**In 1953, Milton Friedman set out to establish “The Methodology of Positive Economics” to improve economics’ contribution to determining policy. He believed that “differences about economic policy among disinterested citizens derive predominant from different predictions about the economic consequences of taking action...rather than from fundamental differences in basic values. (p. 2-3)” While the goal was worthy of the task, it becomes clear quite quickly that Friedman’s normative views will shape the discussion.**

Friedman begins by explaining that “The ultimate goal of a positive science is the development of theory” or “hypothesis” that yields valid and meaningful (i.e., not truistic) predictions about phenomena not yet observed. (p. 3-4)” While few would probably disagree with that statement, a divergence of opinion tends to arise when assumptions upon which a hypothesis rests are unrealistic. However, once one understands the “canons of formal logic (p. 4)” it becomes readily apparent that a hypothesis cannot be rejected solely on these grounds. A basic proposition of formal logic is that x implicates y (x → y) is always true, unless x is true and y is false. Therefore a hypothesis is valid if the predicted outcome (y) occurs, regardless of whether the assumptions (x) are realistic. Although Friedman was only restating the principles of formal logic, arguments of this kind remain prevalent today, hence the importance of his point cannot be understated.

After dispelling the notion that a theory can be rejected based solely on its unrealistic assumptions, Friedman ventures on to the far more difficult task of applying positive economics to policy determination. At this juncture the discussion takes a distinctly normative turn, which Friedman acknowledges but downplays at the same time. The troubles with making this application are primarily three-fold:

1) How does one determine if a theory is valid?

2) How does one choose among hypotheses that are consistent with available evidence?

3) How does one decide the circumstances for a which a theory holds?

Regarding question one, Friedman points out the inherent difficulty (near impossibility) of conducting “controlled experiments” in the social sciences “foster[ed] a retreat into pretty formal or tautological analysis.9 (p. 6)” This path is clearly not suitable if economic theory “is to be able to predict and not merely describe the consequences of action. (p.7)” Empirical evidence seemingly offers the greatest potential for validating theories, but leads to the normative question of “whether [a theory] yields sufficiently accurate predictions. (p. 9)”

Moving beyond questions of validity, Friedman accepts that “The choice among alternative hypotheses equally consistent with the available evidence must to some extent be arbitrary, (p. 5)” but believes “there is general agreement that relevant considerations are suggested by the criteria “simplicity” and “fruitfulness”. (p. 5)” Friedman also mentions “logical completeness and consistency” as criteria which “are relevant but play a subsidiary role. (p. 5)” Establishing the relative weights of these criteria, let alone the relative value of different hypotheses on each scale, clearly extends into normative economics. On the whole I find Friedman’s support for his normative views unconvincing and suggest reordering the criteria as follows: fruitfulness, consistency, logical completeness, and simplicity.

Assuming agreement on a specific, valid hypothesis, one must still decide how “to specify the circumstances under which the formula works. (p. 11)” Friedman argues that “the assumptions [cannot] be used to determine the circumstances for which a theory holds. (p. 11)” Instead, the validity of the hypothesis tells us whether the theory holds for given circumstances. Unfortunately this statement is no more than a tautology and offers little guidance for applying policy to different circumstances in advance of known outcomes. Here we find the importance of a hypothesis’ logical completeness and consistency. Without knowing in advance if, and to what degree, a hypothesis will prove correct, the assumptions (regardless of how unrealistic) become a necessary guide to determining applicable circumstances.

Friedman sets out on the difficult task of developing a positive economics with normative implications. While his efforts to prove that a hypothesis must not be judged solely by the realistic nature of its assumptions were valiant, this strength of his paper has been largely overlooked. Instead, Friedman’s normative views regarding determination of a theory’s validity, choosing among competing valid hypotheses and application to specific circumstances have indoctrinated economists with a means to defend orthodox, mainstream economics from all criticism.

These departures from the “positive”, though not explicitly or (maybe) even implicitly intended by Friedman, have led to the frequent practice of three pitfalls in straight thinking:

1) Fallacies of composition - Ex. Households face a budget constraint based on income and ability to borrow, therefore governments must face similar constraints.

