
2. Tom
3. Sally
we will have either (A) or (B), but never
1. Tom and Sally
2. Bob and Sally
3. Bob and Tom
for example?
“Now comes my question: is it possible to know that, from
1. Bob
2. Tom
3. Sally
we will have either (A) or (B), but never
1. Tom and Sally
2. Bob and Sally
3. Bob and Tom
for example?”
I’m out of my depth here, but I would say no. Tom and Sally might have some awesome comic timing together that makes them really fun to hang out with as a pair, even though they’re kind of boring individually. Bob might dislike Tom and Sally such that hanging out with him plus either of them is awkward and uncomfortable.
The idea of putting goods into bundles and evaulating them based on preferences for those goods individually or in other bundles seems to me like an interesting thought experiment, but utterly useless as a means of predicting or measuring real preferences of real humans. There are simply far too many variables involved to get a meaningful result.
Yeah, thanks. I’ve though about the Tom-and-Sally mojo together after I posted this, and it seems that we can only say the value of the bundles have ABSOLUTELY NO definite relation with value of the singles.
However, I’ve read somewhere, probably in Menger, that goods of higher orders values come from (I don’t remember the exact term) the value of the goods of first-order. Shouldn’t we expect that, for example, if I value:
1. Pizza
2. Hamburguer
I would then value
1. Cheese
2. Bread
If you ignore that this example is bad (because there are way too many variables, for example I could love pure bread more than pizza). The situation is the same, so is Menger wrong? Was even him speaking in the first place?
I don’t know if it was Menger, but the understanding that I have is that this refers to the inputs of a production process. A producer of pizza will base the price he’s willing to pay for cheese, flour, tomato sauce, etc. on the price he believes he can get for the finished pizza. I don’t think it’s meant to address a consumer’s preference for products that can be combined to make a different product.
This is in response to Tel, but I’m doing it stand-alone so it doesn’t get lost in the indentation:
Shoot, Tel, I meant to hit this point but I forgot. When Jim trades 10 pizzas to Sally in exchange for a painting, we can say that the painting has the same market value as 10 pizzas. However, there is no measurement going on here.It’s not that they each had a certain amount of “market value” inherently, and then we put the pizzas up to the painting to figure out how many “pizza market values” were contained in the painting.
E.g. when you put a meter stick against a barn and measure it as 13.4 meters, it works because you assume they both possess units of length and you’re seeing how many of the meter stick’s lengths *equals* the length in the barn.
But when Jim trades 10 pizzas for the painting, he does it because he thinks the painting has MORE value than the 10 pizzas, and vice versa for Sally. They are not measuring market value, they are consulting their subjective value rankings. They are not using equality, they are using inequality.
So there is nothing at all analogous in two people making an exchange and thereby producing a market exchange rate, and a person using a meter stick to measure the length of an object.
But while Jim and Sally are coming to a negotiation of the barter trade they must internally be judging some type of “value” even if it’s not as universal as a meter stick.
I can see two main differences between economic measurement and engineering measurement. Firstly, if I go to the hardware store and buy a tape measure then I can take this with me and use it in a whole variety of circumstances and get a remarkable level of consistency. I could go to the sporting good store and measure a baseball bat, and then I can go into a luggage store and measure a sports bag, and know that if I buy the bat and the bag, take them both home then that bat will neatly fit inside the sports bag.
I might sound headbanging, but so many people take this for granted: why can we be confident that the baseball bat does remain the same length even after we take it home? Now it isn’t *exactly* the same length, on warmer days the metal bat is slightly longer than on cold days, while the wooden bat changes length on wet or dry days… but those effects as small. I can use the same tape measure in Germany, Syria, Australia and have great confidence in a consistent result. I can even send email to Germany and some other guy pulls out his tape measure and emails me back the measurement and it comes out very close to what I would get if I measured it myself.
Now that’s a remarkable achievement when you think about it abstractly. It shouldn’t shock economists that they are unable to match this, because the whole study of physical weights and distance has been worked on for a very long time… and a less complex problem to begin with.
OK, the second difference I can recognize, is that in a good environment (i.e. voluntary, property rights, blah blah) when we see a trade we should presume a gain from trade is happening there, including barter transactions. We cannot measure the gain from trade, but we also believe that although the *concept* of a gain is universal, each instance is only part of the individual. To be more specific, perhaps Jim drives a hard bargain and Sally cannot be bothered haggling so Sally only makes a small gain, while Jim does a bit better. Those things are not portable outside the minds of Jim and Sally, you cannot carry it around like a tape measure. It isn’t even consistent because next time Sally might decide she does want to haggle and gets a better price.
You presume there’s a gain from trade, but outside that presumption the only thing you can observe is the trade itself. That’s why I said that the only empirical result you can come up with is a statistical picture of many transactions. Otherwise you are in the difficult position of insisting that non-events are a part of your reality.
I think I’ve written too many words here…
“But while Jim and Sally are coming to a negotiation of the barter trade they must internally be judging some type of “value” even if it’s not as universal as a meter stick.”
