Primer: Causality In Models — Brian Romanchuk

Since is about identifying regularity in change and developing theories that explain the causation, ideally in terms of variables and linear functions. This is a challenge even in complicated simple systems, e.g, in the natural sciences, and it is a huge challenge when dealing with complex adaptive systems in the life and social sciences. 

Complex adaptive systems are synergistic, meaning that they are greater than the sum of the parts, so that examining the parts alone is insufficient for analyzing the system as a whole, that is, the parts and their relationships.

Complex adaptive systems are also subject to emergence, that is, the appearance of properties that are unforeseeable based on analysis of the existing system. For example, systems with the capability to learn from feedback and change behavior based on learning are unpredictable based on discovery of new knowledge and its application. There is as yet no logic of discovery that formalized the process and no scientific theory that has penetrated the causality so to be be able to influence it.

Furthermore, a framework for approaching causality must be assumed and ideally defined operationally in science, as Brian does in this first sentence following:

One important consideration for indicator construction is the notion of causality (using systems engineering terminology). A non-causal model is a model where the output depends upon the future values of inputs. In the absence of access to a time machine, such a model cannot be directly implemented in the real world. In practice, a non-causal model output is “revised” as new datapoints are added to input series. The result is that we cannot use the latest values of the series to judge the quality of previous “predictions” of the model.

The use of non-causal model might be acceptable for the analysis of a historical episode, or an earlier economic regime (such as various Gold Standard periods). Since new data will not arrive, there will be no revisions....

However, causation is still an open question in the philosophy of science.•

Bond Economics
Primer: Causality In Models
Brian Romanchuk

• Causality can be defined as the "causal" connection between cause and effect, e.g., in terms of conditionality (sufficient condition, necessary condition, necessary and sufficient condition). Causality is established though a scientific theory that accounts for the connection, since "correlation is not causation."

Causation is the entire scope of the subject, which includes "causality" as just defined but is not limited to it. There are ontological and epistemology issues regarding causation that are not settled. 

Generally speaking, modern science assumes 1) ontological monism in assuming naturalism, that is, that "everything" can be explained by natural causes as observables (as in a theory of everything). It also assumes 2) epistemological realism in the sense that the mind (subjectivity) is the mirror reality (objectivity) when the scientific method is correctly applied. 

However, these assumptions regarding the framework for gaining knowledge are more presuppositions than stated assumptions. Philosophy of science attempts to bring clarity to this by examining the various issues that arise.

“Causal Processes in Psychology Are Heterogeneous” — Andrew Gelman

A key difficulty here is that, even though interactions are clearly all over the place, they’re hard to estimate. Remember, you need 16 times the sample size to estimate an interaction than to estimate a main effect. So, along with accepting the importance of interactions, we also have to accept inevitable uncertainty in their estimation. We have to move away from the idea that a statistical analysis will give us effective certainty for the things we care about.

"Representative agents" are homogenous. They serve as a "methodological convenience." This implies that the scope and scale are limited. Model implications cannot be extended beyond the boundaries of the assumptions and data.

This doesn't imply that representative agent models are necessarily useless. But the temptation to overextend them must be avoided to prevent conclusions from falling into fallacy — due to hasty generalization, for instance.

What this means is that it is very difficult to get a binary causal connection of data to result through a linear function where human agents are involved owing to heterogenous causal factors. Causality involves a constellation of causal factors, perhaps including catalysts, whose distribution varies with respect to time and conditions. 

Identifying the constellation of factors and estimating their relative weights in a causal process in which the assumptions also require identification is challenging. Therefore, simple static models are usually overly simplistic regarding events of any degree of complexity, as are most design problems involving social agents affected by systemic relationships.

Getting the "science" right in social science is more difficult in the life sciences than the natural sciences, and much more difficult in the social sciences, in most questions that matter anyway.

