The Lucas critique: A Lucas critique

Download full paper

This paper is closed for comments.

Abstract

The Lucas critique has been and continues to be the cornerstone of modern macroeconomic modelling. In this note we apply the Lucas critique to macroeconomic modelling using deep rational expectations. In conclusion we point out that Lucas’ call for rational expectations models that provide useful economic policy advice has yet to be heeded.

Posted for comments on 21 Feb 2018, 12:13 pm.
Published

Comments (4)

  • Erich Pinzon-Fuchs says:

    In his paper, Christian Müller-Kademann proposes a re-interpretation of the Lucas Critique through the introduction of the concept of deep rational expectations (DRE). The author argues that DRE would take the Lucas Critique to another level, since the rational expectations (RE) framework would be applied not only to a particular model based on which economic agents build their expectations about the consequences of particular policies but also to the problem of how agents select a particular model among others in the first place. In order to apply the Lucas Critique to the model selection problem, the author claims that it is necessary to introduce John Maynard Keynes’s concept of fundamental uncertainty, since “the concept of (fundamental) uncertainty […] is potentially able to reconcile rationality, model consistent expectations and the Lucas Critique” (p. 9). Yet, although the author’s proposition is interesting and quite bold, the author does not discuss it sufficiently, nor does he provide a clear alternative on how to put in place a research agenda based on his idea of DRE. Given that the paper has important normative elements, there is a need for both a more thorough and detailed discussion of the actual use of DRE in macroeconomic modelling, and for a concrete illustration of its use. In conclusion, I think that the paper is interesting and has potentially something important to say. However, in its present form, the paper does not present its arguments in a way that is sufficiently thorough. The author’s over-optimistic and uncritical interpretation of the Lucas Critique makes more harm to the author’s arguments than it helps him in making his point. In this sense, most of the ideas expressed in this paper could benefit from a reflexive examination of the history of the Lucas Critique that does not stem from the “standard narrative” of the history of macroeconomics built by Robert Lucas himself and by other practitioners of macroeconomics.

    Specific comments

    1) One of the main propositions in the paper, namely that macroeconomics should introduce Keynes’s concept of fundamental uncertainty is, in my opinion, insufficiently treated. Indeed, the author refers to Keynes’s (1937) paper as the major source to understand this concept, and yet Keynes does not thoroughly develop this concept in that particular piece. I think that the author’s proposition needs to be thoroughly researched and discussed and that, to do so, the author should study and refer to Keynes (1921) as well as to the secondary literature that also focuses on Keynes’s ideas on uncertainty and probability. Some important references in this sense are Lawson (1985a; 1985b) and Carabelli (1988).

    2)The author presents the Lucas Critique in an over-optimistic and uncritical way. This has to do, in particular, with the assumption of an uncritical stand towards the “standard history” of macroeconomics as defined by Duarte and Lima (2012). The author’s uncritical acceptance of the standard narrative is revealed at different points in the paper. For example, when the author asserts that the Lucas Critique was a “devastating attack on the […] common approach to macroeconometric modelling” (p. 2); that the critique was “convincing” and “successful,” and that macroeconomic models had “achieve[d] consistency” (p. 3) thanks to the Lucas Critique. In addition, the adoption of the standard narrative leads the author to adopt a vision about the macroeconometric models of the 1960s and 1970s that is not necessarily fair. Following this vision, the author does not recognise, as does Lucas (1976, p. 20, footnote 3) for instance, that macroeconometricians were well aware of the fact that the implementation of policies could change the agents’ behaviour and hence the structure of the model, making the model unable to evaluate economic policies. Lucas (1976) explicitly recognises that Jan Tinbergen and Jakob Marschak were aware of this problem since, at least, the 1940s.