But what you are calling the singular “value” there is referring to two distinct valuations relating to the goods. There is only one yard stick set against the barn, but there are two people in the exchange. In that exchange, the two people do not have the same valuation.
“Sally only makes a small gain, while Jim does a bit better.”
You are still presuming a universal valuation there. Sally’s gain cannot be measured against Jim’s gain. There are two different and separate valuators. The gains to Sally can only be compared against the gains she otherwise could have had had she did not something else. Same with Jim. You cannot say Sally gained more than Jim gained.
Yeah, she could have negotiated harder and got a better price, some days she might be willing to put the time and effort into doing that, while on other days she isn’t.
That’s why people do haggle, because they believe there’s an advantage in doing so.
You must surely see that when it comes to price discovery, a gain to Sally would be a loss to Jim. The magnitude of the gain/loss might not be comparable, but if I could have a painting for 10 pizzas or get exactly the same painting for 9 pizzas then the 9 looks like a more attractive price.
I guess I’m presuming that neither side specifically wants to pay a higher price than necessary. Perhaps there are some special cases where that does not apply (e.g. Velben goods). The normal case would be that people prefer the same goods at a lower price. That’s an automatic internal value comparison, not an inter-personal comparison.
“It shouldn’t shock economists that they are unable to match this.”
You say that like it is a flaw in economics. But it isn’t. Amd it has nothing to do with the fact that physical weights and !measures have been around for a very long time. This is a question of logic, not insufficient data or experience. No particular experience or memory of accumulated experiences can change or affect what is logically implicit in all experiences as such.
I guess “flaw” is a value judgement, but you do see plenty of economists trying to make their discipline more like physics, while you don’t see a whole lot of physicists re-working their measurement systems to bring them more into line with economics. This would be a revealed preference in my book.
That’s a brave statement to say that all future economists will be unable to discover any new concepts or any alternative analysis. Ask that “End of History” guy about what happens afterwards.
“That’s a brave statement to say that all future economists will be unable to discover any new concepts or any alternative analysis.”
Well, the scope is incredibly modest. It is just talking about human actions.
Perhaps it is brave. But it is also brave, and possibly braver, to say that *nothing* can be known a priori about reality, and that Empiricism is the only way. I haven’t seen any indication that you are open to the possibility of true synthetic a priori knowledge. Is that not a brave position to take?
Besides, for new discoveries and new analyses, you may not see it but you are by virtue of using those terms, implying a constant unchanging logical structure of thought. Moreover, all discovering and all analyzing would themselves be actions. Whatever new discoveries are made, whatever new analyses are conducted, they could not refute the logical status of the action axioms and propositions, for those realities are engrained in those very same processes of discovery and analysis.
It is impossible to undue to the law of non-contradiction. Even so-called “fuzzy logic” adheres to it. For fuzzy logic is itself presented as “this”, but not “that”. The concept of fuzzy logic presupposes concepts are distinct.
Humans really are capable of discovering absolutely true knowledge that has never been refuted, and never will be refuted. They are modest in scope, but they are incontrovertible.
Hey, this is my comment!
Something else comes to mind. People historically have shown a tendency towards finely divisible items to base their trade around.
For example, the Romans officially used gold and silver coins as money which are obviously valuable but not divisible beyond a single coin. However, the Romans also used salt for many of the smaller transaction, especially in places where the Emperor’s head stamped on a coin might not impress the locals as much as a useful commodity like salt.
As an example, if two Romans are in the midst of a barter transaction, they might get stuck in negotiations and one party says, “OK, how about I offer this small bag of salt just to convince you to go ahead with it?” There’s an automatic presumption here that more salt is going to create a higher incentive than less salt (not necessarily linear). It’s not like one party would accept three ounces of salt but become highly offended at four ounces. With divisibility you can balance out other indivisible items.
That divisibility suggests that at least in a given situation, the overall value of the trade has some sense to it. The way people will always go for those divisible commodities for purpose of exchange, rather than using ox carts as money has got to tell you something.
Now I mention it, the well known physical units have all been carefully chosen to be linear, while human decision making is not linear in any given instance. However, a large and liquid market where many transactions occur will tend to mean this doesn’t matter so much. If you can’t sell your painting to one guy, you can sell it to someone else just as easily.
“That divisibility suggests that at least in a given situation, the overall value of the trade has some sense to it.”
No it doesn’t. What you are seeing as suggesting overall universal value, does not suggest that at all from an economics perspective. To say “cut in two, and then again, and then again, etc” does not “suggest” that you can do this arbitrarily, stop the thought progress there, and then collapse back onto some notion of universal value. You are just abandoning economic science when you do that.
Your actions are pertaining to discrete units, not infinitely divisible units.
A yardstick can in principle be sectioned into an arbitrarily large number of sublengths.
Your valuations cannot and never have and never will.
I never made any such suggestion, indeed a number of times I pointed out that this was a distinction between what you can do with a tape measure and what you can do by looking at markets.
Besides the discussion is about whether individuals measure value, not whether it is universal.
Sorry Tel, what then did you mean by “the overall value of the trade”? To say “the” value is to speak of a universal apart from the two separate instances of valuations of the exchangers.