Statistical Modeling, Causal Inference, and Social Science
“Causal Processes in Psychology Are Heterogeneous”
Andrew Gelman | Professor of Statistics and Political Science and Director of the Applied Statistics Center, Columbia University

Timothy Taylor — Pareidolia: When Correlations are Truly Meaningless

"Pareidolia" refers to the common human practice of looking at random outcomes but trying to impose patterns on them. For example, we all know in the logical part of our brain that there are a roughly a kajillion different variables in the world, and so if we look through the possibilities, we will will have a 100% chance of finding some variables that are highly correlated with each other. These correlations will be a matter of pure chance, and they carry no meaning. But when my own brain, and perhaps yours, sees one of these correlations, I can feel my thoughts start searching for a story to explain what looks to my eyes like a connected pattern.…

Classes in statistics emphasize that "correlation doesn't mean causation." The lesson here is even stronger. Correlation doesn't necessarily mean anything at all.Classes in statistics emphasize that "correlation doesn't mean causation." The lesson here is even stronger. Correlation doesn't necessarily mean anything at all.

Conversable Economist
Pareidolia: When Correlations are Truly Meaningless
Timothy Taylor | Managing editor of the Journal of Economic Perspectives, based at Macalester College in St. Paul, Minnesota

Lars P. Syll — In search of causality

Causality is one of the fundamental problems in philosophy, covering epistemology, philosophy of language, semiotics, and philosophy of science. Since causality is the basis of explanation, it applies to all aspects of understanding and theorizing, as Aristotle pointed out in his Metaphysics millennia ago. Yet, there is still no complete understanding of causality that would end controversy.

There many interrogatives — who, what, when, where, how and why, for example. Description involves the facts — what what, when, where, how much and how long, etc. Explanation involves means and ends — "how" (Greek techné) and "why" (telos).

Natural science deals chiefly with the how. "Speculation" deals with the why. Aristotle opined that all speculation begins with wonder. The Greek word for "speculate" that Aristotle uses is theorein. The root is theo which means god, or divine. Speculation is contemplative rather than active. It involves reflection on experience.

An archaic English term for "to speculate" is "to divine." It means to discern the inner workings. We see the sun rise and set and still speak of the "sunrise" and "sunset," but now we know that the sun is not actually moving at all; the rotational movement of the earth is "causing" the experience.

What we wonder about is a "puzzle" to us. The Greek term is aporia. The root means "impasse." This "causes" us to speculate about how and why in search of an explanation as a "theory."

In ancient time, most of the answers to such foundational questions involved supernatural causes expressed in myths, which were largely anthropomorphisms about natural forces. At the time of the Axial Age, interest shifted toward intellectual (logical) reasoning in place of myth as storytelling became less satisfying intellectually.

Aristotle was the first person in the West to systematize knowledge largely in the form that it has been handed down through the centuries in the West. He understood that a requirement for gaining true knowledge (epistemé) through inquiry was to understand reasoning, so he wrote books on logic as a prerequisite.

Aristotle was also understood that knowledge of the world comes through the senses and so he emphasized the role of observation in gaining knowledge. He was particularly interested in biology as a science understood as theory based on observation rather than storytelling.

Aristotle also recognized the existence of foundational issues that "come before," or are "meta," as we say even today. These are properly the issues for intellectual inquiry, which we still call "philosophy." meaning love of wisdom. Here the Greek term sophia means speculative wisdom rather than practical wisdom. Speculative wisdom is concerned with the way, while practical wisdom is concerned with the how.

Aristotle seems to have gotten off on the wrong foot in some instances, but overall the paradigm of knowledge he set forth still holds sway in the West. In fact, Aristotelianism is now making a comeback.

Today, we are still arguing about causality, what counts as causal explanation, and the degree to such ultimate explanation is possible given bounded rationality.

Lars P. Syll’s Blog
In search of causality
Lars P. Syll | Professor, Malmo University

Bill Black — Mankiw Whiffs on “Learning the Right Lessons from the Financial Crisis

So how does Mankiw answer the question he raises in his first sentence: “What caused the financial crisis of 2008?” He does not answer it. He not even explain why he does not answer his own question.

New Economic Perspectives
Mankiw Whiffs on “Learning the Right Lessons from the Financial Crisis”
William K. Black | Associate Professor of Economics and Law, UMKC

Mike Steiner — Causes in Real Life – How Organizations Perform a Root Cause Analyses (RCA)

Not a priority but of interest if for those who want to know more about how organizations deal with causation by analyzing the concrete in terms of the abstract. 

This is related to what Hegel called "concrete universal, and Marx defined as "concrete abstraction." This is the basis of the dialect for Hegel and Marx's adoption and adaptation of it.

