Economists (and everyone else out there in the econoblogosphere): time to stop hiding behind theory and qualitative analysis and start addressing the empirical data out there.Theory versus Big Data? Maybe not.
Economic theory plays an important role in the analysis of large data sets with complex structure. It can be difficult to organize and study this type of data (or even to decide which variables to construct) without a simplifying conceptual framework, which is where economic models become useful. Better data also allow for sharper tests of existing models and tests of theories that had previously been difficult to assess.Science
Economics in the age of big data
Lira Einavand Jonathan Levin
The Data Revolution and Economic Analysis
Liran Einav, Stanford University and NBER Jonathan Levin, Stanford University and NBER
In A Fistful of Dollars, Clint Eastwood challenges Gian Maria Volonte with the words, “When a man with .45 meets a man with a rifle, you said, the man with a pistol’s a dead man. Let’s see if that’s true. Go ahead, load up and shoot.”
That’s the right words to challenge big data, which recently reappeared in economics debates (Noah Smith, Chris House via Mark Thoma). Big data is a rifle, but not necessary winning. Economists must have special reasons to abandon small datasets and start messing with more numbers.
Unlike business, which only recently discovered the sexiest job of the future, economists do analytics for the last 150 years. They deal with “big data” for half of that period (I count from 1940, when the CPS started). So, how can the new big data be useful to them?
Let’s find out what big data offers. First of all, more information, of course.…Second, big data comes with its own tools, which, like econometrics, are deeply rooted in statistics but ignorant about causation…Economics and Development
There’s nothing ideological in these comments on big data. More data potentially available for research is better than less data. And data scientists do things economists can’t. The objection is the following. Economists mostly deal with the problems of two types. Type One, figuring out how n big variables, like inflation and unemployment, interact with each other. Type Two, making practical policy recommendations for the people who typically read nothing more than executive summaries. While big data can inform top-notch economics research, these two problems are easier to solve with simple models and small data. So, a pistol turns out to be better than a rifle.
How Big Data Informs Economics
So we are back to priors. And no one has any that are shown (empirically) to work under all conditions, like at turning points. Which is the whole point. Otherwise we are like the turkeys before Thanksgiving.
One of the issues is that experiment is not available in social science (including economics) as it is in natural and life sciences. Another is the Lucas Critique, which was anticipated by Keynes. However, Keynes's critique of Tinbergen was arguably more penetrating and robust.
The Lucas Critique and Keynes' Response - Considering the History of Macroeconomics
Elke Muchlinski
One of the issues is that experiment is not available in social science (including economics) as it is in natural and life sciences. Another is the Lucas Critique, which was anticipated by Keynes. However, Keynes's critique of Tinbergen was arguably more penetrating and robust.
Tinbergen has presented a "cryptic method of exposition" (C.W., XIV, 285).22
Keynes stated: "There is first of all the central question of methodology, the logic of applying the method of multiple correlation to unanalysed economic material, which we know to be nonhomogeneous through time. If we are dealing with the action of numerically measurable, independent forces, adequately unanalysed so that we knew we were dealing with independent atomic factors and between them completely comprehensive, acting with fluctuating relative strength on material constant and homogeneous through time, we might be able to use the method of multiple correlation with some confidence for disentangling the laws of their action; (...) In fact we know that every one of these conditions is far from being satisfied by the economic material under investigation. How far does this impair the validity of the method? That seems to me to deserve a most careful preliminary enquiry. The volume which purports to be 'a note on the method' in fact faces none of these difficulties and is in fact mainly occupied, just like the other volume, with elaborate half-explained numerical examples, the method employed in which already begs the question" (C.W., XIV, 285-6). [p. 20]
Keynes criticized Tinbergen's methods because of misguiding modelling:"The pseudo- analogy with the physical sciences leads directly counter to the habit of mind which is most importance for an economist proper to acquire. (...) One has to be constantly on guard against treating the material as constant and homogeneous. It is as though the fall of the apple to the ground depended on the apple's motives, on whether it is worth while falling to the ground, and whether the ground wanted the apple to fall, and on mistaken calculations on the part of the apple as to how far it was from the centre of the earth" (1938, C.W., XIV, 299-300). [p. 22]
In the following quotation one can perceive Keynes's sceptical view. The application of statistical methods and measurements is due to theoretical assumptions which Tinbergen did not make explicitly. This is an impediment to achieving validity and credibility in his work. Keynes asked: "How are these coefficients arrived at? Is it by laborious trail-and-error guessing, or by method? How are the time lags arrived at? Is it by common sense guessing or by method? (...) Is it assumed that the factors investigated are comprehensive and that they are not merely a partial selection out of all the factors at work? How much difference does it make to the method if they are not comprehensive? Is it claimed that there is a likelihood that the equations will work approximately next time? (...) Is it assumed that the future is a determinate function of past statistics? What place is left for expectation and the state of confidence relating to the future? [p. 22]
What place is allowed for non-numerical factors, such as inventions, politics, labour troubles, wars, earthquakes, financial crises? One feels a suspicion that the choice of factors is influenced (...) by what statistics are available, and that many vital factors are ignored because they are statistically intractable or unprocurable. (...). Now I quite agree that it would not be easy to apply the method to these factors. But that seems to me a justification for not using the method in this case rather than for ignoring these matters and telling us what we know alredy with the trimmings of figures which really have no significance" (1938, C.W., XIV, 286-288). [p. 23]
In summary, we then come to the next question Keynes addressed to Tinbergen: "How far are these curves and equations meant to be no more than a piece of historical curve-fitting and description, and how far do they make inductive claims with reference to the future as well as the past?" (C.W., XIV, 315). Keynes presented his investigations of inductive arguments and reasoning in the Treatise on Probability. The validity of an inductive argument definitely depends on the length of the underlying period or sub-periods since the regression coefficient for each period will change with the choice of the period itself.
