LARS P. SYLL

‘Overcontrolling’ in statistics

You see it all the time in studies. “We controlled for…” And then the list starts … The more things you can control for, the stronger your study is — or, at least, the stronger your study seems. Controls give the feeling of specificity, of precision. But sometimes, you can control for too much. Sometimes you end up controlling for the thing you’re trying to measure …

An example is research around the gender wage gap, which tries to control for so many things that it ends up controlling for the thing it’s trying to measure …

Take hours worked, which is a standard control in some of the more sophisticated wage gap studies. Women tend to work fewer hours than men. If you control for hours worked, then some of the gender wage gap vanishes. As Yglesias wrote, it’s “silly to act like this is just some crazy coincidence. Women work shorter hours because as a society we hold women to a higher standard of housekeeping, and because they tend to be assigned the bulk of childcare responsibilities.”

Controlling for hours worked, in other words, is at least partly controlling for how gender works in our society. It’s controlling for the thing that you’re trying to isolate.

Ezra Klein

Trying to reduce the risk of having established only ‘spurious relations’ when dealing with observational data, statisticians and econometricians standardly add control variables. The hope is that one thereby will be able to make more reliable causal inferences. But — as Keynes showed already back in the 1930s when criticizing statistical-econometric applications of regression analysis — if you do not manage to get hold of all potential confounding factors, the model risks producing estimates of the variable of interest that are even worse than models without any control variables at all. Conclusion: think twice before you simply include ‘control variables’ in your models!

When I present this argument … one or more scholars say, “But shouldn’t I control for everything I can in my regressions? If not, aren’t my coefficients biased due to excluded variables?” … The excluded variable argument only works if you are sure your specification is precisely correct with all variables included. But no one can know that with more than a handful of explanatory variables …

A preferable approach is to separate the observations into meaningful subsets—internally compatible statistical regimes … If this can’t be done, then statistical analysis can’t be done. A researcher claiming that nothing else but the big, messy regression is possible because, after all, some results have to be produced, is like a jury that says, “Well, the evidence was weak, but somebody had to be convicted.”

Christopher H. Achen

Kitchen sink econometric models are often the result of researchers trying to control for confounding. But what they usually haven’t understood is that the confounder problem requires a causal solution and not statistical ‘control.’ Controlling for everything opens up the risk that we control for ‘collider’ variables and thereby create ‘back-door paths’ which gives us confounding that wasn’t there to begin with.

Getting causality into statistics

Because statistical analyses need a causal skeleton to connect to the world, causality is not extra-statistical but instead is a logical antecedent of real-world inferences. Claims of random or “ignorable” or “unbiased” sampling or allocation are justified by causal actions to block (“control”) unwanted causal effects on the sample patterns. Without such actions of causal blocking, independence can only be treated as a subjective exchangeability assumption whose justification requires detailed contextual information about absence of factors capable of causally influencing both selection (including selection for treatment) and outcomes. Otherwise it is essential to consider pathways for the causation of biases (nonrandom, systematic errors) and their interactions …

Probability is inadequate as a foundation for applied statistics, because competent statistical practice integrates logic, context, and probability into scientific inference and decision, using causal narratives to explain diverse data. Thus, given the absence of elaborated causality discussions in statistics textbooks and coursework, we should not be surprised at the widespread misuse and misinterpretation of statistical methods and results. This is why incorporation of causality into introductory statistics is needed as urgently as other far more modest yet equally resisted reforms involving shifts in labels and interpretations for P-values and interval estimates.

Sander Greenland

Causality can never be reduced to a question of statistics or probabilities unless you are — miraculously — able to keep constant all other factors that influence the probability of the outcome studied. To understand causality we always have to relate it to a specific causal structure. Statistical correlations are never enough. No structure, no causality.

Statistical patterns should never be seen as anything else than possible clues to follow. Behind observable data, there are real structures and mechanisms operating, things that are  — if we really want to understand, explain and (possibly) predict things in the real world — more important to get hold of than to simply correlate and regress observable variables.

Statistics cannot establish the truth value of a fact. Never has. Never will.

Econometric fictionalism

If you can’t devise an experiment that answers your question in a world where anything goes, then the odds of generating useful results with a modest budget and nonexperimental survey data seem pretty slim. The description of an ideal experiment also helps you formulate causal questions precisely. The mechanics of an ideal experiment highlight the forces you’d like to manipulate and the factors you’d like to hold constant.

Research questions that cannot be answered by any experiment are fundamentally unidentified questions.

One of the limitations of economics is the restricted possibility to perform experiments, forcing it to mainly rely on observational studies for knowledge of real-world economies.