A Philosopher's Take
Causes in Real Life – How Organizations Perform a Root Cause Analyses (RCA)

Mike Steiner | Strategic Initiative Specialist at TransCanada

Daniel Little — New thinking about causal mechanisms

Everyone is familiar with the nostrum, "correlation is not causality." Simply put, correlation can potentially identify input-output relationships with a certain degree of probability. But the relationship is a "black box."

Causal explanation involves opening the box and examining the contents. Correlation shows that something happens; causality in science explains how it happens, elucidating transmission in terms of operations. In formal systems the operators are rules, e.g., expressible by mathematical functions.

Generally speaking correlation is probabilistic, whereas causality is deterministic. Causes are logically antecedent to effects, but arguments based on prior occurrence are post hoc ergo propter hoc fallacies.

There is also probabilistic causation.

Informally, A probabilistically causes B if A's occurrence increases the probability of B. This is sometimes interpreted to reflect imperfect knowledge of a deterministic system but other times interpreted to mean that the causal system under study has an inherently indeterministic nature.

Causality in philosophy involves provision of an account of why something happens based on principles.

Causation is at the heart of the fundamental problems in philosophy of science. It's exploration began in the West in earnest with Aristotle and it has become one of the enduring questions.

Understanding Society
New thinking about causal mechanisms
Daniel Little | Chancellor of the University of Michigan-Dearborn, Professor of Philosophy at UM-Dearborn and Professor of Sociology at UM-Ann Arbor

Noah Smith — "Theory vs. Data" in statistics too

Important.

I think Noah has this right. Fit the tool to the job, rather than the job to the tool.

Aristotle defined speculative knowledge in terms of causal explanation. This definition stuck although Aristotle's analysis of causality did not.

In the Posterior Analytics, Aristotle places the following crucial condition on proper knowledge: we think we have knowledge of a thing only when we have grasped its cause (APost. 71 b 9–11. Cf. APost. 94 a 20). That proper knowledge is knowledge of the cause is repeated in the Physics: we think we do not have knowledge of a thing until we have grasped its why, that is to say, its cause (Phys. 194 b 17–20). Since Aristotle obviously conceives of a causal investigation as the search for an answer to the question “why?”, and a why-question is a request for an explanation, it can be useful to think of a cause as a certain type of explanation. (My hesitation is ultimately due to the fact that not all why-questions are requests for an explanation that identifies a cause, let alone a cause in the particular sense envisioned by Aristotle.) — Stanford Encyclopedia of Philosophy

There is a distinction between reasons and causes. Some types of explanation seek only reasons, while other seek causes. Causation subsequently came to be viewed in terms of articulating mechanisms or lines of transmission (models) that are substantiated in evidence.

Explanation by reasons is different since the strict criterion of articulating mechanisms or lines of transmission that can be checked against evidence is not required.

Explanation by reasons rather than strictly by establishing causation is based on the principle of sufficient reason, which is usually credited to Spinoza and Leibnitz.

In philosophical logic, two negative criteria are foundational. Valid reasoning is vitiated by 1) arguing in a circle and 2) infinite regress.

Without recourse to checking against evidence there is no stopping point in assigning causes other than stipulation, e.g. of a first cause.

However, there may be a reason for a stopping point that doesn't involve causality based on evidence from observation or only stipulation, for example, principles that are "self-evident" based on intuition such as Aristotle's conception of intellectual intuition, or Kant's synthetic a priori propositions as articulated in the Critique of Pure Reason. 

On the other hand, Hume argued that causality is merely over-interpretation of constant correlation, there being no knowledge of the world other than that based on sense data. There is no observable causal link.

Cutting to the chase, scientific explanation based on causality is grounded in models that articulate causal mechanisms or lines of transmission that show how things change invariantly, which is the basis for deterministic functions. Where this is not possible, then there are two other avenues. The first is explanation by giving reasons, which is the domain of speculative philosophy. The second is employing statistics to explore patters of correlation. The question then is to what degree causal models can be gained from statistical methods, or whether it is possible at all. 

This is the issue that Noah Smith's post is getting at.

Noahpinion
"Theory vs. Data" in statistics too
Noah Smith | Bloomberg View columnist

Lars P. Syll — Why testing axioms is necessary in economics

Where do axioms come from and how are they tested?

Axioms are starting points of deductive systems. They are stipulations.