    3) The author considers that the Lucas Critique necessarily implies the use of the rational expectations hypothesis. Yet, a closer look at Lucas’s (1976) paper shows that this is not necessarily the case. In fact, Lucas (1976) argues that the macroeconomic models which have been built to make policy evaluation, should take into account a careful description of the optimising behaviour of individual economic agents and in particular of their reactions to changes in economic policy. The rational expectations hypothesis is only one way to consider these reactions, but it is not the only way (see Goutsmedt et al. 2015). This interpretation of the Critique is quite common and has to do with the spread of a “standard narrative” of the history of macro (and of the Lucas Critique). In this sense, the author should precise that he is not taking the Lucas Critique itself to another level, but rather the rational expectations hypothesis.

    4) The author claims that “the message of the Lucas Critique is an ontological one” (p. 9), meaning that the Lucas Critique, applied at the level of the model selection problem, can tell us something important about the way uncertainty works in the real world. In this sense, and citing John Stuart Mill (1844) hastily, the author claims that the Lucas Critique “seriously challenge[s] if not outright reject[s]” economists’ “relentless search for newer, better models” and their “ontological view of an underlying truth that waits to be discovered” (p. 9). However, hardly any economist would define her/his job as that of seeking for an “underlying truth that waits to be discovered.” Much more evidence and historical work should be undertaken here in order to support the author’s claim.

    Minor comments

    – The abstract is incomplete and should be revised. The paper pretends to do more than announced in the abstract.

    -The author claims that “criticism of the Lucas Critique has become the subject of research agendas in its own right” (p. 4) and cites our paper Goutsmedt et al. (2015). Just to clarify, our “research agenda” is not to criticise the Lucas Critique, but to reflect on the narratives that are built around the history of the Lucas Critique (and of the history of macro in general). The title of our paper “Criticizing the Lucas Critique” must be read together with its subtitle “Macroeconometricians’ response to Robert Lucas,” meaning that it is not us (the historians) who criticise Lucas, but that we want to recover and study the macroeconometricians’ contemporary reactions to the Lucas Critique, which have been left aside in the “standard narrative.” Our research, again, is focused on criticising the “standard narrative” of the history of macroeconomics which has been produced by practitioners of macroeconomics, including Lucas, in order to get a richer and broader narrative that takes into account all the players of particular episodes in the history of macroeconomics. In this project, we follow other historians of macro such as Duarte and Lima (2012) and Forder (2014), among many others.

    Supplementary references:

    Carabelli, Anna M. (1988) On Keynes’s Method. London: The MacMillan Press Ltd.

    Duarte, Pedro and Gilberto Lima (2012) “Introduction: Privileging micro over macro? A history of conflicting positions,” in Duarte and Lima (eds.) Micorfoundations reconsidered: the relationship of micro and macroeconomics in historical perspective. Cheltenham: Edward Elgar.

    Forder, James (2014) Macroeconomics and the Phillips Curve Myth. Oxford University Press.

    Goutsmedt, A. et al. (2015) “Criticizing the Lucas Critique: Macroeconometricians’ Response to Robert Lucas,” CES Working Papers, 2015.59.

    Keynes, John M. (1921) A Treatise on Probability. London: Macmillan and Co., Limited.

    Lawson, Tony (1985a) “Keynes, Prediction and Econometrics.” In Lawson, Tony and Hashem Pesaran (eds.) (1985). Keynes’s Economics: Methodological Issues. London: Croom Helm.

    Lawson, Tony (1985b) “Uncertainty and Economic Analysis.” The Economic Journal, Vol. 95, No. 380, pp. 909-927.

  • Christian Muller-Kademann says:

    Reply to Erich Pinzon-Fuchs Comments on “The Lucas critique: A Lucas critique”

    First off, I would like to thank the reviewer, Erich Pinzon-Fuchs, for his careful reading of the article draft and for taking his time to comment. Erich Pinzon-Fuchs raises several interesting issues and offers suggestions for amending the paper as well as valuable additional input.

    The overall assessment of the reviewer is somewhat critical, however, which is why I would like to take this opportunity to respond in detail.

