Sunday, April 22, 2018

Brian Romanchuk — Why We Should Be Concerned About The Forecastability Of Economic Models

Although it might be possible to find dissenters, the apparent consensus among financial market practitioners is that mathematical economic models provide terrible forecasts. One response is to keep searching through the set of all possible models, hoping to find something that works. The author's suggested response is to accept that forecasting is an inherently impossible task. However, in order to advance beyond nihilism, we need to quantify why mathematical models are terrible. My argument is straightforward: the models that provide the best fit to observed behaviour cannot themselves be forecast with the type of information that we have available in the real world.
(As an aside, I would note that Beatrice Cherrier has written about the preference for tractable (simplicity) of mathematical models in "What is the Cost of 'Tractable' Economic Models," and in a follow up article. Much of what she discusses appears to overlap my thinking about the existing methodology, although I believe that I have a different view. My suggestion is to look at non-forecastable economic models, and tractable (or reduced order) non-forecastable models would be the most interesting. A non-tractable non-forecastable model might provide the best fit to reality, but its complexity would also make it difficult to draw any conclusions from it; this is essentially my concern with agent-based models. Since I was in the middle of laying out my logic, I was not able to work in a longer comparison to her arguments.)
Most economic models are either gadgets or thinking tools that assist in adding rigor. They are not capable of predicting the future any more than sociological models are.

For one thing, there are too many contingent factors involved, even if all were known and could be specified with some degree of precision based on data. 

It is not possible to generate an economic ephemeris. No economic laws of motion have been forthcoming.

The question then becomes how close is is possible to comes, e.g., what is the degree of error. Even that can only be estimated.
The multiplicity of sources of disturbances makes a generalisation of the notion of "small disturbances" difficult. As a result, I would argue that we should instead worry about analysing models with a focus on the properties of their forecast errors rather than the precise nature of the generalisation of exact forecastability....


AXEC / E.K-H said...

Note on ‘Brian Romanchuk — Why We Should Be Concerned About The Forecastability Of Economic Models’

Tom Hickey maintains ‘No economic laws of motion have been forthcoming.’

This is incorrect. See on Wikimedia
Elementary production-consumption economy, Four Basic Economic Laws{{int:filedesc}}

See also
Science does NOT predict the future (only charlatans do)

The Synthesis of Economic Law, Evolution, and History

Egmont Kakarot-Handtke

Matt Franko said...

"not capable of predicting the future "

Well I predict that if the US were to establish a JG then there would no longer exist involuntary unemployment...

I predict that if the US established free university, that there would no longer be any unnecessary student debt...

I predict that if the US implemented a UBI, then there would no longer be those without adequate provision...



(FD: I am not a libertarian...)

Ryan Harris said...
This comment has been removed by the author.
Noah Way said...

You just have to LOVE the way Egmont Kakarot-Handtke blows up other people's arguments by linking to his own website.

Brilliant. Reminds me of Peter Dow

Konrad said...

@Matt Franko: I agree. The only time economics cannot predict anything is when irrefutable facts (like those you mentioned above) are buried by ideologically-driven lies.

Financial speculation is a casino, but fiscal policy is not. When, for example, the federal government taxes too much money out of the economy, we can predict with 100% certainty that the result will be a recession.

When the federal government creates money out of thin air to upgrade the national infrastructure, we can predict with 100% certainty that jobs will be created, and the infrastructure will be improved (unless contractors steal all the money and hide it in offshore accounts).

When a federal government creates its spending money out of thin air, simply by crediting accounts, we can predict with 100% certainty that even if federal taxes were lowered to zero, there would be infinite money for Medicare, Social Security, Food Stamps, etc.

When a “national debt” is denominated in a nation’s own sovereign currency, we can predict with 100% certainty that the “national debt” will not be a problem.

And so on.

Uncertainty in economics only occurs when bullshit enters the picture, or when we speak of market speculation.

Matt Franko said...

“ideologically-driven lies. ”

Yes and I submit that ideology is libertarianism .... not neoliberalism.... the neoliberals are just stupid not lying..

Matt Franko said...

“No economic laws of motion have been forthcoming.”