Axioms must avoid the traps of illogic, circularity, and infinite regress.

Axioms in scientific theories are assumptions derived either from induction as generalization from experience (data) or abduction in C. S. Peirce's sense as discovery through "educated guessing."

The objective of scientific inquiry is not to "prove" axioms, since axioms are the basis of proof in a logical system. Axioms function as criteria for syntactical or logical truth, also called "necessity."

Axioms in hypothetical-deductive systems provide the basis for the scientific method. 

Assumptions of representational models stand or fall with the testing of models they are used to construct. 

Scientific models are used to generate hypotheses that can be tested empirically by deriving hypotheses as theorems from the assumptions that serve as axioms for the system.

Failure of a hypothesis as a theorem of deductive system reveals semantic inconsistency or incompleteness and calls the system into question as a coherent explanation of the data. 

This also calls the explanatory model into question as providing a causal explanation based on a representational model purporting to show causal transmission.

Correlation is not causation. Without a theory explicating transmission, there is no properly scientific explanation.

Lars P. Syll’s Blog

Why testing axioms is necessary in economics

Lars P. Syll | Professor, Malmo University

Ramanan — Public Debt And Current Account Deficits, Part 2

This is a continuation of a recent post at this blog, Public Debt And Current Account Deficits, in which I argued that the current account balance of payments affects the public debt.

The Case for Concerted Action
Public Debt And Current Account Deficits, Part 2
V. Ramanan

Lars P. Syll — RBC models — nonsense on stilts

I don’t think that there is a way to write down any model which at one hand respects the possible diversity of agents in taste, circumstances, and so on, and at the other hand also grounds behavior rigorously in utility maximization and which has any substantive content to it. — James Tobin

Determining causality is a bitch in social science since many factors generally contribute to causality involving social behavior. Studying a single individual and making assumptions about future behavior based on habits and revealed preferences might hold but transferring this to groups of individuals involves the fallacy of composition. This makes the assumption of methodological individualism and microfoundations problematic.

Assuming methodological individualism ignores that regularity in social behavior is more likely induced by stable institutional arrangements than individual factors involving assumptions of homogeneity that rather obviously do not hold in the real world.

Choosing a single variable or a few variables as causal factors operating universally and timelessly to produce regular results is seldom realistic. This is clearly done for convenience, to make the math tractable, rather than as a matter of induction based on empirical data or abduction based on reasoning to the best explanation. 

In many cases the process of identifying assumptions in conventional economics seem to be driven by ideology, with conclusions supported by authority, which is justification by power and gatekeepers rather than either reasoning or evidence.

In addition to the problem identifying assumptions, there is also the issue of assuming ergodicity (time average of a same is equal to the ensemble average) in processes that are conditioned historically and dynamically.

Moreover, the greater the scope the less accurate the solution is likely to be. This is a reason that social sciences have tended to focus on case studies rather than general theories.

Lars P. Syll’s Blog
RBC models — nonsense on stilts
Lars P. Syll | Professor, Malmo University

Sputnik International — Humiliating Billion User Hack Attack on Yahoo 'Likely Launched by a State Actor'

No evidence. Wild inference.

These people seem to have no idea of the issues surrounding identification of causal assumptions.

Conflation of a sufficient condition with a necessary condition.

Beyond lame.

Sputnik International
Humiliating Billion User Hack Attack on Yahoo 'Likely Launched by a State Actor'

Steve Randy Waldman — Attributions of causality

On Brexit, but more important for the thoughts on causes versus conditions and factors.

Interfluidity
Attributions of causality
Steve Randy Waldman

Ramanan — Output At Home And Abroad

Accounting identities are tautologies that say nothing about the world other than that the relevant accounts balance. As identities they are not functions, in which inputs determine outputs in terms of a rule. 

However, accounting identities can be used in theoretical interpretation to arrive at causal explanation, but this requires examining relevant behaviors. For example, one entity's expenditure is a flow that increases another entity's income, which will have a cumulative influence on a stock.

Stock-flow analysis observes stock-flow consistency. Accounting identities are boundary conditions of stock-flow consistency.

It’s fairly common for economists to confuse accounting identities and behavioural relationships.

Question: What is the best way to find it?

Answer: The behaviour of output (at home and abroad) is not discussed in their analysis.