2) Fallacies of analogy - Ex. Greece’s sovereign debt rates are high because of its large public debt-to-GDP and Japan has higher public debt-to-GDP, therefore Japan’s sovereign debt rates should be soaring.

3) Post hoc ergo procter hoc (i.e. "after this, therefore because of this") - Ex. Lehman Brothers went bankrupt right before the financial crisis happened, therefore the failure of Lehman Brothers caused the financial crisis.

The failure of mainstream economics to predict the housing bubble, financial crisis, recession, and weak recovery, has not materially weakened the acceptance of previously held theories. This should not be surprising given Friedman’s claim that “the continued use and acceptance of [a hypothesis] over a long period, and the failure of any coherent, self-consistent alternative to be developed and be widely accepted, is strong indirect testimony to its worth. (p. 14)” Contrary to this view, advances in modeling network systems implies that suboptimal outcomes can actually be self-sustaining.

Friedman begins by explaining that “The ultimate goal of a positive science is the development of theory” or “hypothesis” that yields valid and meaningful (i.e., not truistic) predictions about phenomena not yet observed. (p. 3-4)” While few would probably disagree with that statement, a divergence of opinion tends to arise when assumptions upon which a hypothesis rests are unrealistic. However, once one understands the “canons of formal logic (p. 4)” it becomes readily apparent that a hypothesis cannot be rejected solely on these grounds. A basic proposition of formal logic is that x implicates y (x → y) is always true, unless x is true and y is false. Therefore a hypothesis is valid if the predicted outcome (y) occurs, regardless of whether the assumptions (x) are realistic. Although Friedman was only restating the principles of formal logic, arguments of this kind remain prevalent today, hence the importance of his point cannot be understated.

After dispelling the notion that a theory can be rejected based solely on its unrealistic assumptions, Friedman ventures on to the far more difficult task of applying positive economics to policy determination. At this juncture the discussion takes a distinctly normative turn, which Friedman acknowledges but downplays at the same time. The troubles with making this application are primarily three-fold:

1) How does one determine if a theory is valid?

2) How does one choose among hypotheses that are consistent with available evidence?

3) How does one decide the circumstances for a which a theory holds?

Regarding question one, Friedman points out the inherent difficulty (near impossibility) of conducting “controlled experiments” in the social sciences “foster[ed] a retreat into pretty formal or tautological analysis.9 (p. 6)” This path is clearly not suitable if economic theory “is to be able to predict and not merely describe the consequences of action. (p.7)” Empirical evidence seemingly offers the greatest potential for validating theories, but leads to the normative question of “whether [a theory] yields sufficiently accurate predictions. (p. 9)”

Moving beyond questions of validity, Friedman accepts that “The choice among alternative hypotheses equally consistent with the available evidence must to some extent be arbitrary, (p. 5)” but believes “there is general agreement that relevant considerations are suggested by the criteria “simplicity” and “fruitfulness”. (p. 5)” Friedman also mentions “logical completeness and consistency” as criteria which “are relevant but play a subsidiary role. (p. 5)” Establishing the relative weights of these criteria, let alone the relative value of different hypotheses on each scale, clearly extends into normative economics. On the whole I find Friedman’s support for his normative views unconvincing and suggest reordering the criteria as follows: fruitfulness, consistency, logical completeness, and simplicity.

Assuming agreement on a specific, valid hypothesis, one must still decide how “to specify the circumstances under which the formula works. (p. 11)” Friedman argues that “the assumptions [cannot] be used to determine the circumstances for which a theory holds. (p. 11)” Instead, the validity of the hypothesis tells us whether the theory holds for given circumstances. Unfortunately this statement is no more than a tautology and offers little guidance for applying policy to different circumstances in advance of known outcomes. Here we find the importance of a hypothesis’ logical completeness and consistency. Without knowing in advance if, and to what degree, a hypothesis will prove correct, the assumptions (regardless of how unrealistic) become a necessary guide to determining applicable circumstances.