But still — the idea of performing laboratory experiments holds a firm grip on our wish to discover (causal) relationships between economic ‘variables.’If we only could isolate and manipulate variables in controlled environments, we would probably find ourselves in a situation where we with greater ‘rigour’ and ‘precision’ could describe, predict, or explain economic happenings in terms of ‘structural’ causes, ‘parameter’ values of relevant variables, and economic ‘laws.’

Galileo Galilei’s experiments are often held as exemplary for how to perform experiments to learn something about the real world. Galileo’s heavy balls dropping from the tower of Pisa, confirmed that the distance an object falls is proportional to the square of time and that this law (empirical regularity) of falling bodies could be applicable outside a vacuum tube when e. g. air existence is negligible.

The big problem is to decide or find out exactly for which objects air resistance (and other potentially ‘confounding’ factors) is ‘negligible.’ In the case of heavy balls, air resistance is obviously negligible, but how about feathers or plastic bags?

One possibility is to take the all-encompassing-theory road and find out all about possible disturbing/confounding factors — not only air resistance — influencing the fall and build that into one great model delivering accurate predictions on what happens when the object that falls is not only a heavy ball but feathers and plastic bags. This usually amounts to ultimately stating some kind of ceteris paribus interpretation of the ‘law.’

Another road to take would be to concentrate on the negligibility assumption and to specify the domain of applicability to be only heavy compact bodies. The price you have to pay for this is that (1) ‘negligibility’ may be hard to establish in open real-world systems, (2) the generalization you can make from ‘sample’ to ‘population’ is heavily restricted, and (3) you actually have to use some ‘shoe leather’ and empirically try to find out how large is the ‘reach’ of the ‘law.’

In mainstream economics, one has usually settled for the ‘theoretical’ road (and in case you think the present ‘natural experiments’ hype has changed anything, remember that to mimic real experiments, exceedingly stringent special conditions standardly have to obtain).

In the end, it all boils down to one question — are there any Galilean ‘heavy balls’ to be found in economics, so that we can indisputably establish the existence of economic laws operating in real-world economies?

As far as I can see there are some heavy balls out there, but not even one single real economic law.

Economic factors/variables are more like feathers than heavy balls — non-negligible factors (like air resistance and chaotic turbulence) are hard to rule out as having no influence on the object studied.

Galilean experiments are hard to carry out in economics, and the theoretical ‘analogue’ models economists construct and in which they perform their ‘thought experiments’ build on assumptions that are far away from the kind of idealized conditions under which Galileo performed his experiments. The ‘nomological machines’ that Galileo and other scientists have been able to construct have no real analogues in economics. The stability, autonomy, modularity, and interventional invariance, that we may find between entities in nature, simply are not there in real-world economies. That’s are real-world fact, and contrary to the beliefs of most mainstream economists, they won’t go away simply by applying deductive-axiomatic economic theory with tons of more or less unsubstantiated assumptions.

By this, I do not mean to say that we have to discard all (causal) theories/laws building on modularity, stability, invariance, etc. But we have to acknowledge the fact that outside the systems that possibly fulfil these requirements/assumptions, they are of little substantial value. Running paper and pen experiments on artificial ‘analogue’ model economies is a sure way of ‘establishing’ (causal) economic laws or solving intricate econometric problems of autonomy, identification, invariance and structural stability — in the model world. But they are pure substitutes for the real thing and they don’t have much bearing on what goes on in real-world open social systems. Setting up convenient circumstances for conducting Galilean experiments may tell us a lot about what happens under those kinds of circumstances. But — few, if any, real-world social systems are ‘convenient.’ So most of those systems, theories and models, are irrelevant for letting us know what we really want to know.

To solve, understand, or explain real-world problems you actually have to know something about them — logic, pure mathematics, data simulations or deductive axiomatics don’t take you very far. Most econometrics and economic theories/models are splendid logic machines. But — applying them to the real world is a totally hopeless undertaking! The assumptions one has to make in order to successfully apply these deductive-axiomatic theories/models/machines are devastatingly restrictive and mostly empirically untestable– and hence make their real-world scope ridiculously narrow. To fruitfully analyze real-world phenomena with models and theories you cannot build on patently and known to be ridiculously absurd assumptions. No matter how much you would like the world to entirely consist of heavy balls, the world is not like that. The world also has its fair share of feathers and plastic bags.

Most of the ‘idealizations’ we find in mainstream economic models are not ‘core’ assumptions, but rather structural ‘auxiliary’ assumptions. Without those supplementary assumptions, the core assumptions deliver next to nothing of interest. So to come up with interesting conclusions you have to rely heavily on those other — ‘structural’ — assumptions.