It’s not always the case that it’s true but a good way to find – check whether the economist is talking of the effect of changes in stocks or flows on output.

It’s also of course important to discern what someone is literally saying and what that person is trying to say. Economists aren’t the best communicators.…

The Case for Concerted Action
Output At Home And Abroad
V. Ramanan

Lars P. Syll — Models vs. reality

Lars P. Syll’s Blog
Models vs. reality
Lars P. Syll | Professor, Malmo University

The Arthurian — Antonio Fatas and the Chain of Causality

Another howler from Ken Rogoff, the Thomas D. Cabot Professor of Public Policy and Professor of Economics at Harvard University and one of the Very Serious People of the economics profession (rolling eyes).

Fatas says Rogoff "argues that the world economy is suffering from a debt hangover rather than deficient demand." I had to check that. When I read Fatas, I thought he might be misinterpreting Rogoff. He's not. He's right. Rogoff sees secular stagnation as one possibility, and crushing debt as "another possibility".

Rogoff sees the excessive debt explanation as an alternative to the deficient demand explanation.

That's just plain silly. Excessive debt is not an alternative to the "deficient demand" explanation. Excessive debt is the cause of deficient demand. First, the growing cost of growing debt consumes a growing portion of income. So demand atrophies gradually at first. Then, people suddenly come to think of their debt as excessive, and they suddenly cut their borrowing and spending. Demand falls suddenly -- economists call that a "shock" -- and we have "deficient demand".

The New Arthurian Economics
Antonio Fatas and the Chain of Causality
The Arthurian

Chris Dillow — "Consistent with"

In discussing Paul Romer's wonderful concept of mathiness*, Peter Dorman criticizes economists' habit of declaring a theory successful merely because it is "consistent with" the evidence. His point deserves emphasis. 

If a man has no money, this is "consistent with" the theory that he has given it away. But if in fact he has been robbed, that theory is grievously wrong. Mere consistency with the facts is not sufficient.….

The difference between causes and reasons. Reasons are not necessarily causes. "The dog ate my homework."

There may be different plausible explanations for — reasons consistent with — the same data. 

Science is about establishing causal explanation in terms of "mechanism" or "transmission."

Otherwise, it is handwaving.

So, how can we guard against the "consistent with" error? One thing we need is history: this helps tell us how things actually happened. And - horrific as it might seem to some economists - we also need sociology: we need to know how people actually behave and not merely that their behaviour is "consistent with" some theory. Economics, then, cannot be a stand-alone discipline but part of the social sciences and humanities - a point which is lost in the discipline's mathiness.

Stumbling and Mumbling
"Consistent with"
Chris Dillow | Investors Chronicle

See also

Lars P. Syll’s Blog
Consistency and validity is not enough!
Lars P. Syll | Professor, Malmo University

Brian Romanchuk — Why Chartblogging Is Superior To Mainstream Macro

Orthodox-heterodox economic squabbling has once again erupted on the internet. As always, the mainstream argument is that their methodologies are superior because they are based on mathematical models. My main area of interest is the quantitative end of economics, so I do not pay too much attention to some of the purely literary approaches to economics. But even so, I believe that mathematical and statistical methods are being applied incorrectly by mainstream economists, and so whatever modelling advantage they have is largely illusionary. I illustrate this with a few examples, including an explanation why I believe the mainstream debate about the "natural rate" of interest is largely meaningless.…

Nice brief summary, not wonkish.

Bond Economics
Why Chartblogging Is Superior To Mainstream Macro
Brian Romanchuk

Lars P. Syll — ‘New Keynesianism’ – neat, plausible and wrong

Econometric fol-de-rol. Or fool-de-rubes?

Lars P. Syll’s Blog
‘New Keynesianism’ — neat, plausible and wrong
Lars P. Syll | Professor, Malmo University

Econometrics Beat: Dave Giles' Blog

Econometrics vs. Ad Hoc Empiricism
Dave Giles

Sandwichman — (Other things being elsewhere...) "It’s a ceteris paribus thing,"

The IS-LM "ceteris paribus thing" is an "if p, then q" thing -- a tautology -- not a "since p, therefore q" thing. One cannot draw conclusions about the real world from it.

EconoSpeak
(Other things being elsewhere...) "It’s a ceteris paribus thing,"

Sandwichman