    According to the reviewer, the key deficit of the paper is a lack of sufficient discussion of its main proposition. This main proposition is, in the opinion of the reviewer, that “the concept of (fundamental) uncertainty […] is potentially able to reconcile rationality, model consistent expectations and the Lucas Critique” (quoted from the paper). Mr. Pinzon-Fuchs writes: “the author does not discuss [his proposition] sufficiently, nor does he provide a clear alternative on how to put in place a research agenda based on his idea of [deep rational expectations]”, and further explains “Given that the paper has important normative elements, there is a need for both a more thorough and detailed discussion of the actual use of DRE in macroeconomic modelling, and for a concrete illustration of its use.”

    Second to this main objection, Mr. Pinzon-Fuchs points out that the “author’s over-optimistic and uncritical interpretation of the Lucas Critique makes more harm to the author’s arguments than it helps him in making his point”, and hence recommends “most of the ideas expressed in this paper could benefit from a reflexive examination of the history of the Lucas Critique that does not stem from the `standard narrative’ of the history of macroeconomics built by Robert Lucas himself and by other practitioners of macroeconomics.”

    I will address these major two points here and discuss the reviewer’s other comments in an extended version of this reply.

    There is no way of not agreeing with the referee that the main proposition of my paper is indeed that fundamental uncertainty holds the key for reconciling rationality, model consistent expectations and the Lucas Critique. In contrast to what Mr. Pinzon-Fuchs suggests, however, I do not start with this proposition nor do I aim at proving it. My objective, rather, is, to apply the Lucas critique (its analytical, or positive and also less original element) to the solution of the Lucas critique (its normative, very original and highly influential part). The novel feature, therefore, is to apply the Lucas critique to itself. In order to meaningfully do so, I recur to what I call deep rational expectations (or DRE when using the referee’s abbreviation).

    In other words, my paper is at the same time less ambitious than the referee thinks (proving a proposition) but also more ambitious in that it turns Lucas’ criticism against his solution.

    When studying the history of the Lucas critique, it becomes almost immediately apparent that its positivist part, which is concerned with highlighting the inconsistency of “naive” macromodelling of economic policy conduct, cannot be considered really original because very similar arguments had already been around for quite some time (Goutsmedt, A., Pinzon-Fuchs, E., Renault, M. and Sergi, F., 2016, p.6). Instead, the normative part, which is about offering a solution to the known issues, really made all the difference.

    This observation indicates that “market” success of economic arguments is more likely when a solution to a known problem can be offered. It does not follow, however, that the discovery and the description of the problem itself is unimportant. Quite to the contrary, it is Lucas’ and his followers’ — not mine — main selling proposition that hardly any other solution but his offers mathematical elegance, flexibility and overall appeal. I rather think that we should not err again in hastily ranking a solution higher than a proper analysis of the problem just because it seems to be a solution.

    We should not do so even though the history of the Lucas critique has shown that the “solution” seems to be more important than the critique. Rather, rushing in another answer bears the risk of getting it wrong again and wasting (again) countless resources on a flawed approach.

    My arguments clearly address Lucas’ solution, the normative part, which I show to be invalid by the standards of the Lucas critique’s positivist part. Therefore, what is considered a solution to the inconsistency problem of economic policy modelling is, in fact, not a solution. This is the key insight that I humbly ask to be accepted. I am sorry to say that I cannot — as yet — offer a solution to the modelling inconsistency arising from DRE. (In fact, I sense an impossibility theorem there but, regrettably, I am not (yet) able to prove any.) Therefore, I conclude my abstract “Lucas’ call for rational expectations models that provide useful economic policy advice has yet to be heeded.”

    To sum up my answer to the first main criticism, I certainly agree with the referee that the implications of DRE for macromodelling needs to be thoroughly discussed. However, before doing so we first have to understand the effect DRE has on the available solutions to the Lucas critique (positivist part). Enhancing this understanding is the main purpose of the paper.