In physics, we have theories and centuries of experience and experiments that show how gravity makes bodies move. In economics, we know there is nothing equivalent. So instead mainstream economists necessarily have to load their theories and models with sets of auxiliary structural assumptions to get any results at all in their models.

So why then do mainstream economists keep on pursuing this modelling project?

Mainstream ‘as if’ models are based on the logic of idealization and a set of tight axiomatic and ‘structural’ assumptions from which consistent and precise inferences are made. The beauty of this procedure is, of course, that if the assumptions are true, the conclusions necessarily follow. But it is a poor guide for real-world systems.

The way axioms and theorems are formulated in mainstream economics often leaves their specification without almost any restrictions whatsoever, safely making every imaginable evidence compatible with the all-embracing ‘theory’ — and theory without informational content never risks being empirically tested and found falsified. Used in mainstream ‘thought experimental’ activities, it may, of course, ​be very ‘handy,’ but totally void of any empirical value.

Some economic methodologists have lately been arguing that economic models may well be considered ‘minimal models’ that portray ‘credible worlds’ without having to care about things like similarity, isomorphism, simplified ‘representationality’ or resemblance to the real world. These models are said to resemble ‘realistic novels’ that portray ‘possible worlds’. And sure: economists constructing and working with those kinds of models learn things about what might happen in those ‘possible worlds’. But is that really the stuff real science is made of? I think not. As long as one doesn’t come up with credible export warrants to real-world target systems and show how those models — often building on idealizations with known to be false assumptions — enhance our understanding or explanations about the real world, well, they are just nothing more than just novels.  Showing that something is possible in a ‘possible world’ doesn’t give us a justified license to infer that it therefore also is possible in the real world. ‘The Great Gatsby’ is a wonderful novel, but if you truly want to learn about what is going on in the world of finance, I would recommend rather reading Minsky or Keynes and directly confronting real-world finance.

Different models have different cognitive goals. Constructing models that aim for explanatory insights may not optimize the models for making (quantitative) predictions or deliver some kind of ‘understanding’ of what’s going on in the intended target system. All modelling in science has tradeoffs. There simply is no ‘best’ model. For one purpose in one context model A is ‘best’, for other purposes and contexts model B may be deemed ‘best’. Depending on the level of generality, abstraction, and depth, we come up with different models. But even so, I would argue that if we are looking for what I have called ‘adequate explanations’ (Syll, Ekonomisk teori och metod, Studentlitteratur, 2005) it is not enough to just come up with ‘minimal’ or ‘credible world’ models.

The assumptions and descriptions we use in our modelling have to be true — or at least ‘harmlessly’ false — and give a sufficiently detailed characterization of the mechanisms and forces at work. Models in mainstream economics do nothing of the kind.

Coming up with models that show how things may possibly be explained is not what we are looking for. It is not enough. We want to have models that build on assumptions that are not in conflict with known facts and that show how things actually are to be explained. Our aspirations have to be more far-reaching than just constructing coherent and ‘credible’ models about ‘possible worlds’. We want to understand and explain ‘difference-making’ in the real world and not just in some made-up fantasy world. No matter how many mechanisms or coherent relations you represent in your model, you still have to show that these mechanisms and relations are at work and exist in society if we are to do real science. Science has to be something more than just more or less realistic ‘story-telling’ or ‘explanatory fictionalism.’ You have to provide decisive empirical evidence that what you can infer in your model also helps us to uncover what actually goes on in the real world. It is not enough to present your students with epistemically informative insights about logically possible but non-existent general equilibrium models. You also, and more importantly, have to have a world-linking argumentation and show how those models explain or teach us something about real-world economies. If you fail to support your models in that way, why should we care about them? And if you do not inform us about what are the real-world intended target systems of your modelling, how are we going to be able to value or test them? Without giving that kind of information it is impossible for us to check if the ‘possible world’ models you come up with actually hold also for the one world in which we live — the real world.

The Keynes-Tinbergen debate on econometrics

It is widely recognized but often tacitly neglected that all statistical approaches have intrinsic limitations that affect the degree to which they are applicable to particular contexts … John Maynard Keynes was perhaps the first to provide a concise and comprehensive summation of the key issues in his critique of Jan Tinbergen’s book Statistical Testing of Business Cycle Theories …

Keynes’s intervention has, of course, become the basis of the “Tinbergen debate” and is a touchstone whenever historically or philosophically informed methodological discussion of econometrics is undertaken. It has remained the case, however, that Keynes’s concerns with the “logical issues” regarding the “conditions which the economic material must satisfy” still
gain little attention in theory and practice.