Finally Keynes objected to the hypothesis of independence. "Must we push our preliminary analysis to the point at which we are confident that the different factors are substantially independent of one another? This is not discussed. Yet I think it is important. For, if we are using factors which are not wholly independent, we lay ourselves open to the extraordinarily difficult and deceptive complications of 'spurious' correlation" (C.W., XIV, 309). [ p. 24]
Having introduced central objections Keynes made against Tinbergen's work, it is quite obvious that expectation is a key notion in Keynes's economic thinking. He came to this view in his fundamental critique on the classical view of Benthamite calculation. [p. 24] In contrast to that, he outlined his metatheoretical or methodological view rejecting natural sciences as an inappropriate approach to economics. Moral science in modern terms could be expressed as a social science.
Surprisingly none of the authors who devotionally follow Lucas and Lucasiansm refer to Keynes's criticism. This could have avoided deep misunderstanding of what Keynes said. The common sense in economics nowadays is to accuse Keynesian macroeconometrics and macroeconomics of having neglected the importance of expectations. The widespread accepted identification of Tinbergen's model of macropolicy with Keynes's view caused a misunderstanding of the core of his criticism of econometrics. His proviso with the presented methods was in no way an expression of model nihilism, since he has already explained significant aspects of model building in economics. One has to be aware that economic material is non-linear, non-reversible, non-homogeneous, not independent of one another and which has to be judged on the basis of time lags. It is necessary to choose those variables which are not only suitable but also important to the purpose in question. [p. 25]
Blinder (1998) elaborates in some arguments having contextual importance. Macro- policy relies on models, both macroeconomics and macroeconometrics are concerned with special problems of model building. These are in short:
(1) The unknown "true model". In fact any choice of a model implies uncertainty if the emphasized proposition is a "true" one. There is no way out of uncertainty. Blinder refers to the distinction of uncertainty and risk in Knight's terminology (1921) explaining why uncertainty rather than risk is a problem in model building. "Risk arises when a random variable has a known probability distribution; uncertainty arises when the distribution is unknown" (1998, 77). Since the distribution of uncertainty is persistently and categorically unknown (because there is no such distribution), there is no chance of eliminating uncertainty through model building. [p. 26]
(2) Uncertainty in the forecast. This problem can be methodically solved by replacing the unknown future variables with "their expected values" defined as the "certainty equivalence" principle. This operation looks easier than it is because of the great amount of unknown exogenous variables which have to be integrated in the model. Furthermore this method is problematic because of the non-linearity of the economy and serious doubt that the objective function is quadratic. But who knows the objective function? Since no one have ever found a justification of "an objective" function, nor given it, practical economists or central bankers must create such a function. Again, the purpose of model building is due to problems of the contemporary world therefore "policymakers almost always will be contemplating changes in policy instruments that can be expected to lead to small changes in macroeconomic variables" (Blinder 1998, 10).
(3) Uncertainty about parameters. Blinder refers to Brainard (1967) to describe that uncertainty about parameters should lead to a more or less conservative behavior of the central bankers, i.e. assuming the lowest movement of parameters.
(4) Uncertainty about model selection. This problem is connected with the first three aspects and there is no chance to avoid a choice of models among many others. Relying on a universal model would be the worst case, rather than using a wide variety of different models with a reasonable critique.
(5) Finally, Blinder emphasized the "long and variable lags" macroeconometrics and macroeconomics have to recognize. "It is essential, in my view, for central bankers to realize that, in a dynamic economy with long lags in monetary policy, today's monetary policy decision must be thought of as the first step along a path. The reason is simple: Unless you have thought through your expected future actions, it is impossible to make today's decision rationally "(1998, 14). [p. 27]
Some authors refer to the Lucas critique in a way that shows the importance of recognizing that economics is a social science which is not guided by certain laws or by the law of gravity or by numbers. Any regime change has to be implemented on the basis of changes in expectations and responses by economic agents. This common knowledge once implemented by Keynes, then used by Lucas and Lucasianism to substitute Keynesian macroeconomics, regardless of what Keynes did say, is now applied to aspects of model building. [p. 29]
Elke Muchlinski
No comments:
Post a Comment