Friedman sets out on the difficult task of developing a positive economics with normative implications. While his efforts to prove that a hypothesis must not be judged solely by the realistic nature of its assumptions were valiant, this strength of his paper has been largely overlooked. Instead, Friedman’s normative views regarding determination of a theory’s validity, choosing among competing valid hypotheses and application to specific circumstances have indoctrinated economists with a means to defend orthodox, mainstream economics from all criticism.

These departures from the “positive”, though not explicitly or (maybe) even implicitly intended by Friedman, have led to the frequent practice of three pitfalls in straight thinking:

1) Fallacies of composition - Ex. Households face a budget constraint based on income and ability to borrow, therefore governments must face similar constraints.

2) Fallacies of analogy - Ex. Greece’s sovereign debt rates are high because of its large public debt-to-GDP and Japan has higher public debt-to-GDP, therefore Japan’s sovereign debt rates should be soaring.

3) Post hoc ergo procter hoc (i.e. "after this, therefore because of this") - Ex. Lehman Brothers went bankrupt right before the financial crisis happened, therefore the failure of Lehman Brothers caused the financial crisis.

The failure of mainstream economics to predict the housing bubble, financial crisis, recession, and weak recovery, has not materially weakened the acceptance of previously held theories. This should not be surprising given Friedman’s claim that “the continued use and acceptance of [a hypothesis] over a long period, and the failure of any coherent, self-consistent alternative to be developed and be widely accepted, is strong indirect testimony to its worth. (p. 14)” Contrary to this view, advances in modeling network systems implies that suboptimal outcomes can actually be self-sustaining.

**“The Methodology of Positive Economics”**is ultimately both a positive and normative method for discerning real implications of economic hypotheses, although the line has become blurred over time. If economics is to regain its grandeur, it must recognize the strict limits of positive economics before readdressing the normative decisions posed by Friedman. In this manner, economics may come to study and accept a much broader range of hypotheses that are not only more fruitful, but also more logically complete and consistent.

**Update - Credit should be given to Ernest Nagel’s paper, “Assumptions in Economic Theory”, for inspiring a number of my thoughts in this post.**

"Friedman begins by explaining that “The ultimate goal of a positive science is the development of theory” or “hypothesis” that yields valid and meaningful (i.e., not truistic) predictions about phenomena not yet observed. (p. 3-4)” While few would probably disagree with that statement, a divergence of opinion tends to arise when assumptions upon which a hypothesis rests are unrealistic. However, once one understands the “canons of formal logic (p. 4)” it becomes readily apparent that a hypothesis cannot be rejected solely on these grounds. A basic proposition of formal logic is that x implicates y (x → y) is always true, unless x is true and y is false. Therefore a hypothesis is valid if the predicted outcome (y) occurs, regardless of whether the assumptions (x) are realistic. Although Friedman was only restating the principles of formal logic, arguments of this kind remain prevalent today, hence the importance of his point cannot be understated."

ReplyDeleteFriedman missed a sitter. x → y is true if x is false. Therefore, all you have to do to get valid hypotheses is to start with false assumptions. Piece of cake! ;)

He actually attempts this:

Delete"In so far as a theory can be said to have “assumptions” at all, and in so far as their “realism” can be judged independently of the validity of predictions, the relation between the significance of a theory and the “realism” of its “assumptions” is almost the opposite of that suggested by the view under criticism. Truly important and significant hypotheses will be

found to have “assumptions” that are wildly inaccurate descriptive representations of reality, and, in general, the more significant the theory, the more unrealistic the assumptions (in this sense).12 (p. 8)"

Obviously this is another normative view of Friedman's about which hypotheses are "truly important and significant."

Well, I guess that Friedman would have regarded Velikovsky's theory of Venus quite highly. ("Worlds in Collision" was published in 1950.) At a time when astronomers thought that Venus was wet and cold, Velikovsky correctly predicted that it would be dry and hot, based on truly preposterous assumptions. ;)

DeleteIf you read Friedman closely at page 8-9, it does not look like he is restating the principles of formal logic at all, he is actually giving a definition of 'validity' for a hypothesis (that its 'predictions' are not contradicted by 'observation') that is entirely non-standard, and he calls it the only relevant notion of validity. The standard definition would be what Friedman calls a truism. Trying to relate these two notions seems far from a canonical account of logic. 'Unrealistic' is a non-standard term too, and I would say part of the issue here is that formal logic is quite silent on the distinction between what we might like to call 'immediate observations' (which is a good definition of reality for some perhaps) and assumptions.