    When studying the impact of DRE, I consider it most efficient to first focus on the mainstream solutions because, whether we like it or not, these are defining the scientific and public economic discussions at large. Therefore, although I very much endorse the idea of a holistic approach to the history of the Lucas critique, I think it justified to start with “the `standard narrative’ of the history of macroeconomics built by Robert Lucas himself and by other practitioners of macroeconomics” (Mr. Pinzon-Fuchs). My choice of starting with the standard interpretation should not, however, be mistaken for a wholehearted support. I simply state the facts without taking sides.

    This pragmatic decision notwithstanding, I certainly subscribe to the view that the Lucas critique deserves a critical rather than an over-optimistic interpretation. I do so because the more widely a method is accepted the more scrutiny it should face if we want to spend scarce scientific resources wisely. In fact, applying the Lucas critique to the Lucas critique itself is, in my view, a contribution to the desired critical approach.

    The reviewer also comments on some specific issues not yet mentioned such as the definition of DRE and the question of whether or not the Lucas critique yields an ontological message and what the ultimate goal of macroeconomic research is.

    I will refrain from posting my answers to these issues here also in order to save space. The full reply is available at https://www.s-e-i.ch/Projects/FiscalPolicy/LucasReply2comment1.html (including a printable version).

    Summing up my responses to Mr. Pinzon-Fuchs’ report, I would like to offer once again my gratitude for the thorough review and helpful suggestions. The paper draft will be amended by adding explanation and corrections according to the above discussion. I am sure that the reviewer’s input will thus significantly benefit the readers of the article.

    References

    Goutsmedt, A., Pinzon-Fuchs, E., Renault, M. and Sergi, F. (2016). Criticizing the Lucas Critique: Macroeconometricians’ Response to Robert Lucas, Universite´ Paris 1 Pantheon-Sorbonne (Post-Print and Working Papers) halshs-01364814, HAL.

    Keynes, J. M. (1921). A treatise on probability, Cambridge University Press, Cambridge.

    Keynes, J. M. (1937). The General Theory of Employment, The Quarterly Journal of Economics 51(2): 209 – 223.

    Lucas, R. J. (1976). Econometric policy evaluation: A critique, Carnegie-Rochester Conference Series on Public Policy 1(1): 19 – 46.

  • Lefteris Tsoulfidis says:

    This is a well written critical review on the so-called Lucas Critique. The main argument is that that the parameters of the econometric models used for policy analysis and of course predictions should account more carefully for expectations. As a result, these parameters are not necessary given but variable. All this is well known and the article is a good summary of the issues at hand.
    For more the author would improve the argument by addressing two questions: First, to give more information about the properties of the chosen model in which, one way or another, expectations must be accounted for and in this sense the author must say more about the properties and consequences of the rational expectations. Second one must have some knowledge about the size of variation in the parameters of the preferred model, because if the change in parameters is relatively small then the model can be used and judged on the basis of other considerations. In this respect the author must bring into the discussion some more results from the empirical macro-econometric literature.

    Finally, the paper will improve if the Lucas critique is contrasted with the Goodhart’s laws which appeared simultaneously and independently of the Lucas Critique and essentially they are not very different.

    References.
    Tsoulfidis L. (2010) Competing Schools of Economic Thought. Berlin Heidelberg: Springer.

    Recommendation: Accept subject to the above revisions

    • Christian Muller-Kademann says:

      It has been a pleasure for me to receive Mr. Tsoulfidis’ comments and recommendations. I do generally agree with his views and am confident that according ammendments to the paper manuscript are rather straightfoward. At this stage, I would only caution against the expectation that a specific model would be able to remedy the Lucas critique in its entirety. It could very well be that this critique (its positive part that largely coincides with Goodhart’s law) implies that a model solution may be impossible. If so, we would rather need an epistemological or even ontologic response to the Lucas critique. My paper is, however, focussed on the internal contradiction of the Lucas critique. In any case, I am very grateful for Mr. Tsoulfidis’ careful considerations.