Muhammad Ali Nasir & Jamie Morgan

Mainstream economists often hold the view that Keynes’ criticism of econometrics was the result of a sadly misinformed and misguided person who disliked and did not understand much of it.

This is, however, as Nasir and Morgan convincingly argue, nothing but a gross misapprehension.

To be careful and cautious is not the same as to dislike. Keynes did not misunderstand the crucial issues at stake in the development of econometrics. Quite the contrary. He knew them all too well — and was not satisfied with the validity and philosophical underpinning of the assumptions made for applying its methods.

Keynes’ critique of the “logical issues” regarding the conditions that have to be satisfied if we are going to be able to apply econometric methods, is still valid and unanswered in the sense that the problems he pointed at are still with us today and largely unsolved. Ignoring them — the most common practice among applied econometricians — is not to solve them.

To apply statistical and mathematical methods to the real-world economy, the econometrician has to make some quite strong assumptions. In a review of Tinbergen’s econometric work — published in The Economic Journal in 1939 — Keynes gave a comprehensive critique of Tinbergen’s work, focusing on the limiting and unreal character of the assumptions that econometric analyses build on:

Completeness: Where Tinbergen attempts to specify and quantify which factors influence the business cycle, Keynes maintains there must be a complete list of all the relevant factors to avoid misspecification and spurious causal claims. Usually, this problem is ‘solved’ by econometricians assuming that they somehow have a ‘correct’ model specification. Keynes is, to put it mildly, unconvinced:

It will be remembered that the seventy translators of the Septuagint were shut up in seventy separate rooms with the Hebrew text and brought out with them, when they emerged, seventy identical translations. Would the same miracle be vouchsafed if seventy multiple correlators were shut up with the same statistical material? And anyhow, I suppose, if each had a different economist perched on his a priori, that would make a difference to the outcome.

Homogeneity: To make inductive inferences possible — and be able to apply econometrics — the system we try to analyse has to have a large degree of ‘homogeneity.’ According to Keynes most social and economic systems — especially from the perspective of real historical time — lack that ‘homogeneity.’ As he had argued already in Treatise on Probability (ch. 22), it wasn’t always possible to take repeated samples from a fixed population when we were analysing real-world economies. In many cases, there simply are no reasons at all to assume the samples to be homogenous. Lack of ‘homogeneity’ makes the principle of ‘limited independent variety’ non-applicable, and hence makes inductive inferences, strictly seen, impossible since one of its fundamental logical premises is not satisfied. Without “much repetition and uniformity in our experience” there is no justification for placing “great confidence” in our inductions (TP ch. 8).

And then, of course, there is also the ‘reverse’ variability problem of non-excitation: factors that do not change significantly during the period analysed, can still very well be extremely important causal factors.

Stability: Tinbergen assumes there is a stable spatio-temporal relationship between the variables his econometric models analyze. But as Keynes had argued already in his Treatise on Probability it was not really possible to make inductive generalisations based on correlations in one sample. As later studies of ‘regime shifts’ and ‘structural breaks’ have shown us, it is exceedingly difficult to find and establish the existence of stable econometric parameters for anything but rather short time series.

Measurability: Tinbergen’s model assumes that all relevant factors are measurable. Keynes questions if it is possible to adequately quantify and measure things like expectations and political and psychological factors. And more than anything, he questioned — both on epistemological and ontological grounds — that it was always and everywhere possible to measure real-world uncertainty with the help of probabilistic risk measures. Thinking otherwise can, as Keynes wrote, “only lead to error and delusion.”

Independence: Tinbergen assumes that the variables he treats are independent (still a standard assumption in econometrics). Keynes argues that in such a complex, organic and evolutionary system as an economy, independence is a deeply unrealistic assumption to make. Building econometric models from that kind of simplistic and unrealistic assumptions risks producing nothing but spurious correlations and causalities. Real-world economies are organic systems for which the statistical methods used in econometrics are ill-suited, or even, strictly seen, inapplicable. Mechanical probabilistic models have little leverage when applied to non-atomic evolving organic systems — such as economies.

It is a great fault of symbolic pseudo-mathematical methods of formalising a system of economic analysis … that they expressly assume strict independence between the factors involved and lose all their cogency and authority if this hypothesis is disallowed; whereas, in ordinary discourse, where we are not blindly manipulating but know all the time what we are doing and what the words mean, we can keep “at the back of our heads” the necessary reserves and qualifications and the adjustments which we shall have to make later on, in a way in which we cannot keep complicated partial differentials “at the back” of several pages of algebra which assume that they all vanish.