ReplyDeleteStill, I do think Friedman's views on page 13-15 about 'realism' are interesting and I think he is basically correct on how social science should currently proceed. Although, there has to be some tension between his idea that 'unrealistic assumptions tend to be significant' and 'theories are validated by failing to refute them with observations'. If our observations are faithful to reality (otherwise what is the point of using them to test theories), after a long enough period of observation shouldn't these start crowding out and making invalid many of these unrealistic assumptions, and all you really have left to (honestly) reason about are realistic observations? Friedman seems to imply that this would never happen, but I am not so sure.

If 'predictions' are not contradicted by 'observation', then presumably Y is valid. This is irrespective of whether the 'assumptions' are true or realistic. However, Friedman does deviate from formal canons of logic in the quote from my comment above and the following:

Delete"To put this point less paradoxically, the relevant question to ask about the “assumptions” of a theory is not whether they are descriptively “realistic,” for they never are, but whether they are sufficiently good approximations for the purpose in hand. And this question can be answered only by seeing whether the theory works, which means whether it yields sufficiently accurate predictions. (p. 9)"

The problem here is that one can only determine if the 'assumptions' are good enough based on future outcomes, if the purpose at hand is informing policy. Since we cannot know the future, it is unknown whether or not the theory works until it has been tested. But even then, if the predictions are not verified, one can claim the model was not a good enough approximation of the real world at that given time. In this manner, a hypothesis cannot be invalidated by observation either.

This gets to your second point about crowding out unrealistic assumptions. If theories cannot be invalidated by observation (which Friedman accepts) and simplicity is promoted over realism, then invalid, unrealistic theories can actually crowd out valid, realistic theories. I believe this has happened, to some degree, since Friedman wrote the paper.

"A basic proposition of formal logic is that x implicates y (x → y) is always true, unless x is true and y is false."

ReplyDeleteWow, I'm not comfortable with that.

In computer programming one would say "If X then Y" so that X implies Y, but only if X is true. Not always.

x → y is poorly translated into English as "if x then y". As Quine points out, it should be "x only if y".

DeleteMy logic lacks the black tie, no doubt.

Deletex = "the sun is up"

y = "it is daytime"

1.

if"the sun is up"then"it is daytime"Ok.

2. "the sun is up"

only if"it is daytime"Ok.

3. "it is daytime" is always true, unless "the sun is up" is true and "it is daytime" is false.

I still don't like this one.

Min -- Hey, a long time ago at Worthwhile Canadian Initiative, on Nick Rowe's Why has (private) debt increased?, someone named Min wrote:

"Haven't we seen this before, at least in the run up to the Great Depression? I saw some commentary about that last year or the year before. i'll try to find it.

The U. S. also had a run-up of private debt before the depression of 1837 - 1843. That was fueled by land speculation, not consumerism."

Same Min?

I am most interested to see links or references on this "run-up of private debt before the depression of 1837 - 1843".

Art

Hi, Art. Yup. same old Min. :)

DeleteI had done a good bit of online search about the Depression of 1837 and paying off the national debt by Jackson in 1835. But I didn't keep track of most of it. Some people blamed Jackson for the depression, largely because of the fact that he drained hard money from the economy by selling gov't land only for specie under the Specie Circular. Jackson blamed paper money, issued by private banks, largely to fund speculation.

Here is a reference, Summer's "History of American Currency" at http://oll.libertyfund.org/index.php?option=com_staticxt&staticfile=show.php?title=1653&layout=html

The SPECULATION section and following sections goes into this. :)

Thank you, Min. That link is massive, and fascinating.

DeleteAnd massive!

Looked at it a couple times for an hour or so. But from it already I have ideas for a post or two of my own. So, thanks again.