The laws in this regard are created by us Tom... they don’t evolve from the apes by random chance...

We pass a law authorizing $50k per high school graduate for college education then that is what is going to happen ...

Tom Hickey said...

I updated this post with Brian's next. Here is a clickable link.

AXEC / E.K-H said...

Infantile model bricolage, or, How many economists can dance on a non-existing pinpoint?
Comment on Brian Romanchuk on ‘Forecastability And Economic Modelling’*

“The highest ambition an economist can entertain who believes in the scientific character of economics would be fulfilled as soon as he succeeded in constructing a simple model displaying all the essential features of the economic process by means of a reasonably small number of equations connecting a reasonably small number of variables. Work on this line is laying the foundations of the economics of the future . . .” (Schumpeter, 1946)

The future is now and economists still do NOT have the paradigmatic simple core model but a heap of in incommensurable and contradicting constructions. Pluralism may have its merits elsewhere but is the worst thing that can happen in science. As the ancient Greeks already observed: “There are always many different opinions and conventions concerning any one problem or subject-matter…. This shows that they are not all true. For if they conflict, then at best only one of them can be true.” (Popper)

Fact is that, in economics, ALL models are axiomatically false. It holds: “When the premises are certain, true, and primary, and the conclusion formally follows from them, this is demonstration, and produces scientific knowledge of a thing.” (Aristotle, 300 BC) Fact is that the premises of current model are neither certain, true, nor primary.

Brian Romanchuk’s SIM model is a case in point. He enumerates his key premises as follows.
• The model is a straightforward three sector model, with a household sector, business sector, and government.
• The household consumption function is defined in terms of a pair of propensity to consume parameters (out of income, out of wealth). …
• The business sector is constrained to break even, …
• Government policy is specified in terms of government consumption and a fixed tax rate.

Brian Romanchuk starts with macrofoundations, which is correct. But then he assumes a consumption function, which is a NONENTITY, and break even for the business sector, which kills the model already at this early stage because a zero profit economy is the most idiotic NONENTITY of them all.

Let us contrast this with the standard microfoundations approach. The whole analytical superstructure of Orthodoxy is based upon this set of hardcore propositions a.k.a. axioms:
• HC1 There exist economic agents.
• HC2 Agents have preferences over outcomes.
• HC3 Agents independently optimize subject to constraints.
• HC4 Choices are made in interrelated markets.
• HC5 Agents have full relevant knowledge.
• HC6 Observable economic outcomes are coordinated, so they must be discussed with reference to equilibrium states. (Weintraub)

HC3 introduces marginalism which is the all-pervasive principle of Orthodoxy. HC3, though, and HC5 and HC6 are plain NONENTITIES.#1

See part 2

AXEC / E.K-H said...

Part 2

In order to be applicable HC3, requires a lot of auxiliary assumptions, most prominently a well-behaved/differentiable production function.#2 Taken together, all axioms and auxiliary assumptions then crystallize to supply-function/demand-function/equilibrium or what Leijonhufvud famously called the Totem of Micro.#3

The methodological fact of the matter is that ALL models that take just one NONENTITY into the premises are a priori false.#4

So, because these premises are NOT “certain, true, and primary” they cannot be used for model building: expected utility, rationality/bounded rationality/animal spirits, constrained optimization, well-behaved production functions, supply/demand functions, simultaneous adaptation, equilibrium, first/second derivatives, total income=value of output, I=S, real-number quantities/prices, ergodicity. Every theory/model that contains just one NONENTITY goes straight into the wastebasket.

The standard microfoundations approach with all its variants and derivatives up to DSGE is methodologically false. The same holds for Keynes’ macrofoundations and all After-Keynesian variants.

To put NONENTITIES into the premises is the defining characteristic of fairy tales, science fiction, theology, Hollywood movies, politics, proto-science, and the senseless model bricolage of scientifically incompetent economists.#5

Egmont Kakarot-Handtke


#1 The solemn burial of marginalism

#2 Putting the production function back on its feet

#3 Equilibrium and the violation of a fundamental principle of science

#4 The future of economics: why you will probably not be admitted to it, and why this is a good thing

#5 How to restart economics