Building econometric models can’t be a goal in itself. Good econometric models are means that make it possible for us to infer things about the real-world systems they ‘represent.’ If we can’t show that the mechanisms or causes that we isolate and handle in our econometric models are ‘exportable’ to the real world, they are of limited value to our understanding, explanations or predictions of real-world economic systems.

The kind of fundamental assumption about the character of material laws, on which scientists appear commonly to act, seems to me to be much less simple than the bare principle of uniformity. They appear to assume something much more like what mathematicians call the principle of the superposition of small effects, or, as I prefer to call it, in this connection, the atomic character of natural law. The system of the material universe must consist, if this kind of assumption is warranted, of bodies which we may term (without any implication as to their size being conveyed thereby) legal atoms, such that each of them exercises its own separate, independent, and invariable effect, a change of the total state being compounded of a number of separate changes each of which is solely due to a separate portion of the preceding state …

The scientist wishes, in fact, to assume that the occurrence of a phenomenon which has appeared as part of a more complex phenomenon, may be some reason for expecting it to be associated on another occasion with part of the same complex. Yet if different wholes were subject to laws qua wholes and not simply on account of and in proportion to the differences of their parts, knowledge of a part could not lead, it would seem, even to presumptive or probable knowledge as to its association with other parts.

Linearity: To make his models tractable, Tinbergen assumes the relationships between the variables he studies to be linear. This is still standard procedure today, but as Keynes writes:

It is a very drastic and usually improbable postulate to suppose that all economic forces are of this character, producing independent changes in the phenomenon under investigation which are directly proportional to the changes in themselves; indeed, it is ridiculous.

To Keynes, it was a ‘fallacy of reification’ to assume that all quantities are additive (an assumption closely linked to independence and linearity).

The unpopularity of the principle of organic unities shows very clearly how great is the danger of the assumption of unproved additive formulas. The fallacy, of which ignorance of organic unity is a particular instance, may perhaps be mathematically represented thus: suppose f(x) is the goodness of x and f(y) is the goodness of y. It is then assumed that the goodness of x and y together is f(x) + f(y) when it is clearly f(x + y) and only in special cases will it be true that f(x + y) = f(x) + f(y). It is plain that it is never legitimate to assume this property in the case of any given function without proof.

J. M. Keynes “Ethics in Relation to Conduct” (1903)

And as even one of the founding fathers of modern econometrics — Trygve Haavelmo — wrote:

What is the use of testing, say, the significance of regression coefficients, when maybe, the whole assumption of the linear regression equation is wrong?

Real-world social systems are usually not governed by stable causal mechanisms or capacities. The kinds of ‘laws’ and relations that econometrics has established, are laws and relations about entities in models that presuppose causal mechanisms and variables — and the relationship between them — being linear, additive, homogenous, stable, invariant and atomistic. But — when causal mechanisms operate in the real world they only do it in ever-changing and unstable combinations where the whole is more than a mechanical sum of parts. Since statisticians and econometricians — as far as I can see — haven’t been able to convincingly warrant their assumptions of homogeneity, stability, invariance, independence, and additivity as being ontologically isomorphic to real-world economic systems, Keynes’ critique is still valid. As long as — as Keynes writes in a letter to Frisch in 1935 — “nothing emerges at the end which has not been introduced expressively or tacitly at the beginning,” I remain doubtful of the scientific aspirations of econometrics. Especially when it comes to using econometrics for making causal inferences, it is still often based on counterfactual assumptions that have outrageously weak grounds.

In his critique of Tinbergen, Keynes points us to the fundamental logical, epistemological and ontological problems of applying statistical methods to a basically unpredictable, uncertain, complex, unstable, interdependent, and ever-changing social reality. Methods designed to analyse repeated sampling in controlled experiments under fixed conditions are not easily extended to an organic and non-atomistic world where time and history play decisive roles.

Econometric modelling should never be a substitute for thinking. From that perspective, it is really depressing to see how much of Keynes’ critique of the pioneering econometrics in the 1930s-1940s is still relevant today.

The general line you take is interesting and useful. It is, of course, not exactly comparable with mine. I was raising the logical difficulties. You say in effect that, if one was to take these seriously, one would give up the ghost in the first lap, but that the method, used judiciously as an aid to more theoretical enquiries and as a means of suggesting possibilities and probabilities rather than anything else, taken with enough grains of salt and applied with superlative common sense, won’t do much harm. I should quite agree with that. That is how the method ought to be used.

J. M. Keynes, letter to E.J. Broster, December 19, 1939